System and method for protein corona sensor array for early detection of diseases

Abstract

To provide sensor arrays for detecting biomolecules and methods of use.SOLUTION: Recognition of a biomolecular fingerprint from a sample of a subject is combined with ability to determine a disease state of the subject on a continuum of health care. In some aspect, the present invention provides a sensor array comprising a plurality of sensor elements, where the plurality of sensor elements differ from each other in at least one physicochemical property and the plurality of sensor elements comprises at least two sensor elements. In some aspects, each sensor element is bindable to a plurality of biomolecules in a sample to produce a biomolecule corona signature, where each sensor element has a distinct biomolecule corona signature different from others.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of U.S. Provisional Patent Application No. 62 / 435,409, filed December 16, 2016, the entire contents of which are incorporated herein by reference.

[0002] <Statement Regarding Federally Sponsored Research> Not applicable.

[0003] <Incorporated by reference> All publications, patents, and patent applications mentioned in this specification are incorporated by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference.

[0004] The field of the invention relates to sensor arrays for the detection and diagnosis of different disease states, and in particular the invention relates to the ability to diagnose or prognose a disease or disorder. [Background technology]

[0005] The earlier a disease is diagnosed, the greater the chance that it can be cured or successfully managed, resulting in a better prognosis for the patient. Treating a disease early can prevent or delay problems caused by the disease and may improve the patient's prognosis by extending their lifespan and / or improving their quality of life.

[0006] Early diagnosis of cancer is important because many of these cancers can be successfully treated at an early stage. For example, the 5-year survival rates after early diagnosis and treatment of breast, ovarian, and lung cancer are 90%, 90%, and 70%, respectively, compared to 15%, 5%, and 10% when patients are diagnosed at the most advanced stages of the disease. Once cancer cells leave their tissue of origin, it becomes very difficult to treat successfully with available established therapies. Although early diagnosis is possible with recognition of warning signs of cancer and prompt action, many cancers (e.g., lung cancer) only show symptoms after the cancer cells have already invaded surrounding tissues and metastasized throughout the body. For example, more than 60% of patients with breast, lung, colon, and ovarian cancer have hidden or metastatic colonies at the time the cancer is discovered. Thus, there is an urgent need to develop effective approaches for early detection of cancer. Such an approach should have the sensitivity to identify cancer at various stages and the specificity to give a negative result when the person being tested does not have cancer. Although extensive efforts have been made to develop methods for early detection of cancer, and a significant number of risk factors and biomarkers have been proposed, a broadly relevant platform for the early detection of a wide range of cancers remains elusive.

[0007] As various cancers can alter the composition of plasma even in their early stages, one promising approach for early detection is molecular blood analysis for biomarkers. Although this strategy has already been successful for some cancers (e.g. PSA for prostate cancer), there are still no specific biomarkers for the early detection of the majority of cancers. For such cancers (e.g. lung cancer), none of the defined candidate blood biomarkers have been clinically validated and only a few of them have reached late-stage clinical development. Thus, there is a need for new methods to improve our ability to detect cancer at a very early stage. Novel approaches are urgently needed. Summary of the Invention

[0008] The present invention provides a highly sensitive and versatile sensor array for the detection of a wide range of diseases and disorders and for the determination of disease status in a subject. What is unique about the present invention is that it combines the recognition of biomolecular fingerprints from samples from subjects with the ability to determine disease status in said subjects across the health care continuum.

[0009] In some aspects, the present invention provides a sensor array comprising a plurality of sensor elements, the plurality of sensor elements differing from one another in at least one physicochemical property, the plurality of sensor elements comprising at least two sensor elements. In some aspects, each of the sensor elements is capable of binding to a plurality of biomolecules in a sample to generate a biomolecular corona signature, each of the sensor elements having a biomolecular corona signature that is distinct from the others. In some aspects, the sensor elements are nanoscale sensor elements.

[0010] In some aspects, the plurality of sensor elements are configured to generate a plurality of biomolecular corona signatures upon contact with the sample, the combination of the plurality of biomolecular corona signatures generating a biomolecular fingerprint of the sample.

[0011] In some aspects, the sensor element is coupled to a substrate.

[0012] In some aspects, the sensor elements are discrete elements (regions) on a substrate, plate or chip that differ in form and function, with each different element (region) producing a different biomolecular corona signature. Together, the substrate, chip or array itself forms a sensor array, providing a biomolecular fingerprint for the sample.

[0013] In some aspects, the invention provides a method of detecting a disease state in a subject, comprising obtaining a sample from the subject, contacting the sample with a sensor array, and determining a biomolecular fingerprint associated with the sample, the sensor array comprising a plurality of sensor elements, the plurality of sensor elements differing from one another in at least one physicochemical property, the plurality of sensor elements comprising at least two sensor elements, the biomolecular fingerprint distinguishing disease states in the subject. In some aspects, the method further comprises comparing the biomolecular fingerprint of the sample to a panel of biomolecular fingerprints associated with a plurality of disease states to determine a disease state associated with the sample.

[0014] In another aspect, the invention provides a method for determining a biomolecular fingerprint associated with at least a disease state or at least one disease or disorder, comprising: (a) obtaining samples from at least two subjects diagnosed with the disease state or at least one disease or disorder; (b) contacting each sample with a sensor array described herein; and (c) determining a biomolecular fingerprint for the sensor array associated with the disease state or at least one disease or disorder. In some aspects, step (c) further comprises detecting the composition of the biomolecular corona of each sensor element, where the combination of the compositions of each biomolecular corona between the different sensor elements generates the biomolecular fingerprint associated with the sample. In other aspects, step (c) separates the biomolecular corona from each sensor element and analyzes a plurality of biomolecules of each biomolecular corona, where the combination of analyzed biomolecules generates a biomolecular fingerprint.

[0015] In yet another aspect, the invention provides a method for diagnosing or prognosing a disease or disorder in a subject. The method includes obtaining a sample from a subject, contacting the sample with a sensor array described herein to generate a biomolecular fingerprint, comparing the biomolecular fingerprint to a panel of biomolecular fingerprints associated with a plurality of diseases or disorders, and diagnosing or prognosing the disease or disorder.

[0016] In yet another aspect, the invention provides a method for identifying a pattern of biomarkers associated with a disease or disorder, comprising: (a) obtaining samples from at least two subjects diagnosed with the disease or disorder and at least two control subjects; (b) contacting each sample with a sensor array to generate a plurality of biomolecular coronas for each subject; and (c) comparing the composition of the plurality of biomolecular coronas of the subjects having the disease or disorder with the composition of the plurality of biomolecular coronas of the control subjects to determine a pattern of biomarkers associated with the disease or disorder.

[0017] In some aspects, the disease or disorder is cancer, a cardiovascular disease, an endocrine disorder, an inflammatory disease, or a neurological disease.

[0018] In one aspect, the disease or disorder is cancer.

[0019] In another aspect, the disease or disorder is cardiovascular disease, hi one aspect, the cardiovascular disease is coronary artery disease (CAD).

[0020] In a further aspect, the disease or disorder is a neurological disease, hi one aspect, the neurological disease is Alzheimer's disease.

[0021] In some aspects, the invention provides a kit for diagnosing or prognosing a disease or disorder, the kit comprising a sensor array comprising a plurality of nanoscale sensor elements, the plurality of sensor elements differing from one another in at least one physicochemical property, the plurality of sensor elements comprising at least two sensor elements.

[0022] In another aspect, the invention provides a kit for determining and / or detecting at least one biomarker associated with a disease or disorder, the kit comprising at least one sensor array comprising a plurality of sensor elements, the plurality of sensor elements differing from one another in at least one physicochemical property, the plurality of sensor elements comprising at least two sensor elements.

[0023] In yet another aspect, the present invention provides a method for distinguishing states of a complex biological sample using a plurality of particles having surfaces with different physicochemical properties, comprising the steps of contacting the complex biological sample with the plurality of particles to bind proteins of the complex biological sample to the plurality of particles, determining a biomolecular fingerprint indicative of proteins that bind to the plurality of particles, and correlating the biomolecular fingerprint with a biological state of the subject, wherein the pattern of protein binding between the plurality of particles differs depending on the physicochemical properties of the surfaces of the particles.

[0024] In another aspect, the present invention provides a sensor array comprising a plurality of particles having surfaces with different physicochemical properties, configured to contact a complex biological sample with the plurality of particles such that proteins of the complex biological sample bind to the plurality of particles, and a binding pattern of the proteins between the plurality of particles is configured to depend on the physicochemical properties of the surfaces of the particles.

[0025] In yet another aspect, the invention provides a sensor array comprising a plurality of liposomes differing in at least one protein binding property defined by a lipid-based surface of each liposome, the lipid-based surface of each liposome configured to contact a subset of proteins of a sample at a lipid-protein interface and thereby bind to the subset of proteins to generate a pattern of protein binding, the pattern of protein binding of a first liposome differing in the at least one protein binding property from a pattern of protein binding of a second liposome different from the first liposome.

[0026] In another aspect, the invention provides a method for identifying a biological state of a subject using a plurality of liposomes differing in at least one protein binding property defined by the lipid-based surface of each liposome, comprising the steps of contacting a sample with a plurality of liposomes to bind proteins of the sample to the plurality of liposomes, separating the proteins from the liposomes, determining a biomolecular fingerprint of the proteins separated from the liposomes, and correlating the biomolecular fingerprint with the state of the complex biological sample of the subject, wherein the pattern of protein binding differs among the liposomes having different protein binding properties.

[0027] These and other aspects and advantages of the present invention will become apparent from the following description. The specification is described with reference to the accompanying drawings, which form a part hereof and which illustrate preferred embodiments of the present invention. Such embodiments do not necessarily represent the full scope of the invention, and therefore the scope of the invention should be interpreted based on the claims and the description of this specification.

[0028] The patent or application file contains at least one color drawing. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee. [Brief description of the drawings]

[0029] [Figure 1] FIG. 1 shows an example of a study design of a protein corona pattern approach for cancer detection. Three types of liposomes were incubated with plasma from healthy and cancer patients, and the protein corona patterns formed on each liposome in each subject's plasma (healthy and different cancers) were characterized by liquid chromatography-tandem mass spectrometry (LC-MS / MS). The formation of protein coronas on the three types of liposomes resulted in the enrichment of overlapping but distinct pools of selected plasma proteins, and the enriched proteins served as the basis for subsequent multivariate analyses. Through a classification approach, significant proteins in the corona patterns were identified and used to predict cancer using both blind plasma and cohort samples to test the accuracy of the multi-nanoparticle protein corona nanosystem. 29 human subjects (25 cancer patients, i.e., 5 patients for 5 cancer types and 4 healthy subjects) representing 261 separate runs of LC-MS / MS (3 liposomes, 29 subjects, 3 replicates) were used to train the classification model. 16 human subjects (3 patients per 5 cancer types and 1 healthy subject) from blind plasma and 144 separate runs of LC-MS / MS (3 liposomes, 16 subjects, 3 replicates) were used for cancer prediction, i.e. to test the classification model. 15 human subjects (5 patients per 3 cancer types) from cohort plasma representing 135 separate runs of LC-MS / MS (3 liposomes, 15 subjects, 3 replicates) were also used for ultra-early cancer prediction. [Figure 2A] TEM images of liposomes and their size distribution profile. [Figure 2B]Physicochemical properties of different liposomes before and after interaction with human plasma from patients with different diseases. DLS and zeta potential data (Pdi: molecular weight dispersity from cumulant fitting) for various liposomes before interaction with human plasma and corona composites without excess plasma obtained after incubation with plasma from healthy and cancer patients. [Figure 2C] Classification of identified corona proteins from liposomes according to their physiological functions in human plasma of healthy individuals and patients with different types of cancer (data shown reflects calculations from 5 biological plasmas for each group and 3 technical replicates for each plasma). [Figure 3A1] Classification of identified coronas by the sensor array element based on physiological functions including acute phase in human plasma of healthy subjects and patients with different types of cancer. [Figure 3A2] Classification of identified coronas by the sensor array element based on physiological functions including acute phase in human plasma of healthy subjects and patients with different types of cancer. [Figure 3A3] Classification of identified coronas by the sensor array element based on physiological functions including acute phase in human plasma of healthy subjects and patients with different types of cancer. [Figure 3B1] Classification of identified coronas by the sensor array element based on physiological functions including clotting in human plasma of healthy subjects and patients with different types of cancer. [Figure 3B2] Classification of identified coronas by the sensor array element based on physiological functions including clotting in human plasma of healthy subjects and patients with different types of cancer. [Figure 3B3] Classification of identified coronas by the sensor array element based on physiological functions including clotting in human plasma of healthy subjects and patients with different types of cancer. [Figure 3C1]Classification of identified coronas by the sensor array element based on physiological functions including immunoglobulins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3C2] Classification of identified coronas by the sensor array element based on physiological functions including immunoglobulins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3C3] Classification of identified coronas by the sensor array element based on physiological functions including immunoglobulins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3D1] Classification of identified coronas by the sensor array element based on physiological functions involving lipoproteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3D2] Classification of identified coronas by the sensor array element based on physiological functions involving lipoproteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3D3] Classification of identified coronas by the sensor array element based on physiological functions involving lipoproteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3E1] Classification of coronas identified by the sensor array element based on physiological functions including tissue leakage in human plasma of healthy subjects and patients with different types of cancer. [Figure 3E2] Classification of coronas identified by the sensor array element based on physiological functions including tissue leakage in human plasma of healthy subjects and patients with different types of cancer. [Figure 3E3] Classification of coronas identified by the sensor array element based on physiological functions including tissue leakage in human plasma of healthy subjects and patients with different types of cancer. [Figure 3F1] Classification of identified coronas by the sensor array element based on physiological functions involving complement proteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3F2]Classification of identified coronas by the sensor array element based on physiological functions involving complement proteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3F3] Classification of identified coronas by the sensor array element based on physiological functions involving complement proteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3G1] Classification of identified coronas by the sensor array element based on physiological functions including other plasma proteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3G2] Classification of identified coronas by the sensor array element based on physiological functions including other plasma proteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 3G3] Classification of identified coronas by the sensor array element based on physiological functions including other plasma proteins in human plasma of healthy subjects and patients with different types of cancer. [Figure 4A] Finding predictors for separation of each class by PLS discriminant analysis and contribution of individual predictors. Weighted VIP predictor exploration was performed by adding ranked variables one by one to the PLS-DA model and calculating the classification error for internal cross-validation (10-fold). A minimal set of 69 predictors was found with a reduction in classification error, indicating the highest potential importance for separating each class from the others. Contribution of individual markers to separation of each class based on PLS discriminant analysis. VIP plot ranking markers of 69 selected variables for their contribution to separation of each class from PLS discriminant analysis. A VIP score >1 indicates important proteins that give good prediction of class membership, while variables with a VIP score <1 indicate insignificant proteins for each class. [Figure 4B] The results of 69 variables for glioblastoma are shown. [Figure 4C] The VIP values ​​of 69 variables for meningiomas are shown. [Figure 4D]The VIP values ​​of 69 variables for lung cancer are shown. [Figure 4E] The VIP values ​​of 69 variables for myeloma are shown. [Figure 4F] The VIP values ​​of 69 variables for pancreatic cancer are shown. [Figure 5A] Figure 1 shows a PLS-DA plot showing the separation of different cancer samples from each other and from controls.Figure 2 shows a PLS score plot obtained using the PLS toolbox, projecting the subjects into the subspace created by the first, second and third latent variables of the model. [Figure 5B] Figure 14 is a PLS-DA plot showing the separation of different cancer samples from each other and from controls. Subjects are displayed where the fourth and fifth latent variables of the model are shown. As can be seen, meningioma and glioblastoma cases were not well separated in three dimensions, but can be separated in the fourth and fifth dimensions of the PLS model. [Figure 5C-D] Assignment maps obtained by CPANN with all and selected variables. On the left, the assignment map obtained by training a CPANN network (8x8 neurons) with the entire dataset (1823 variables). The quality of the mapping is poor and there are conflicts in the mapping for different cancer types. On the right, the assignment map achieved by training a CPANN network (8x8 neurons) with 69 variables. High-dimensional input vectors (samples) are mapped onto a 2D network of neurons to preserve similarity and topology. The color indicates the similarity of neurons to a particular type of input vector (class type). The map also shows the importance of the predictor selection process and the impact on the model quality of removing uninformative and irrelevant predictors. [Figure 5E] Dendrogram depicting 51 proteins identified as discriminating between six cancer groups. [Figure 5F]The 51 proteins identified as discriminating between the six cancer groups are shown in a "heatmap" created in conjunction with an unsupervised clustering algorithm (agglomerative HCA using longest distance linkage). Visual inspection of both the dendrogram (Figure 5E) and the heatmap (Figure 5F) showed clear clustering of the cancer-specific protein corona signatures and the six sample groups (five groups of cancer samples plus normal samples), as well as the expected similarity of the five patients for each group. The heatmap also shows striking differences in the patterns of variables (markers) for the different cancers (each example represents a patient, each row represents a protein). Higher and lower protein levels are shown in red and green, respectively. [Figure 6A] PLS score plot obtained by considering 69 significant markers, with the cohort subjects projected in the subspace created by the first and second latent variables of the model. [Figure 6B] A PLS-DA model was created using the eight variables, with cohort subjects projected into the subspace created by the first and second latent variables of the model with good statistics. [Figure 6C] The assignment map was obtained by training a CPANN network (8 × 8 neurons) with 69 significant markers. [Figure 6D] Assignment map obtained by training a CPANN network (8x8 neurons) with only 8 markers, with no misclassifications. The number of samples is indicated on each neuron. [Figure 7A] FIG. 1 is a schematic diagram of the study outline, showing the selection of informative variables and the construction of a classification model. [Figure 7B] Protein names and 69 selected variable IDs are listed. Some proteins were present in the protein corona of more than one liposome (DOPG, DOTAP, and CHOL are shown in italic and underlined font, bold font, and plain font, respectively). [Figure 7C] These are disease-specific biomarkers covered as significant variables by the proposed model. [Figure 8A] Receiver operating characteristic (ROC) plots obtained from PLS-DA based on the top 69 ranked variables for controls. ROC plots of sensitivity (true positive rate, Y-axis) versus 1-specificity (false positive rate, X-axis) based on PLS-DA constructed from the 69 markers with the highest contributions for six classes (control, glioblastoma, meningioma, myeloma, pancreatic cancer, and lung cancer). [Figure 8B] Receiver operating characteristic (ROC) plots obtained from PLS-DA based on the top 69 ranked variables for glioblastoma. ROC plots of sensitivity (true positive rate, Y-axis) versus 1-specificity (false positive rate, X-axis) based on PLS-DA constructed from the 69 markers with the highest contributions for six classes (control, glioblastoma, meningioma, myeloma, pancreatic cancer, and lung cancer). [Figure 8C] Receiver operating characteristic (ROC) plots obtained from PLS-DA based on the top 69 ranked variables for meningioma. ROC plots of sensitivity (true positive rate, Y-axis) versus 1-specificity (false positive rate, X-axis) based on PLS-DA constructed from the 69 markers with the highest contributions for six classes (control, glioblastoma, meningioma, myeloma, pancreatic cancer, and lung cancer). [Figure 8D] Receiver operating characteristic (ROC) plots obtained from PLS-DA based on the top 69 ranked variables for myeloma. ROC plots of sensitivity (true positive rate, Y-axis) versus 1-specificity (false positive rate, X-axis) based on PLS-DA constructed from the 69 markers with the highest contributions for six classes (control, glioblastoma, meningioma, myeloma, pancreatic cancer, and lung cancer). [Figure 8E]Receiver operating characteristic (ROC) plots obtained from PLS-DA based on the top 69 ranked variables for pancreatic cancer. ROC plots of sensitivity (true positive rate, Y-axis) versus 1-specificity (false positive rate, X-axis) based on PLS-DA constructed from the 69 markers with the highest contributions for six classes (control, glioblastoma, meningioma, myeloma, pancreatic cancer, and lung cancer). [Figure 8F] Receiver operating characteristic (ROC) plots obtained from PLS-DA based on the top 69 ranked variables for lung cancer. ROC plots of sensitivity (true positive rate, Y-axis) versus 1-specificity (false positive rate, X-axis) based on PLS-DA constructed from the 69 markers with the highest contributions for six classes (control, glioblastoma, meningioma, myeloma, pancreatic cancer, and lung cancer). [Figure 9A] FIG. 1 is a schematic diagram showing the expansion of a three-way data matrix into a two-way matrix. [Figure 9B] Assignment maps obtained by CPANN (14 × 14) trained with all 1823 variables using 90 samples (replicates). The number of samples is indicated on each neuron. The color of the neuron (assigned label) was determined based on the similarity between the class label (6 × 1 binary vector) and the weight vector in the output layer of the corresponding neuron. Despite using all biomarkers, there were different similarities between samples of the same cancer class. Replicate samples were also mapped onto adjacent or identical neurons. [Figure 9C] The classification error of the CPANN maps was calculated using 10-fold cross-validation at different map sizes. [Figure 9D] The CPANN network has 69 weight layers, equal to the number of variables used to train the model. The i-th weight layer reflects the influence of the i-th variable (biomarker) on the pattern of the assignment map. [Fig. 9EF]Correlations between the assignment map and the 69 weight layers (weight maps) can be calculated to help identify biomarkers associated with each cancer class, and correlations can also be determined visually and the similarity monitored by the absolute value of the correlation coefficients of the two maps. [Figure 10A] Figure 1 shows the importance of proteins in the classification versus the percentage of proteins adsorbed onto protein corona nanosystems. Panels (a)-(c) illustrate the importance of observed protein-liposome interactions ("variables") in predicting specific cancers. Proteins are grouped by their physiological function. Panels (d)-(f) illustrate the percentage of proteins adsorbed onto each liposome. The protein-liposome groups that emerge as relevant for cancer prediction are distinct depending on the cancer (panels (a)-(c)). Moreover, this difference is substantially more pronounced than the variability in the percentage of proteins adsorbed onto liposomes in these cancers (panels (d)-(f)). [Figure 10B] Venn diagram showing the number of unique proteins identified in the corona composition of each liposome and their combinations (the table on the right shows the same data numerically). [Figure 10C] Variable importance in classification is shown. Each column indicates the importance of a specific protein. The three dots in each column correspond to the importance of the observed interactions with each of the three liposomes. The horizontal lines spanning between the dots indicate the 25th and 75th percentiles of importance for a classifier trained on 1000 random draws of the training set from the data. These confidence intervals indicate the "stability" of the trained model to random draws of the data in terms of the protein-liposome interactions on which it depends. [Figure 11]Weighted average of protein-liposome interaction classification cancer. Distribution of absolute Z scores for each patient group, histogrammed for the 100 most abundant proteins (grey) and biomarkers identified so far (white with black dots). Long black bars correspond to Z scores for linear combinations of protein-liposome interactions. A large Z score for a specific protein-liposome interaction indicates that the group "separates" from the rest of the patients in that particular interaction. Thus, the figure shows that while no individual protein-liposome interaction is sufficient to classify the cancer, the weighted combination causes a separation of 2 to 2.5 standard deviations. [Figure 12] Multiliposomes concentrate low abundance and rare proteins. Protein corona contribution vs known plasma concentration plotted on log-log scale. Each point represents a single protein and liposome with corona contribution for each diseased and healthy individual, corona contribution and plasma concentration normalized to albumin. Plasma concentrations vary over 10^10 and liposome arrays detect these same proteins over 10^4-10^5. Corona contributions of proteins with unknown / unreported plasma concentrations are plotted in the red area on the right. [Figure 13] Examples of types of nanoscale sensor elements that can be used in some embodiments of a sensor array are shown below. Different types of nanoparticles (e.g., organic, inorganic and polymeric nanoparticles) with various physicochemical properties (e.g., different surface properties, sizes and shapes) can be used as sensor elements. Sensor arrays can be made with a minimum of two elements to an unlimited number of elements. [Figure 14] One method of collecting corona-coated particles in in vitro, ex vivo and in vivo conditions is illustrated below: The particles are incubated with biological fluids (e.g., plasma from patients with different types of diseases) and the corona-coated particles are collected and stored for analysis. [Figure 15A]We provide an example of binding nano object materials (with different physicochemical properties) to a substrate (with different physicochemical properties) to create a protein corona sensor array chip before interaction with a protein source (e.g. human plasma with various diseases). Specific protein coronas are formed on the surface of nano object materials with different physicochemical properties. The substrate may also be coated with several kinds of proteins, which has little effect on the detection effectiveness of the chip. [Figure 15B] Examples are given of the binding of nano object materials (with different physicochemical properties) to a substrate (with different physicochemical properties) to create a protein corona sensor array chip after interaction with a protein source (e.g. human plasma with various diseases). [Figure 16A] We present examples of protein corona sensor array chips (fabricated by a wide range of available approaches such as lithography and mold casting) with nanocurvatures with different physicochemical properties before interaction with protein sources (e.g. human plasma with various diseases). Specific protein coronas are formed on the surface of nano object materials with different physicochemical properties. The substrate may also be coated with several kinds of proteins, which has little effect on the detection effectiveness of the chip. [Figure 16B] We present examples of protein corona sensor array chips (fabricated by a wide range of available approaches such as lithography and mold casting) with nanocurvatures with different physicochemical properties after interaction with protein sources (e.g. human plasma with various diseases). Specific protein coronas are formed on the surface of nano-object materials with different physicochemical properties. The substrate may also be coated with several kinds of proteins, which has little effect on the detection effectiveness of the chip. [Figure 17A]We provide examples of binding of nano object materials (with different physicochemical properties) to substrates (with different physicochemical properties) to create protein corona sensor array micro / nanofluidic chips before interaction with protein sources (e.g. human plasma with various diseases). Specific protein coronas are formed on the surface of nano object materials with different physicochemical properties. Also, the substrate may be coated with several kinds of proteins, which has little effect on the detection effectiveness of the chip. [Figure 17B] We present an example of the conjugation of nano-object materials to a substrate to create a protein corona sensor array micro / nanofluidic chip after interaction with a protein source (e.g. human plasma with various diseases). Specific protein coronas are formed on the surface of nano-object materials with different physicochemical properties. [Figure 18A] We present examples of protein corona sensor array micro / nanofluidic chips (fabricated by a wide range of available approaches such as lithography and mold casting) with nanocurvatures having different physicochemical properties prior to interaction with a protein source (e.g. human plasma with various diseases). [Figure 18B] We present examples of protein corona sensor array micro / nanofluidic chips (fabricated by a wide range of available approaches such as lithography and mold casting) with nanocurvatures having different physicochemical properties after interaction with a protein source (e.g. human plasma with various diseases). [Figure 19A] An example is shown of the attachment of random-order nano-object materials (with different physicochemical properties) to a substrate (with different physicochemical properties) to create a protein corona sensor array chip prior to interaction with a protein source (e.g., human plasma with various diseases). [Figure 19B] An example is shown of the binding of random-order nano-object materials (with different physicochemical properties) to a substrate (with different physicochemical properties) to create a protein corona sensor array chip after interaction with a protein source (e.g., human plasma with various diseases). [Figure 20A]We present examples of protein corona sensor array chips (fabricated by a wide range of available approaches such as lithography and mold casting) with random order nanocurvatures with different physicochemical properties prior to interaction with a protein source (e.g., human plasma with various diseases). [Figure 20B] We present examples of protein corona sensor array chips (fabricated by a wide range of available approaches such as lithography and mold casting) with random order nanocurvatures with different physicochemical properties after interaction with a protein source (e.g., human plasma with various diseases). [Figure 21A] We provide examples of binding of random-order nano-object materials (with different physicochemical properties) to a substrate (with different physicochemical properties) to create a protein corona sensor array micro / nanofluidic chip prior to interaction with a protein source (e.g. human plasma with various diseases). [Figure 21B] Examples of binding of random-order nano-object materials to substrates for creating protein corona sensor array micro / nanofluidic chips after interaction with a protein source (eg, human plasma for various diseases) are shown. [Figure 22A] We present examples of protein corona sensor array micro / nanofluidic chips (fabricated by a wide range of available approaches such as lithography and mold casting) with random order nanocurvatures with different physicochemical properties prior to interaction with a protein source (e.g. human plasma with various diseases). [Figure 22B] We present examples of protein corona sensor array micro / nanofluidic chips (fabricated by a wide range of available approaches such as lithography and mold casting) with random order nanocurvatures with different physicochemical properties after interaction with a protein source (e.g. human plasma with various diseases). [Figure 23] An example is shown of the attachment of nanoparticles (NPs) to the surface of a silica substrate (as a representative substrate) via an amidation reaction between amino groups on the silica substrate surface and carboxylic acid groups on the nanoparticle surface. [Figure 24]An example is shown of the attachment of nanoparticles to the surface of a silica substrate (as a representative substrate) via a ring-opening reaction between an epoxy group on the silica substrate surface and an amino group on the nanoparticle surface. [Diagram 25] An example is shown of the attachment of nanoparticles to the surface of a silica substrate (as a representative substrate) via a Michael addition reaction between maleimide groups on the silica substrate surface and thiol or amino groups on the nanoparticle surface. [Figure 26] An example is shown of nanoparticles being bonded to the surface of a silica substrate (as a representative substrate) via a urethanization reaction between an isocyanate group on the silica substrate surface and a hydroxyl or amino group on the nanoparticle surface. [Figure 27] An example is shown of the binding of nanoparticles to the surface of a silica substrate (as a representative substrate) via an oxidation reaction between a thiol group on the surface of the silica substrate and a thiol group on the surface of the nanoparticle. [Figure 28] An example is shown of the attachment of nanoparticles to the surface of a silica substrate (as a representative substrate) via "click" chemistry between an azide group on the silica substrate surface and an alkyne group on the nanoparticle surface. [Figure 29] An example is shown of the attachment of nanoparticles to the surface of a silica substrate (as a representative substrate) via a thiol exchange reaction between 2-pyridinethiol groups on the silica substrate surface and thiol groups on the nanoparticle surface. [Diagram 30] An example is shown of the attachment of nanoparticles to the surface of a silica substrate (as a representative substrate) via a coordination reaction between a boronic acid group on the silica substrate surface and a diol group on the nanoparticle surface. [Diagram 31] An example of binding of nanoparticles to the surface of a silica substrate (as a representative substrate) via a UV light irradiation addition reaction between a C=C bond on the silica substrate surface and a C=C bond on the nanoparticle surface is shown. [Diagram 32] An example of nanoparticle attachment to a gold substrate surface (as a representative substrate) via Au-thiol bonds is shown. [Diagram 33] An example is shown of the attachment of nanoparticles to a gold substrate surface (as a representative substrate) via an amidation reaction between carboxylic acid groups on the gold substrate surface and amino groups on the nanoparticle surface. [Diagram 34] An example is shown of the attachment of nanoparticles to a gold substrate surface (as a representative substrate) via "click" chemistry between azide groups on the gold substrate surface and alkyne groups on the nanoparticle surface. [Diagram 35] An example is shown of the attachment of nanoparticles to a gold substrate surface (as a representative substrate) via a urethanization reaction between NHS groups on the gold substrate surface and amino groups on the nanoparticle surface. [Diagram 36] An example is shown of the attachment of nanoparticles to a gold substrate surface (as a representative substrate) via a ring-opening reaction between an epoxy group on the gold substrate surface and an amino group on the nanoparticle surface. [Figure 37] An example is shown of the attachment of nanoparticles to a gold substrate surface (as a representative substrate) via a coordination reaction between boronic acid groups on the gold substrate surface and diol groups on the nanoparticle surface. [Figure 38] An example is shown of the bonding of nanoparticles to a gold substrate surface (as a representative substrate) via a UV light irradiation addition reaction between a C=C bond on the gold substrate surface and a C=C bond on the nanoparticle surface. [Figure 39] An example is shown of the binding of nanoparticles to a gold substrate surface (as a representative substrate) via a "ligand-receptor" interaction between biotin on the gold substrate surface and avidin on the nanoparticle surface. [Diagram 40] An example is shown of the binding of nanoparticles to a gold substrate surface (as a representative substrate) via "host-guest" interactions between a-cyclodextrin (a-CD) on the gold substrate surface and adamantine (Ad) on the nanoparticle surface. [Diagram 41] 1 shows the release of proteins from the surface of nanoparticles and their corona composition analysis. 2 shows the analysis of protein corona sensor array data using supervised and unsupervised approaches for disease identification and differentiation. [Diagram 42] (a)-(d) are examples of sensor arrays with random-order nanoscale sensor elements for fluorescence and luminescence readout. [Diagram 43](a)-(d) are examples of sensor arrays with on-the-order nanoscale sensor elements for fluorescence and luminescence readout. [Figure 44A] Characterization of bare polystyrene and silica nanoparticles with different functional groups (unmodified, amine-modified (NH2) and carboxyl-modified (COOH)). Three different polystyrene nanoparticles were used (unmodified, P-NH2 and P-COOH), along with their size, DLS and zeta potential of the bare particles. [Figure 44B] Characterization of uncoated polystyrene and silica nanoparticles with different functional groups (unmodified, amine-modified (NH2) and carboxyl-modified (COOH)). Three different silica nanoparticles were used (unmodified, S-NH2 and S-COOH) along with their size, DLS and zeta potential. [Figure 44C] Characterization of uncoated polystyrene and silica nanoparticles with different functional groups (unmodified, amine-modified (NH2) and carboxyl-modified (COOH)). TEM of uncoated polystyrene nanoparticles. [Fig.44D] Characterization of uncoated polystyrene and silica nanoparticles with different functional groups (unmodified, amine-modified (NH2) and carboxyl group-modified (COOH)). TEM of uncoated silica nanoparticles. [Figure 45A] Characterization of protein corona coated polystyrene and silica nanoparticles with different functional groups (unmodified, amine modified (NH2) and carboxyl group modified (COOH)). DLS and zeta potential of polystyrene nanoparticles topped with protein corona. [Figure 45B] Characterization of protein corona coated polystyrene and silica nanoparticles with different functional groups (unmodified, amine modified (NH2) and carboxyl group modified (COOH)). DLS and zeta potential of silica nanoparticles topped with protein corona. [Figure 45C]Characterization of protein corona coated polystyrene and silica nanoparticles with different functional groups (unmodified, amine modified (NH2) and carboxyl group modified (COOH)). TEM of polystyrene nanoparticles with protein corona. [Figure 45D] Characterization of protein corona coated polystyrene and silica nanoparticles with different functional groups (unmodified, amine modified (NH2) and carboxyl group modified (COOH)). TEM of silica nanoparticles topped with protein corona. [Diagram 46] FIG. 1. Types of cancer plasma samples screened using polystyrene and silica nanoparticles. [Figure 47] Protein corona profiles of polystyrene and silica nanoparticles (100 nm) with unmodified, amine-modified and carboxyl-modified surfaces after incubation with plasma of patients with different cancers analyzed using SDS PAGE. [Figure 48] Protein corona profiles of polystyrene and silica nanoparticles (100 nm) with unmodified, amine-modified and carboxyl-modified surfaces after incubation with plasma from healthy individuals analyzed using SDS-Page. [Figure 49] 1 is a plot showing separation of healthy individuals and cancer patients using a sensor array of the present invention. [Figure 50A] 1 shows the characterization of polystyrene and silica nanoparticles used for CAD screening, showing profiles of uncoated, CAD, no CAD and control treated nanoparticles. [Figure 50B] 1 shows the characterization of polystyrene and silica nanoparticles used for CAD screening, showing the zeta potential for different groups of nanoparticles. [Figure 50C] FIG. 1 shows the characterization of polystyrene and silica nanoparticles used for CAD screening, showing TEM for different groups of nanoparticles in the CAD screen. [Figure 51A]FIG. 1 shows the protein concentrations of the different protein coronas obtained from the analysis of nanoparticles with CAD, no CAD and no risk of CAD (control), and Bradford assay of the protein concentrations of the different protein coronas. [Figure 51B] FIG. 1 shows the protein concentrations of the different protein coronas obtained from the analysis of CAD, no CAD and no risk of CAD (control) nanoparticles, indicating that individualized protein corona profiles were analyzed and compared by 1D-SDS-PAGE. [Figure 51C] FIG. 1 shows the protein concentrations of the different protein coronas obtained from the analysis of CAD, no CAD and no risk of CAD (control) nanoparticles, and shows that the differences in the amount of protein in the protein corona (PC) of CAD, no CAD and control were determined by gel analysis by densitometry. [Figure 52-1] 13 is a bar showing the difference in percentage contribution of the top 20 abundant proteins in the PC. [Figure 52-2] 13 is a bar showing the difference in percentage contribution of the top 20 abundant proteins in the PC. [Figure 52-3] 13 is a bar showing the difference in percentage contribution of the top 20 abundant proteins in the PC. [Figure 52-4] 13 is a bar showing the difference in percentage contribution of the top 20 abundant proteins in the PC. [Figure 52-5] 13 is a bar showing the difference in percentage contribution of the top 20 abundant proteins in the PC. [Figure 52-6] 13 is a bar showing the difference in percentage contribution of the top 20 abundant proteins in the PC. [Figure 53] 1 is a plot showing classification of subjects into CAD, no CAD, and control by analysis of fingerprints produced by protein corona-coated nanoparticles. [Figure 54A]Synthetic and biological identity of nanoparticles after incubation in Alzheimer's plasma. Nanosight™ nanoparticle tracking analysis (size). Polystyrene nanoparticles before and after coating with AD protein corona. Uncoated nanoparticles are 90-100 nm and uniform in size. Alzheimer's disease (AD) protein corona (PC) coated nanoparticles are larger and more heterogeneous in size. Intensity profiles and scatter plots of each measurement are reported. Values ​​are mean ± SD (n=3). [Figure 54B] Synthetic and biological identity of nanoparticles after incubation in Alzheimer's plasma. Nanosight™ nanoparticle tracking analysis (size). Silica nanoparticles before and after coating with Alzheimer's protein corona. Uncoated nanoparticles are 90-100 nm and uniform in size. Alzheimer's protein corona coated nanoparticles are larger and more heterogeneous in size. Intensity profiles and scatter plots of each measurement are reported. Values ​​are mean ± SD (n=3). [Figure 55-1] TEM analysis. Nanoparticles before and after coating with AD protein corona were analyzed by transmission electron microscopy to evaluate potential changes in morphology and size. P: polystyrene; PN: polystyrene-NH2; PC: polystyrene-COOH; S: silica; SN: silica-NH2; SC: silica-COOH. All nanoparticles showed an increase in size after incubation in plasma. [Figure 55-2] TEM analysis. Nanoparticles before and after coating with AD protein corona were analyzed by transmission electron microscopy to evaluate potential changes in morphology and size. P: polystyrene; PN: polystyrene-NH2; PC: polystyrene-COOH; S: silica; SN: silica-NH2; SC: silica-COOH. All nanoparticles showed an increase in size after incubation in plasma. [Figure 56]SDS-PAGE gels and densitometric analysis of binding. The loading order is P, P-NH2, P-COOH, S, S-NH2, S-COOH, where P: polystyrene; PN: polystyrene-NH2; PC: polystyrene-COOH; S: silica; SN: silica-NH2; SC: silica-COOH. Individualized protein corona profiles were analyzed and compared using SDS-PAGE. Four representative gels of Alzheimer's disease protein corona and one healthy protein corona are shown. The intensity of binding to plasma proteins adsorbed on the nanoparticles was analyzed with ImageJ (y-axis: intensity, x-axis: molecular weight). [Figure 57] Healthy and AD disease classification. White dots are AD and black dots are healthy samples. [Figure 58] SDS-Page gel analysis of silica nanoparticles (NPs) of different sizes loaded with equal volumes (left, 10 ul) or with equal amounts (right, 10 ug). [Figure 59] FIG. 1 is a schematic showing an experiment performed to demonstrate the presence of nucleic acids in the biomolecular corona. [Figure 60] Agarose gel analysis of nucleic acids bound to three different nanoparticles. [Figure 61-1] Analysis of nucleic acid content in plasma. [Figure 61-2] Analysis of nucleic acid content in plasma. [Figure 62-1] Analysis of the amount of nucleic acid associated with the biomolecular corona of nanoparticles upon dissociation of proteins from the corona by urea. [Figure 62-2] Analysis of the amount of nucleic acid associated with the biomolecular corona of nanoparticles upon dissociation of proteins from the corona by urea. [Figure 63-1] Analysis of the amount of nucleic acid associated with the biomolecular corona of nanoparticles when the corona proteins do not dissociate from the particle surface. [Figure 63-2] Analysis of the amount of nucleic acid associated with the biomolecular corona of nanoparticles when the corona proteins do not dissociate from the particle surface. [Figure 64-1]Nucleic acids were first purified from plasma using a purification kit and then assayed for the amount of nucleic acid associated with the biomolecular corona of the nanoparticles upon incubation with the nanoparticles. [Figure 64-2] Nucleic acids were first purified from plasma using a purification kit and then assayed for the amount of nucleic acid associated with the biomolecular corona of the nanoparticles upon incubation with the nanoparticles. [Figure 65] FIG. 1 is a schematic diagram of a method for distinguishing states of a complex biological sample. [Figure 66] FIG. 1 is a schematic diagram of a computer system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0030] While the invention has been described in terms of one or more preferred embodiments, many equivalents, alternatives, variations, and modifications, aside from those expressly stated, are possible and are within the scope of the invention.

[0031] The present invention provides a sensor array and method for use in prognosis, diagnosis and detection of disease conditions in a subject. The sensor array of the present invention is different from known sensor arrays that include individual sensors that detect specific biomolecules. In the sensor array of the present invention, the biomolecules do not need to be known as the system does not rely on the presence or absence of a specific biomolecule or the amount of a specific disease marker. The novel sensor array is capable of detecting changes in the composition of the biomolecular corona associated with different sensor elements. This ability to detect relative changes or patterns (of the actual biomolecules associated with the different sensor elements or of the amount and / or combination of each biomolecule associated with each sensor element) allows a unique biomolecular fingerprint to be determined for each array. This biomolecular fingerprint allows for stratification of different health and disease states of subjects. In some embodiments, the biomolecular fingerprint can not only distinguish healthy subjects from subjects at various different stages of a disease or disorder, but can also determine a pre-disease state in a subject that later develops a disease or disorder. This is significantly different and novel from systems in the art that measure or detect specific biomarkers associated with a disease or disorder to provide trends (e.g., changes or likelihood) of disease progression. The sensors and methods of the present invention are capable of detecting disease before any signs or symptoms appear, or in other words, of pre-diagnosing disease before any concrete signs or symptoms appear.

[0032] What is unique about the present invention is the combination of the recognition of biomolecular fingerprints from samples from such subjects and the ability to determine the disease status of these subjects in the health care continuum.

[0033] The present invention is based on the inventors' research showing that the surface of a sensor element, e.g. a nanomolecule, when in contact with a biological sample is rapidly covered by a layer of different biomolecules, including proteins, forming a "biomolecular corona". The type, amount and category of biomolecules that compose these biomolecular coronas are strongly related to the physicochemical properties of the sensor element itself and to the complex interactions between the different biomolecules themselves and the sensor element. These interactions result in This generates a unique biomolecular corona signature for each sensor element. In other words, which biomolecule interacts with a sensor element not only affects the composition of the biomolecular corona, but may also alter which other different biomolecules can interact with that particular sensor element.

[0034] Different sensor elements, each with its own biomolecular corona signature, can contact a sample to generate a unique biomolecular fingerprint for that sample. This fingerprint can then be used to determine the disease state of a subject. Embodiments of the invention are described in more detail below.

[0035] The present invention provides a sensor array comprising, consisting essentially of, or consisting of a plurality of sensor elements, the plurality of sensor elements differing from one another in at least one physicochemical property. In some embodiments, each sensor element is capable of binding a plurality of biomolecules in a sample to generate a biomolecular corona signature. In some embodiments, each sensor element has a distinct biomolecular corona signature.

[0036] The plurality of sensor elements, when in contact with a sample, produces a plurality of biomolecular corona signatures, which collectively form a biomolecular fingerprint. A "biomolecular fingerprint" is a combined composition or pattern of biomolecules that is comprised of at least two biomolecular corona signatures for the plurality of sensor elements. do.

[0037] As used herein, the term "sensor element" refers to an element capable of binding to a plurality of biomolecules when contacted with a sample, and encompasses the term "nanoscale sensor element." In one embodiment, the sensor element is an element that is between about 5 nanometers and about 50,000 nanometers in at least one dimension.Suitable sensor elements include, for example and without limitation, sensor elements of about 5 nm to about 50,000 nm in at least one direction, and include about 5 nm to about 40,000 nm, or about 5 nm to about 30,000 nm, or about 5 nm to about 20,000 nm, or about 5 nm to about 10,000 nm, or about 5 nm to about 5000 nm, or about 5 nm to about 1000 nm, or about 5 nm to about 500 nm, or about 5 nm to about 50 nm, or about 10 nm to about 100 nm, or about 20 nm to about 200 nm, or about 30 nm to about 300 nm. or about 40 nm to about 400 nm, or about 50 nm to about 500 nm, or about 60 nm to about 600 nm, or about 70 nm to about 700 nm, or about 80 nm to about 800 nm, or about 90 nm to about 900 nm, or about 100 nm to about 1000 nm, or about 1,000 nm to about 10,000 nm, or about 10,000 nm to about 50,000 nm, and any combination or amount therebetween (e.g., 5 nm, 10 nm, 15 nm, 20 nm, 25 nm, 30 nm, 35 nm, 40 nm, 45 nm, 5 0nm, 55nm, 60nm, 65nm, 70nm, 80nm, 90nm, 100nm, 125nm, 150nm, 175nm, 2 00nm, 225nm, 250nm, 275nm, 300nm, 350nm, 400nm, 450nm, 500nm, 550nm, 6 00nm, 650nm, 700nm, 750nm, 800nm, 850nm, 900nm, 1000nm, 1200nm, 1300 nm, 1400nm, 1500nm, 1600nm, 1700nm, 1800nm, 1900nm, 2000nm, 2500nm, 3 (000nm, 3500nm, 4000nm, 4500nm, 5000nm, 5500nm, 6000nm, 6500nm, 7000nm, 7500nm, 8000nm, 8500nm, 9000nm, 1000nm, 11000nm, 12000nm, 13000nm, 14000nm, 15000nm, 16000nm, 17000nm, 18000nm, 19000nm, 20000nm, 25000nm, 30000nm, 35000nm, 40000nm, 45000nm, 50000nm and any numbers in between). A nanoscale sensor structure is one micrometer in at least one direction. Suitable examples of ranges for nanoscale sensor elements include, but are not limited to, elements that are, for example, from about 5 nm to about 1000 nm in one direction, such as from about 5 nm to about 500 nm, or from about 5 nm to about 400 nm, or from about 5 nm to about 300 nm, or from about 5 nm to about 200 nm, or from about 5 nm to about 100 nm, or from about 5 nm to about 50 nm, or from about 10 nm to about 1000 nm, or from about 10 nm to about 750 nm, or from about 10 nm to about 500 nm, or from about 10 nm to about 250 nm, or from about 10 nm to about 200 nm, or from about 10 nm to about 100 nm, or from about 50 nm to about 1000 nm, or from about 50 nm to about 500 nm, or from about 50 nm to about 250 nm, or from about 50 nm to about 200 nm, or from about 50 nm to about 100 nm, and any combination, range, or amount therebetween. (e.g., 5 nm, 10 nm, 15 nm, 20 nm, 25 nm, 30 nm, 35 nm, 40 nm, 45 nm, 50 nm, 55 nm, 60 nm, 65 nm, 70 nm, 80 nm, 90 nm, 100 nm, 125 nm, 150 nm, 175 nm, 200 nm, 225 nm, 250 nm, 275 nm, 300 nm, 350 nm, 400 nm, 450 nm, 500 nm, 550 nm, 600 nm, 650 nm, 700 nm, 750 nm, 800 nm, 850 nm, 900 nm, 1000 nm, etc.). With respect to the sensor arrays described herein, use of the term sensor element includes the use of nanosensor elements with respect to sensors and related methods.

[0038] The term "plurality of sensor elements" refers to two or more, e.g., at least two sensor elements. In some embodiments, the plurality of sensor elements comprises at least two sensor elements to at least 1000 sensor elements, preferably from about two sensor elements to about 100 sensor elements. In preferred examples, the array comprises at least two to at least 100 sensor elements, alternatively at least two to at least 50 sensor elements, alternatively at least two to at least 30 sensor elements, alternatively at least two to at least 20 sensor elements, alternatively at least two to at least 10 sensor elements, alternatively at least three to at least 50 sensor elements, alternatively at least three to at least 30 sensor elements, alternatively at least three to at least 20 sensor elements, alternatively at least three to at least 10 sensor elements, alternatively at least four to at least 50 sensor elements, alternatively at least four to at least 30 sensor elements, alternatively at least four to at least 40 sensor elements, alternatively at least four to at least 50 sensor elements, alternatively at least four to at least 30 sensor elements, alternatively at least four to at least 40 sensor elements, alternatively at least four to at least 50 sensor elements, alternatively at least four to at least 30 sensor elements, alternatively at least four to at least 40 sensor elements, alternatively at least four to at least 50 sensor elements, alternatively at least four to at least 50 sensor elements, alternatively at least four to at least 3 ... At least 20 sensor elements, or at least 4 to at least 10 sensor elements, including any number of sensor elements possible therebetween (e.g., at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 5 2, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, etc.). In some embodiments, the sensor array comprises at least 6 sensor elements to at least 20 sensor elements, or at least 6 sensor elements to at least 10 sensor elements.

[0039] The term "plurality of nanoscale sensor elements" refers to more than one, e.g., at least two nanoscale sensor elements. In some embodiments, the plurality of nanoscale sensor elements comprises at least two nanoscale sensor elements to at least 1000 nanoscale sensor elements, preferably from about two nanoscale sensor elements to at least 100 nanoscale sensor elements. In preferred embodiments, the array comprises at least two to at least 100 nanoscale sensor elements, alternatively at least two to at least 50 nanoscale sensor elements, alternatively at least two to at least 30 nanoscale sensor elements, alternatively at least two to at least 20 nanoscale sensor elements. Alternatively, the sensor may include at least two to at least ten nanoscale sensor elements, alternatively at least three to at least fifty nanoscale sensor elements, alternatively at least three to at least thirty nanoscale sensor elements, alternatively at least three to at least twenty nanoscale sensor elements, alternatively at least three to at least ten nanoscale sensor elements, alternatively at least four to at least fifty nanoscale sensor elements, alternatively at least four to at least thirty nanoscale sensor elements, alternatively at least four to at least twenty nanoscale sensor elements, alternatively at least four to at least ten nanoscale sensor elements, and any number of nanoscale sensor elements possible therebetween (e.g., at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36 , 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 225, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 750, 800, etc.).

[0040] As used herein, the term "biomolecular corona" refers to a number of different biomolecules that can bind to a sensor element. The term "biomolecular corona" encompasses "protein corona," a term used in the art to refer to proteins, lipids, and other plasma components that bind to nanoparticles upon contact with a biological sample or biological system. As used herein, the term "biomolecular corona" also encompasses both soft and hard protein coronas, as used in the art. Regarding this, see, for example, Milani et al. "Reversible versus Irreversible Binding of Transferring to Polystyrene Nanoparticles: Soft and Hard Corona" ACS NANO, 2012, 6(3), pp. 2532-2541; Mirshafiee et al. "Impact of protein pre-coating on the protein corona composition and nanoparticle cellular uptake" Biomaterials vol. 75, Jan 2016 pp. 295-304, Mahmoudi et al. "Emerging understanding of the protein corona at the nano-bio interfaces" Nanotoday 11(6) Dec. 2016, pp. 817-832, and Mahmoudi et al. "Protein-Nanoparticle Interactions: Opportunities and Challenges" Chem. Rev., 2011, 111(9), pp. 5610-5637, the contents of which are incorporated herein by reference in their entirety. As described in these prior art documents, the adsorption curves show the construction of a strongly bound monolayer up to a point of monolayer saturation (at a geometrically defined protein to NP ratio), beyond which a secondary weakly bound layer is formed. The first layer is irreversibly bound (hard corona), whereas the second layer (soft corona) shows dynamic exchange. Proteins that adsorb with high affinity form a so-called "hard" corona consisting of strongly bound proteins that do not easily desorb, whereas proteins that adsorb with low affinity form a so-called "soft" corona consisting of weakly bound proteins. Soft and hard coronas can also be defined based on their exchange times. Hard coronas usually show significantly larger exchange times, on the order of several hours. See, for example, M. Rahman et al. Protein-Nanoparticle Interactions, Spring Series in Biophysics 15, 2013, which is incorporated herein by reference in its entirety.

[0041] The term "biomolecular corona signature" refers to the composition, signature or pattern of different biomolecules bound to each individual sensor. This signature refers not only to different biomolecules, but also to differences in the amount, level or quantity of biomolecules bound to the sensor element or the conformational state of the biomolecules bound to the sensor element. The biomolecular corona signature of each sensor element may contain some of the same biomolecules and / or may contain different biomolecules relative to other sensor elements and / or may differ in the level or quantity, type or confirmation of the biomolecules. The biomolecular corona signature may depend on the physicochemical properties of the sensor element as well as the nature of the sample and the exposure time. In some embodiments, the biomolecular corona signature may be determined by the physicochemical properties of the sensor element, the amount, the type or the confirmation of the biomolecules ... The signature is a protein corona signature. In other cases, the biomolecular corona signature is a polysaccharide corona signature. In yet other cases, the biomolecular corona signature is a metabolite corona signature. In some embodiments, the biomolecular corona signature is a lipidomic corona signature.

[0042] In some embodiments, the biomolecular corona signature comprises biomolecules found in a soft corona and a hard corona. In some embodiments, the soft corona is a soft protein corona. In some embodiments, the hard corona is a hard protein corona.

[0043] The term "biomolecule" refers to biological components that may be involved in the corona formation, including but not limited to proteins, polypeptides, polysaccharides, carbohydrates, lipids, lipoproteins, metabolites, oligonucleotides, metabolomes, or combinations thereof. The biomolecular corona signature of each sensor element may include some of the same biomolecules and / or may include different biomolecules relative to other sensor elements, and / or the level or quantity, type or certainty of biomolecules bound to each sensor element may differ. In one embodiment, the biomolecule is selected from the group of proteins, nucleic acids, lipids, and metabolomes.

[0044] In some embodiments, the sensor array comprises, consists essentially of, or consists of a first sensor element that generates a first biomolecular corona signature and at least one second sensor element that generates at least one second biomolecular corona signature when the sensor array contacts a biological sample. The biomolecular fingerprint is a combination of the first biomolecular signature and at least one second biomolecular corona signature. The biomolecular signature can be composed of at least two biomolecular corona signatures to many different biomolecular signatures, for example at least 1000 different biomolecular corona signatures. The biomolecular coronas may be analyzed (assayed) separately for each sensor element to determine the biomolecular corona signature for each element and combined to determine the biomolecular fingerprint, or two or more biomolecular coronas may be analyzed simultaneously to generate a biomolecular fingerprint at once.

[0045] In some embodiments, the biomolecular fingerprint comprises at least two biomolecular corona signatures. In some embodiments, the biomolecular fingerprint comprises at least two biomolecular corona signatures to at least 1000 biomolecular corona signatures, preferably from about two biomolecular corona signatures to about 100 biomolecular corona signatures. In preferred examples, the biomolecular fingerprint comprises at least two to at least 100 biomolecular corona signatures, alternatively at least two to at least 50 biomolecular corona signatures, alternatively at least two to at least 30 biomolecular corona signatures, alternatively at least two to at least 20 biomolecular corona signatures, alternatively at least two to at least 10 biomolecular corona signatures, alternatively at least three to at least 50 biomolecular corona signatures, alternatively at least three to at least 30 biomolecular corona signatures, alternatively at least three to at least 20 Alternatively, the corona signature may include at least three to at least ten biomolecular corona signatures, alternatively, the corona signature may include at least four to at least fifty biomolecular corona signatures, alternatively, the corona signature may include at least four to at least thirty biomolecular corona signatures, alternatively, the corona signature may include at least four to at least twenty biomolecular corona signatures, alternatively, the corona signature may include at least four to at least ten biomolecular corona signatures, and any number of biomolecular corona signatures possible therebetween (e.g., at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23 ,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,60,70,80,90,100,110,120,130,140,150,160,170,180,190,200,225,250,300,350,400,450,500,550,600,650,700,750,800 etc.).

[0046] Advances in proteomic analysis using mass spectrometry have provided new insights into changes occurring across the spectrum of health and disease, including early cancer. However, the sensitivity and specificity of mass spectrometry approaches are still insufficient for robust early cancer detection, in part due to the large noise generated by the 10,000 proteins that comprise the human proteome, with estimated concentrations ranging from 35-50 mg / ml for albumin to 1-10 pg / ml for some cytokines. Existing techniques require a trade-off between coverage and throughput of plasma protein processing. Several attempts have been made to significantly increase the current low level of protein detection, including depletion of highly abundant proteins, isobaric labeling at the peptide level for multiplexed relative quantification, post-depletion plasma fractionation strategies, biomarker harvesting techniques, mathematical approaches for the analysis of high-quality data sets, and multiplexed workflows (i.e., combination of approaches). Despite such efforts, mass spectrometry approaches to plasma proteomics have not achieved robust success for early cancer detection. Indeed, no previous studies, proteomic or otherwise, have reported accurate prediction and classification of the spectrum of cancer, including the earliest pre-symptomatic stages. The sensor array of the present invention provides the first detection system that accurately predicts and classifies disease states, including pre-symptomatic disease states, for a number of diseases.

[0047] Previous attempts have attempted to use "disease-specific protein coronas" to find a type of cancer using gel electrophoresis and changes in nanoparticle aggregate size. However, as shown in the following examples, subtle differences in the protein coronas on the surface of a type of nanoparticle are not sufficient for stable and accurate cancer identification and differentiation with acceptable prediction accuracy, mainly due to the persistent problem of insufficient proteomic coverage. The sensor array described in the present invention is capable of accurately classifying disease states. The sensor array described in the present invention is not only capable of predicting disease states, but also capable of classifying patients before disease onset (e.g., Alzheimer's patients) or classifying patients according to disease type (e.g., cancer type).

[0048] To substantially advance the capabilities of protein coronas for stable and accurate cancer detection with superior predictive capabilities, the inventors have developed a sensor array (sometimes referred to herein as a protein corona nanosystem or a sensor array nanosystem). Compared with previous approaches limited to the surface of a single nanoparticle, the sensor array provides significantly more comprehensive proteomic data over a wider dynamic range of plasma protein concentrations. To significantly increase the number and range of low-abundance and high-abundance proteins identified without protein depletion, the sensor array allows complex biological samples (e.g., human plasma samples) to be sampled using multiple nanoparticles with different physicochemical properties. This effectively reduces noise in the vast amount of available proteomic information, allowing for more accurate and early differentiation of disease-specific proteomic signatures. Furthermore, due to the combination of protein-nanoparticle and protein-protein interactions uniquely obtained using the sensor array, each type of protein can be present at different concentrations on the surface of different nanoparticles, thereby providing additional proteomic information. The use of multiple nanoparticles with different physicochemical properties is driven primarily by our recent discovery that even small changes in the physicochemical properties of the nanoparticles can induce dramatic yet reproducible changes in the protein corona composition.

[0049] The sensor array of the present invention has a sensitivity and dynamic range of 10 for protein detection using mass spectrometry approaches. The assay of the present invention is capable of detecting proteins found in the sub-nanogram range in a sample. This assay or approach has a much larger dynamic range than current assays for measuring proteins in a sample. For example, mass spectrometry only has a dynamic range of 104-106. This novel sensor array has the ability to sample a larger dynamic range than previously achievable. The sensor array of the present invention allows the detection and determination of low abundance and rare proteins that were previously undetectable.

[0050] The term "sample" refers to a biological sample or complex biological sample obtained from a subject. Suitable biological samples include, but are not limited to, biological fluids including, but not limited to, systemic blood, plasma, serum, lung lavage, cell lysate, menstrual blood, urine, processed tissue samples, amniotic fluid, cerebrospinal fluid, tears, saliva, semen, etc. In a preferred embodiment, the sample is a blood or serum sample. Plasma contains thousands of different proteins, with concentrations of these proteins varying by orders of magnitude of 12. The sensor array of the present invention can detect changes in these blood samples over time or across disease states of a subject.

[0051] In some embodiments, biological fluids or complex biological samples are prepared by methods and kits well known in the art. For example, some biological samples (e.g., menstrual blood, blood samples, semen, etc.) are first centrifuged at low speed to remove cellular debris, blood clots, and other cellular components that may interfere with the array. In other embodiments, tissue specimens may be processed, for example, mincing or homogenizing the tissue sample, treating with enzymes to break up the tissue, and / or centrifuging to remove cellular debris to allow analysis and extraction of biomolecules within the tissue sample. Methods for separating and / or properly preparing and storing blood samples are well known in the art, including, but not limited to, the addition of anticoagulants.

[0052] Suitable sensor elements include, but are not limited to, particles such as organic particles, inorganic particles, or combinations thereof. In some embodiments, the particles are, for example, nanoparticles, microparticles, micelles, liposomes, iron oxide, graphene, silica, protein-based particles, polystyrene, silver particles, gold particles, particle dots, palladium, platinum, titanium, and combinations thereof. In some embodiments, the nanoparticles are liposomes. One of ordinary skill in the art would be able to select and prepare suitable particles. Suitable nanoscale sensor elements are less than 1 micrometer in at least one dimension. In some aspects, the nanoscale sensor elements are less than about 100 nm in at least one dimension.

[0053] [overview] The present disclosure provides a method for detecting a disease state using a biomolecular corona nanosystem, hi one embodiment, the method comprises detecting a disease-specific protein corona.

[0054] <Sensor array> FIG. 65 shows an exemplary scheme of the array system of the present disclosure. As shown in step 1 of FIG. 65, a complex biological sample (e.g., blood 704) can be taken from a subject 702 exhibiting a biological state 703 (e.g., a disease state, such as before any physical symptoms of the disease and / or during early and intermediate stages of the disease). Suitable biological samples 704 include, but are not limited to, biological fluids, including, but not limited to, systemic blood, plasma, serum, lung lavage, cell lysate, menstrual blood, urine, processed tissue samples, amniotic fluid, cerebrospinal fluid, tears, saliva, semen, and the like. In a preferred embodiment, the sample is a blood or serum sample. In some embodiments, plasma 706 can be separated from blood cells 708 of subjects exhibiting a biological state 703 (e.g., healthy subjects (disease-free) and cancer patients (disease state)) as shown in step 2 of FIG. 65.

[0055] Next, as shown in step 3 of Figure 65, a complex biological sample (e.g., plasma 706) may be incubated with a sensor array 710 including a plurality of particles 712 having different physicochemical properties. The plurality of particles 712 may be incubated with the plasma 706 such that biomolecules in the plasma 706 (e.g., proteins in the plasma 706) may bind to one or more of the particles 712. The biomolecules (e.g., proteins) bound to the particles 712 may then be separated in a protein solution 714 for further analysis, e.g., to determine the composition of proteins bound to each type of particle 712 (e.g., anionic, neutral, and cationic particles 712), as shown in step 4 of Figure 65.

[0056] The protein solution 714 may be characterized by, for example, liquid chromatography-tandem mass spectrometry (LC-MS / MS) 716, as shown in step 5 of Figure 65. Proteins identified using LC-MS / MS 716 can then be analyzed in step 6 of Figure 65 to determine biomolecular fingerprints 718 associated with the biological state 703 (e.g., representative of proteins, nucleic acids, lipids, and polysaccharides that are bound to one or more particles 712).

[0057] In step 7 of Figure 65, a computer 720 (e.g., computer system 101 of Figure 66) can be used to associate the biomolecular fingerprints 718 (e.g., protein fingerprints) with a biological state 703 (e.g., health state, disease state). For example, the computer system 720 can be used to perform an analysis of the biomolecular fingerprints 718 (e.g., protein fingerprints) of at least two samples (e.g., complex biological samples such as plasma 706) to create an association 722 between the biological state 703 and the biomolecular fingerprints 718 (e.g., protein fingerprints) (step 8 of Figure 65). The associations 722 can be made by association analysis or statistical classification using methods well known in the art, including, but not limited to, various supervised and unsupervised data analysis and clustering approaches such as hierarchical cluster analysis (HCA), principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), machine learning (also known as random forests), logistic regression, decision trees, support vector machines (SVM), K-nearest neighbors, naive Bayes, linear regression, polynomial regression, SVM for regression, K-means, and hidden Markov models, among others. In other words, the biomolecular fingerprints 718 of each sample (e.g., plasma 706) are compared / analyzed to each other (e.g., using computer 720) to determine what patterns are common between the individual fingerprints with statistical significance in order to determine a biological state associated with the biomolecular (e.g., protein) fingerprints 718.

[0058] Associations 722 can link biomolecular fingerprints 718 (e.g., protein fingerprints) to various biological states 703. For example, by comparing biomolecular fingerprints 718 between subjects 702 diagnosed with a disease (i.e., biological state 703 is a disease state) and subjects 702 not diagnosed with a disease (i.e., biological state 703 is a disease-free state), associations 722 can be generated between the biomolecular fingerprints 718 of diseased subjects 702 and disease states. Such associations 722 between biomarker fingerprints 718 and disease states (i.e., biological states 703) can, in some embodiments, be determined very early in the progression of the disease (i.e., before any physical symptoms of the disease appear and / or before the disease is diagnosed) or late in the progression of the disease.

[0059] Other examples of biological states 703 that may be associated with a biomolecular fingerprint 718 include responsiveness or non-responsiveness to a drug or pharmaceutical agent, the level of immune system activation (e.g., due to a subject being exposed to a foreign antigen), and resistance to side effects associated with drug administration. This includes verifying the sensitivity of a subject to the administration of a particular ingredient or substance and the likelihood of the subject exhibiting an allergic reaction to the administration of the particular ingredient or substance.

[0060] <Computer Control System> The present disclosure provides a computer control system programmed to carry out the methods of the present disclosure. Figure 66 shows a computer system 100 programmed or otherwise configured to associate a biomolecular fingerprint 718 with a biological state 703. This determination, analysis or statistical classification is performed by methods well known in the art, including, but not limited to, various supervised / unsupervised data analysis and clustering approaches, such as, for example, hierarchical cluster analysis (HCA), principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), machine learning (also known as random forests), logistic regression, decision trees, support vector machines (SVM), K nearest neighbors, naive Bayes, linear regression, polynomial regression, SVM for regression, K-means, and hidden Markov models, among others. The computer system 100 can perform various aspects of the biomolecular fingerprint 718 analysis of the present disclosure, such as comparing / analyzing the biomolecular coronas of numerous samples to determine with statistical significance what patterns are common among individual biomolecular coronas to determine a biomolecular fingerprint 718 associated with a biological state 703. A computer system can be used to develop classifiers to detect and distinguish between different biomolecular fingerprints 718 (e.g., characteristics of the composition of protein coronas). Data collected from the sensor arrays disclosed herein can be used to train machine learning algorithms, particularly algorithms that receive array measurements from patients and output a specific biomolecular corona composition from each patient. Prior to training the algorithm, the raw data from the arrays can first be denoised to reduce variability in individual variables.

[0061] Machine learning can be generalized as the ability of a learning machine to perform accurately on new unseen examples / tasks after going through a training dataset. Machine learning may include the following concepts and methods: Supervised learning concepts include AODE; artificial neural networks such as backpropagation, autoencoders, Hopfield networks, Boltzmann machines, restricted Boltzmann machines, and spiking neural networks; Bayesian statistics such as Bayesian networks and Bayesian knowledge bases; case-based reasoning; Gaussian process regression; gene expression programming; group methods of data manipulation (GMDH); inductive logic programming; learning from examples; lazy learning; learning automata; learning vector quantization; logistic model trees; minimum message length such as nearest neighbor methods and analogical modeling (decision trees, decision graphs, etc.); probabilistically correct learning (PAC) learning; ripple down rules, knowledge acquisition, and so on. These may include: acquisition methods; symbolic machine learning algorithms; support vector machines; random forests; ensembles of classifiers such as bootstrap aggregating (bagging) and boosting (hyperalgorithms); ordinal regression; information fuzzy networks (IFNs); conditional random fields; ANOVA; linear classifiers such as Fisher's Linear Discriminant Analysis, linear regression, logistic regression, multinomial logistic regression, naive Bayes classifier, perceptron, support vector machines; quadratic classifiers; k-nearest neighbors; boosting; decision trees such as C4.5, random forest, ID3, CART, SLIQ, SPRINT; Bayesian networks such as naive Bayes; and hidden Markov models. Unsupervised learning concepts may include: expectation maximization algorithm; vector quantization; generative topological maps; artificial neural networks such as information bottleneck methods and self-organizing maps; association rule learning such as the Apriori algorithm, the Eclat algorithm, and the FP-growth algorithm; hierarchical clustering such as single-linkage clustering and concept clustering; cluster analysis such as K-means, fuzzy clustering, DBSCAN, and the OPTICS algorithm; and outlier detection such as local outlier factor methods. Semi-supervised learning concepts may include generative models; sparse separation; graph vectors; The learning concepts may include: deep belief networks; deep Boltzmann machines; deep convolutional neural networks; deep recurrent neural networks; and hierarchical temporal memories. The reinforcement learning concepts may include: temporal difference learning; Q-learning; learning automata; and SARSA. The deep learning concepts may include: deep belief networks; deep Boltzmann machines; deep convolutional neural networks; deep recurrent neural networks; and hierarchical temporal memories.

[0062] A computer system 100, as shown in FIG. 66, is configured to perform the methods described herein. The system 100 includes a central computer server 101 programmed to perform the exemplary methods described herein. The server 101 includes a central processing unit (CPU, also referred to as a "processor") 105, which may be a single-core processor, a multi-core processor, or multiple processors for parallel processing. The server 101 also includes memory 110 (e.g., random access memory, read-only memory, flash memory); electronic storage 115 (e.g., a hard disk); a communication interface 120 (e.g., a network adapter) for communicating with one or more other systems; and peripherals 125, which may include cache, other memory, data storage devices, and / or electronic display adapters. The memory 110, storage 115, interface 120, and peripherals 125 communicate with the processor 105 via a communication bus (solid lines), such as a motherboard. The storage 115 may be a data storage device for storing data. The server 101 is operatively connected to a computer network ("network") 130 using the communication interface 120. Network 130 may be the Internet, an intranet and / or an extranet, an intranet and / or an extranet in communication with the Internet, a telecommunications or data network. In some cases, network 130 may implement a peer-to-peer network with server 101, allowing devices connected to server 101 to function as clients or servers.

[0063] The storage device 115 may store files such as subject reports and / or communications involving data regarding individuals or any aspect of data related to the present disclosure.

[0064] The computer server 101 can communicate with one or more remote computer systems via a network 130. The one or more remote computer systems may be, for example, a personal computer, a laptop, a tablet, a phone, a smartphone, or a mobile terminal.

[0065] In some applications, computer system 100 includes a single server 101. In other situations, the system includes multiple servers that communicate with each other via an intranet, an extranet, and / or the Internet.

[0066] The server 101 may be configured to store measurement data or databases as described herein, patient information from the subject, such as medical history, family history, demographic data, and / or other clinical or personal information as may be relevant to a particular application. Such information may be stored on the storage device 115 or the server 101, and such data may be transmitted over a network.

[0067] The methods described herein may be implemented using machine (or computer processor) executable code (or software) stored on an electronic storage location of the server 101, such as on memory 110 or electronic storage 115. During use, the code may be executed by the processor 105. In some cases, the code may be collected from storage 115 and stored on memory 110 for ready access by the processor 105. In some circumstances, electronic storage 115 may be omitted and the machine executable instructions may be stored on memory 110. Alternatively, the code may be stored on a second computer system. The system 140 can be implemented.

[0068] Aspects of the systems and methods described herein, such as the server 101, may be incorporated into programming. Various aspects of the technology may be generally considered as an "article of manufacture" or "product" in the form of machine (or processor) executable code and / or associated data carried on or embodied in such type of machine-readable medium. The machine executable code may be stored in electronic storage such as memory (e.g., read-only memory, random access memory, flash memory) or hard disk. A "storage" type medium may include tangible memory such as any and all computers or processors or their associated modules such as various semiconductor memories, tape drives, disk drives, etc. that provide non-transitory storage at any point in time for the software programming. All or portions of the software may at times be communicated over the Internet or various other telecommunications networks. Such communication may allow the software to be loaded, for example, from one computer or processor to another, for example, from a management server or host computer to the computer platform of an application server. Thus, other types of media that may carry software elements may include light waves, radio waves, and electromagnetic waves, such as those used across physical interfaces between local equipment via wired or optical landline networks and across various air links. Physical elements such as wired or wireless links or optical links that propagate such waves are also considered to be software-bearing media. As used herein, unless limited to non-transitory tangible "storage" media, the term computer or machine "readable medium" refers to any medium that participates in providing instructions to a processor for execution.

[0069] The computer systems described herein may comprise computer executable code for executing the algorithms or algorithm-based methods described herein. In some applications, the algorithms described herein use a memory device that comprises at least one database.

[0070] Data related to the present disclosure can be transmitted over a network or connection for receipt and / or review by a recipient. A recipient may be, but is not limited to, the subject to which the report pertains, or his / her caregiver, such as a healthcare provider, manager, other healthcare professional or other caretaker, or the individual or organization that performed and / or commissioned the analysis. A recipient may also be a local or remote system (e.g., a server or other "cloud computing" architecture system) for storing such reports. In one embodiment, the computer readable medium comprises a medium suitable for transmitting the results of an analysis of a biological sample using the methods described herein.

[0071] Aspects of the systems and methods described herein, such as the computer system 101 depicted in FIG. 66, may be embodied in programming. Various aspects of the technology may be thought of as an "article of manufacture" or "product" generally in the form of machine (or processor) executable code and / or associated data carried on or embodied in any type of machine-readable medium. The machine-executable code may be stored in electronic storage such as memory (e.g., read-only memory, random access memory, flash memory) or a hard disk. A "storage" type medium may include tangible memory such as any and all computers or processors or their associated modules such as various semiconductor memories, tape drives, disk drives, etc. that provide non-transitory storage at any point in time for the software programming. All or portions of the software may at times be communicated over the Internet or various other telecommunications networks. Such communication may be, for example, from one computer or processor to another, for example, The software may be loadable, for example, into the computer platform of the application server from a management server or host computer. Thus, other types of media that may carry software elements may include light waves, radio waves, and electromagnetic waves, such as those used across physical interfaces between local equipment via wired or optical landline networks and across various air links. Physical elements such as wired or wireless links or optical links that propagate such waves are also considered to be software-bearing media. As used herein, unless limited to non-transitory tangible "storage" media, the term computer or machine "readable medium" refers to any medium that participates in providing instructions to a processor for execution.

[0072] Thus, a machine-readable medium such as a computer executable code may take many forms, including but not limited to a tangible storage medium, a propagating wave medium, or a physical transmission medium. Non-volatile memory includes optical or magnetic disks, such as any of the storage devices in any computer(s) that may be used for execution, such as the databases shown in the figures. Volatile storage media includes dynamic memory, such as the main memory of such a computer platform. Tangible transmission media include coaxial cables, copper wire and fiber optics, including the wires that comprise a bus within a computer system. Propagating wave transmission media may be in the form of electric or electromagnetic signals or acoustic or light waves, such as those generated in radio frequency (RF) and infrared (IR) data communications. Thus, common forms of computer readable media include, for example, floppy disks, flexible disks, hard disks, magnetic tape, any other magnetic media, CD-ROMs, DVDs or DVD-ROMs, any other optical media, punch cards paper tape, any other physical storage media having a pattern of holes, RAM, ROM and EPROMs, FLASH-EPROMs, any other memory chips or cartridges, a propagating wave carrying data or instructions, a cable or link carrying such a propagating wave, or any other medium from which programming code and / or data can be read by a computer. Many of these forms of computer readable media may be used in carrying one or more sequences of one or more instructions to a processor for execution.

[0073] <Physicochemical properties> In some embodiments, the multiple sensor elements comprise, consist essentially of, or consist of multiple particles, each particle distinguished from one another by at least one physicochemical property such that each sensor element has a unique biomolecular corona signature when contacted with the same sample.

[0074] The physicochemical properties of the sensor elements within the array refer to, for example, composition, size, surface charge, hydrophobicity, hydrophilicity, surface functionality (surface functional groups), surface topography, surface curvature and shape. The term composition encompasses the use of different types of materials as well as differences in the chemical and / or physical properties of materials, such as the conductivity of a selected material between sensor elements.

[0075] The surface curvature is usually determined by the size of the nanoparticle. Thus, on the nanometer scale, as the size of the nanoparticle changes, the surface curvature of the particle changes, and this change in surface curvature affects the binding selectivity of the surface. For example, at a certain curvature, the surface of the particle has a binding affinity for a certain type of biomolecule, while at a different curvature, it has another binding affinity and / or a binding affinity for a different biomolecule. The curvature can be adjusted to create multiple sensor elements with altered affinities for different biomolecules. Sensor arrays can be created with multiple sensor elements with different curvatures (e.g., different sizes), resulting in multiple sensor elements, each with a different biomolecule corona signature.

[0076] Surface morphology can also be modified by methods such as patterning the surface to provide different affinities or designing the surface curvature at multiple length scales. Surface patterning can be achieved, for example, by forming the sensor element by block polymerization, where at least two blocks have different chemical nature, by using at least two different polymers and phase separating the polymers during polymerization to form nanoparticles, and / or by cross-linking the separated polymers following phase separation. Designed surface ratios at multiple length scales can be achieved, for example, by employing Pickering emulsions stabilized by finely divided particles (Sacanna et al., 2007) for the synthesis of nanoparticles. In some embodiments, the finely divided particles are selected from, for example, metal oxides such as silicates, aluminates, titanates, aluminum, silicon, titanium, nickel, cobalt, iron, manganese, chromium, or vanadium oxides, carbon black, and nitrides or carbides, such as, for example, boron nitride, boron carbide, silicon nitride, or silicon carbide, among others.

[0077] For example, sensor elements, including nanoscale sensor elements, may be functionalized to have different physicochemical properties. The optimal method for functionalizing a sensor element is well known in the art and depends on the composition of the sensor element (e.g., gold, iron oxide, silica, silver, etc.), and may include, for example, aminopropyl group modification, amino group modification, boronic acid modification, carboxylic acid modification, methylation modification, N-succinimidyl ester modification, PEG modification, streptavidin modification, methyl ether modification, triethoxylpropylaminosilane modification, thiol modification, PCP modification, citrate modification, lipoic acid modification, etc. These include, but are not limited to, BPEI modification, carboxyl group modification, and hydroxyl group modification. In one embodiment, the nanoparticles may be modified with amine groups (-NH 2 or carboxyl groups (COOH). In some embodiments, the nanoscale sensors The sensor element is functionalized with a polar functional group. Non-limiting examples of polar functional groups include carboxyl, hydroxyl, thiol, cyano, nitro, ammonium, imidazolium, sulfonium, pyridinium, pyrrolidinium, phosphonium, or any combination thereof. In some embodiments, the functional group is an acidic functional group (e.g., sulfonic acid, carboxyl, etc.), a basic functional group (e.g., amino, cyclic secondary amino (pyrrolidyl and piperidyl), pyridyl, imidazole, guanidine), carbamoyl, hydroxyl, aldehyde, etc. In some embodiments, the polar functional group is an ionic functional group. Non-limiting examples of ionic functional groups include ammonium, imidazolium, sulfonium, pyridinium, pyrrolidinium, phosphonium. In some embodiments, the sensor element is functionalized with a polymerizable functional group. Non-limiting examples of polymerizable functional groups include vinyl and (meth)acrylic groups. In some embodiments, the functional groups are pyrrolidyl acrylate, acrylic acid, methacrylic acid, acrylamide, 2-(dimethylamino)ethyl methacrylate, hydroxyethyl methacrylate, and the like.

[0078] In other embodiments, the physicochemical properties of the sensor element can be modified by modifying the surface charge. For example, the surface can be modified to give a net neutral charge, a net positive surface charge, a net negative surface charge, or a zwitterionic charge. The charge of the surface can be controlled during elemental synthesis or by post-synthetic modification (modification) of the charge during surface functionalization. In polymeric nanoparticles, charge differences can be obtained during synthesis using different synthesis methods and differently charged comonomers, as well as in inorganics with mixed oxidation states.

[0079] <Nanoparticles> In some embodiments, the particle is a nanoparticle. In some embodiments, the particle is a liposome. The liposome may comprise any lipid capable of forming a particle. The term "lipid" refers to a lipid that is an ester of fatty acids and is insoluble in water and many organic solvents. Lipids generally fall into at least three classes: (1) "simple lipids," including fats and oils, (2) "complex lipids," including phospholipids and glycolipids, and (3) "derived lipids," such as steroids. In one embodiment, the liposome comprises one or more cationic or anionic lipids and one or more stabilizing lipids. Stabilizing liposomes are well known in the art and include, but are not limited to, DOPG (1,2-dioleoyl-sn-glycero-3-phospho-(1'-rac-glycerol)), DOTAP (1,2-dioleoyl-3-trimethylammonium-propane)-DOPE (dioleoylphosphatidylethanolamine), CHOL (DOPC-cholesterol), and combinations thereof.

[0080] The lipid-based surface of the liposome can contact a subset of biomolecules (e.g., proteins) of a complex biological sample (e.g., plasma or any sample containing a complex mixture of biomolecules such as proteins, nucleic acids, and / or polysaccharides and lipids) at the lipid-biomolecule (e.g., protein) interface, thereby binding a subset of proteins to generate a pattern of biomolecule (e.g., protein) binding.

[0081] In one embodiment, the liposome comprises a cationic lipid. As used herein, the term "cationic lipid" refers to a lipid that is cationic or that becomes cationic (protonated) as the pH decreases below the pK of the lipid's ionizable group, but becomes increasingly more neutral at higher pH values. At pH values ​​below the pK, the lipid can associate with negatively charged nucleic acids. In some embodiments, the cationic lipid can comprise a zwitterionic lipid, which is predicted to become positively charged as the pH decreases. In some embodiments, the liposome comprises a cationic lipid. In some embodiments, the cationic lipid comprises any of a number of lipid species that have a net positive charge at a selected pH, such as physiological pH. Such lipids include N,N-dioleyl-N,N-dimethylammonium chloride (DODAC); N-(2,3-dioleyl)propyl)-N,N,N-trimethylammonium chloride (DOTMA); N,N-distearyl-N,N-dimethylammonium bromide (DDAB); N-(2,3-dioleoyloxy)propyl)-N,N,N-trimethylammonium chloride (DOTAP); 3-(N-(N'N'-dimethylaminoethane)-carbamoyl)cholesterol (DC-Chol), N-(1-(2,3-dioleoyloxy)propyl)-N-2-(sperminecarboxamido)ethyl)-N,N-dimethyl-(ammonium trifluoroacetate) (DOSPA), dioctadecanoate, and dimethyloctadecanoate. Decylamidoglycyl carboxyspermine (DOGS), 1,2-dioleoyl-3-dimethylammonium propane (DODAP), N,N-dimethyl-2,3-dioleoyloxy)propylamine (DODMA), N-(1,2-dimyristyloxyprop-3-yl)-N,N-dimethyl-N-hydroxyethylammonium bromide (DMRIE), 1,2-dioleoyl-sn-3-phosphoethanolamine (DOPE), N-(1-(2,3-dioleyloxy)propyl)-N-(2-(sperminecarboxamido)ethyl)-N,N-dimethyl-ammonium trifluoroacetate (DOSPA), dioctadecylamidoglycyl carboxyspermine (DOGS) and 1,2-ditetradecanoyl-sn-glyceryl These lipids include, but are not limited to, 3-phosphocholine (DMPC). The following lipids are cationic and have a positive charge at less than physiological pH: DODAP, DODMA, DMDMA, 1,2-dilinoleyloxy-N,N-dimethylaminopropane (DLinDMA), 1,2-dilinolenyloxy-N,N-dimethylaminopropane (DLenDMA). In some embodiments, the lipid is an amino lipid.

[0082] In some embodiments, the liposomes contain one or more additional lipids that stabilize the particle during its formation. Suitable stabilizing lipids include neutral lipids and anionic lipids. The term "neutral lipid" refers to any one of a number of lipid species that exist in an uncharged or neutral zwitterionic form at physiological pH. Exemplary neutral lipids are diacylphosphatidyl ... Choline (diacylphosphatidylcholines), diacylphosphatidylethanolamine, Exemplary neutral lipids include, for example, distearoylphosphatidylcholine (DSPC), dioleoylphosphatidylcholine (DOPC), dipalmitoylphosphatidylcholine (DPPC), dioleoylphosphatidylglycerol (DOPG), dipalmitoylphosphatidylglycerol (DPPG), dioleoyl-phosphatidylethanolamine (DOPE), palmitoyloleoylphosphatidylcholine (POPC), palmitoyloleoyl-phosphatidylethanolamine (POPE), dioleoyl-phosphatidylethanolamine 4 -(N-maleimidomethyl)-cyclohexane-1-carboxylate (DOPE-mal), dipalmitoyl phosphatidylethanolamine (DPPE), dimyristoyl phosphoethanolamine (DMPE), distearoyl-phosphatidylethanolamine (DSPE), 16-O-monomethyl PE, 16-O-dimethyl PE, 18-1-trans PE, 1-stearoyl-2-oleoyl-phosphatidylethanolamine (SOPE) and 1,2-dielideyl-sn-glycero-3-phosphoethanolamine (transDOPE). In one embodiment, the neutral lipid is 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC).

[0083] The term "anionic lipid" refers to any lipid that is negatively charged at physiological pH. These lipids include phosphatidylglycerol, diacylphosphatidylserine, diacylphosphatidic acid, N-dodecanoylphosphatidylethanolamine, lysylphosphatidylglycerol, palmitoyloleoylphosphatidylglycerol (POPG) and other anionic modified groups attached to neutral lipids. In some embodiments, the liposomes include glycolipids (e.g., monosialoganglioside GM.sub.1). In some embodiments, the liposomes include a sterol, such as cholesterol. In some embodiments, the liposomes include an additional stabilizing lipid that is a polyethylene glycol lipid. Suitable polyethylene glycol lipids include PEG-modified phosphatidylethanolamines, PEG-modified phosphatidic acids, PEG-modified ceramides (e.g., PEG-CerC14 or PEG-CerC20), PEG-modified dialkylamines, PEG-modified diacylglycerols, and PEG-modified dialkylglycerols. Representative polyethylene glycol lipids include PEG-c-DOMG, PEG-c-DMA, and PEG-s-DMG. In one embodiment, the polyethylene glycol lipid is N-[(methoxypoly(ethylene glycol).sub.2000)carbamyl]-1,2-dimyristyloxylpropyl-3-amine (PEG-c-DMA). In one embodiment, the polyethylene glycol lipid is PEG-c-DOMG.

[0084] Suitable liposomes may be solid lipid nanoparticles (SLN) consisting of solid lipids, emulsifiers and / or water / solvent. SLN may include, but are not limited to, combinations of triacylglycerols (tristearin), partial glycerides (Invitol®), fatty acids (stearic acid, palmitic acid), steroids (cholesterol) and waxes (cetyl palmitate). Various emulsifiers and their combinations (Pluronic® F68, F127) have been used to stabilize lipid dispersions. Suitable ingredients for use in preparing the SNL sensor element include, for example, phospholipids, glycerol, poloxamer 188, soy phosphatidylcholine, Compritol®, cetyl palmitate, PEG 2000, PEG 4500, Tween® 85, ethyl oleate, sodium alginate, ethanol / butanol, tristearic glycerides, PEG 400, isopropyl myristate, Pluronic® F68, Tween® 80, trimyristin, tristearin, trilaurin, stearic acid, glyceryl caprate such as Capmul® MCM C10, cocoa butter, triglyceride coconut butter, 1-octadecanol, glyceryl behenate such as Compritol® 888 ATO, glyceryl palmitostearate such as Precirol® ATO5, and the like. These include, but are not limited to, serine and cetyl palmitate wax.

[0085] In some embodiments, the plurality of sensor elements comprises, consists essentially of, or consists of a plurality of half particles having different geometric shapes that can be produced by casting techniques, 3D printing, or 4D printing. Suitable sub-particles are well known in the art and include, but are not limited to, sub- and partial particles of any geometric shape, such as spheres, rods, triangles, cubes, and combinations thereof. Preferably, in one embodiment, the sub-particles have different physicochemical properties obtained by 3D printing.

[0086] In some embodiments, the sensor elements, including nanoscale sensor elements, are created by 3D or 4D printing. Suitable methods for 3D and 4D printing of sensor elements, including nanoscale sensor elements, are well known in the art. Suitable materials for 3D and 4D printing include, but are not limited to, for example, plastics and synthetic polymers (e.g., polyethylene glycol diacrylate (PEG-DA), poly(e-caprolactone (PCL), poly(propylene oxide (PPO), poly(ethylene oxide) (PEO) and the like), metals, powders, glasses, ceramics and hydrogels. Suitable shapes created by 3D or 4D printing include, but are not limited to, for example, full or partial spheres (e.g., 3 / 4 or hemispheres), rods, cubes, triangles or other geometric or non-geometric shapes.

[0087] 3D printing methods include, but are not limited to, microextrusion printing, inkjet bioprinting, laser-assisted bioprinting, stereolithography, omnidirectional printing, and stamp printing.

[0088] In some embodiments, the nanoscale sensor element is a nanoparticle. Suitable nanoparticles are well known in the art and include, but are not limited to, natural or synthetic polymers, copolymers, and terpolymers (wherein the core is composed of a metal or inorganic oxide, including a magnetic core). Suitable polymeric nanoparticles include, but are not limited to, polystyrene, poly(lysine), chitosan, dextran, poly(acrylamide) and its derivatives, such as N-isopropylacrylamide, N-tert-butylacrylamide, N,N-dimethylacrylamide, polyethylene glycol, poly(vinyl alcohol), gelatin, starch, degradable biopolymers, silica, and the like.

[0089] In various embodiments, the core of the nanoparticle may comprise an organic particle, an inorganic particle, or a particle that includes both organic and inorganic materials. For example, the particle may have a core structure that is or includes a metal particle, a quantum dot particle, a metal oxide particle, or a core-shell particle. For example, the core structure may be or include a polymer particle or a lipid-based particle, and the linker may include a lipid, a surfactant, a polymer, a hydrocarbon chain, or an amphiphilic polymer. For example, the linker may include a polyethylene glycol or a polyalkylene glycol, e.g., a first end of the linker may include a lipid attached to a polyethylene glycol (PEG) and a second end may include a functional group attached to the PEG. In these methods, the first or second functional group may include an amine group, a maleimide group, a hydroxy group, a carboxyl group, a pyridylthiol group, or an azide group.

[0090] In certain embodiments, the nanoparticles may be prepared from a polymeric material selected from the group consisting of poly(styrenesulfonate), polyvinylpyrrolidone, polyvinyl acetate, polyvinylpyrrolidone-, poly(vinylpyrrolidone-), ... The composition may comprise a polymer comprising vinyl acetate copolymer, polyvinyl alcohol (PVA), acrylates, polyacrylic acid (PAA), vinyl acetate, crotonic acid copolymer, polyacrylamide, polyethylene phosphonic acid, polybutene phosphonic acid, polystyrene, polyvinyl phosphonic acid, polyalkylenes, carboxyvinyl polymers, sodium alginate, carrageenan, xanthan gum, acacia gum, gum arabic, guar gum, pullulan, agar, chitin, chitosan, pectin, karaya gum, locust bean gum, maltodextrin, amylose, corn starch, potato starch, rice starch, tapioca starch, bean starch, sweet potato starch, barley starch, wheat starch, hydroxypropylated high amylose starch, dextrin, levan, elsinan, gluten, collagen, whey protein isolate, casein, milk protein, soy protein, keratin, gelatin or copolymers, derivatives or mixtures thereof.

[0091] In other embodiments, the polymer may be or include polyethylene, polycarbonate, polyanhydride, polyhydroxy acid, polypropyl fumarate, polycaprolactone, polyamide, polyacetal, polyether, polyester, poly(orthoester), polycyanoacrylate, polyvinyl alcohol, polyurethane, polyphosphazene, polyacrylate, polymethacrylate, polycyanoacrylate, polyurea, polystyrene, polyamine, or copolymers, derivatives, or mixtures thereof.

[0092] In some embodiments, the present disclosure provides nanoparticles comprising biodegradable polymers. Non-limiting examples of biodegradable polymers may be poly-β-amino-esters (PBAEs), poly(amidoamines), polyesters including polylactic-co-glycolic acid (PLGA), polyanhydrides, bioreducible polymers, and other biodegradable polymers. In some embodiments, the biodegradable polymers may be 2-(3-aminopropylamino)ethanol end-modified poly(1,4-butanediol diacrylate-co-4-amino-1-butanol), (1-(3-aminopropyl)-4-methylpiperazine end-modified poly(1,4-butanediol diacrylate-co-4-amino-1-butanol), 2-(3-aminopropylamino)ethanol end-modified poly(1,4-butanediol diacrylate-co-5-amino-1-pentanol), or 1-(3-aminopropyl)-4-methylpiperazine end-modified poly(1,4-butanediol diacrylate-co-5-amino-1-pentanol). (1-(3-aminopropyl)-4-methylpiperazine end-modified poly(1,4-butanediol diacrylate-co-5-amino-1-pentanol), 2-(3-aminopropylamino)ethanol end-modified poly(1,5-pentanediol diacrylate-co-3-amino-1-propanol) and (1-(3-aminopropyl)-4-methylpiperazine end-modified poly(1,5-pentanediol diacrylate-co-3-amino-1-propanol).

[0093] <Array board> In some embodiments, the sensor array comprises a substrate. Regardless of the identity of the sensor elements, the invention can be implemented with a matrix of sensor elements fixed, connected and / or bonded to a solid substrate. The substrate may comprise, consist essentially of or consist of polydimethylsiloxane (PDMS), silica, gold or a gold-coated substrate, silver or a silver-coated substrate, platinum or a platinum-coated substrate, zinc or a zinc-coated substrate, carbon-coated substrate, etc. One skilled in the art will be able to select an appropriate substrate for the sensor array. In some embodiments, the sensor elements and the substrate may be made of the same element, e.g., gold. In some embodiments, the substrate and the sensor elements (e.g., nanoparticles) form a chip.

[0094] In some embodiments, the multiple sensor elements comprise a single surface, plate or chip that includes two or more discrete sensor elements (regions) having a topological difference such that a discrete biomolecular corona is formed at each discrete element (region). The methods described herein may be fabricated to include two or more discrete elements (regions), which may be raised surfaces with different geometric shapes, different sizes, different charges, or other topological differences resulting in discrete sensor elements with the ability to form discrete biomolecular coronas.

[0095] In some embodiments, the sensor element is non-covalently attached to the substrate. Suitable methods of non-covalent attachment are well known in the art and include, but are not limited to, for example, metal coordination, charge interactions, hydrophobic-hydrophobic interactions, chelation, etc. In other embodiments, the sensor element is covalently attached to the substrate. Suitable methods of covalently attaching the sensor element to the substrate include, but are not limited to, for example, click chemistry, irradiation, etc.

[0096] For illustrative purposes, methods of attaching a sensor element, e.g., a nanoscale sensor element, to a substrate are shown in Figures 23-40. For example, the sensor element can be attached to a substrate via an amidation reaction between an amino group on the silica substrate surface and a carboxylic acid group on the nanoparticle surface (Figure 23), via a ring-opening reaction between an epoxy group on the silica substrate surface and an amino group on the nanoparticle surface (Figure 24), via a Michael addition reaction between a maleimide group on the silica substrate surface and a thiol or amino group on the nanoparticle surface (Figure 25), via a urethane reaction between an isocyanate group on the silica substrate surface and a hydroxyl or amino group on the nanoparticle surface (Figure 26), via an oxidation reaction between a thiol group on the silica substrate surface and a thiol group on the nanoparticle surface (Figure 27), via "click" chemistry between an azide group on the silica substrate surface and an alkyne group on the nanoparticle surface (Figure 28), or via a "click" chemistry between an azide group on the silica substrate surface and an alkyne group on the nanoparticle surface (Figure 29). Figure 28) 2-pyridyldithiol groups on the silica substrate surface and thiol groups on the nanoparticle surface (Figure 29), through a coordination reaction between a boronic acid group on the silica substrate surface and a diol group on the nanoparticle surface (Figure 30), and through a UV light irradiation addition reaction between a C=C bond on the substrate surface and a C=C bond on the nanoparticle surface (Figure 31). The sensor element may be attached to a gold substrate, for example a silica substrate. Suitable methods for attaching the sensor element to a gold substrate are well known in the art, for example via Au-thiol binding (Figure 32), via an amidation reaction between a carboxylic acid group on the gold substrate surface and an amino group on the nanoparticle surface (Figure 33), via "click" chemistry between an azide group on the gold substrate surface and an alkyne group on the nanoparticle surface (Figure 34), via a urethanization reaction between an NHS group on the gold substrate surface and an amino group on the nanoparticle surface (Figure 35), or via ring-opening reaction between an epoxy group on the gold substrate surface and an amino group on the nanoparticle surface. reaction (Figure 36), through a coordination reaction between a boronic acid group on the gold substrate surface and a diol group on the nanoparticle surface (Figure 37), through a UV light irradiation addition reaction between a C=C bond on the gold substrate surface and a C=C bond on the nanoparticle surface (Figure 38), through a "ligand-receptor" interaction between biotin on the gold substrate surface and avidin on the nanoparticle surface (Figure 39), through a "host-guest" interaction between a-cyclodextrin (a-CD) on the gold substrate surface and adamantine (Ad) on the nanoparticle surface (Figure 40), etc. Includes.

[0097] In another example, modified surface groups may be attached to the core structure of nanoparticles using so-called "click chemistry" (see, for example, the Sigma Aldrich catalog and U.S. Pat. No. 7,375,234, the entire contents of which are incorporated herein by reference). Among the reactions that comprise the click chemistry field, one example is the Huisgen 1,3-dipolar cycloaddition of alkynes to azides to form 1,4-disubstituted-1,2,3-triazoles. The copper(I)-catalyzed reaction is mild and highly efficient, does not require protecting groups, and in many cases, no purification is required. Azide and alkyne modifying groups are usually inert to biological molecules and aqueous environments. Triazoles have similarities to the ubiquitous amide moiety found in nature, but unlike amides, they are difficult to cleave. In addition, they are nearly impossible to oxidize or reduce.

[0098] The multiple sensor elements are attached to the substrate either randomly or in different (distinguishable) patterns. The sensor elements may be arranged in a substantially uniform pattern. The pattern of the arrayed sensor elements may vary depending on the pattern in which the sensor elements are attached to the substrate. Each sensor element is spaced apart by a distance. The distance between the sensor elements (e.g., nanoparticles) arranged on the substrate may vary depending on the length of the linker used for attachment or other manufacturing conditions. According to various embodiments, the plurality of sensor elements on the array can be fabricated with a desired inter-element distance and pattern. Suitable distinguishable patterns are well known in the art and include, but are not limited to, parallel lines, squares, circles, triangles, and the like. Furthermore, the sensor elements may be arranged in rows or columns. In some embodiments, the substrate is a flat substrate, while in other embodiments, the substrate may be in the form of a microchannel or nanochannel. For illustrative purposes, suitable embodiments are shown in Figures 15A-22B. The sensor elements may be contained within a microchannel or nanochannel that regulates or controls the flow of the sample through the sensor array. Suitable microchannels are 10 μm to about 100 μm in size.

[0099] In some embodiments, non-limiting examples of multiple sensor elements include, but are not limited to, among others: (a) multiple sensor elements made from the same material but with different physicochemical properties; (b) multiple sensor elements where one or more sensor elements are made from different materials with the same or different physicochemical properties; (c) multiple sensor elements made from the same material with different sizes; (d) multiple sensor elements made from different materials with relatively the same size; (e) multiple sensor elements made from different materials and with different sizes; (f) multiple sensor elements where each element is made from a different material; (g) multiple sensor elements with different charges. The multiple sensor elements can be any suitable combination of two or more sensor elements such that each sensor element provides a unique biomolecular corona signature. For example, the multiple sensor elements may include one or more liposomes or one or more nanoparticles as described herein. In one embodiment, the multiple sensor elements can be multiple liposomes with different lipid content and / or different charges (cations / anions / neutral ions). In another embodiment, the multiple sensors may include one or more nanoparticles made from the same material but with different sizes and physicochemical properties. In another embodiment, the sensors are made from different materials (e.g., silica and polystyrene) and have similar or different sizes and / or physicochemical properties (e.g., modifications, e.g., -NH 2 , -COOH functionalized), or a combination thereof. are given purely by way of example and are not intended to limit the scope of the invention.

[0100] The angle of curvature of the particle's surface can change depending on the size of the particle. This change in curvature changes the surface area that proteins can bind and interact with on the particle. Increasing the size of the particle changes the amount of protein bound as well as the pattern of protein binding to different sized nanoparticles, as shown in Figure 58 (in this example, SDS-PAGE analysis of proteins on nanoparticles with diameters of 0.1 μm, 3 μm, and 4 μm is shown).

[0101] The novelty of the sensor array is that it can detect different proteins among different sensor elements, but also compare the levels of the same protein among different sensor elements. For example, without being bound by any theory, a theoretical example is described to illustrate the specificity of the sensor array of the present invention. In some embodiments, a sample is contacted with a first sensor element A, a second sensor element B, and a third sensor element C, each of which generates a remote protein corona signature (i.e., A', B', and C'). The composition of each protein corona signature A', B', and C' can be different from each other. In some embodiments, A', B', and C' may comprise different amounts of the same protein, which can provide additional proteomic information that cannot be obtained by characterizing the sample using previously known approaches. In other words, each unique corona protein information from each nanoparticle acts as a unique variable, and therefore more accurate. provides more data proteomics data. For example, albumin may be found in the protein corona signature of only one sensor element A, in the signatures of two sensors, e.g. A and B, B and C, or A and C, or in the biomolecular signatures of all three sensors (A, B, C). Furthermore, the sensor array can determine not only the presence or absence of a protein, e.g. albumin, but also the comparison of the amount of the protein in one sensor to the other sensors. For example, albumin may be found at a concentration X in the signature of A, at a concentration of 1 / 3X in sensor B, and at a concentration of 2X in sensor C for a particular biomolecular fingerprint. In another biomolecular fingerprint, the same protein, albumin, may be found at three different concentrations, e.g. 1 / 8X in sensor D, 3X in sensor E, and 1 / 4X in sensor F. Thus, multiple sensors can provide not only a data point of the concentration of a protein, but also a comparison of the concentration of a protein between two or more sensors. Furthermore, comparing the concentration of rare or low abundance proteins to the concentration of known proteins can provide further data on the protein corona. For example, to illustrate, the concentration of an unknown protein, e.g., protein Z, may be compared to the amount of a known protein, e.g., albumin, in different biomolecular coronas. For example, Z may be found in a ratio to albumin of 1:8 on sensor A, 1:50 on sensor B, and absent on sensor C. FIG. 12 provides a comparison of rare protein vs. albumin concentration in the analyzed protein corona. Thus, by statistical analysis, one can obtain both the presence of proteins and their relative concentrations between each sensor element, as well as the concentration of rare or low abundance proteins when compared to a particular concentration of a known protein when analyzing the data.

[0102] In some embodiments, the channels are formed by lithography, etching, embossing, or molding of a polymer surface. Typically, the manufacturing process may use any one or more of these processes, and different parts of the array may be manufactured and assembled or bonded together using different methods.

[0103] Lithography uses light or other forms of energy, such as electron beams, to change materials. Typically, a polymer surface or precursor (e.g., photoresist, a light-blocking material) is coated on a substrate and selectively exposed to light or other forms of energy. Depending on the photoresist, exposed areas of the photoresist are either retained or dissolved in a subsequent process step, commonly known as "development." This process results in a pattern of photoresist on the substrate. In some embodiments, the photoresist is used as a master mold in a molding process. In some embodiments, a polymer precursor is poured onto the substrate with the photoresist, polymerized (i.e., cured), and then peeled off.

[0104] In some embodiments, photoresist is used as a mask for the etching process. For example, after patterning photoresist on a silicon substrate, channels can be etched into the substrate using a deep reactive ion etching (DRIE) process or other chemical etching processes known in the art (e.g., plasma etching, KOH etching, HF etching, etc.). The photoresist is removed, and the substrate is bonded to another substrate using one of any bonding methods known in the art (e.g., anodic bonding, adhesive bonding, direct bonding, eutectic bonding, etc.). Multiple lithography, etching, and machining steps such as drilling may be included as needed.

[0105] In some embodiments, a polymer substrate may be heated and embossed against a master mold. The master mold may be formed by a variety of processes including lithography and machining. The polymer substrate is then bonded to another substrate to form channels and / or mixing devices. Machining processes may also be included as needed.

[0106] In some embodiments, for the injection molding process, a molten polymer or metal or alloy is injected into a suitable mold and cooled to solidify. The mold usually consists of two parts to remove the molded part. The part thus produced is bonded to the substrate and mounted on the substrate.

[0107] In some embodiments, sacrificial layer etching may be used to form the channels. Lithographic techniques may be used to pattern a material on a substrate. This material is covered by another material with a different chemical property. This material may be subjected to lithographic and etching processes or other machining processes. The substrate is then exposed to a chemical agent that selectively removes the first material. Channels are formed in the second material, leaving voids where the first material was present before the etching process.

[0108] In some embodiments, the microchannels are machined directly into the substrate by laser machining or CNC machining, and multiple machined layers are then bonded together to obtain the final substrate.

[0109] In some embodiments, the width or height of each channel is about 1 μm to about 1000 μm. In some embodiments, the width or height of each channel is about 5 μm to about 500 μm. In some embodiments, the width or height of each channel is about 10 μm to about 100 μm. In some embodiments, the width or height of each channel is about 25 μm to about 100 μm. In some embodiments, the width or height of each channel is about 50 μm to about 100 μm. In some embodiments, the width or height of each channel is about 75 μm to about 100 μm. In some embodiments, the width or height of each channel is about 10 μm to about 75 μm. In some embodiments, the width or height of each channel is about 10 μm to about 50 μm. In some embodiments, the width or height of each channel is about 10 μm to about 25 μm.

[0110] In some embodiments, the maximum width or height of the channel is about 1 μm, about 5 μm, about 10 μm, about 20 μm, about 30 μm, about 40 μm, about 50 μm, about 60 μm, about 70 μm, about 80 μm, about 90 μm, about 100 μm, about 250 μm, about 500 μm, or about 1000 μm.

[0111] In some embodiments, the width of each channel is about 5 μm to about 100 μm, hi some embodiments, the width of the channel is about 5 μm, about 10 μm, about 15 μm, about 20 μm, about 25 μm, about 30 μm, about 35 μm, about 40 μm, about 45 μm, about 50 μm, about 60 μm, about 70 μm, about 80 μm, about 90 μm, or about 100 μm.

[0112] In some embodiments, the height of each channel is about 10 μm to about 1000 μm. In some embodiments, the height of the channel is about 10 μm, about 100 μm, about 250 μm, about 400 μm, about 500 μm, about 600 μm, about 750 μm, or about 1000 μm. In certain embodiments, the height of the channel(s) through which the sample flows is about 500 μm. In certain embodiments, the height of the channel(s) through which the sample flows is about 500 μm.

[0113] In some embodiments, the length of each channel is about 100 μm to about 10 cm. In some embodiments, the length of the channel is about 100 μm, about 1.0 mm, about 10 mm, about 100 mm, about 500 mm, about 600 mm, about 700 mm, about 800 mm, about 900 mm, about 1.0 cm, about 1.1 cm, about 1.2 cm, about 1.3 cm, about 1.4 cm, about 1.5 cm, about 5 cm, or about 10 cm. In certain embodiments, the length of the channel(s) through which the sample flows is about 1.0 cm. In certain embodiments, the length of the channel(s) through which the sample flows is about 1.0 cm.

[0114] <Biomolecular Corona Nanosystem> Provided herein is a biomolecular corona nanosystem or sensor array comprising, consisting essentially of, or consisting of a plurality of sensor elements, said plurality of sensor elements differing from one another in at least one physicochemical property. In some embodiments, the plurality of sensor elements are a plurality of nanoparticles. In some embodiments, the plurality of nanoparticles are a plurality of liposomes. In some embodiments, each sensor element is capable of binding a plurality of biomolecules in a complex biological sample to generate a biomolecular corona signature. In some embodiments, each sensor element has a distinguishable biomolecular corona signature.

[0115] A biomolecular corona signature refers to the composition, signature, or pattern of different biomolecules bound to each individual sensor element or each nanoparticle. In some embodiments, the biomolecular corona signature is a protein corona signature. In another embodiment, the biomolecular corona signature is a polysaccharide corona signature. In yet another embodiment, the biomolecular corona signature is a metabolite corona signature. In some embodiments, the biomolecular corona signature is a lipidomic corona signature. Signature refers not only to different biomolecules, but also to differences in the amount, level, or quantity of biomolecules bound to a sensor element or nanoparticle, or differences in the conformation of biomolecules bound to a sensor element or nanoparticle. It is contemplated that the biomolecular corona signature of each sensor element may include some of the same biomolecules, may include distinguishable biomolecules with respect to other sensor elements or nanoparticles, and / or may differ in the level or quantity, type, or confirmation of biomolecules. The biomolecular corona signature may depend not only on the physicochemical properties of the sensor element or nanoparticle, but also on the nature of the sample and the exposure time, hi some embodiments, the biomolecular corona signature comprises biomolecules found in the soft corona and the hard corona.

[0116] In some embodiments, the sensor array comprises, consists essentially of, or consists of a first sensor element that generates a first biomolecular corona signature and at least one second sensor element or at least one nanoparticle that generates at least one second biomolecular corona signature when the sensor array is contacted with a complex biological sample. The biomolecular fingerprint is a combination of the first biomolecular signature and at least one second biomolecular signature. It is contemplated that the biomolecular signature can be created from many different biomolecular signatures, from at least two biomolecular corona signatures, for example at least 1000 different biomolecular corona signatures. The biomolecular corona can be analyzed separately for each sensor element to determine a biomolecular corona signature for each element, each nanoparticle, or each liposome, which can be combined to determine the biomolecular fingerprint, or two or more biomolecular coronas can be analyzed simultaneously to generate a biomolecular fingerprint at once.

[0117] A biomolecular fingerprint can distinguish between different potential biological states (e.g., disease states) in a subject. In some embodiments, a biomolecular fingerprint can be associated with the progression of a disease or disorder and / or associated with a disease state in a subject.

[0118] In some embodiments, the biomolecular fingerprint can determine the disease state of a subject. As used herein, the term "disease state" of a subject refers to the ability of the sensor array of the present technology to distinguish between different states of a disease in a subject. This term includes pre-symptomatic or prodromal states of a disease or disorder (a stage in which the subject does not have any outward signs or symptoms of the disease or disorder but will develop the disease or disorder in the future), and stages in which the subject is afflicted by a disease or disorder. The biomolecular fingerprints may be associated with a pre-disease or prodromal disease state, where the subject appears healthy when there are no outward signs or symptoms of the disease, but will develop the disease in the future. In another example, the biomolecular fingerprints may indicate that the subject has a disease, and if the disease is at an early, intermediate, or end stage, the unique biomolecular fingerprints associated with each stage may be distinguished.

[0119] As mentioned above, a disease state also includes a precursor state of a disease or disorder. This precursor state is a state in which a subject does not have any outward signs or symptoms of the disease or disorder (although there may be sub-macro changes in the subject's biomolecules found in the subject's blood or other biological fluids) but will develop the disease or disorder in the future. This precursor state can also be described as a state in which the first pathological changes of the disease are observed, such as changes in the biomolecular fingerprint of a biological sample from a patient.

[0120] Another example of a biological state is a healthy, disease-free state. The array can also determine a specific biomolecular fingerprint associated with a healthy, disease-free state in which the subject will not develop a disease in the future. In this case, there is no evidence that the subject is disease-free at the time of testing and will not develop a disease in the future.

[0121] As used herein, the term "biological state" encompasses any biological characteristic of a subject that can be revealed in a biological sample as defined herein. Biological states can be detected using the methods disclosed herein, and two subjects with different biological states will reveal these differences in the composition of the samples. For example, biological states include disease states of a subject. A disease state can be detected when a change occurs in the molecular composition (e.g., the level of one or more proteins) of a sample of a subject that is indicative of the disease state relative to a sample of a subject that does not have the disease state (i.e., the biological state is healthy or disease-free).

[0122] Another example of a biological state is the level of responsiveness of a subject to a particular therapeutic treatment (e.g., administration of one or a combination of drugs or pharmaceuticals). In some embodiments, the biological state is the responsiveness of a subject (e.g., with respect to a particular threshold in an assay) to a particular drug. In another embodiment, the biological state is the non-responsiveness of a subject (e.g., with respect to a particular threshold in an assay) to a particular drug. In some embodiments, the level of responsiveness of a subject to a drug (i.e., a biological state of responsiveness to a drug or a biological state of non-responsiveness to a drug) is related to factors such as variability in metabolism or pharmacokinetics of a drug between subjects.

[0123] Another example of a biological state is the level of immune response exhibited by a subject. In some embodiments, the biological state can be an increased immune response. In another embodiment, the biological state can be a decreased immune response. Immune responses can differ between subjects as a result of a number of variables. For example, immune responses can differ between subjects as a result of differing degrees of exposure to exogenously introduced antigens (e.g., those associated with viruses or bacteria), as a result of differing susceptibility to autoimmune diseases or disorders, or secondarily as a result of response to other biological states in the subject (e.g., disease states such as cancer).

[0124] Other examples of biological states that can be associated with a biomolecular fingerprint include a subject's susceptibility to side effects associated with drug administration and a subject's allergy to the administration of a particular component or substance. This includes confirmation of the possibility of an electron transport reaction.

[0125] The innovation in this sensor array and related methods differs from current methods that detect or measure the presence or absence or level of specific biomarkers to predict whether a subject may have a propensity or likelihood to develop a disease or disorder at a very early stage of the disease, before signs or symptoms can be monitored or analyzed. Advantageously, the array can distinguish between no disease or disorder, having a precursor to a disease or disorder, and having a disease or disorder in the subject's health. However, the invention is not limited to these embodiments and covers a spectrum of other disease states that may arise in the subject's health care and disease continuum.

[0126] Further, the innovation of the sensor array of the present invention can be illustrated by the following examples. The sensor array of the present invention is not only very sensitive and can detect small changes in the amount of biomolecules in a sample, but also depends on the interactions between the biomolecules. For example, without being bound by any theory, a theoretical example is described to illustrate the uniqueness of the sensor array of the present invention. For example, when biological samples are taken from a subject over time (e.g., before the appearance of signs and symptoms of a disease state, pre-symptomatic state, and during the early and middle stages of a disease). In these samples, one may only measure the level (e.g., quantify the amount) of biomolecule X in the sample and find that the concentration does not change across different disease states. Thus, the measurement of biomolecule X may not be a useful marker of disease. However, with the sensor array of the present invention, the level of biomolecule X may be the same, but the composition of the protein corona signature associated with the sample may change over time and across samples due to the interaction of biomolecule X with other biomolecules Y, Z. For example, biomarker X may change its association from biomolecule Y to biomolecule Z, or the interaction of X with Y and Z over time leads to a conformational change in X. This type of change, where the total concentration does not change but the unique biomolecular fingerprint (including biomolecule X) associated with the disease state changes, allows for differentiation and association with other disease states. This is not seen using traditional methods that simply measure the amount of biomolecule X in a sample because this level remains the same for all diseases. This is a new and unexpectedly highly specific and useful assay. Using this assay, biomarkers for diseases that have not previously been characterized as biomarkers can be determined, including biomarker patterns, because these markers do not change in absolute amount in the sample but have consistent and measurable changes in their interaction with the sensor elements of the sensor array. The ability to differentiate patterns by analysis of the biomolecular corona provides the ability to associate these patterns with different disease states.

[0127] As described above, the sensor array of the present invention has the ability to detect proteins over ten orders of magnitude, which is more sensitive than previous methods. The uniqueness of the sensor array of the present invention lies in the ability of the array to detect proteins regardless of their concentration levels in the sample. The assay relies on the ability of the biomolecules (e.g., proteins), known or unknown, in the sample to interact with the sensor elements, and also interact with different sensor elements in different amounts and different ways, as well as with other biomolecules associated with the sensor elements. This allows low abundance and rare proteins to be detected even in the presence of high abundance proteins (e.g., albumin) in the sample. As shown in FIG. 12 of Example 1, the sensor array detects concentrations over 10 orders of magnitude even in the presence of high abundance proteins. The sensor array is also capable of detecting proteins at previously unknown / unreported plasma concentrations not previously known using current methods. One unique feature of the sensor array of the present invention is its ability to detect low abundance and rare proteins. It is the power of the system of the present invention that the sensor array of the present invention can be used not only to determine disease state (disease-free, pre- or early-stage and late-stage disease), but also in some cases to differentiate between disease subtypes (e.g., differentiating between types of cancer, such as lung cancer, breast cancer, myeloma, etc.).

[0128] <Detection method> The sensor array of the present invention may be used in various methods described herein. The ability to determine a unique biomolecular fingerprint for a sample provides a novel and innovative means for measuring a particular disease state in a subject. These biomolecular fingerprints can be used to determine the disease state of a subject, to diagnose or prognose a disease in a subject, or to identify a unique pattern of biomarkers associated with a disease state of a disease or disorder. For example, changes in a biomolecular fingerprint in a subject over time (days, months, years) provide the ability to track a disease or disorder (e.g., disease state) in a subject, which can be broadly applied to determining a biomolecular fingerprint that can be associated with an early stage of a disease or any other disease state. As discussed above, for example, for cancer, the ability to detect a disease early, possibly before the cancer has fully progressed or metastasized, provides a better outcome for these patients, provides the ability to extend life expectancy, and reduces the mortality associated with the disease. Thus, the sensor array of the present invention provides a unique opportunity to obtain a biomolecular fingerprint associated with a pre-onset or precursor state of a disease.

[0129] It is understood that even before a disease has progressed to the point where it shows visible signs or symptoms, at a macroscopic level, changes are occurring within the subject's body and biological systems. It is possible to recognize these early presymptomatic signs that are expected by using the sensor array of the present invention. The inventors have found that by comparing the biomolecular fingerprints of a subject at different points in a disease state (e.g., before the disease shows signs and develops, after early signs or symptoms of the disease appear, and / or at a late stage of the disease), unique biomolecular fingerprints associated with different disease states related to the progression of the disease can be obtained. In other words, a biomolecular fingerprint can be determined that allows identifying subjects who will develop the disease in the future. This allows for early monitoring and early treatment, resulting in a better outcome for subjects diagnosed with the disease.

[0130] In some embodiments, a method of detecting a disease or disorder in a subject is provided, the method comprising: (a) obtaining a sample from the subject; (b) contacting the sample with a sensor array as described herein; and (c) determining a biomolecular fingerprint associated with the sample, the biomolecular fingerprint distinguishing a health state of a subject in a disease state, for example, being free of the disease or disorder, having a precursor state to the disease or disorder, or having the disease or disorder.

[0131] Determining a biomolecular fingerprint associated with the sample includes detecting a biomolecular corona signature for at least two sensor elements, where the combination of the at least two biomolecular corona signatures produces the biomolecular fingerprint. In some embodiments, the biomolecular corona signatures of the at least two sensor elements are analyzed separately and the results are combined to determine the biomolecular fingerprint. In some embodiments, the biomolecular corona signatures of the at least two elements are analyzed simultaneously or in the same sample.

[0132] In some embodiments, a method for determining a biomolecular fingerprint includes detecting and determining a biomolecular corona signature of at least two sensor elements, which in some embodiments includes separating a plurality of biomolecules bound to each sensor element (e.g., separating the biomolecular corona from the sensor element) to determine the biomolecular fingerprint. This can be done by analyzing multiple biomolecules to determine the composition of multiple biomolecular coronas. Depending on the array design, the composition of each biomolecular corona signature for each sensor element can optionally be analyzed separately and the results combined to generate a biomolecular fingerprint (e.g., each sensor element is in a separate channel or compartment and the specific composition of the biomolecular corona for that particular sensor element can be analyzed separately (e.g., the biomolecules are separated and analyzed by mass spectrometry or chromatography, or the multiple biomolecules that remain bound to the sensor element are detected by fluorescence, luminescence, or other means)). In another embodiment, the at least two sensor elements are on the same array and the composition of the biomolecular corona for the at least two sensor elements is analyzed simultaneously by separating the biomolecular coronas from both sensor elements in a solution and analyzing the solution to determine the biomolecular signatures.

[0133] Methods for analyzing the plurality of biomolecules that make up a biomolecular corona signature or biomolecular fingerprint are well known in the art and include, but are not limited to, gel electrophoresis, liquid chromatography, mass spectrometry, nuclear magnetic resonance spectroscopy (NMR), Fourier transform infrared spectroscopy (FTIR), circular dichroism, Raman spectroscopy, and combinations thereof. In a preferred embodiment, the analysis is performed by liquid chromatography, mass spectrometry, or a combination thereof.

[0134] In a preferred embodiment, the sensor array is a label-free array.

[0135] In some embodiments, a label array may be used, and the corona signature could be determined as a change in signal (e.g., fluorescence, luminescence, charge, colorimetric dye). Suitable examples are shown in Figures 42-43. For example, the array may include chemically responsive dyes in a printable formulation to detect and identify biomolecules based on equilibrium interactions between the biomolecules and the responsive dyes.

[0136] In some embodiments, the sensor elements comprise a complex of a first component and a polymeric fluorophore or other quencher component that chemically complements the first component, such complex having an initial background or reference fluorescence. When the first component contacts the biomolecular corona, it affects the quenching of the fluorophore, and this change in fluorescence can be measured. After illuminating and / or exciting the sensor with a laser, the effect and / or change in fluorescence for each sensor element can be measured and compared to the background fluorescence to generate a biomolecular fingerprint.

[0137] The sensor arrays and methods described herein can be used to determine a disease state and / or a prognosis or diagnosis of a disease or disorder, including, but not limited to, cancer, cardiovascular disease, endocrine disorders, inflammatory diseases, neurological diseases, and the like.

[0138] In one embodiment, the disease or disorder is cancer. In a preferred embodiment, the sensor array and methods described herein can diagnose cancer (e.g., determine whether a subject has (a) no cancer, (b) a pre-cancerous stage, (c) early stage cancer, or (d) late stage cancer), as well as, in some embodiments, determine the type of cancer. As shown in the following examples, a sensor array with six sensor elements was able to accurately determine the disease state of cancer or no cancer. In addition, these examples show that a sensor array with six sensor elements was able to distinguish between different types of cancer (e.g., lung cancer, glioblastoma, meningioma, myeloma, pancreatic cancer).

[0139] As used herein, the terms "cancer" and "tumor" are used interchangeably and are intended to encompass any cancer, neoplastic, or pre-neoplastic disease characterized by abnormal cell growth. Examples of cancers include lung cancer, pancreatic cancer, myeloma, myeloid leukemia, meningioma, glioblastoma, breast cancer, esophageal squamous cell carcinoma, gastric cancer, prostate cancer, bladder cancer, ovarian cancer, thyroid cancer, neuroendocrine tumors, colorectal cancer, ovarian cancer, head and neck cancer, Hodgkin's lymphoma, non-Hodgkin's lymphoma, rectal cancer, urinary tract cancer, uterine cancer, oral cancer, skin cancer, stomach cancer, brain tumors, liver cancer, laryngeal cancer, and esophageal cancer. Cancer, breast tumor, fibrosarcoma, myxosarcoma, liposarcoma, chondrosarcoma, osteosarcoma, chordoma, angiosarcoma, endothelial sarcoma, Ewing's sarcoma, squamous cell carcinoma, basal cell carcinoma, adenocarcinoma, sweat gland carcinoma, sebaceous gland carcinoma, papillary carcinoma, papillary adenocarcinoma, cystadenocarcinoma, medullary carcinoma, bronchogenic carcinoma, renal cell carcinoma, hepatoma, bile duct carcinoma, choriocarcinoma, seminoma, embryonal carcinoma, Wilms' tumor, cervical cancer, testicular tumor, The cancer may be selected from the group consisting of uterine cancer, lung cancer, small cell lung cancer, bladder cancer, epithelial cancer, glioblastoma, neuroma, craniopharyngioma, schwannoma, glioma, astrocytic tumor, meningioma, melanoma, neuroblastoma, retinoblastoma, leukemia and lymphoma, acute lymphoblastic leukemia, acute myeloid polycythemia vera, multiple myeloma, Waldenstrom's macroglobulinemia, heavy chain disease, acute nonlymphocytic leukemia, chronic lymphocytic leukemia, chronic myelogenous leukemia, childhood malignant acute lymphocytic leukemia (ALL), thymic ALL, B-cell ALL, acute megakaryoblastic leukemia, Burkitt's lymphoma, T-cell leukemia, small cell, large cell and non-small cell lung cancer, acute granulocytic leukemia, germ cell tumor, uterine cancer, gastric cancer, hairy cell leukemia, thyroid cancer and other cancers known in the art. In preferred embodiments, the cancer is selected from the group consisting of lung cancer, pancreatic cancer, myeloma, myeloid leukemia, meningioma, glioblastoma, breast cancer, esophageal squamous cell carcinoma, gastric cancer, prostate cancer, bladder cancer, ovarian cancer, thyroid cancer, and neuroendocrine tumors.

[0140] As used herein, the term "cardiovascular disease" (CVD) or "cardiovascular disorder" is used to classify a number of conditions that affect the heart, heart valves, and the body's vascular system (e.g., veins and arteries) and encompasses diseases and conditions including, but not limited to, atherosclerosis, myocardial infarction, acute coronary syndrome, angina pectoris, congestive heart failure, aortic aneurysm, aortic dissection, iliac or femoral aneurysm, pulmonary embolism, atrial fibrillation, stroke, transient ischemic attack, systolic heart failure, diastolic heart failure, myocarditis, atrial tachycardia, ventricular fibrillation, endocarditis, peripheral artery disease, and coronary artery disease (CAD). Furthermore, the term cardiovascular disease refers to a subject who will eventually have a cardiovascular event or complication, and refers to adverse symptoms in a subject due to cardiovascular disease such as sudden cardiac death or acute coronary syndrome, including, but not limited to, myocardial infarction, unstable angina, aneurysm, stroke, heart failure, non-fatal myocardial infarction, stroke, angina, transient ischemic attack, aortic aneurysm, aortic dissection, cardiomyopathy, abnormal cardiac catheterization, abnormal cardiac imaging, stent or graft revascularization, risk of undergoing abnormal stress tests, risk of undergoing myocardial perfusion, and death.

[0141] As used herein, the ability to detect, diagnose or prognose cardiovascular disease, such as arteriosclerosis, can include determining whether a patient is at a presymptomatic stage of cardiovascular disease, has early, moderate or severe cardiovascular disease, or has suffered one or more cardiovascular events or complications associated with cardiovascular disease.

[0142] Atherosclerosis (atherosclerotic cardiovascular disease or ASVD) is a cardiovascular disease in which arterial walls thicken and arteries narrow and harden as a result of the invasion, accumulation and deposition of arterial plaque containing white blood cells into the innermost layer of the arterial wall. Arterial plaque is an accumulation of macrophage cells or cellular debris and contains lipids (cholesterol and fatty acids), calcium and variable amounts of fibrous connective tissue. Diseases associated with atherosclerosis include, but are not limited to, atherothrombosis, coronary heart disease, deep vein thrombosis, carotid artery disease, angina pectoris, peripheral arterial disease, chronic kidney disease, acute coronary syndromes, vascular stenosis, myocardial infarction, aneurysm or stroke.

[0143] To illustrate, in one embodiment, the sensor array includes different degrees of stenosis of the subject. Without being limited thereto, different stages of atherosclerosis may be distinguished.

[0144] For further illustration, the following example shows the use of the sensor array to detect different states of coronary artery disease. A sensor array containing six sensor elements was able to distinguish subjects diagnosed with CAD by coronary angiography, symptomatic patients with healthy coronary arteries (no CAD), patients with restenosis (recurrence of CAD after treatment), and healthy subjects with no risk factors (Figure 53). The sensor array of the present invention was sensitive enough to detect differences between people with symptoms of coronary artery disease but no arterial stenosis (e.g., no CAD vs. CAD group). This provides a novel diagnostic CAD test that can be used as a non-invasive screen for at-risk patients.

[0145] The term "endocrine disorder" refers to a disorder associated with dysregulation of a subject's endocrine system. Endocrine disorders can occur when glands over- or under-produce endocrine hormones, disrupting hormone balance, or due to the development of lesions in the endocrine system (such as nodules or tumors) that may or may not affect hormone levels. Suitable endocrine disorders that may be treated include, but are not limited to, acromegaly, Addison's disease, adrenal cancer, adrenal gland disorders, aplastic thyroid cancer, Cushing's syndrome, de Quervain's thyroiditis, diabetes, follicular thyroid cancer, gestational diabetes, goiter, Graves' disease, growth disorders, growth hormone deficiency, chronic thyroiditis (Hashimoto's disease), eosinophilic cell thyroid cancer, hyperglycemia, hyperparathyroidism, hyperthyroidism, hypoglycemia, hypoparathyroidism, hypothyroidism, low testosterone, medullary thyroid cancer, MEN1, MEN2A, MEN2B, menopause, metabolic syndrome, obesity, osteoporosis, papillary thyroid cancer, parathyroid disease, pheochromocytoma, pituitary disorders, pituitary tumors, polycystic ovary syndrome, prediabetes, asymptomatic, thyroiditis, thyroid cancer, thyroid disease, nodular goiter, thyroiditis, Turner's syndrome, type 1 diabetes, type 2 diabetes, and the like.

[0146] As referred to herein, inflammatory disease refers to a disease caused by uncontrollable inflammation in the body of a subject. Inflammation is a biological response of a subject to painful external or internal stimuli, such as pathogens, dead cells and tissues, irritants, etc. However, when the inflammatory response becomes abnormal, it can cause soft tissue damage, resulting in various diseases and disorders. Inflammatory diseases include, but are not limited to, asthma, glomerulonephritis, inflammatory bowel disease, rheumatoid arthritis, hypersensitivity, pelvic inflammatory disease, autoimmune diseases, arthritis; necrotizing enterocolitis (NEC), gastroenteritis, pelvic inflammatory disease (PID), emphysema, pleuritis, pyelitis, pharyngitis, angina, acne vulgaris, urinary tract infection, appendicitis, bursitis, colitis, cystitis, dermatitis, phlebitis, rhinitis, tendonitis, tonsillitis, vasculitis, autoimmune diseases; celiac disease; chronic prostatitis, hypersensitivity, reperfusion injury; sarcoidosis, transplant rejection, vasculitis, interstitial cystitis, hay fever, periodontitis, atherosclerosis, psoriasis, ankylosing spondylitis, juvenile rheumatoid arthritis, Behcet's disease, spondyloarthritis, uveitis, systemic lupus erythematosus, and cancer. For example, arthritis includes rheumatoid arthritis, psoriatic arthritis, osteoarthritis, or juvenile idiopathic arthritis.

[0147] Neuropathy or neurological disease are used interchangeably and refer to diseases of the brain, spine and the nerves connecting them. Neurological diseases include, but are not limited to, brain tumors, epilepsy, Parkinson's disease, Alzheimer's disease, ALS, arteriovenous malformations, cerebrovascular disease, cerebral aneurysms, epilepsy, multiple sclerosis, peripheral neuropathy, post-herpetic neuralgia, stroke, frontotemporal dementia, demyelinating diseases, multiple sclerosis, Devic's disease (i.e., neuromyelitis optica), central pontine myelinolysis, progressive multifocal leukoencephalopathy, leukodystrophies, Guillain-Barre syndrome, progressive inflammatory neuropathies, Charcot-Marie-Tooth disease, chronic inflammatory demyelinating polyneuropathy and anti-MAG peripheral neuropathy. Neuropathies also include immune-mediated neuropathies (IMNDs), which include diseases in which at least one component of the immune system reacts against host proteins present in the central or peripheral nervous system and contributes to the disease pathology. IMNDs include demyelinating diseases, paraneoplastic neurological syndromes, immune-mediated encephalomyelitis, immune-mediated autonomic neuropathy, myasthenia gravis, and other conditions. These include, but are not limited to, acute myelitis, autoantibody-associated encephalopathy, and acute disseminated encephalomyelitis.

[0148] In a non-limiting example, the following example provides a method for diagnosing Alzheimer's disease in a patient using the sensor array and methods described herein. The sensor array was not only able to accurately distinguish between patients with and without Alzheimer's disease, but was also able to detect pre-symptomatic patients who developed Alzheimer's disease (as determined by cohort plasma) several years after screening. This provides the advantage of being able to treat the disease at a very early stage before the disease develops.

[0149] The sensor arrays and methods of the present invention in some embodiments are capable of detecting pre-symptomatic stages of a disease or disorder. A pre-symptomatic stage is one in which a patient does not exhibit any signs or symptoms of the disease. A pre-symptomatic stage of cancer is one in which no cancer or tumor or cancer cells have been identified in a subject. A pre-symptomatic stage of a neurological disease is one in which a person does not exhibit one or more symptoms of a neurological disease. The ability to diagnose a disease before one or more signs or symptoms of the disease appear allows the subject to be closely monitored and the disease to be treated at a much earlier stage, which can provide a better prospect of halting the progression of the disease or reducing the severity of the disease.

[0150] The sensor array and method of the present invention in some embodiments can detect early stages of disease or disorders. An early stage of disease refers to when the first signs or symptoms of the disease become evident in a subject. In most cases, diseases that can be detected before or at an early stage are easier to treat and result in a more positive outcome for the patient. For example, for cancer, early stages of disease include stage 0 and stage 1 cancer. Stage 0 is cancer in situ, which means that the cancer is still in the location where it originated and has not spread to nearby tissues. Cancers at this stage can often be cured by surgically removing the entire tumor in most cases. Stage 1 cancers are often small cancers or tumors that have not grown deep into nearby tissues and have not spread to lymph nodes or other parts of the body. Additionally, the color stage of a disease can be a stage where there are no outward signs or symptoms. For example, for Alzheimer's disease, an early stage is the pre-Alzheimer's stage, where symptoms are not yet detectable and the patient develops Alzheimer's disease months or years later.

[0151] In some embodiments, the sensor array and method can detect intermediate stages of disease, where the patient is experiencing one or more symptoms of the disease after the initial signs and symptoms appear. For example, for cancer, stage II or III cancer is considered intermediate stage and indicates a larger cancer or tumor that has grown deeper into nearby tissues. In some cases, stage II or III cancer may also have spread to lymph nodes but not to other parts of the body.

[0152] Additionally, the sensor arrays and methods are capable of detecting late or advanced stages of disease, which may also be referred to as "severe" or "advanced," and often indicate that the patient is suffering from multiple symptoms and effects of the disease. For example, severe stages of cancer include stage IV, where the cancer has spread to other organs or parts of the body, and is sometimes referred to as advanced or metastatic cancer.

[0153] In some embodiments, the methods of the present technology include comparing the protein fingerprint of the sample to a panel of protein fingerprints associated with multiple diseases and / or multiple disease states to determine whether the sample is indicative of a disease and / or disease state. For example, samples can be taken from a population of subjects over time. When a subject develops a disease or disorder, the invention can characterize and detect changes in the biomolecular fingerprint in the subject over time by comparing the biomolecular fingerprint of a sample from the same subject before the subject developed the disease to the biomolecular fingerprint of the subject after the subject developed the disease. In some embodiments, samples can be obtained from a cohort of patients who all developed the same disease, allowing the biomolecular fingerprints associated with different stages of the disease (e.g., from a pre-symptomatic state to a diseased state) for these patients to be analyzed and characterized.

[0154] To illustrate, examples have shown that the methods and sensor arrays of the present invention are not only capable of distinguishing between different types of cancer, but also different stages of cancer (eg, early stages of cancer).

[0155] A method of determining a biomolecular fingerprint associated with at least one disease or disorder and / or disease state is contemplated, comprising obtaining samples from at least two (2) subjects diagnosed with at least one disease or disorder or having the same disease state, contacting each sample with a sensor array as described herein to determine a biomolecular fingerprint for each sensor array, and analyzing the fingerprints of the at least two samples to determine a biomolecular fingerprint associated with the at least one disease or disorder and / or disease state.

[0156] <Classification of biomolecular corona> The method for determining a biomolecular fingerprint associated with a disease or disorder and / or disease state includes analyzing the biomolecular fingerprints of at least two samples. This determination, analysis or statistical classification is performed by methods well known in the art, including, but not limited to, various supervised and unsupervised data analysis, machine learning, deep learning and clustering approaches including hierarchical cluster analysis (HCA), principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), random forests, logistic regression, decision trees, support vector machines (SVM), K-nearest neighbors, naive Bayes, linear regression, polynomial regression, SVM for regression, K-means clustering and hidden Markov models. In other words, the biomolecular fingerprints of each sample are compared and analyzed with each other to determine common patterns in the individual fingerprints with statistical significance to determine the biomolecular fingerprint associated with a disease or disorder or disease state.

[0157] Typically, machine learning algorithms are used to build models that accurately assign class labels to examples based on input features that describe the examples. In some cases, it may be advantageous to employ machine learning and / or deep learning approaches in the methods described herein. For example, machine learning can be used to associate biomolecular fingerprints with various disease states (e.g., no disease, prodromal disease, early or late stage disease, etc.). For example, in some cases, one or more machine learning algorithms are employed in conjunction with the methods of the invention to analyze the data detected and obtained by the biomolecular coronas and biomolecular fingerprints derived therefrom. For example, in one embodiment, machine learning can be coupled with the sensor arrays described herein to determine whether a subject is pre-cancer, has cancer, or does not have or has cancer, as well as to determine the type of cancer.

[0158] The following examples demonstrate the ability of the sensor arrays described herein to determine, with statistical significance, disease states for a number of different diseases, including cancer, cardiovascular disease, and neurological diseases (e.g., Alzheimer's disease). As described below, the sensor arrays can be used to determine disease states for a variety of diseases and disease states. The assay is not limited to these particular embodiments, as it can be applied to any of the above.

[0159] In some embodiments, the method includes obtaining samples from control subjects that are contacted with the sensor array to generate control biomolecular fingerprints. These control biomolecular fingerprints can then be used to compare with the biomolecular fingerprints of subjects with a disease or disorder and / or a particular disease state to determine a biomolecular fingerprint that is characteristic of the disease or disorder and / or a particular disease state.

[0160] The method may, for example, include obtaining a control sample from at least one control subject, contacting the control sample with a sensor array to generate a plurality of control biomolecular coronas, and analyzing the plurality of biomolecules of each control biomolecular corona. The method may further include comparing the plurality of biomolecules of the plurality of control biomolecular coronas to the plurality of biomolecules of the plurality of biomolecular coronas from a subject having a disease or disorder to determine a biomolecular fingerprint associated with at least one disease or disorder.

[0161] Methods for diagnosing or prognosing a disease or disorder are also contemplated, comprising obtaining a sample from a subject, contacting the sample with a sensor array to generate a biomolecular fingerprint, comparing the biomolecular fingerprint to a panel of biomolecular fingerprints associated with a plurality of diseases or disorders, and diagnosing or prognosing the disease or disorder.

[0162] In some embodiments, a method of identifying a pattern or specific biomarker associated with a disease or disorder is contemplated. Suitable methods include, for example, performing the methods described above, e.g., obtaining samples from at least two subjects diagnosed with a disease or disorder and at least two control subjects, contacting each sample with a sensor array to generate a biomolecular fingerprint, and comparing the biomolecular fingerprint of the subjects with the disease or disorder to the biomolecular fingerprint of the control subjects to determine at least one pattern and / or biomarker associated with the disease or disorder. Suitably, the method comprises at least two diseased subjects and at least two control subjects, or at least five diseased subjects and at least five control subjects, at least ten diseased subjects and at least ten control subjects, at least fifteen diseased subjects and at least fifteen control subjects, at least twenty diseased subjects and at least twenty control subjects, and all variations therebetween (e.g., at least two to one hundred diseased subjects and at least two to one hundred control subjects).

[0163] In some embodiments, the arrays and methods can determine a pattern of biomarkers associated with a disease state or a disease or disorder, or in some embodiments, a specific biomarker associated with a disease or disorder. For example, not only can biomarkers that may be associated with disease states listed herein be identified, but new biomarkers or patterns of biomarkers associated with a disease state or a disease or disorder can also be determined. As discussed above, some biomarkers or patterns of biomarkers for a particular disease or disorder can be changes in biomolecules associated with the sensor array of the invention and can be different from what is typically referred to in the art as a biomarker, for example, increased expression of a particular biomolecule associated with the disease. As discussed above, interactions between a biomolecule (e.g., biomolecule X) and other biomolecules (e.g., biomolecules Y and Z) provide the ability to associate with a particular disease state, and are unrelated to the absolute concentration of biomarker X in a sample over time or across disease states. Thus, a molecule that would not be considered a biomarker in the traditional sense because its absolute concentration in a sample does not change from pre-symptomatic to disease states, in the context of this disclosure, can be associated with a disease state because its relative change measured by the array of the invention. In other words, an increase or decrease in the interaction between biomolecule X and the array (due to interactions of X with the sensor elements and other biomolecules in the sample) provides a signal that the biomarker is associated with a disease state.

[0164] Suitable cancer biomarkers include, for example, AHSG (α2-HS-glycoprotein), AKR7A2 (aflatoxin B1 aldehyde reductase), AKT3 (PKBγ), ASGR1 (ASGPR1), BDNF, BMP1 (BMP-1), BMPER, C9, CA6 (carbonic anhydrase VI), CAPG (CapG), CDH1 (cadherin-1), CHRDL1 (Chordin-Like1), CKB-CKM- (CK-MB), CLIC1 (intracellular chloride channel 1), CMA1 (chymase), CNTN1 (contactin-1), COL18A1 (endostatin), CRP, CTSL2 (cathepsin V), DDC (dopa decarboxylase), EGFR (ERBB1), FGA-FGB-FGG (D-dimer), FN1 (fibronectin FN1.4), GHR (growth hormone receptor), GPI (glucose phosphate isomerase), HMGB1 (HMG-1), HNRNPAB (hnRNP A / B), HP (haptoglobin, mixed type), HSP90AA1 (HSP90α), HSPA1A (HSP70), IGFBP2 (IGFBP-2), IGFBP4 (IGFBP-4), IL12B-IL23A (IL-23), ITIH4 (inter-α-trypsin inhibitor heavy chain H4), KIT (SCF sR), KLK3-SERPINA3 (PSA-ACT), L1CAM (NCAM-L1), LRIG3, MMP12 (MMP-12), MMP7 (MMP-7), NME2 (NDP kinase B), PA2G4 (ErbB3 binding protein Ebp1), PLA2G7 (LpPLA2 / PAFAH), PLAUR (suPAR), PRKACA (PRKA C-α), PRKCB (PKC-βII), PROK1 (EGTA-VEGF), PRSS2 (trypsin-2), PTN (pleiotrophin), SERPINA1 (α1-antitrypsin), STC1 (stanniocalcin-1), STX1A (syntaxin1A), TACSTD2 (GA733-1 protein), TFF3 (trefoil factor 3), TGFBI (βIGH3), TPI1 (triosephosphate isomerase), TPT1 (fortilin), YWHAG (14-3-3 protein gamma), YWHAH (14-3-3 protein eta), prostate cancer biomarkers such as PSA, Pro-PSA, PHI, PCA3, TMPRSS3:ERG, PCTMT, MTEN, breast cancer markers such as epidermal growth factor receptor 2 (HER2) oncogene, melanoma biomarker BRAF, lung cancer biomarkers EML4-ALK, A2ML1, BAX, C10orf47, Clorfl62, CSDA, EIFC3, ETFB, GABARAPL2, GUKI, GZMH, HIST1H3B, HLA-A, HSP90 AA1, NRGN, PRDX5, RABACI, RABAGAP1L, RPL22, SAP18, SEPW1, SOX1, EGFR, EGFRvIII, apolipoprotein AI, apolipoprotein CIII, myoglobin, tenascin C, MSH6, claudin-3, claudin-4, caveolin-1, coagulation factor III, CD9, CD36, CD37, CD53, CD63, C D81, CD136, CD147, Hsp70, Hsp90, Rabl3, Desmocollin-1, EMP-2, CK7, CK20, GCDF15, CD82, Rab-5b, Annexin V, MFG-E8, HLA-DR, miR200 microRNA, MDC, NME-2, KGF, PIGF, Flt-3L, HGF, MCP1, SAT-1, NIP-1-b, GCLM, OPG, TNFBiomarkers include, but are not limited to, RII, VEGF-D, ITAC, MMP-10, GPI, PPP2R4, AKR1B1, Amy1A, MIP-1b, P-cadherin, EPO, etc. For example, biomarkers for breast cancer include, but are not limited to, ER / PR, HER-2 / neu, etc. Biomarkers for colon cancer include, but are not limited to, EGFR, KRAS, UGT1A1, etc. Biomarkers related to leukemia / lymphoma include, but are not limited to, for example, CD20 antigen, CD30, FIP1L1-PDGFRalpha, PDGFR, Philadelphia chromosome (BCR / ABL), PML / RARalpha, TPMT, UGT1A1, etc. Biomarkers related to lung cancer include, but are not limited to, for example, ALK, E These include, but are not limited to, GFR, KRAS, etc. Biomarkers are known in the art and are described, for example, in Bigbee W, Herberman RB. Tumor markers and immunodiagnosis. In: Bast RC Jr., Kufe DW, Pollock RE, et al., editors. Cancer Medicine. 6th ed. Hamilton, Ontario, Canada: BC Decker Inc., 2003; Andriole G, Crawford E, Grubb R, et al. Mortality results from a randomized prostate-cancer screening trial. New England Journal of Medicine 2009; 360(13):1310-1319; Schroder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-cancer mortality in a randomized European study. New England Journal of Medicine 2009; 360(13):1320-1328; Buys SS, Partridge E, Black A, et al. Effect of screening on ovarian cancer mortality: the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Randomized Controlled Trial. JAMA 2011; 305(22):2295-2303; Cramer DW, Bast RC Jr, Berg CD, et al. Ovarian cancer biomarker performance in prostate, lung, colorectal, and ovarian cancer screening trial specimens. Cancer Prevention Research 2011; 4(3):365-374; Sparano JA, Gray RJ, Makower DF, et al. Prospective validation of a 21-gene expression assay in breast cancer.New England Journal of Medicine 2015; First published online September 28, 2015. doi: 10.1056 / NEJMoa1510764. the entire contents of which are incorporated herein by reference.

[0165] Advantageously, these methods can be used to determine biomarkers associated with cancer. For example, in one embodiment, the cancer is glioblastoma, and the biomarkers are selected from the group consisting of HABP1, VTNC, CO3, ITIH2, ITIH1, CO7, FHR5, CBPN, ALBU, PLMN, CO4A, PRDX2, VWF, C4BPA, APOB, HBB, CNDP1, CRP, SAA4, APOE, CSCL7, and combinations thereof. In another embodiment, the cancer is meningioma, and the biomarkers are selected from the group consisting of FCN3, RET4, HABP2, CBPN, and combinations thereof. In another embodiment, the cancer is pancreatic cancer, and the biomarkers are selected from the group consisting of KNG1, IC1, CBPB2, TRFE, GELS, CXCL7, HPTR, PGK1, AACT, LUM, APOE, FIBB, APOA2, A1BG, A1AT, LBP, APOA1, H4, FIBG, and combinations thereof. In another embodiment, the cancer is lung cancer and the biomarker is selected from the group consisting of CO0, CRP, SAA4, APOA1, A1AT, GELS, and combinations thereof. In another embodiment, the cancer is myeloma and the biomarker is ALBU.

[0166] Biomarkers can also be associated with cardiovascular disease, which are well known in the art and include, but are not limited to, physiological biomarkers based on measurements of lipid profiles, glucose and hormone levels, and levels of important biomolecules such as serum ferritin, triglyceride to HDLp (high density lipoprotein) ratio, lipophorin to cholesterol ratio, lipid to lipophorin ratio, LDL cholesterol levels, HDLp and apolipoprotein levels, lipophorin and LTP ratio, sphingolipids, omega-3 index, and ST2 levels, among others. Suitable biomarkers for cardiovascular disease are described, for example, but not limited to, van Holten et al. "Ciculating Biomarkers for Predicting Cardiovascular Disease Risk; a Systemic Review and Comprehensive Overview of Meta-Analyses" PLoS One, 2013 8(4): e62080, the entire contents of which are incorporated herein by reference. No. 6,313,935, filed on Oct. 23, 2003, and is incorporated herein by reference.

[0167] Biomarkers are also associated with neurological diseases. Suitable biomarkers are well known in the art and include, but are not limited to, Aβ1-42, t-tau and p-tau181, α-synuclein, among others. See, for example, Chintamaneni and Bhaskar "Biomarkers in Alzheimer's Disease: A Review" ISRN Pharmacol. 2012. 2012: 984786. Published online 2012 Jun 28, the entire contents of which are incorporated herein by reference.

[0168] Biomarkers for inflammatory diseases are well known in the art and include, for example, cytokines / chemokines, immune-related effectors, acute phase proteins [C-reactive protein (CRP) and serum amyloid A (SAA)], reactive oxygen species (ROS), and other inflammatory diseases, such as C-reactive protein (CRP), S100, LIF, CXCL1, CXCL2, CXCL4, CXCL5, CXCL8, CXCL9, CXCL10, CCL2, CCL23, IL-Iβ, IL-IRa, TNF, IL-6, IL-10, IL-17A, IL-17, among others. These include, but are not limited to, mediators such as transcription factors and growth factors including F, IL-21, IL-22, IFNγ, CXCR1, CXCr4, CXCR5, GM-CSF, GM-CSFR, G-CSF, G-CSF, G-CSFR, EGF, VEGFA, LEP, SAA1, VCAM1, CRP, MMP1MMP3, TNFRSF1A, RETN, CHI3L1, antinuclear antibodies (ANA), rheumatoid factor (RF), antibodies against cyclic citrullinated peptide (anti-CCP) and antibodies against chronic IBD (fecal calprotectin). For example, suitable biomarkers for inflammatory bowel disease include CRP, ESR, pANCA, ASCA and fecal calprotectin. See, for example, Yi Fengming and Wu Jianbing, "Biomarkers of Inflammatory Bowel Disease," Disease Markers, vol. 2014, Article ID 710915, 11 pages, 2014. doi:10.1155 / 2014 / 710915, the entire contents of which are incorporated herein by reference. Can be enjoyed.

[0169] The terms "individual," "subject," and "patient" are used interchangeably herein, regardless of whether the subject has been or is undergoing any form of treatment. As used herein, the term "subject" generally refers to any vertebrate, including, but not limited to, mammals. Examples of mammals include primates, including monkeys and humans, equines (e.g., horses), canines (e.g., dogs), cats, various domesticated livestock (e.g., ungulates, such as pigs, hogs, goats, sheep, etc.), and domesticated pets (e.g., cats, hamsters, mice, and guinea pigs). Preferably, the subject is a human.

[0170] The arrays and methods described herein can be used under many different conditions to provide a desired biomolecular fingerprint. For example, the size of the sensor element, the flow rate of the sample through the sensor, the time the sensor array is incubated with the sample, and the temperature at which the sensor array is incubated can all be varied to provide a reproducible biomolecular fingerprint. Suitable sizes of sensor elements include nanoscale sensor elements that are less than a micrometer in at least one dimension.

[0171] Suitable times for incubating the array or plurality of sensor elements include at least a few seconds, e.g., at least 10 seconds to about 24 hours, e.g., at least about 10 seconds, at least about 15 seconds, at least about 20 seconds, at least about 25 seconds, at least about 30 seconds, at least about 40 seconds, at least about 50 seconds, at least about 60 seconds, at least about 90 seconds, at least about 2 minutes, at least about 3 minutes, at least about 4 minutes, at least about 5 minutes, at least about 6 minutes, at least about 7 minutes, at least about 8 minutes, at least about 9 minutes, at least about 10 minutes, at least about 15 minutes, at least about 20 minutes, at least about 25 minutes, at least about 30 minutes, at least about 45 minutes, at least about 50 minutes, at least about 60 minutes, at least about 90 minutes, at least about 2 hours, at least about 3 hours, at least about 4 hours, at least about 5 hours, at least about 6 hours, at least about 7 hours, at least about 8 hours, at least about 9 hours, at least about 10 hours, at least about 12 hours, at least about 14 hours, at least about 15 hours, at least about 16 hours, at least about 17 hours, at least about 18 hours, at least about 19 hours, at least about 20 hours, including all times and increments therebetween (e.g., 10 seconds, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49 , 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60 seconds etc; 1 minute, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60 etc; 1 hour, 2 hours, 3 hours, 4 hours, 5 hours, 6 hours, 7 hours, 8 hours etc).

[0172] Furthermore, the temperature at which the analysis is performed can be determined by one of skill in the art and includes temperatures of about 4°C to about 40°C, alternatively about 4°C to about 20°C, alternatively about 10°C to about 15°C, alternatively about 10°C to about 40°C, such as about 4°C, about 5°C, about 6°C, about 7°C, about 8°C, about 9°C, about 10°C, about 11°C, about 12°C, about 13°C, about 14°C, about 15°C, about 16°C, about 17°C, about 18°C, about 19°C, about 20°C, about 21°C, about 22°C, about 25°C, about 30°C, about 35°C, about 37°C, etc. Suitably, the analysis may be performed at room temperature (e.g., around about 37°C, e.g., about 35°C to about 40°C).

[0173] The method may include contacting the sample with the sensor array. Contacting the sample with the sensor array may use any flow rate that allows the sample to flow through the sensor array. In some aspects, the flow rate of the stream, the Reynolds number, or the relative cross-sectional areas of the flow streams can be varied to provide adequate contact between the sample and the sensor array.

[0174] For example, in embodiments using nanochannels or microchannels, the cross-sectional area of ​​the first stream may be greater than 1% of the cross-sectional area of ​​the channel. In another example, the cross-sectional area of ​​the first stream may be less than 90% of the cross-sectional area of ​​the channel. The cross-sectional area ratio may be 10:1 to 1:10, 1:5 to 5:1, 1:3 to 3:1, 1:2 to 2:1, or 1:1.

[0175] In certain other circumstances, the flow of the sample through the array may have a Reynolds number at the sample introduction location of between 300 and 1,000,000. In some cases, the sample introduction location may be a nanochannel or a microchannel.

[0176] <Kit> Aspects of the disclosure described with respect to methods can be used in conjunction with the sensor arrays or kits discussed in this disclosure, and similarly, aspects of the disclosure described with respect to sensor arrays and methods can be used in conjunction with kits, and aspects of the disclosure described with respect to kits can be used in conjunction with methods and sensor arrays.

[0177] The present disclosure provides a kit. The kit is suitable for use in the methods described herein. A suitable kit includes a kit for determining a biomolecular fingerprint for a sample, comprising a sensor array as described herein. In one aspect, the kit provides a sensor array comprising at least two sensor elements having different physicochemical properties. In some aspects, the kit provides a comparative panel of biomolecular fingerprints for using the biomolecular fingerprints to determine a disease state of a subject. In some aspects, the kit includes instructions for determining the biomolecular fingerprint. In some preferred embodiments, the sensor array is provided as a chip array in the kit.

[0178] In another aspect, a kit for determining a disease state in a subject or diagnosing or prognosing a disease in a subject is provided. A suitable kit comprises a sensor array with at least two sensor elements having different physicochemical properties for determining a biomolecular fingerprint. Further, the kit may comprise a sensor array for detecting biomolecules of different disease states or different diseases or disorders. Further included is a comparison panel of fingerprints. Instructions for the determination and analysis of biomolecular fingerprints are provided.

[0179] It will be apparent to those skilled in the art that many additional modifications beyond those already described can be made without departing from the concept of the present invention. In interpreting this disclosure, all terms should be interpreted in the broadest possible manner consistent with the context. Variations of the term "comprising" should be interpreted to refer inclusively to elements, components, or steps, such that a referenced element, component, or step may be combined with other elements, components, or steps not specifically referenced. An embodiment that is described as "comprising" certain elements may also be considered to be "consisting primarily of" and "consisting of" those elements. The terms "consisting primarily of" and "consisting of" should be interpreted in accordance with the interpretations of the MPEP and the relevant Federal Circuit Court. The transitional phrase "consisting primarily of" limits the scope of the claim to the specified materials or steps and "those that do not materially affect the basic and novel characteristics of the claimed invention." "Consisting of" is a term that excludes any element, step, or ingredient not recited in the claim.

[0180] The following non-limiting examples are provided for illustrative purposes only and are not intended to limit the scope of the breadth of techniques and protocols which may be utilized with the compositions and methods of the present invention, which will be readily understood and practiced by one of ordinary skill in the art.

[0181] [example] <Example 1A: Label-free sensor array for early cancer detection> This example provides a label-free sensor array for the early detection of various cancers. The sensor array is composed of three different cross-reactive liposomes with different surface charges (i.e., cationic (DOPG (1,2-dioleoyl-sn-glycero-3-phospho-(1'-rac-glycerol)), anionic (DOTAP (1,2-dioleoyl-3-trimethylammonium-propane)-DOPE (dioleoylphosphatidylethanolamine)) and intermediate (CHOL (DOPC-cholesterol))) whose protein corona composition changes in response to interaction with plasma from patients with different types of various cancers, namely lung cancer, pancreatic cancer, myeloma, meningioma and glioblastoma. Although there was no single protein corona composition specific for any type of cancer, the change in the corona composition pattern provides a unique "fingerprint" for each type of cancer. Hard Coronas are profiled on the sensor array elements with plasma from patients with early, intermediate and advanced cancers. The protein coronas formed on the surface of the sensor array elements (nanoparticles) are strongly dependent on the physicochemical properties of those nanoparticles and can also be strongly influenced by the type of disease present in the donor of the human plasma used for incubation. The size and charge of corona-coated nanoparticles after incubation with plasma from patients with five different types of cancer (i.e. glioblastoma multiforme, lung cancer, meningioma, multiple myeloma and pancreatic cancer) and healthy subjects (see Table 1) were investigated using dynamic light scattering (DLS / Nanosight™) and transmission electron microscopy (TEM), and the results showed that the physicochemical properties of the corona-coated nanoparticles are substantially dependent on the type of cancer (Figures 2A, 2B).

[0182] [Table 1] TIFF2024170513000003.tif114170

[0183] Quantitative assessment of total proteins adsorbed on the nanoparticles was performed via BCA or NanoOrange® assays and showed significant differences in the amount of adsorbed proteins after incubation in plasma from patients with different types of cancer (Figure 2B). Quantitative assessment of total proteins adsorbed on the surface of liposomes showed that the amount of protein was highly dependent on the type of cancer (Figure 2B). Protein corona composition on the surface of three liposomes was evaluated by liquid chromatography-mass spectrometry (LC-MS / MS) revealing a high abundance of 1,800 known proteins. The contribution of individual proteins and their categories (i.e., complement, coagulation, tissue leakage, lipoproteins, acute phase, immunoglobulins and other plasma proteins) to the corona composition was revealed (Figure 2C, Figures 3A-3F). These results showed a significant association between protein composition and type of cancer and also with the type of sensor element (i.e., type of nanoparticle).

[0184] Extensive literature has shown that there is a significant relationship between cancer development and variations in complement, coagulation, tissue leakage, lipoproteins, acute phase, and immunoglobulins. Therefore, cross-reactive interactions of these protein categories with nanoparticles can provide a unique "fingerprint" for each type of cancer, facilitating cancer identification and differentiation. It is therefore anticipated that the protein corona sensor array will cross-reactively adsorb a wide range of proteins involved in cancer induction and development, which can be used for cancer identification and differentiation.

[0185] Supervised classification is deployed to identify and differentiate cancers using the results from the protein corona sensor array. A focused classification approach is applied to the proteomic data of the protein corona composition (cationic, anionic, and neutral) of three liposomes to investigate whether the protein corona fingerprints (PCFs) of different sensor elements can be utilized as biosensors and form unique patterns for different diseases. The details of the method are described in the Methods section. A variable importance in weighted projection (VIP) score was introduced and applied to the ranking of variables based on partial least squares discriminant analysis (PLS-DA) as a linear projection method. The selection of the most important variables (proteins) in building the classification model can be guided by the obtained set of ranked variables. In this regard, the top ranked variables were added one by one to the model and the classification error of the PLS-DA model was monitored. We found that the error of the classification model was minimized by using only the top 69 features (Figure 4A). The new 69-dimensional feature space was successfully used to discriminate 30 samples belonging to six classes by PLS-DA with a high classification accuracy (0.97) using leave-one-out and 10-fold cross-validation (Figures 4A, 4B). The classification parameters are shown in Table 2. To provide a visual representation of the relative specificity of the findings, the contribution of each single selected protein (VIP) to the separation of each cancer group was plotted on the y-axis and x-axis, respectively (Figures 4B-4F). Proteins with higher VIP scores can be considered the most informative set or the best diagnostic set in discriminating each disease from controls and among all cancer categories.

[0186] [Table 2]

[0187] PLS-DA and Counterpropagation Artificial Neural Networks (CPANN) were then applied to the selected variables and the entire sample as supervised classification, linear and non-linear approaches, respectively. Note that the primary dataset with all variables before variable selection was poorly discriminant and could not be separated into six groups.

[0188] Next, we decided to use a nonlinear classification method to further examine and analyze the data. Visualizing the feature space can help understand the hidden structure and topological relationships among the patterns. To reduce the dimensionality of the feature space while preserving the topological relationships in the data structure, CPANN (supervised variant of self-organizing map / SOM) was used to learn and predict the class membership of the patterns and at the same time create a 2D map of neurons to provide valuable information about the data structure (from the nonlinear approach). Details of CPANN are described in the Methods section. CPANN maps of different sizes were checked using 10-fold cross-validation and a map containing 64 (8 × 8) neurons was selected with the minimum classification error (Figure 9C). Furthermore, the topological structure of the data in the high-dimensional space is reflected in the assignment map created by CPANN (Figures 5C, 5D). Considering the similarity of neurons to the input vector, the map can be partitioned into six distinct zones associated with different types of cancer and control samples. Samples with the same class label are mapped onto nearby or same neurons, which means that the selected variables provide information that helps discriminate samples in the feature space. The relative positions and orientations of the six zones on the map can contribute qualitative information about the similarity between cancer types. To illustrate the impact of variable selection on the quality of the mapping, another CPANN was trained with all 1823 variables, and the resulting maps showed that the selected biomarkers (variables) play an important role in discriminating between cancer types and classifying them accurately (Figures 5C and 5D).

[0189] Based on the obtained results, linear and nonlinear models showed high accuracy as estimated from acceptable specificity, sensitivity and class error values. In agreement with these findings, unsupervised clustering (HCA) based on 69 markers allowed clear separation of various cancer and control samples (Figures 5E-5F). As shown in Figures 5A-5F, there is a considerable similarity between the glioblastoma and meningioma groups of samples, which is difficult to discriminate, probably closely related to the similar plasma proteomic patterns in these two brain cancers. These results reflect the fact that the plasma concentrations of many proteins in the corona vary widely not only between patients with different types of cancer, but also between healthy individuals.

[0190] To illustrate the pattern recognition capabilities of the biosensor, we performed a series of analyses on the data obtained from individual nanoparticles. As expected, cancer-specific fingerprint patterns could not be extracted from the PCF of each nanoparticle alone. We found that pattern recognition techniques applied to the abundance of proteins in the protein corona formed on three different liposomes (cationic, anionic, neutral) not only correctly distinguished cancer from control samples, but also each type of cancer under consideration from the others.

[0191] Identification of proteins with important roles in cancer detection and differentiation as potential biomarkers for certain types of cancer The use of biomarkers both before cancer diagnosis (in risk assessment and screening / early detection) and after cancer diagnosis (in treatment monitoring, selection of additional treatments, and recurrence detection) could generate significant therapeutic and health economic benefits. To understand the potential biological relevance of the 69 selected proteins that differentiate cancer samples, we manually searched previously published reports in PubMed about protein biomarkers of specific types of cancer that were up- or down-regulated according to different disease stages. The resulting data were compared with the selected proteins in the model for the identification of matched markers and the determination of the biological relevance of the proposed model. Interestingly, we found that the specific cancer biomarkers were significantly correlated with the 69 selected proteins. We noticed a significant number of biomarkers specific to the five cancer study groups among the selected predictors reported as markers ( Figure 7B ).

[0192] The high specificity of the selected markers for distinguishing the five groups of cancers obtained from the introduced protein corona sensor array approach showed significant correlation with ongoing work in the complex cancer proteomic space. Thus, this strategy not only provides a basis for cancer prediction but also translates its promise into reality. Notably, the discrimination between the different groups occurs as a result of multiple predictors (not individual biomarkers) that vary systematically simultaneously, which results in a unique pattern for each specific type of cancer. Based on this evidence, the most informative predictors selected by the proposed model, which have not been reported as cancer-related biomarkers so far, have great potential as novel diagnostic biomarker candidates. The variations and functions of these promising candidates in cancer patients should be closely monitored to define their role in cancer development. By focusing on the unique patterns obtained from a huge number of subjects through a set of informative predictors, researchers will be able to predict cancer at different stages more accurately than with current methods.

[0193] Cohort data analysis To investigate the capability of our sensor array for ultra-early cancer detection, we used a cohort of plasma samples from healthy patients who were diagnosed with one of five types of cancer several years after plasma collection. We used the cohort samples to evaluate the applicability of our proposed models, linear and nonlinear, with 69 selected predictors for cancer prediction.

[0194] To achieve this goal, values ​​associated with the 69 variables were entered into the model to predict class membership for each cohort subject at a given fixed optimal parameter value. Notably, 19 variables (proteins) out of the 69 variables were absent in the proteomic profile of the protein corona sensor array of the cohort samples, and therefore their quantities in the validation data matrix were zero. Interestingly, both the linear and nonlinear models yielded good predictions for all five samples. The observed distances between the training and cohort samples (Figures 6A and 6B) were probably influenced by the zero values ​​added to the data matrix for these 19 absent variables. As shown in Figures 6C and 6D, the cohort samples were placed in the correct neurons related to glioblastoma in the CPANN map method.

[0195] Liposomes Cholesterol (Chol) was purchased from Sigma Aldrich (St. Louis, MO, USA). DOPC (dioleoylphosphatidylcholine), DOPE (dioleoylphosphatidylethanolamine), DOPG (1,2-dioleoyl-sn-glycero-3-phospho-(1'-rac-glycerol)) and DOTAP (1,2-dioleoyl-3-trimethylammonium-propane) were purchased from Avanti Polar. Lipids (Alabaster, AL, USA). DOPG, DOTAP-DOPE (1:1 molar ratio) and DOPC-Chol (1:1 molar ratio) liposomes were prepared by dissolving the appropriate amount of lipid in chloroform:methanol = 9:1 (v / v). The chloroform:methanol mixture was evaporated by rotary evaporation. The lipid film was placed under vacuum overnight and hydrated with phosphate buffered saline (PBS) 10 mmol / L (pH 7.4) to a final lipid concentration of 1 mg / ml. The resulting liposome suspension was sized by extrusion through a 50 nm polycarbonate filter with an Avanti mini extruder (Avanti Polar Lipids, Alabaster, AL, USA).

[0196] Human plasma collection, preparation and storage Human plasma (HP) was collected from healthy subjects and from cancer patients diagnosed with glioblastoma multiforme, lung cancer, meningioma, multiple myeloma, or pancreatic cancer. The study was approved by the Ethics Committee of the University of Rome La Sapienza (glioblastoma multiforme, meningioma, multiple myeloma), the University of Naples Federico II (lung cancer), and the University of Rome Biomedical Sciences (pancreatic cancer). Briefly, blood was collected by venipuncture from healthy subjects and cancer patients using a BD® P100 blood collection system (Franklin Lakes, NJ, USA) equipped with push-button technology to reduce blood waste while minimizing the risk of contamination. After clot formation, samples were centrifuged at 1000 × g for 5 min to pellet blood cells and the supernatant was removed. After ensuring the absence of hemolysis, plasma (1 ml) collected from each donor was divided into 200 microliter aliquots and stored in labeled Protein LoBind tubes at -80 °C until use. For analysis, aliquots were thawed at 4°C and then warmed to room temperature (RT).

[0197] Cohort plasma samples We used human plasma collected from healthy individuals who were diagnosed with brain, lung, and pancreatic cancer within 8 years of plasma collection. The plasma samples were collected through the NIH-funded Golestan Cohort Study conducted by the National Cancer Institute (NCI) in the United States, the International Agency for Research on Cancer (IARC) in France, and Tehran University of Medical Sciences (TUMS) in Iran. In this study, plasma was collected and stored from 50,000 healthy subjects, of which more than 1,000 developed various types of cancer in the following years. The study used samples from five individuals for each cancer. These important plasma samples provide us with a unique opportunity to investigate the capabilities of our inventive protein corona sensor array for early cancer detection.

[0198] Size and Zeta Potential Uncoated liposomes were incubated with HP (1:1 v / v) for 1 h at 37 °C. Afterwards, samples were centrifuged at 14000 rpm for 15 min at 4 °C to pellet the liposome-HP (human plasma) complexes. The resulting pellets were washed three times with phosphate-buffered saline (PBS) and resuspended in ultrapure water. For size and zeta potential measurements, 10 μL of each sample was diluted with 990 μL of distilled water. All size and zeta potential measurements were performed at room temperature using a Zetasizer Nano ZS90 (Malvern, UK) equipped with a 5 mW HeNe laser (wavelength λ = 632.8 nm) and a digital logarithmic correlator. To obtain the distribution of the particle diffusion coefficient D, the normalized intensity autocorrelation function was analyzed by the CONTIN method. The Stokes-Einstein equation (R H =K B T / 6πηD) to define D as the effective hydrodynamic radius R H Here, K B T is the thermal energy and η is the solution viscosity. The electrophoretic mobility, u, of the samples was measured using laser Doppler electrophoresis. The zeta potential was calculated using the Smoluchowski relation (zeta potential = uη / ε), where η and ε are the viscosity and dielectric constant of the solvent, respectively. The size and zeta potential of the liposome-HP complexes are given as the mean ± standard deviation (SD) of five individual measurements.

[0199] Protein analysis The liposome preparations were incubated with HP (1:1 v / v) for 1 h at 37°C. The liposome-HP complexes were then pelleted at 15,000 × g for 15 min at 4°C and washed three times with PBS. The washed pellets were resuspended in 8 mol / L urea, NH 4 CO 3 The samples were resuspended in 50 mmol / L. Ten microliters of each sample was added to five wells of a 96-well plate. Protein quantification was performed by adding 150 microliters / well of Protein Assay Reagent (Pierce®, Thermo Scientific, Waltham, MA, USA). The multiwells were shaken to allow for rapid centrifugation. The plates were incubated at room temperature for 5 min. Absorbance was measured at 660 nm using the GloMax® detection system (Promega, Madison, WY, USA). Protein concentrations were calculated using a standard curve with appropriate background correction. Results are given as the mean ± SD of five individual replicates.

[0200] Protein identification and quantification. The incubation was performed as described in another publication (Label-free quantitative analysis for studying the interactions between nanoparticles and plasma proteins. Analytical and Bioanalytical Chemistry, 2013, 405, 2-3, 635-645, and the contents of the publication). The procedure was as described previously (Shotgun proteomic analytical approach for studying proteins adsorbed onto liposome surface. Analytical and Bioanalytical Chemistry, 2013, 401, 4, 1195-1202, the entire contents of which are incorporated herein by reference). The liposome formulations were incubated with HP (1:1 v / v) for 1 hour at 37°C. The samples were centrifuged at 15000 x g for 15 minutes to pellet the liposome-HP complexes. The pellets were washed three times with 10 mmol / L Tris-HCl (pH 7.4), 150 mmol / L NaCl, and 1 mmol / L EDTA. After washing, the pellets were air-dried and resuspended in digestion buffer. Digestion and peptide desalting were performed as previously described (Shotgun proteomic analytical approach for studying proteins adsorbed onto liposome surface. Analytical and Bioanalytical Chemistry, 2013, 401, 4, 1195-1202, the entire contents of which are incorporated herein by reference). Briefly, the pellets were washed with 40 mL of 100 mL of 10% ethanol and 10% ethanol. 8 mol / l urea and NH 4 CO 3The peptides were resuspended in 50 mmol / l and digested by adding 2 micrograms of trypsin. The digested peptides were desalted using an SPE C18 column, reconstituted with an appropriate volume of 0.1% formic acid solution, and stored at -80°C until analysis. The digested peptides were analyzed by nano-high performance liquid chromatography (HPLC) coupled to a tandem mass spectrometer (MS / MS). Nano-HPLC MS / MS analysis was performed using a Dionex Ultimate® 3000 (Dionex, Sunnyvale, CA, USA) directly connected to a hybrid linear ion trap and orbitrap mass spectrometer (Orbitrap LTQ-XL®, Thermo Scientific, Bremen, Germany) by a nano-electrospray ion source. The peptide mixture was concentrated on a 300 μm ID × 5 mm Acclaim PepMap® 100 C18 precolumn (Dionex, Sunnyvale, CA, USA) with a premixed mobile phase consisting of ddH2O / ACN, 98 / 2 (v / v) with 0.1% (v / v) HCOOH at a flow rate of 10 microliters / min. The peptide mixture was then separated by reversed-phase (RP) chromatography. A 3-h optimized LC gradient consisting of mobile phase A of ddH2O / HCOOH (99.9 / 0.1, v / v) and mobile phase B of ACN / HCOOH (99.9 / 0.1, v / v) was used to detect the largest set of peptides. MS spectra of the eluting peptides were collected over the m / z range of 350-1700 using a resolution setting of 60,000 (full width at half maximum at m / z 400) in data-dependent mode of operation. MS / MS spectra were collected for the five most abundant ions in each MS scan. Further details can be found in another publication (Shotgun proteomic analytical approach for studying proteins adsorbed onto liposome surface. Analytical and Bioanalytical Chemistry, 2013, 401, 4, 1195-1202).Three individual samples (biological replicates) were run for each experimental condition. Biological replicates were prepared, each of which was measured in triplicate ( Nine measurements were obtained for each experimental condition (technical replicates). Raw data files were submitted to Mascot (v2.3, Matrix Science, London, UK) using the Thermo-Finningan LCQ / DECA raw file data import filter to perform a database search against the non-redundant Swiss-Prot database (09-2014, 546000 sequences, Homo sapiens taxonomic limit). For the database search, trypsin was identified as the protease, with a maximum of two missed cleavage. Carbamidomethylation was set as the fixed modification of cysteines. In addition, oxidation of methionine was selected as a variable modification. The monoisotopic mass tolerance for precursor and fragmentation ions was set at 10 ppm and 0.8 Da, respectively. Charge states of +2 or +3 were selected as precursor ions. The output files of the proteomic analysis were sent to the commercial software Scaffold (v3.6, Proteome Software, Portland, OR, USA). Peptide identifications were considered valid when they exceeded a probability threshold of 95%, set by the Peptide Prophet algorithm. Protein identifications were accepted when they could be made with a probability of greater than 99.0% and contained at least two unique peptides. Proteins that contained shared peptides and could not be distinguished by MS / MS analysis alone were grouped to satisfy Occam's razor principle. Unweighted spectral counts (USC) were used to assess the consistency of biological replicates in the quantitative analysis, and normalized spectral counts (NCS) were used to collect protein content.

[0201] statistical analysis All statistical analyses were performed using PLS, Kohonen and CPANN toolboxes and graphs were generated using Microsoft Excel®, XLSTAT® and MATLAB®.

[0202] Data matrix Each row of the prediction matrix (X) relating to an individual was derived from the total protein content obtained from the three-protein corona sensor array (Figure 4A). In a pre-processing step, the normalized data in matrix X, i.e., the relative protein abundance (RPA), was autoscaled.

[0203] Classification and Clustering Partial Least Squares Discriminant Analysis (PLS-DA) Partial least squares discriminant analysis is a well-known multivariate approach that can be considered as a linear analysis and dimensionality reduction method consisting of two principal and structural parts, which searches for latent variables as linear combinations of the original independent variables (i.e., data matrix X) that have the maximum covariation with the corresponding dependent variable (i.e., class membership, Y). The measured components include latent variables such as scores and loadings, which indicate how the latent variables and their original variables are related. Based on the ability of PLSDA to reduce the dimensionality of the data, linear mapping of different data patterns and visualization as graphs can be performed. PLS-DA is particularly suitable when dealing with fairly collinear and noisy patterns. The main problem associated with large theta sets in proteomics is the large number of variables (i.e., proteins) to monitor and the relatively small number of samples. Hence, there may be a high redundancy in the variables, making them uninformative and irrelevant for the classification. In this way, the predictive performance of the classification can be improved by eliminating uninformative variables or finding new uncorrelated ones. In biological applications such as this study, where we not only have to decide whether a sample belongs to one of many known groups, but also which variables are most relevant in best discriminating between classes, methods such as PLS-DA are good candidate approaches to find new uncorrelated latent variables while preserving the variation in the data. The important powers of the original variables for generating latent predictions can also be calculated by variable importance in projection (VIP) analysis and play an important role in making decisions about the variables.

[0204] Identification of optimal variables based on weighted VIP. Partial least squares discriminant analysis (PLS-DA) was used to find the VIP value for the variables. VIP is a joint measure of how well a variable contributes to describing two sets of data, dependent (Y) and independent variables (X). Weights in a PLS model reflect the covariance between the independent and dependent variables, and by including weights, VIP can not only reflect how well the dependent variable is described, but also its It can also reflect how important the information is to the model of the independent variables.

[0205] To identify the best subset of variables, a VIP score based approach was deployed. The VIP score can be calculated by running PLS-DA on the dataset. In this approach, the VIP scores of variables were calculated 50 times, with random permutations of the training and validation sets used each time (random training sets are selected in each iteration by considering 80 percent coverage of each class of interest). Considering the most important variables, those with large VIP score values ​​(>2), the top 200 variables can be selected in each iteration and added to the top variable pool. Then, the frequency of occurrence (Freq) for each variable is calculated. i ) and average VIP score

number

number

number

number

[0206] Counterpropagation Artificial Neural Network (CPANN) Counterpropagation artificial neural network (CPANN) is a supervised variable self-organizing map consisting of two layers of neurons arranged on a prescribed N × N grid. CPANN can be used to map data from a high-dimensional feature space of neurons to a low-dimensional (usually 2) discrete space and to predict the class membership of unknown samples. Input vectors (sample feature vectors) and corresponding class membership vectors (binary vectors) were presented to the input and output layers of CPANN, respectively. The weighting of neurons in both layers is based on competitive learning and neuronal cooperation (see Table 3). Thus, similar input vectors can be mapped onto the same or adjacent neurons and vice versa. The final assignment map properly reveals the structure of the data in the feature space and preserves the distance of the patterns in the low-dimensional grid of neurons. Figure 9C shows a high-quality assignment map of CPANN with top-ranked biomarkers. The distinct regions for each class reduce the classification error. The risk is minimized. The appropriate size of the map can be determined by performing 10-fold cross-validation at different map sizes. The trained CPANN can be used to assign class membership to unlabeled samples. The presence of redundant and uninformative variables in the training data affects the quality of the map and increases the risk of classification errors (Figure 5C). The process is a nonlinear mapping that helps to visualize high-dimensional input objects on a two-dimensional neuronal grid. It is a self-organizing procedure and solves the classification problem in an unambiguous way. More information on the CPANN method can be found in the following references:

[0207] Hierarchical Cluster Analysis (HCA) Hierarchical cluster analysis is a widely used unsupervised method to explore and visualize heterogeneous and large datasets into distinct homogeneous clusters, such as those frequently used in proteomics. Hierarchical cluster strategies can be classified into two categories: agglomerative and divisive. Agglomerative methods first separate each object into its own individual cluster, then combine the clusters in turn, and similar objects or clusters are merged until all objects belong to only one cluster. Meanwhile, divisive methods start with all objects in one large cluster, and partition these into progressively smaller clusters until each object belongs to an individual cluster. Finally, the objects are organized into a dendrogram with defined clusters on the branches. In cluster analysis, two important concepts should be defined to identify homogeneous subgroups: similarity (determine a numerical value for the similarity between objects and construct a similarity matrix) and linkage (connect or not connect objects into groups). Here, we apply agglomerative hierarchical clustering together with a longest distance linkage algorithm for unsupervised analysis based on selected variables.

[0208] Cohort Sample Prediction The predictive performance of linear and non-linear variables with 69 selected predictors was evaluated by cohort sample analysis. To achieve this, values related to the 69 variables were input into the model, and the class membership of each cohort subject was predicted at fixed optimal parameter values. Nineteen variables (proteins) within the proteome profile of our protein corona sensor array for the cohort samples were not detected and were considered zero for these variables in the validation data matrix. Interestingly, the linear and non-linear models all gave good predictions for all five samples. The distances between the training and cohort samples shown in Figure 6A are probably related to the zero values added to the data matrix for these 19 absent variables. We examined the effect of the zero values substituted for the absent variables by removing these 19 proteins from both data matrices of the training and prediction sets, and then constructed a PLS-DA model to predict the cohort samples. As expected, the long distance between the cohort and training samples was not observed.

[0209]

Table 3

[0210] For each variable, the correlation coefficient of the corresponding weight map with the assignment map pattern can be calculated. CC = 0: Indicates no correlation between the biomarker and the associated class. 1 > CC(i) > 0: Agreement between the biomarker intensity and the cancer-associated class. 0 < CC < -1: Inverse correlation between the biomarker value and the cancer-associated class. Also, CC values > 0.5 or < -0.5 were colored. For example, biomarker 128 The weight map of 2 has a fairly high correlation with the 4 patterns of the cancer class on the assignment map and could be an important biomarker for samples from patients with myeloma.

[0211] Example 1B: Multi-nanoparticle protein corona characterization and deep analysis of the human proteome using machine learning enables accurate identification and differentiation of early stage cancers.

[0212] In a second embodiment, the collective protein corona data for a given plasma sample, derived from the protein corona profiles of the individual nanoparticles, can identify and discriminate between different types of cancer by using machine learning (e.g., random forest approach) to analyze the data. The collected data was analyzed in more detail by machine learning using the sensor array described in Example 1A. This sensor array (i.e., protein corona nanosystem) unambiguously and stably identifies cancer and allows discrimination between different types of cancer. This system can be used to predict the type of cancer using blind plasma samples. Ability for ultra-early detection by using plasma of healthy individuals in a survival cohort who were diagnosed with cancer several years after plasma collection to determine pre-cancer biomolecular fingerprints.

[0213] FIG. 1 depicts an overview of the use of the sensor array (nanoparticle system) for using multi-protein corona proteomics using both blinded plasma and cohort samples for cancer detection in known cancer patients and for cancer prediction.

[0214] [result] <Protein corona profiling of nanosystems using plasma from patients with early, intermediate and advanced stages of cancer>

[0215] Our protein corona nanosystem was composed of three different cross-reactive liposomes with different surface charges [i.e., cationic (DOPG (1,2-dioleoyl-sn-glycero-3-phospho-(1'-rac-glycerol))), anionic (DOTAP (1,2-dioleoyl-3-trimethylammonium-propane)-DOPE (dioleoylphosphatidylethanolamine))) and intermediate (CHOL (DOPC-cholesterol))] and their protein corona profiles were measured after exposure to plasma from healthy patients or patients with one of five cancers, namely lung cancer, pancreatic cancer, myeloma, meningioma and glioblastoma.

[0216] For each patient (5 patients / group), we performed proteomic analysis in triplicate for each of the three liposomes (see Figures 2A and 2B for details of liposome size and charge) using our protein corona nanosystem (i.e., total number of studies: 3*(29)*3=261). Although there was no single protein corona composition specific for any cancer type as described in Example 1A, the collective protein corona composition for a given plasma sample obtained from different liposomes provided a unique "fingerprint" for each cancer type.

[0217] The composition of the protein corona formed on the surface of liposomes depends largely on their physicochemical properties, but can also be greatly influenced by the unique types, concentrations, and conformations of proteins and other biomolecules present in a given patient's plasma. The size and charge of corona-coated liposomes were investigated using dynamic light scattering (DLS / Nanosight™) after incubation with plasma from patients with five types of cancer (see Table 1) and healthy individuals. The results confirmed that the physicochemical properties of corona-coated nanoparticles differed for the different types of cancer (Figures 2A, 2B).

[0218] Quantitative assessment of the total proteins adsorbed on the liposomes was performed using BCA (bicinchoninic acid) or NanoOrange® assays, and the results confirmed that the amount of adsorbed proteins was significantly different after incubation in plasma from patients with various cancers (Figures 3A1-3F3). Quantitative assessment of the total proteins adsorbed on the surface of liposomes showed the dependence of the protein amount on the type of cancer (Figures 3A1-3F3). The protein corona composition on the surface of the three liposomes was evaluated by liquid chromatography-tandem mass spectrometry (LC-MS / MS), revealing 876 known proteins. The contribution of individual proteins and their categories (i.e., complement, coagulation, tissue leakage, lipoproteins, acute phase, immunoglobulins and other plasma proteins) to the corona composition was revealed (Figures 3A1-3F3). The results showed an association between the protein composition and the type of cancer, as well as between the protein composition and the type of liposome. The mechanism behind this variation is that the high surface-to-volume ratio of nanoparticles provides a unique opportunity for a wide range of human plasma proteins (high abundance proteins, low abundance plasma proteins, and extremely rare proteins) to participate in the corona composition, without a direct correlation with the plasma protein concentration. Moreover, in our system, conformational changes in plasma proteins can also alter the protein corona composition via significant modification of the protein-nanoparticle interaction sites. These characteristics make the protein corona profile unique among other approaches developed for the analysis of proteins in human plasma.

[0219] The protein corona composition at the surface of liposomes contains a wide range of human plasma proteins, including high abundance proteins (e.g., albumin, transferrin, complement proteins, apolipoproteins, and α-2-macroglobulin) and extremely rare proteins (defined as <100ng / ml), such as transforming growth factor beta-1-induced transcript 1 protein (~10ng / mg), fructose bisphosphate aldolase A (~20ng / ml), thioredoxin (~18ng / ml), and L-selectin (~92ng / ml). We also identified 388 proteins without known previous plasma concentrations (http: / / plasmaproteomedatabase.org / ). The obtained protein information was then analyzed by machine learning approaches to investigate the capability of our protein corona nanosystem for stable and accurate cancer detection.

[0220] <Development of a classifier for cancer detection and discrimination using results obtained from protein corona nanosystems> To evaluate the cancer-detecting capabilities of the various protein corona nanosystems, we trained a classifier using proteomic data for 29 plasma samples (five patients for each of five cancer types; and four healthy samples) collected from three distinct liposomes. In particular, we trained an algorithm that takes array measurements from patients and outputs one of six labels (one of the five cancer types or healthy). Before training the classifier, the raw data from the three nanoparticles was first “denoised” via low-rank tensor decomposition, described in a later section. This denoising implicitly reduces the large variability observed in the individual corona elements. We then trained a random forest classifier, a well-known nonlinear classification algorithm, on the resulting denoised data.

[0221] We tested the accuracy of this classifier on 16 blind samples (3 patients for each of the 5 cancer types; and 1 healthy sample). We measured the overall classification accuracy for correctly assigning these blind samples to one of the 6 labels, along with the sensitivity and specificity separately for each of the 5 cancer types (Table 4). Due to the relatively small number of plasma samples (45 in total), and to ensure the robustness of our results, we performed a randomized multivariate analysis of this procedure (29 samples). We trained a random forest classifier and performed a total of 1000 iterations (measurements of accuracy, sensitivity, and specificity) on the remaining 16 samples. The training and test sets in each of the 1000 iterations were randomly selected from among all the class stratified data. This approach allows us to calculate unbiased estimates of the p-values ​​for the classification accuracy that we report (see Table 4). In particular, the last column of Table 4 shows that the overall accuracy was 96.2% for the 1000 iterations, or equivalently, the overall error was 3.8%. We also observed that the p-value was 0.04 for the null hypothesis that the overall classification error was less than 87.5%, and therefore we can reject this null hypothesis with 95% confidence. Sensitivity for the five individual cancer types ranged from 87.4% to 100.0%, and individual specificities ranged from 97.9% to 100.0%.

[0222] Table 4 shows the overall classification accuracy, sensitivity and specificity. The overall classification accuracy (column 2) for protein corona nanosystems with 1, 2 and 3 liposomes is shown. Both the classification accuracy and the associated p-value improved with the addition of liposomes. The individual sensitivity and specificity for glioblastoma, lung cancer, meningioma, myeloma and pancreatic cancer (columns 3-12) also showed that the sensitivity and specificity improved with the addition of liposomes. The experimental results were averaged over 1000 individual extractions from a training set of 29 plasmas and evaluated on the remaining 16 plasmas. The p-values ​​are for the null hypothesis that the classification error is less than 87.5%.

[0223] [Table 4]

[0224] <The value of multi-liposomes in protein corona nanosystems> To evaluate the importance of including multiple liposomes in the protein corona nanosystem, we repeated the entire classification procedure using 1000 partitions of the data for training and testing samples, this time using data measured from only one liposome. This was done for each of the three liposomes, and the results are shown in the first row of Table 4. Compared to the complete array containing all three liposomes, the single liposome array showed significantly lower accuracy, sensitivity, and specificity. We also performed this procedure again for the set of two liposomes, and the results are shown in the second row of Table 4. The two liposome system was more accurate than the one liposome system, but less accurate than the complete array of three liposomes. Overall, this demonstrated the value of the variation in corona composition patterns between different liposomes and the necessity of including multiple such observations.

[0225] <Variables and Protein Importance and Stability> The Random Forest model also produces an importance score for each variable (i.e., each liposome-protein pair). This score primarily measures how important that variable is in discriminating between patients with different types of cancer. For each "tree" in the Random Forest, the importance score of every variable used to construct the tree is calculated based on the importance of that variable. The importance score for a variable is defined as the proportion of the training set that belongs to the "leaf" nodes that use the tree weighting (variables not used to construct a tree are assigned a score of zero), and thus the overall importance score for a variable is the average of its importance scores in each tree.

[0226] For different biological families of proteins (same as those used in Figures 3A1-3F3), we calculated the overall importance in discriminating different cancer types (Figure 10A, (a)-(c)). The results suggested that different families of proteins are important in the detection of different cancers. For example, acute phase proteins were relatively important in the detection of meningiomas, and lipoproteins were relatively important in the detection of glioblastomas. Notably, these variations were different from the variability in the proportion of each category adsorbed on the liposomes (Figure 10A, (d)-(f)). Interestingly, these results were in good agreement with the biological functions of these protein categories. For example, it is generally accepted that lipid metabolism is highly altered in glioblastomas compared to healthy tissues, which may be the main reason for the large changes in the observed interactions of lipoproteins with liposomes (Figure 2C and Figure 10A). The number of identified and unique proteins in the corona composition of different liposomes as well as their combinations are shown in Figure 10B.

[0227] We also calculated the total most important proteins (Figure 10C and Table 5). These proteins were detected in the combination of all three liposomes, again indicating the important role of multiliposomes in the nanosystem. We also evaluated the robustness of this set of "important" proteins for the classifiers estimated on the training data divided into 1000 folds mentioned above. In particular, Figure 10B shows the 25th to 75th percentiles of importance scores for the 30 most important proteins on average. These indicate that the set of important proteins is stable to the folds of the data used to train the model. Among these most important proteins (see Table 5), some have been recognized to play important roles in cancer development. For example, ficolins (both ficolin 2 and ficolin 3) are serum pattern recognition molecules with opsonic properties that have a significant ability to regulate complement activation. Serum concentrations of ficolin 3 have been shown to be higher in ovarian cancer patients than in healthy subjects. Furthermore, ficolin 3 was identified in a stratified proteomic analysis of prostate cancer serum, suggesting that this protein also has a role in prostate cancer. On the other hand, it has been demonstrated that apolipoprotein A2 and its isoforms are overexpressed in prostate cancer serum and that the concentrations of acute phase proteins (e.g., complement proteins) can change by ≥25% in the presence of inflammatory diseases such as cancer. A clear link between cancer and the hemostatic system has long been well documented. Hemostasis regulates blood flow by regulating platelet adhesion and fibrin deposition. Several proteins involved in hemostasis have been linked to the regulation of angiogenesis. Among these, fibrinogen is a major protein in the hemostatic process and is found in many tumors, where it regulates angiogenesis and tumor growth and has been linked to metastasis formation. Indeed, plasma levels of fibrinogen have been used to predict clinical outcomes in patients with nonmetastatic renal cell carcinoma and to predict distant metastasis in pancreatic cancer.

[0228] Table 5 provides information regarding the total most significant proteins found to be detected in all three liposome combinations. * Concentrations were obtained using spectral counts from the Plasma Proteome Database (www.plasmaproteomedatabase.org).

[0229] [Table 5] TIFF2024170513000013.tif58170

[0230] To further explore the role of the key proteins identified by our machine learning approach, we searched them in the Open Targets database, a platform for the identification and validation of therapeutic targets. This database calculates a disease association score for each protein based on evidence from various other databases (including the GWAS catalog, UniProt, Gene2Phenotype, Cancer Gene Census, IntOGen, Europe PMC, and Reactome) to derive a score on a scale of 0 (lowest) to 1.0 (highest) for disease association. Of the proteins listed, three have strong associations with cancer. Hyaluronan binding protein has a very strong general association with cancer (1.0), fibrinogen β chain has a moderately strong association with lung cancer (0.4), while keratin, type 1 cytoskeleton 14 has a strong association with prostate cancer (0.72). Nearly all of the other proteins in Table 5 have some weak associations with various cancers (0.05-0.4). Thus, collectively, the protein has been linked to known cancer associations.

[0231] Eliminating variability in single measurements using tensor decomposition We found that due to the large variability in the patient population and the noise caused by measurement errors, any one protein was insufficient for classification. For each of the 100 most abundant proteins (resulting in 3*100=300 variables), we calculated the mean absolute Z-score of that protein concentration in the corona observed for patients with each type of cancer. A higher Z-score for a given protein for a type of cancer indicates that this protein is useful for detecting this type of cancer. Figure 11 (blue bars) shows a histogram of these mean absolute Z-scores for each type of cancer and the healthy group. We also measured the mean absolute Z-scores for proteins previously linked to these specific types of cancer, and these are shown in the same histogram in Figure 10 (light gray bars). Of these hundreds of variables, only one has an absolute Z-score above 2.0, and only one type of cancer (myeloma), suggesting that no single protein is sufficient for accurate classification.

[0232] The tensor decomposition that precedes the construction of our random forest effectively serves to calculate a weighted average of a small number of observed corona protein compositions for a given sample. This average is over all proteins and nanoparticles in a given plasma sample, and serves to reduce the variability in the observed concentration of any given protein. For example, in Figure 11 (long black bars), the mean absolute Z-score for one of these weighted averages is plotted for each cancer type. The absolute Z-scores of these weighted averages are significantly higher than the absolute Z-scores of any one variable. This shows how our approach reduces the variability of a protein across samples with the same indication. In addition to the intuition that it removes noise, this denoising has a significant impact on our classification ability. Without denoising, simply training a Random Forest classifier on the raw corona data results in a classification accuracy of 94% (p-value 0.07), which is lower than the 96.2% (p-value 0.04) reported in Table 4. For the cohort data reported in Table 7, denoising allowed us to increase the classification accuracy from 92% (p-value 0.04) to 94.1% (p-value 0.01).

[0233] <Data dependency of classification accuracy> Our final assessment of the 45 non-cohort patients was the value of the data. Our original classifier was trained on five samples for each cancer type. To gauge the value of including more or less training data, we repeated the classification procedure (splitting the data into training and test samples, training the classifier on the training samples, measuring performance on the test samples, all done 1000 times) on different numbers of training samples per class, from four to six (Table 6). Classification accuracy, sensitivity, and specificity all increased with increasing number of training samples. With six training samples per class, the overall accuracy reached 96.8% (i.e., the overall error rate was 3.2%). This strongly suggests that accuracy, sensitivity, and specificity will continue to increase by including more data in training the classifier.

[0234] Table 6 shows that classification accuracy improves with more data. Column 2 shows the total classification error and associated p-value when the training set consisted of 4, 5 and 6 samples from each cancer. Both the classification error and the associated p-value improved with the addition of samples from each cancer. Columns 3-12 show the sensitivity and specificity for glioblastoma, lung cancer, meningioma, myeloma and pancreatic cancer when the training set consisted of 4-6 samples. Experimental results were averaged over 1000 separate extractions of training sets for 4 healthy patients and 4-6 patients with each type of cancer. The p-values ​​are for the null hypothesis of at least 2 classification errors.

[0235] [Table 6]

[0236] <Sampling of low- and high-abundance plasma proteins using liposomes without the need for protein removal> The multi-liposome protein corona nanosystem uses machine learning methods to generate unique "fingerprint" protein patterns for each type of cancer and for healthy individuals. The mechanisms underlying the unique ability of the protein corona in identifying and differentiating cancers are In pursuit of this algorithm, we thoroughly analyzed the corona composition for the contributions of both high- and low-abundance proteins and compared these results with the concentrations of corona proteins in human plasma. In this regard, the corona composition and contribution of each protein (e.g., protein X) to plasma were normalized to albumin using (total albumin peptides in protein corona) / (total protein X peptides in protein corona) and (albumin concentration in plasma) / (protein X concentration in plasma), respectively. It should be noted that we manually surveyed the concentrations of identified corona proteins in plasma using an online proteome database (http: / / plasmaproteomedatabase.org / ). Since the total albumin peptides in the protein corona are proportional to the total amount of albumin in the protein corona, dividing these peptides by protein X peptides corresponds to the albumin to protein X concentration ratio in plasma. As shown in Figure 12, we found that plasma concentrations varied over 10^10 (on a double logarithmic scale) while liposomes detected these same proteins over 10^4-10^5. In other words, we demonstrated that the protein corona composition has a good ability to assemble a wide range of low abundance proteins (≦100 ng / mg) and very rare proteins (≦10 ng / mg). This means that obtaining data on high abundance proteins in the protein corona does not preclude the detection of peptides derived from low abundance and very rare proteins. It is noteworthy that many of the detected proteins are low abundance, rare or unknown / unreported proteins. The involvement of low abundance proteins in the corona composition is mainly due to the exchange of low affinity corona proteins with high affinity proteins and slower adsorption kinetics.

[0237] Detection of low abundance proteins in human plasma requires depletion of high abundance proteins and post-depletion plasma fractionation strategies, but even with these depletion strategies, plasma proteomics has not been consistently successful in early cancer detection. Results obtained with our system show that, in contrast to human plasma proteins, the protein corona composition contains a wide range of low abundance and extremely rare plasma proteins without the need for depletion (which may result in the unintended removal of low abundance proteins / biomarkers). In addition, we found several proteins in the protein corona whose concentrations in human plasma are unknown / unreported due to their extremely low concentration in human plasma. More specifically, cationic, neutral and anionic liposomes consist of 323, 189 and 155 proteins with unknown / unreported plasma concentrations, respectively. The contribution of these proteins to the protein corona is shown in the square boxes on the right side of Figure 12.

[0238] The use of multiple liposomes (with different surface properties) offered a unique opportunity to increase the detection depth of low abundance and extremely rare proteins (see Figure 12), thereby significantly improving the sensitivity, specificity and predictive accuracy of the protein corona nanosystem. Each liposome provided a different pattern of contributed proteins that was highly dependent on the cancer type.

[0239] <Cancer detection and discrimination in cohort samples> Finally, to investigate the ability of the protein corona nanosystem to detect very early cancers, we used cohort plasma (obtained from the NIH-funded Golestan Cohort Study, details of which are given in the Methods section) from healthy individuals who had been diagnosed with pancreatic, lung and brain cancer several years after the plasma collection. We performed 1000 experiments following the same procedure. In each experiment, we split the data into 12 training samples (4 for each of the three cancer types) and 3 test samples (1 for each of the three cancer types). The observations were denoised as described above and then a random forest classifier was trained on the training samples. Finally, the overall accuracy, sensitivity and specificity were measured on the test samples (Table 7). The overall accuracy was 94.1% (error rate was 0.01). The sensitivity ranged from 83.2% to 100.0%, and the specificity ranged from 91.6% to 100.0% for the three cancer types. These relatively high values ​​suggest that the protein corona array is well suited to detecting cancer at its earliest stages.

[0240] Table 7 shows the overall classification accuracy, sensitivity and specificity for the cohort samples. Column 2 shows the overall classification accuracy for 1, 2 and 3 liposomes. Both the classification error and the associated p-value improved with the addition of liposomes to the nanosystem. Columns 3-8 show the sensitivity and specificity for brain, lung and pancreatic cancer, respectively. Again, sensitivity and specificity improved with the addition of liposomes. Experimental results were averaged over 1000 individual extractions from a training set containing 12 patients and evaluated on the remaining 3 patients. p-values ​​are for the null hypothesis that the classification accuracy is <66.7%.

[0241] [Table 7]

[0242] We also performed the same experiment with smaller subsets of liposomes containing each liposome individually, and with all three unique sets of two liposomes. Again, the overall precision, sensitivity and specificity improved significantly with increasing numbers of liposomes, again demonstrating the value of the array system.

[0243] [Consideration] Previous studies using other approaches for protein corona or plasma proteome fractionation have not allowed for simultaneous multiple cancer screening with acceptable specificity, sensitivity and predictive accuracy. This is the first time that a sensor array has been developed with specificity, sensitivity and predictive accuracy not only for cancer but also for specific cancer subtypes.

[0244] We have shown that fractionation of the plasma proteome by the protein corona is independent of the abundance of a particular plasma protein, but rather through a number of factors including protein affinity to the nanoparticle surface through a wide range of forces including Coulombic forces, London dispersion forces, hydrogen bonding acidity and basicity, polarizability, lone pairs, and protein-protein interactions between proteins involved in the corona structure. The protein corona composition is dynamic and is initially dominated by abundant proteins including albumin, immunoglobulins, and fibrinogen, which together with 19 other proteins constitute more than 99% of the protein mass in the plasma proteome. The remaining 1% of the plasma proteome is composed of more than 10,000 proteins, and a subset of these proteins with higher affinity and / or lower adsorption kinetics for the nanoparticle surface compete with the high abundance proteins for inclusion in the corona composition. In addition to proteins with higher binding affinity being exchanged for proteins with lower binding affinity at the nanoparticle surface, other low affinity proteins in the outer protein corona layer are favorably protein-adsorbed with the already formed protein layer. There may also be a contribution from these other low affinity proteins due to protein interactions, meaning that the exchanged low abundance proteins may be able to direct the formation of the protein corona towards the adsorption of the lower abundance proteins.

[0245] In this disclosure, we demonstrate for the first time the ability to enrich low-abundance proteins (<100n The relative concentration capacity of the protein corona (defined as 10 ng / ml) was analyzed, which showed that many of these proteins were present at concentrations <10 ng / ml. We show that the protein corona is a 10-fold increase in abundance, with concentrations approaching 1 pg / ml. More importantly, we show that many of these proteins play a key role in cancer identification and discrimination via machine learning approaches. The role of the protein corona in assembling a wide range of low-abundance and extremely rare proteins goes a long way to addressing a major cause behind the limited success of current mass spectrometry methods (including LC-MS / MS) in early cancer detection, as they can only detect proteins in a dynamic range of 10^4-10^6. In other words, the protein corona composition allows sampling of the vast dynamic range of the plasma proteome, thereby greatly increasing protein coverage without protein depletion. Since the protein corona obtained from a single nanoparticle consists of at most hundreds of different proteins (a small subset of the entire proteome), we hypothesized that using multi-nanoparticles with different physicochemical properties would provide additional dimensions of proteomic information. These aspects include: 1) each additional nanoparticle with different physicochemical properties may be able to recruit additional unique low abundance proteins (Figure 12); 2) overlapping corona proteins on two or more nanoparticle surfaces may participate with different corona contributions, thus altering the concentration and identity of other participating corona proteins; and 3) the unique and overlapping corona protein information from each nanoparticle acts as a unique variable, thus providing more data for our machine learning approach. Combining more nanoparticles for plasma fractionation and proteomic analysis provides more information with superior sensitivity, specificity and predictive accuracy compared to fewer nanoparticles for cancer detection and discrimination (Table 4).Conceptually, our multi-nanoparticle approach is similar to the olfactory system, where the specificity of odor recognition comes from the response patterns from hundreds of highly cross-reactive odorant receptors, where no single receptor provides complete information, but in combination is highly specific for the identification of a given odorant (i.e., ~400 active receptors in humans can detect and distinguish ~10,000 odorants, and ~1200 active receptors in dogs can detect ~1,000,000 odorants). Similar approaches have been used by others to detect and distinguish among diverse families of analytes, various foods and beverages, pathogens and fungi, biomolecules, and nanoparticles themselves.

[0246] Three different cross-reactive liposomes (with negative, neutral and positive surface charges) were used and their protein corona profiles were measured after contacting with plasma from individual patients with one of five cancers: lung cancer, pancreatic cancer, myeloma, meningioma or glioblastoma. To identify and discriminate cancers using the results of the protein corona nanosystem, we used a well-defined random forest machine learning approach. To ensure that our protein corona nanosystem stably and accurately detects cancers with good predictive accuracy, we designated 1000 different sample sets as training samples (i.e. plasma with known cancer and healthy plasma) or testing samples (i.e. blind plasma). Although the protein corona composition from any one nanoparticle was not specific for any particular cancer type with acceptable predictive accuracy (i.e., 86.0%, p-value 0.43 for classification error less than 87.5%), we found that the pattern of corona composition obtained from multiliposomes provided a unique “fingerprint” for each cancer type with excellent predictive accuracy (i.e., 96.2%, p-value 0.04 for classification error less than 87.5%). These results, based on deep analysis of the human proteome using genomic protein corona characterization and machine learning, confirm the promise of this system for unambiguous identification and discrimination of cancer as well as error-free discrimination from healthy subjects with excellent specificity (97.0%-100.0%).

[0247] To investigate the ability of the protein corona nanosystem in the ultra-early detection of cancer, cohort plasma samples were used. These samples were collected from healthy individuals who were diagnosed with lung, pancreatic or brain cancer 8 years after the plasma collection. The results of the protein corona nanosystem with plasma from 15 patients in the cohort study revealed that our approach accurately identified and differentiated cancers even in these pre-diagnostic cohort samples where cancer detection was not possible using current alternatives. Consistent with our findings for fresh plasma samples, multi-liposome plasma sampling provided significantly better classification accuracy (94.1% vs. 75.4%) and specificity [i.e., brain cancer (100.0% vs. 86.1%), lung cancer (96.6% vs. 90.0%) and pancreatic cancer (91.6% vs. 86.9%)] compared to single liposome sampling. Notably, the protein corona profiles of the cohort samples are distinct compared to previous fresh cancer samples. This is mainly due to the fact that the cohort samples were collected at the time of screening of healthy individuals and therefore stored frozen for a long period of time (approximately 10 years). It is increasingly recognized that long-term storage of plasma significantly affects the concentration and integrity of a subset of proteins, which alters the protein corona composition and reduces the sensitivity of our protein corona nanosystem. Thus, the maximum sensitivity of the nanosystem may be achieved when fresh plasma is used as part of cancer screening or post-diagnosis cancer workup. The latter setting is valuable in that changes in the protein corona pattern over time may provide useful information related to early tumor recurrence or the presence of residual disease after tumor removal.

[0248] In summary, we have presented a proof of concept that the multi-nanoparticle protein corona nanosystem has the potential to increase the depth of protein detection and therefore the unique ability to accurately detect and differentiate cancer even at its earliest stages when existing techniques fail to detect the disease. The multi-nanoparticle protein corona pattern obtained from our protein corona nanosystem provides a unique and multivariate "fingerprint" for cancer detection that is not possible by analyzing the protein corona of only one nanoparticle. Furthermore, the protein corona pattern represents the collective enrichment of a wide range of plasma proteins (including both high-abundance and rare proteins) using different nanoparticles, which differentiates it from other multivariate whole plasma proteomic approaches that have failed to produce suitable results for early cancer detection. The depth of protein detection (the main limitation of protein analysis in human plasma) and the predictive accuracy of the nanosystem are further improved by using additional nanoparticles with different physicochemical properties. The favorable predictive results of our machine learning and protein corona characterization approach on both blind fresh plasma and retrospective cohort plasma provide a suitable substrate for subsequent prospective cancer studies. Furthermore, the multi-nanoparticle protein corona nanosystem can be utilized for other important human diseases where early detection can significantly extend life span and improve quality of life.

[0249] Liposomes were prepared as described in Example 1A.

[0250] Human plasma collection, preparation and storage were performed as described in Example 1A.

[0251] Cohort plasma samples were prepared as described in Example 1A.

[0252] Transmission Electron Microscopy (TEM). Liposome formulations were characterized by TEM as previously reported. Briefly, 10 μl of each sample was loaded onto formvar-coated grids negatively stained with 1% uranyl acetate, washed with ultrapure water, and air-dried. Measurements were performed using a Zeiss Libra® 120 and image analysis was performed using ImageJ software.

[0253] The size and zeta potential were determined as detailed in Example 1A.

[0254] Protein assays were performed as described in Example 1A.

[0255] Protein identification and quantification were performed as in Example 1A. Unweighted spectral counts (USC) were used to assess the consistency of biological replicates in quantitative analysis, and normalized spectral counts (NCS) were used to collect protein content.

[0256] Statistical Analysis. Statistical analyses herein were performed in Python using the scikit-learn, numpy and scipy packages, and figures and graphs were generated using the bokeh package in Python, Microsoft Excel, XLSTAT and MATLAB.

[0257] Data matrix For all 60 plasma samples (45 non-cohort, 15 cohort), labeled i = 1,...,60, we have the data matrix X i (3 rows and ~900 columns), each row of the matrix is was generated to correspond to the protein content of a single nanoparticle obtained from the protein corona nanosystem. As a preprocessing step, we converted the protein content to relative protein abundance (RPA) by row normalizing the entire matrix.

[0258] Classification and Clustering Tensor Decomposition We treat the data as a three-mode tensor, with the first two modes corresponding to the nanoparticles and proteins, and the third mode corresponding to the plasma samples, which can be expressed as a function of each sample, X i , is equivalent to stacking on top of each other the observation matrices corresponding to ,. The data were denoised via low Tucker rank tensor decomposition (135, 136, 137) using code implemented in Python for this project (available for research and other use upon request). Each matrix X i X i ≒US i V T where U is a matrix with rows that can be viewed as latent features corresponding to each nanoparticle, and similarly V is a matrix with columns that can be viewed as latent features corresponding to each protein, and these latent features are shared across the entire data matrix. Finally, for each S i Nanoparticles and proteins is a matrix that encodes interactions between features, which can be made unique across samples. We estimate this decomposition in two steps: (a) we estimate U and V by truncated singular value decomposition for the mode 1 and mode 2 expansions of the tensor, and then we consider these estimates and (b) we use the least squares method to estimate each S i Fit the matrices separately I did.

[0259] Random Forest Classification The Random Forest model is a well-known machine learning algorithm for classification. A Random Forest is composed of multiple decision trees, each of which makes a simple classification decision based on a relatively small number of variables. The trees are created (or "trained") with different, randomly drawn subsets of the variables so that no two trees are identical. For a new sample, each tree is trained on a different set of variables until the set of training samples reaches the bottom. , up and down. The entire forest is used for classification weighting to allow multiple decision trees to "vote" for a label (in this case one of five types of cancer or health), where each tree votes for a set of training examples at the bottom. In our proprietary algorithm, each random forest was composed of 1000 decision trees and trained using the scikit-learn package. Importance scores were also calculated using the same package.

[0260] <Example 2: Building additional sensor arrays for disease detection> A sensor array consisting of 12 different cross-reactive nanoparticles was fabricated, including three liposomes, three superparamagnetic iron oxide nanoparticles, and six gold nanoparticles. These types of liposomes (DOPG (1,2-dioleoyl-sn-glycero-3-phospho-(1'-rac-glycerol)), DOTAP (1,2-dioleoyl-3-trimethylammonium-propane)-DOPE (dioleoylphosphatidylethanolamine) and CHOL (DOPC-cholesterol)) with negative, neutral and positive surface charges were synthesized based on our previous report. Ultra-uniform PEG-coated superparamagnetic iron oxide nanoparticles with a size of 20 nm and various PEG molecular weights (i.e., 300, 3000 and 6000) were obtained from Micromod®. Gold nanoparticles with a core size of ~2 nm and different surface functionalities were synthesized.

[0261] The nanoparticles are all the same size but have different surface properties, which are expected to result in significantly different corona compositions in response to the plasma of patients with various cancers. The sensor array explores the potential of protein corona sensor arrays to identify and differentiate cancer types via patient plasma proteins, as described in Example 1.

[0262] Samples from a wide range of cancers were examined, including lung cancer, pancreatic cancer, meningioma, myeloma, glioblastoma, esophageal squamous cell carcinoma, and gastric adenoma. Protein corona arrays based on different nanoparticles (liposomes, iron oxide, and gold) were designed to examine the variations in biomolecular fingerprints for various cancer types. The sensor arrays can also identify proteins that are markers for specific types of cancer, including new unknown biomarkers for different cancer cell types. Pattern recognition can be determined based on different types of cancer or cancer stages.

[0263] We will analyze a wide range of human plasma from healthy individuals who were diagnosed with various cancers several years after the plasma collection. The plasma samples were collected through an NIH-funded cohort study called the Golestan Cohort Study, conducted by the National Cancer Institute (NCI) in the United States, the International Agency for Research on Cancer (IARC) in France, and the Tehran University of Medical Sciences (TUMS) in Iran. In this study, we collected and stored plasma from 50,000 healthy subjects. More than 1,000 of these subjects developed various cancers in the following years. The samples were stored at IARC and used for analysis by our team. These important plasma samples provide us with a unique opportunity to investigate the early cancer detection capabilities of our inventive protein corona sensor array. In addition, we will also evaluate and identify proteins that are useful in the identification and differentiation of these cancers.

[0264] We believe that the results obtained by applying the sensor array of the present invention to cohort plasma samples will not only be useful for early cancer detection and screening, but also help identify novel protein markers involved in the development of cancer. The selection of components of our sensor array has a unique ability over other developed methods to provide fingerprints for a wide range of proteins in a non-specific and cross-reactive manner for cancer identification and differentiation.

[0265] The hard corona profile of the sensor array element was examined using plasma from patients with intermediate and advanced stage cancer. The protein corona formed on the surface of the sensor array elements strongly depends on the physicochemical properties of those nanoparticles and can be strongly influenced by the type of disease present in the donor of the human plasma used for the incubation. To prepare the corona-coated nanoparticles, 12 prepared nanoparticles were (separately) incubated with human plasma of nine cancer types (lung cancer, pancreatic cancer, myeloma, myeloid leukemia, meningioma, glioblastoma, breast cancer, esophageal squamous cell carcinoma and gastric adenoma) and separated from the uninvolved free proteins using a well-defined centrifugation approach. Centrifugation is usually performed at 13000g for 30 minutes at 15°C. The supernatant is removed and the recovered particles may be redispersed in 500 microliters of PBS. This procedure is repeated to remove loosely bound proteins. To remove loosely bound proteins from the surface of the nanoparticles, the recovered nanoparticles are redispersed in chilled PBS (15°C) and collected by centrifugation. The size and charge of the corona-coated nanoparticles are then determined using DLS / Nanosight® and compared to these initial values ​​obtained in buffer. Quantitative assessment of the total proteins adsorbed on the nanoparticles is performed using BCA or NanoOrange® assays, while a qualitative shotgun proteomic analysis identifies the proteins adsorbed on the surface of 12 nanoparticles. Briefly, after separation of proteins from the surface of the nanoparticles (based on the protocol of Saha, K.; Rahimi, M.; Yazdani, M.; Kim, ST; Moyano, DF; Hou, S.; Das, R.; Mout, R.; Rezaee, F.; Mahmoudi, M. ACS nano 2016, 10, (4), 4421-4430, which is incorporated herein by reference), the proteins are injected into a liquid chromatography-mass spectrometry (LC-MS / MS) instrument. Proteins are identified from the data obtained by screening of relevant databases. To obtain the total number of LC-MS / MS spectra for all peptides attributed to the matched proteins, the semi-polydispersity index (SPI) obtained from the cumulant fitting quantitative evaluation of protein abundance was used. aPolydispersity index (PpSpC) is used for spectral counts (SpC). The normalized SpC (NpSpC) amount of each protein identified in the LC-MS / MS spectrum is calculated using the following formula:

number

[0266] Supervised and unsupervised clustering analyses are deployed to identify and differentiate cancer using the results obtained from the sensor array. To investigate whether the protein corona fingerprints (PCFs) of different sensor elements could be used as biosensors and form unique patterns (biomolecular corona signatures) for different diseases, we applied a focused classification approach as described in Example 1 to the proteomic data from the protein corona composition of three liposomes (cationic, anionic, and neutral).

[0267] Example 3: Attaching nanoparticles to a substrate to create a sensor array Different types of particles may be used as nanoscale sensor elements in the practice of the present invention. Additionally, different methods are provided for bonding the nanoscale sensor elements to the substrate. In addition, the nanoscale sensor elements can be attached to the substrate in different patterns.

[0268] Specific examples of the configurations of different sensor elements on different substrates are shown in Figs. 15A to 43.

[0269] Example 4: Sensor array with silica and polystyrene nanoparticles for cancer screening This example shows that the sensor array of the present invention having different nanoparticles than those used in Example 1 is also capable of detecting cancer samples among healthy patient samples.

[0270] Using the experimental protocol of Example 1, a sensor array was designed using modified silica and polystyrene particles. In this example, a total of six nanoparticles were fabricated, namely nanoparticles with three different surface functional groups: unmodified, amine-modified, and carboxyl-modified (P-NH). 2 , P-COOH, S-NH 2 Two types of nanoparticles, namely polystyrene (P) and silica (S), having the same molecular weight as the carboxyl group (C-COOH) were used.

[0271] Different functional groups (unmodified, amine-modified (NH 2 ) and carboxyl group modification (COOH) Characterization of uncoated polystyrene and silica nanoparticles with different functional groups is shown in Figures 44A-44D, showing their size, DLS and zeta potential of uncoated particles, and TEM images. Characterization of protein corona coated polystyrene and silica nanoparticles with different functional groups is shown in Figures 45A-45D, showing their size, DLS, zeta potential, and TEM images of polystyrene and silica nanoparticles topped with protein corona.

[0272] Six nanoparticle sensor arrays were contacted with plasma from healthy individuals or patients with rectal, breast, bladder, thyroid, uterine, ovarian, and kidney cancer (five patients per cancer) as shown in Figure 46, using the methods described in Example 1. The protein corona profiles of polystyrene and silica nanoparticles (100 nm) were analyzed by SDS PAGE. A comparison of the protein coronas of unmodified, amine-modified, and carboxyl-modified particles by SDS PAGE is shown in Figure 47.

[0273] Figure 48 shows the protein corona profile for healthy plasma for polystyrene and silica nanoparticles (100 nm) analyzed by SDS-PAGE. A comparison of the protein corona of unmodified, amine-modified and carboxyl-modified particles is shown.

[0274] The data was analyzed as described in Example 1A. The results of statistical analysis and clustering are shown in Figure 49. As shown, it was possible to identify and classify healthy individuals compared to cancer patients (healthy controls are in an orthogonal space to cancer patient samples as shown in Figure 49).

[0275] This example shows that this sensor array using six nanoparticle arrays can distinguish and detect patients with cancer from healthy individuals.

[0276] Materials: Silica particles with three different modifications were purchased from Kisker-Products (https: / / www.kisker-biotech.com / ), and polystyrene particles with three different modifications were Polysilicon particles were purchased from Polyscience (http: / / www.polysciences.com / ). All particles were of the same size (100 nm). The morphology, mean size, polydispersity index (PDI) and zeta potential of these particles were characterized by TEM, DLS and zeta potential measurements.

[0277] Experimental information: Incubated in 50% human healthy plasma for 1 hour. SDS P AGE: 4-20% acrylamide, 45 min, 40 mA / gel. Staining: Colloidal Blue Comassie® overnight. Nanoparticles used: 0.5 mg.

[0278] Example 5: Protein corona sensor array nanosystem identifies coronary artery disease Coronary artery disease (CAD) is the most common type of heart disease and the leading cause of death in both men and women. Early detection of CAD is essential to prevent death, prolong survival, and improve patients' quality of life. This example describes a non-invasive six-nanoparticle sensor array nanosystem for ultra-precise detection of CAD using specific protein corona (PC) pattern recognition. Although the protein corona of a single nanoparticle does not provide the required specificity, the multivariate protein corona of six different nanoparticles with different surface chemistry provides the desired information to selectively discriminate each cardiovascular condition under examination.

[0279] CAD is a chronic condition that progresses gradually throughout a person's life with onset during adolescence. It is characterized by the presence of atherosclerotic plaques in the coronary arteries. The origin of atherosclerosis lies in the dysfunction of the vascular endothelium, where upon exposure to stress stimuli and inflammatory factors (e.g., oxidative stress and hemodynamic forces), endothelial cells express surface adhesion molecules that trigger the recruitment of circulating leukocytes and cholesterol-laden low-density lipoproteins (LDL). 5 These events lead to the formation of atherosclerotic plaques. This narrows the coronary arteries and reduces blood flow. Depending on how quickly the plaque develops and how severely the artery is blocked, the condition can culminate in a myocardial infarction. 5

[0280] Accurate and early diagnosis of CAD in risk subjects is crucial to promptly initiate ad-hoc treatment and avoid further complications. Coronary angiography has been the most accurate and reliable method of CAD diagnosis until now. However, inserting a catheter from an artery in the arm (or neck or upper thigh) to the heart is invasive, costly, and also causes many side effects including infection, damage to the catheterized artery, allergy, and excessive bleeding. Therefore, there is an urgent need to develop new tests for CAD detection. Although several inflammatory biomarkers have been reported to be useful for diagnosis, unfortunately, none of these have been used in clinical practice, which highlights the still need for new diagnostic tests that can be commonly performed. We have developed a nano-based blood test as a new tool for CAD diagnosis.

[0281] As shown in the previous example for cancer, the personalized protein corona serves as a fingerprint of a given plasma state. This example uses the same approach to pinpoint the formation of atherosclerotic plaques through the resulting changes in plasma composition. Indeed, plaque-associated cells (e.g. foam cells, macrophages, mast cells, monocytes and T cells) shed various biomolecules (e.g. cytokines, proteolytic enzymes and vasoactive biomolecules) into the blood, 22 It may therefore be possible to induce changes in the pattern of corona protein composition on the surface of various nanoparticles relative to the protein corona of patients without plaque formation.

[0282] In this example, the protein corona formed around nanoparticles was analyzed using plasma obtained from i) patients diagnosed with CAD after coronary angiography (CAD), ii) symptomatic patients who underwent coronary angiography but whose coronary vessels were deemed healthy (no CAD), iii) restenosis (recurrence of CAD after treatment), and iv) healthy volunteers (controls) with no risk factors for vascular disease (e.g. family history, smoking, obesity, hypertension).

[0283] To obtain a broader spectrum of adsorbed proteins, this example uses six commercially available nanoparticles as sensor array elements, giving them different compositions, surface chemistries and / or functional groups, resulting in a six-nanoparticle array that can be used as a simple and non-invasive diagnostic test. We have fabricated a protein corona-based sensor array that can detect CAD signals and detect vascular endothelial cell death. This novel diagnostic CAD test can be used as a pre-emptive non-invasive screening for at-risk patients. Remarkably, these results show that the protein corona patterns can highly accurately discriminate between CAD, non-CAD, restenosis, and control patients, providing a novel, accurate, non-invasive, and previously undeveloped tool for vascular-based CAD diagnosis.

[0284] <Result> In this example, a total of six nanoparticles were prepared, namely nanoparticles with three different surface functional groups (i.e., unmodified, amine-modified, and carboxyl-functionalized (P-NH)), as shown in Figures 50A-50C. 2 , P-CO OH, S-NH 2 and S-COOH) (i.e., polystyrene (P) and Nanoparticles of Lica (S), were used. To compare the differences in results between the synthetic new identity of the uncoated nanoparticles and their corresponding biological identity (protein corona coated nanoparticles), the size, zeta potential and morphology of the nanoparticles were measured before and after incubation in plasma. Dynamic light scattering showed that all the uncoated nanoparticles were largely monodisperse, as indicated by a polydispersity index of ≦0.02 and a uniform size of about 100 nm (range of about 93 nm to 120 nm) (Figure 50A). After 1 hour of incubation with patient plasma, the size of all nanoparticles was shown to increase due to the presence of a layer of adsorbed protein (PC), and the thickness and composition of this layer were dependent on the protein concentration, surface properties and size of the nanoparticles. All the uncoated nanoparticles had a negative surface charge (Figure 50B), with the amine modified ones being slightly less negative than the others due to the contribution of the positive amine groups. These results are consistent with the specifications provided by the supplier and other studies. Amine group modification was insufficient to switch the surface charge of silica and polystyrene nanoparticles, which are characterized by all negatively charged residues on the surface at physiological pH.

[0285] Once exposed to plasma, the total surface charge becomes less negative (-5mV to -25mV) due to the charge of many plasma proteins at physiological pH. Overall, protein corona-coated nanoparticles tend to have similar physicochemical properties, being larger and less negative than their uncoated counterparts, regardless of the plasma used for incubation. However, when incubated with plasma without CAD, P and S nanoparticles increased in size to ≈85nm, which is larger than that shown with plasma in other conditions (40-50nm). On the other hand, S-NH nanoparticles incubated with CAD plasma showed a large increase in size to ≈85nm, which is larger than that shown with plasma in other conditions (40-50nm). 2 Proteins in nanoparticles The corona thickness was greater (≈40 nm) than that obtained by incubating the same nanoparticles with other types of plasma (≈30 nm). Transmission electron microscopy showed that the nanoparticles did not change their morphology and structure after incubation in plasma (Figure 50C). Furthermore, an increase in size was observed after coating, thus confirming the results obtained by dynamic light scattering. The protein concentration of the different protein coronas was evaluated by Bradford assay, which showed that all silica nanoparticles adsorbed less protein than polystyrene nanoparticles in the protein corona (Figure 51A). This observation was supported by the analysis of the protein corona by 1D-SDS PAGE. Plasma was collected from five patients with CAD, without CAD and without CAD risk (control), and the protein corona obtained for all nanoparticles was analyzed by Comassie stained SDS-PAGE. The proteins were separated on a PAGE gel (Figure 51B). The gel was analyzed by densitometry and differences in the amount of proteins in the protein corona of CAD, no CAD and control of the same nanoparticles were detected (Figure 51C, arrows). In some cases, we also noted the presence or absence of some proteins in the specific protein corona (Figure 51C, blue arrows). These results confirmed that the formation of atherosclerotic plaques alters the plasma composition and thus gives rise to differences in the protein corona. The protein corona was analyzed by LC-MS / MS analysis. More than 150 proteins were identified in each of the protein-coated nanoparticle samples. In addition, to obtain information on the protein content, Therefore, to obtain information on the contribution of each of the identified proteins in the protein corona, the spectral count value, which represents the total number of fragmentation spectra of all peptides attributed to a specific protein, was used. The difference in the contribution of the top 20 abundant proteins in the protein corona is reported in Figures 52-1 to 52-6.

[0286] Data from all samples were collected, analyzed and sorted, and a key for each measurement was created by concatenating nanoparticles, surface modification, plasma type and label. Proteins identified with less than two peptides were excluded from consideration, and the TIB5000 library was used to pivot the data so that rows were protein accessions and columns were values ​​containing the keys and percent contributions. CO Spotfire® Anslyst 7.6.1 was used.

[0287] In addition to the analytics, we analyzed the presence of exclusive proteins in the corona of a given condition. For this, we constructed a Venn diagram to facilitate the search for common and unique proteins in a big data set of proteins or genes. The protein corona of each nanoparticle was analyzed separately for all six classes. We looked for exclusive proteins by comparing the common proteins in all the protein coronas of CAD, no CAD and controls obtained from the five patients. As a result, various exclusive proteins were exclusively identified in the exclusive protein corona patterns of each nanoparticle used in this study. Among them, several proteins involved in the regulation of complement activation (Factor H-related protein 3, complement component C8γ) were exclusively identified in the protein corona patterns of CAD patients. This result is consistent with the recently described role of complement activation in the pathogenesis of cardiovascular diseases (development of atherosclerosis, plaque rupture and thrombosis). Another example is shown by apolipoprotein(a), which could be considered as an attractive biomarker candidate for use in clinical practice for CAD, which we detected exclusively in the protein corona patterns of CAD patients. Apolipoprotein(a) is the major component of lipoprotein(a) and is known to undergo proteolytic cleavage and its fragments are deposited in atherosclerotic plaques.

[0288] We confirmed the accuracy of our approach in discriminating between CAD and CAD-free patients by analyzing nine blinded plasma samples (three for each condition and three controls). Protein coronas were formed around six nanoparticles of our protein corona sensor array nanosystem. Proteins in the protein corona were then identified by LC-MS / MS (Supplementary Information) and the results were analyzed using the same classification and clustering approach.

[0289] The statistics and data analysis are presented graphically in Figures 54A, 54B, showing a discrete separation of the CAD, no CAD and control (no CAD risk) categories.

[0290] In conclusion, we have demonstrated the accuracy of the 6-nanoparticle protein corona sensor array for the detection of CAD. The protein corona sensor array nanosystem developed in this study was sensitive and accurate in discriminating between CAD, restenosis, no-CAD, and healthy individuals, further indicating its great potential value as a technology platform to be used in clinical settings. Indeed, despite the presence of many symptoms typically associated with atherosclerotic plaques, the patients in the no-CAD group investigated in this study did not have any blockages in their arteries.

[0291] The protein corona patterns obtained from this platform represent unique multivariate fingerprints that are more accurate and comprehensive than those obtained using the protein corona of a single nanoparticle. The approach described herein may be of great use in the detection of CAD as well as various other human diseases, improving the quality of life of many patients. Especially due to the non-invasive nature of the test, we envision that such tests will be used more willingly and more frequently by patients than angiography. And unlike angiography, this test is easy to perform and has no side effects, allowing patients to check the status of their arteries from the earliest symptoms, and early detection can greatly reduce CAD complications.

[0292] <Method> Nanoparticles Silica particles with three different modifications were purchased from Kisker-Products (https: / / www.kisker-biotech.com / ), and polystyrene particles with three different modifications were were purchased from Polyscience (http: / / www.polysciences.com / ). All particles were of the same size (100 nm). The morphology, average size, polydispersity index (PDI) and zeta potential of these particles were characterized by TEM, DLS and zeta potential measurements.

[0293] Protein corona formation Protein coronas were generated by incubating 0.5 mg of nanoparticles with an equal volume of human plasma in deionized water. Incubation was performed for 1 h at 37°C with stirring. Immediately after incubation, centrifugation was performed at 14,000 rpm for 30 min at 10°C to form a pellet. The pellet was then washed and suspended in 200 μl of phosphate-buffered saline (PBS) at 4°C. The centrifugation measurements were repeated three times under the same conditions as before. The pellets of protein corona-coated nanoparticles were resuspended in 8 M urea, 50 mM ammonium bicarbonate for subsequent SDS-PAGE gel and LC-MS / MS analysis or resuspended in deionized water for subsequent size and zeta potential analysis.

[0294] Nanoparticle characterization The size and zeta potential of uncoated and protein-coated nanoparticles were characterized by diluting 10 μl of each sample in a total of 1 ml distilled water. Measurements were performed using a Zetasizer Nano ZS90 (Malvern, UK). Size and surface charge values ​​were presented as the mean ± SD of three independent measurements.

[0295] Protein concentration assay The amount of protein in the corona was determined by a five-point standard curve (R 2 =0.99) Protein concentrations were assessed by Bradford assay (Bio-Rad) using bovine serum albumin at known concentrations as data. Protein concentrations were recorded as the mean ± SD of three experiments.

[0296] 1D-SDS PAGE gel Proteins in the corona were dissolved in 8M urea, 50mM ammonium bicarbonate. An equal volume of Laemmli buffer 2X was added to the pellet and heated to 90°C for 5 min, then loaded onto a 4-20% Mini-PROTEAN™ TGX® precast gel (Bio-Rad Laboratories, Hercules, CA) and separated at 120V for 1 h. Proteins were stained overnight with Coomassie™ brilliant blue (Fisher Scientific, Fair Lawn, NJ, USA) followed by extensive washing in ultrapure water.

[0297] Mass spectrometry / statistics / PLS-DA Mass spectrometry, statistical analysis and PLS-DA were performed as described in Example 1.

[0298] Example 6: Multi-nanoparticle protein-corona test for early Alzheimer's disease detection The pathology of Alzheimer's disease (AD) begins decades before the discovery of clinical symptoms. The need for accurate and non-invasive early diagnosis of AD is rapidly increasing. In this paper, a state-of-the-art sensor array nanosystem has been developed that uses clustering techniques to detect subtle changes in plasma protein patterns and determine the presence of Alzheimer's disease. The developed technology can also be applied to the diagnosis of other diseases in the future, as the sensor array generates a unique fingerprint for each plasma proteomic change caused by the disease and is sensitive enough to capture such changes, unlike current technologies.

[0299] As discussed in the previous examples, the composition of the protein corona varies based on the plasma proteome, which may be altered as a result of disease, resulting in a “personalized protein corona” (PPC). However, PPCs cannot provide a robust and accurate strategy for early disease detection, mainly due to the large overlap of similar proteins in the protein corona composition. In this study, by using a PPC formed around a multi-nanoparticle platform containing six nanoparticles, we provide a fingerprint pattern for robust and accurate detection of Alzheimer’s disease with unprecedented predictive accuracy and specificity. The developed sensor array test successfully distinguished between patients with Alzheimer’s disease and those without, and also between patients who developed Alzheimer’s disease several years later (using cohort plasma). These examples provide a feasible, non-invasive alternative to current Alzheimer’s disease detection tools, providing unparalleled capabilities for early detection and treatment.

[0300] <Results and Discussion> The sensor array test involves incubating plasma samples with six different nanoparticles, the latter being 100 nm polystyrene and silica nanoparticles, each with tunable surface chemistries (unmodified, -amino and -carboxyl conjugated, hereafter P, P), as described for CAD analysis in Example 5. -NH 2 , P-COOH, S, S-NH 2, referred to as S-COOH) and a narrow size distribution. As a first proof of concept, the size, surface charge and morphology of nanoparticles before and after protein corona formation were analyzed to investigate the differences between Alzheimer's plasma and control plasma. Nanoparticle tracking analysis (Nanosight) was used to characterize the size of nanoparticles. In such analysis, a laser beam illuminates the nanoparticles, from which the scattered light is visualized via an optical microscope. Meanwhile, a video is recorded by a camera aligned with the beam, showing the nanoparticles' movement (30-60 frames / sec). Before incubation in plasma, all nanoparticles were uniform in size (Figure 58, polystyrene nanoparticles 90-100 nm; silica nanoparticles 80-100 nm) and the negative surface charge was consistent with that provided by the manufacturer (Table 8). After incubation in plasma with stirring for 1 h at 37 °C, the protein corona-coated nanoparticles were collected by centrifugation and then washed extensively to remove unbound or loosely bound proteins. In all cases, we found that the size of the protein corona-coated nanoparticles increased and that the size distribution became broader, indicating a decrease in the homogeneity of the population (Figures 54A and 54B, scatter plots). The average increase in size was 30 nm, thus indicating a thickness of the protein corona layer of 15 nm, which is consistent with the literature 1 Reported in This is in agreement with the data and with the observations made with plasma from healthy volunteers (data not shown). The presence of plasma proteins on the surface of the nanoparticles changes its charge, i.e. makes it less negative, reaching the normal value for plasma proteins (Table 8, -20 mV to -0 mV).

[0301] [Table 8]

[0302] A slight difference was observed in the surface charge of the same nanoparticles incubated with Alzheimer's and healthy plasma (Table 8). However, this difference was not large and therefore not useful for distinguishing between the different plasma states. Transmission electron microscopy (TEM) was also used to evaluate the morphology of the nanoparticles, which remained unchanged after the formation of the protein corona. In fact, all nanoparticles were round in morphology both before and after incubation with plasma (Figures 55-1 and 55-2). The presence of a thin layer of protein associated with an increase in the size of the protein corona-coated nanoparticles was observed, which confirms the results obtained by Nanosight (Figures 55-1 and 55-2).

[0303] The protein coronas associated with the particles were separated by gel electrophoresis (SDS-PAGE) and later visualized by staining with Coomassie® Brilliant Blue (Figure 56). In general, the six nanoparticles had distinct protein corona patterns by visual evaluation of the lanes, which was expected and desired for the best performance of our approach. Silica nanoparticles appeared to have less protein adsorbed on their surface than polystyrene nanoparticles. In many samples, the protein corona of unmodified polystyrene was more concentrated in proteins than the others, with molecular weights of ≈60 kDa, most likely due to albumin. This was true for both the plasma of Alzheimer's patients and the plasma of controls (Figure 56, arrows). Densitometric analysis of the bands associated with each protein corona confirmed that silica nanoparticles adsorbed less protein than normal, but also revealed that the number of proteins that made up their protein corona was greater. For example, there were more than two bands at the level of albumin for the protein corona of polystyrene nanoparticles (Figure 56, arrows). On the other hand, several differences were recorded in the PPC (personalized protein corona) of Alzheimer's patients and healthy individuals, especially in the case of silica nanoparticles (Figure 56, arrows). However, these differences were not statistically significant and were insufficient to accurately distinguish between the two groups of patients under study. To investigate the protein corona more deeply and to know the exact identity and amount of proteins that compose the different protein coronas, all samples were analyzed by mass spectrometry. To determine the contribution of each protein in the protein corona, spectral counting label-free analysis, which is widely adopted for quantitative profiling of the protein corona around nanoparticles, was used. 2-4 This calculation was performed for each plasma patient (separately with each nanoparticle). The Alzheimer's disease specific protein corona profile was thus generated from the combined contribution of the protein coronas associated with the six nanoparticles. Table 9 shows the patient populations used.

[0304] [Table 9]

[0305] <Conclusion> This is a proof-of-concept study for the development of a MNPC blood test for the diagnosis of Alzheimer's disease. The MNPC test showed unprecedented predictive accuracy and specificity. While individual biomarker blood-based tests often give false positives and therefore require further analysis to confirm the diagnosis, our approach is able to record the interactions of a set of six nanoparticles with all proteins for which they have high affinity, thus generating a highly specific protein corona fingerprint for a given disease. The nanoparticles act as nano-concentrators of plasma proteins on their surface (each nanoparticle...

Claims

1. 1. A method of identifying a state of a complex biological sample of a subject using a plurality of particles, comprising: (a) contacting the complex biological sample with a plurality of particles to allow binding of a plurality of proteins of the complex biological sample to a surface of the plurality of particles to form distinct biomolecular corona signatures; (b) isolating and identifying at least a subset of proteins from said plurality of particles to define a biomolecular fingerprint representative of said combination of biomolecular corona signatures; (c) correlating the biomolecular fingerprint with a biological state of the subject; the plurality of particles having surfaces with different physicochemical properties; A method, wherein the pattern of binding of a plurality of proteins between said plurality of particles differs depending on said physicochemical properties of said surface of said plurality of particles.

2. The method of claim 1 , wherein the complex biological sample is plasma.

3. 10. The method of claim 1, wherein identifying the subset of proteins comprises performing at least one of gel electrophoresis, liquid chromatography, mass spectrometry, NMR, FTIR, circular dichroism, and Raman spectroscopy.

4. The method of claim 1 , wherein the plurality of particles comprises nanoparticles.

5. The method of claim 1 , wherein the plurality of particles is unlabeled.

6. The method of claim 1 , wherein the plurality of particles enriches low abundance proteins from the biological sample.

7. The method of claim 1 , wherein the complex biological sample comprises at least one of proteins, nucleic acids, lipids, and polysaccharides.

8. 2. The method of claim 1, wherein the biomolecular fingerprint is distinct from a second biomolecular fingerprint of a complex biological sample from a second subject not exhibiting the biological condition.

9. The method of claim 1 , wherein the biological condition is a disease state of the subject.

10. 10. The method of claim 1, wherein the subject is free of physical symptoms of a disease or disorder associated with the biological condition.

11. The method of claim 1 , wherein the physicochemical properties of the surface of the particle include at least one of charge, composition, size, hydrophobicity, hydrophilicity, surface functionality, surface topography, and shape.

12. The method of claim 1 , wherein the physicochemical property of the surface of the particle comprises an electric charge.

13. 10. The method of claim 9, wherein the subject is diagnosed with a disease or disorder associated with the disease state.

14. The disease state is cancer, cardiovascular disease, endocrine disorder, inflammatory disease, and neurological disease. The method of claim 13, wherein the at least one of

15. 15. The method of claim 14, wherein the disease state is a cancer selected from the group consisting of lung cancer, pancreatic cancer, colon cancer, myeloma, myeloid leukemia, meningioma, glioblastoma, breast cancer, esophageal squamous cell carcinoma, gastric cancer, prostate cancer, bladder cancer, ovarian cancer, thyroid cancer, neuroendocrine tumors, and combinations thereof.

Images