Identification and use of glycopeptides as biomarkers for diagnosis and treatment monitoring

Abstract

To provide methods of identifying and using glycopeptides as biomarkers for diagnosis and treatment monitoring.SOLUTION: Provided herein are methods of identifying new biomarkers for various diseases using proteomics, peptidomics, metabolics, proteoglycomics, glycomics, mass spectrometry and machine learning. The present disclosure also provides glycopeptides as biomarkers for various diseases such as cancer and autoimmune diseases.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates generally to the fields of multi-omics, particularly glycomics and glycoproteomics, advanced instrumentation big data, machine learning, and artificial intelligence, to identify biomarkers for disease diagnosis and treatment monitoring. [Background technology]

[0002] Protein glycosylation and other post-translational modifications play important structural and functional roles in all aspects of human growth and development. Defective protein glycosylation is associated with several diseases. Identifying glycosylation changes at an early stage of disease provides an opportunity for early detection, intervention, and higher chances of survival in affected subjects. There are currently methods to detect early cancer and identify biomarkers that can distinguish certain types of cancer from other diseases. These methods include proteomics, peptidomics, metabolism, proteoglycomics, and glycomics using mass spectrometry (MS).

[0003] Although protein glycosylation provides useful information about cancer and other diseases, one drawback of this method is that glycans cannot be traced back to the site on the protein from which they originate. To gain more knowledge about cancer biology and early cancer detection, it is important to identify not only the glycans but also their attachment sites on the protein. Glycoprotein analysis is generally challenging for several reasons. For example, a single glycan composition in a peptide may contain many isomeric structures due to different glycosidic linkages, branches, and many monosaccharides with the same mass. Furthermore, the presence of multiple glycans sharing the same peptide backbone splits the MS signal into various glycoforms, lowering their individual abundance compared to unglycosylated peptides. Therefore, it has been difficult to develop algorithms that can identify glycans and their peptides from tandem MS data. It is also difficult to obtain comprehensive fragmentation for both glycans and peptides due to the different fragmentation efficiencies of glycans and peptides.

[0004] Therefore, there is a demand to provide a method of site-specific glycoprotein analysis to obtain important and detailed information on protein glycosylation patterns that provide accurate quantitative information on the heterogeneity of glycosylation sites in diseased cells, tissues or biological fluids compared to non-diseased cells. Such a method would lead to the identification of disease biomarkers, especially for diseases such as cancer. There is also a demand to combine site-specific glycoprotein analysis data with deep learning and advanced LC / MS instrumentation to identify and validate new disease targets, such as glycan-based drug targets for diseases such as cancer, thereby reducing the time to identify new biomarkers. Summary of the Invention

[0005] The present disclosure relates to methods for identifying biomarkers for various diseases. These biomarkers are glycosylated peptide fragments obtained by fragmentation of glycosylated proteins from biological samples. The methods for identifying biomarkers rely on the use of advanced mass spectrometry techniques that allow accurate mass measurement of glycosylated peptide fragments and site-specific glycosylation analysis. The mass spectrometry methods of the present disclosure advantageously serve to analyze a large number of glycosylated proteins at once from a biological sample.

[0006] In one embodiment, the disclosure provides a method for identifying glycosylated peptide fragments as potential biomarkers, the method comprising: fragmenting glycosylated proteins with one or more proteases in each of a plurality of biological samples isolated from a subject to generate glycosylated peptide fragments; Quantifying the glycosylated peptide fragments by liquid chromatography and mass spectrometry (LC-MS), providing a quantitative result; analyzing the quantitative results together with the classification of interest using machine learning methods to select glycosylated peptide fragments that are useful for predicting the classification; and determining the identity of the glycosylated peptide fragment.

[0007] In another embodiment, the method includes subjects having a disease or condition and subjects not having the disease or condition, hi a further embodiment, the subjects include subjects receiving treatment for a disease or condition and subjects having a disease or condition but not receiving treatment.

[0008] In another embodiment, the method of the present disclosure is applicable to any disease or condition that can be detected by analyzing glycosylated peptide fragments from a biological sample of a subject. In one embodiment, the disease is cancer. In another embodiment, the disease is an autoimmune disease. In another embodiment, the method of the present disclosure provides glycosylated peptide fragments that are O-glycosylated or N-glycosylated. In another embodiment, the method of the present disclosure provides glycosylated peptide fragments having an average length of 5 to 50 amino acid residues.

[0009] In another embodiment, the method of the disclosure uses a glycosylated protein that is one or more of alpha-1-acid glycoprotein, alpha-1-antitrypsin, alpha-1B-glycoprotein, alpha-2-HS-glycoprotein, alpha-2-macroglobulin, antithrombin III, apolipoprotein B-100, apolipoprotein D, apolipoprotein F, beta 2 glycoprotein 1, ceruloplasmin, fetuin, fibrinogen, immunoglobulin (Ig) A, IgG, IgM, haptoglobin, hemopexin, histidine-rich glycoprotein, kininogen-1, serotransferrin, transferrin, vitronectin, and zinc-alpha-2-glycoprotein.

[0010] In another embodiment, the method of the present disclosure comprises fragmenting a glycosylated protein using at least two proteases. In another embodiment, the method of the present disclosure uses LC-MS technology using multiple reaction monitoring mass spectrometry (MRM-MS).

[0011] In another embodiment, the present disclosure provides a method for identifying glycosylated peptide fragments as potential biomarkers for various diseases described herein, wherein the biological sample is body tissue, saliva, tears, sputum, spinal fluid, urine, synovial fluid, whole blood, serum or plasma obtained from a subject. In one embodiment, the subject is a mammal. In another embodiment, the subject is a human.

[0012] In another embodiment, the disclosure provides a method for identifying glycosylated peptide fragments as potential biomarkers, the method comprising: fragmenting glycosylated proteins with one or more proteases in each of a plurality of biological samples isolated from a subject to generate glycosylated peptide fragments; Quantifying the glycosylated peptide fragments by liquid chromatography and mass spectrometry (LC-MS), providing a quantitative result; analyzing the quantitative results together with the classification of interest using machine learning methods to select glycosylated peptide fragments that are useful for predicting the classification; determining the identity of the glycosylated peptide fragment, wherein the machine learning approach is deep learning, neural network, linear discriminant analysis, quadratic discriminant analysis, support vector machine, random forest, nearest neighbor, or a combination thereof. In another embodiment, the machine learning approach is deep learning, neural network, or a combination thereof.

[0013] In another embodiment, the present disclosure provides a method for identifying glycosylated peptide fragments as potential biomarkers for various diseases described herein, wherein the analysis further comprises genomic data, proteomic, metabolic, lipidomic data, or a combination thereof. [Brief description of the drawings]

[0014] [Figure 1] Schematic showing the integration of glycomics, LC / MS and machine learning, which can be further combined with proteomics, genomics, lipidomics and metabolism. [Diagram 2] Changes in immunoglobulin G (IgG) glycopeptide ratios in plasma samples from breast cancer patients and controls are shown. [Diagram 3] Figure 2 shows changes in IgG glycopeptide ratios in primary sclerosing cholangitis (PSC) and primary biliary cirrhosis (PBC) samples and plasma samples from healthy donors. [Figure 4]Discriminant analysis data for IgG, IgA and IgM glycopeptides in plasma samples from healthy donors and PSC and PBC samples are shown separately. [Diagram 5] Combined discriminant analysis data of IgG, IgA and IgM glycopeptides in plasma samples of PSC and PBC patients and healthy donors are shown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] definition As used herein, the following words and phrases are generally intended to have the meanings set forth below, unless the context in which they are used indicates otherwise.

[0016] It should be noted that, as used in this specification and the claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise.

[0017] The term "biological sample" refers to any biological fluid, cell, tissue, organ, or portion thereof. It also includes, but is not limited to, tissue sections obtained by biopsy, or cells in tissue culture, or cells adapted for tissue culture. It also includes, but is not limited to, saliva, tears, sputum, sweat, mucus, feces, gastric juice, abdominal fluid, amniotic fluid, cyst fluid, peritoneal fluid, spinal fluid, urine, synovial fluid, whole blood, serum, plasma, pancreatic juice, breast milk, lung lavage fluid, bone marrow, etc.

[0018] The term "biomarker" refers to a characteristic biological indicator or biologically derived indicator of a process, event, or condition. A biomarker also indicates a particular biological state, such as the presence of a disease or condition, or the risk of a disease or condition. It includes biological molecules, or fragments of biological molecules, the change or detection of which can be correlated with a particular physical state or condition. Examples of biomarkers include, but are not limited to, biomolecules including nucleotides, amino acids, fatty acids, steroids, antibodies, hormones, steroids, peptides, proteins, carbohydrates, and the like. Further examples include glycosylated peptide fragments, lipoproteins, and the like.

[0019] The term "comprising" is intended to mean that the compositions and methods include the recited methods, but do not exclude other methods.

[0020] The term "glycan" refers to the carbohydrate moiety of a glycoconjugate such as a glycopeptide, glycoprotein, glycolipid, or proteoglycan.

[0021] The term "glycoform" refers to the unique primary, secondary, tertiary, and quaternary structures of a protein to which glycans of a particular structure are attached.

[0022] The term "glycosylated peptide fragment" refers to a glycosylated peptide (or glycopeptide) having an amino acid sequence that is the same as some, but not all, of the amino acid sequence of a glycosylated protein, and which is obtained by fragmentation (e.g., by one or more proteases) to obtain the glycosylated peptide.

[0023] The term "multiple reaction monitoring mass spectrometry (MRM-MS)" refers to a sensitive and selective method for targeted quantification of proteins / peptides in biological samples. Unlike traditional mass spectrometry, MRM-MS is highly selective (targeted), allowing researchers to fine-tune the instrument to look specifically for peptides / protein fragments of interest. MRM increases the sensitivity, specificity, speed, and quantitation of peptides / protein fragments of interest, including promising biomarkers. MRM-MS uses triple quadrupole (QQQ) mass spectrometers or quadrupole time-of-flight (qTOF) mass spectrometers.

[0024] The term "protease" refers to an enzyme that proteolyzes or breaks down proteins into smaller polypeptides or amino acids. Examples of proteases include serine proteases, threonine proteases, cysteine ​​proteases, aspartic acid proteases, glutamic acid proteases, metalloproteases, aspartic peptide lyases, and combinations thereof.

[0025] The term "subject" refers to a mammal. Non-limiting examples of mammals include humans, non-human primates, mice, rats, dogs, cats, horses, or cows. Non-human mammals can be advantageously used as subjects that represent animal models of disease, pre-disease, or pre-disease conditions. The subject can be male or female. The subject can be a person previously identified as having a disease or condition, and optionally has undergone or is undergoing a therapeutic intervention for the disease or condition. Alternatively, the subject can be a subject that has not been previously diagnosed as having a disease or condition. For example, the subject can be a subject that exhibits one or more risk factors for a disease or condition, or a subject that does not exhibit risk factors for a disease, or a subject that is asymptomatic of a disease or condition. The subject can also be a subject that is suffering from or at risk of developing a disease or condition.

[0026] The term "treatment" or "treating" refers to any treatment of a disease or condition in a subject, such as a mammal, and includes: 1) preventing or protecting against a disease or condition, i.e., not causing clinical symptoms to develop; 2) inhibiting the disease or condition, i.e., preventing or inhibiting the development of clinical symptoms; and / or 3) alleviating the disease or condition, causing regression of clinical symptoms.

[0027] method The present disclosure relates in some embodiments to glycoproteomics, glycoproteomics for biomarker discovery, target discovery, and validation using advanced LC / MS instruments. The present disclosure utilizes machine learning methods to process molecular data. The analysis further includes utilizing genomic data, proteomics, metabolic, lipidomics data, or a combination thereof in discovering new biomarkers for various diseases. A general schematic diagram of the method of the present disclosure is shown in FIG. 1.

[0028] The present disclosure provides a method for site-specific glycosylation analysis that leads to the identification of new types of biomarkers with higher sensitivity and specificity. The method involves the quantification of glycosylated peptides, thus facilitating the differential analysis of different glycoforms associated with a particular protein at different sites. The method provides information on the protein amount and site-specific glycosylation profile, and therefore provides insight into whether changes in glycosylation profile are due to changes in protein glycosylation or changes in protein concentration. Site-specific glycosylation analysis combined with machine learning methods provides for the identification of new biomarkers for various diseases or conditions.

[0029] The quantitative glycoproteomics methods of the present disclosure are used to discover biomarkers for various diseases. These methods are based on the fact that certain glycoforms are elevated and others are downregulated in some diseases, and the LC / MS methods of the present disclosure distinguish disease versus no disease by analyzing significant glycosylation changes. In one embodiment, site-specific glycosylation analysis involves identifying the glycoprotein of interest, the modification site, what the modification is, and then measuring the relative abundance of each modification. In some embodiments, the disease is cancer. In other embodiments, the disease is an autoimmune disease.

[0030] Using the methods of the present disclosure, biological samples from thousands of subjects are digitized, generating vast amounts of data that are then subjected to deep machine learning analysis to discover new targets for various diseases. Specifically, deep learning is used to compare the clustering of known and unknown peptides with glycosylation signatures seen by LC / MS in disease and control conditions. Such discriminant analysis of glycosylated peptides leads to the identification of disease biomarkers.

[0031] The identification of biomarkers and their corresponding functions, such as their expression levels, are then used to develop diagnostic test methods for diseases or conditions. The methods rely, at least in part, on measuring one or more selected biomarkers and analyzing the results for association with a disease or condition. The methods can further be used to select one or more treatments, determine treatment regimens, or monitor response to treatment of a particular disease or condition. Thus, the present disclosure provides methods for the prevention, diagnosis, treatment, monitoring, and prognosis of diseases or conditions. In some embodiments, the methods are useful for distinguishing subjects with a disease or condition from healthy subjects. In some embodiments, the methods are useful for distinguishing subjects with cancer from healthy subjects. In some embodiments, the methods are useful for aiding in the diagnosis of cancer or for monitoring cancer.

[0032] Targeted and Non-Targeted Approaches The biomarker discovery methods of the present disclosure use both targeted and / or non-targeted approaches. The methods generally consist of three distinct phases: a discovery phase, a pre-validation phase, and a validation phase.

[0033] Discovery Phase The targeted approach involves identifying and monitoring known glycoproteins with known glycoforms in a subject's biological sample. FDA-approved glycoprotein biomarkers are known for various diseases, which can be monitored using the disclosed methods to identify a subject's classification. Typically, changes in the glycosylation of the biomarkers are tumor-specific and help identify likely risks of disease or disease stages. The targeted approach focuses on known glycoproteins and their glycoforms, whose biological significance has been established at the beginning of the study before data collection is performed, chemically characterized and biologically annotated. Quantification is performed using internal and chemical standards.

[0034] Specifically, in a targeted approach, site-specific glycosylation analysis is performed in biological samples from a case-control study of a large number of subjects with a disease or condition and an equal number of matched control subjects without the disease or condition. Glycoproteins of interest, such as disease-associated glycoproteins or glycoproteins with biological activity, are first identified in the biological sample. LC / MS is then used to analyze the site of modification, the nature of the modification, the identity of the modification, and the relative abundance of each modification, leading to identification and quantification of peptide fragments. In this approach, a triple quadrupole (QQQ) mass spectrometer is used to quantitate glycosylated peptide fragments and analyze them with respect to subject classification.

[0035] The non-targeted approach involves learning the glycosylation patterns of known and unknown peptide fragments, providing detailed information about changes in glycosylation patterns that help identify the classification of the subject. The non-targeted approach is based on relative quantification techniques that provide the "up or down regulation" of glycoproteins. Specifically, the up or down regulation of glycoproteins is monitored in relation to the classification of the subject. For example, glycoprotein fragments are monitored for subjects with a disease or pathology and for subjects without a disease or pathology. In this approach, the chemical identity of each glycoprotein fragment before data is acquired is unknown. In one embodiment, the non-targeted approach uses a quadrupole time-of-flight (qTOF) mass spectrometer for the analysis of glycosylated peptide fragments. In this approach, the instrument is used to accurately measure the masses of components in a sample without any preconceptions about what these components are.

[0036] Candidates that are differentially expressed between groups (disease vs. no disease) are selected for further evaluation using machine learning methods that allow prediction of important clinical features and classification by feature selection techniques. Performance is evaluated using internal cross-validation, which selects features and builds a model using the training set. The resulting model is evaluated on a test set that was not used to build the model. False positive rates are controlled using the false discovery rate (FDR) approach introduced by Benjamin and Hochberg.

[0037] Pre-validation phase Candidate biomarkers thus identified in the discovery phase are then tested in an independent test set of biological samples obtained from a number of subjects with the disease or condition and their matched controls without the disease or condition to determine the performance of the candidate biomarkers. The selected biomarkers, their ranking, and the parameter estimates of the model developed in the discovery phase are all part of the modeling and are tested in this independent pre-validation phase. Depending on the signal of the candidate biomarkers, the diagnostic test classifies the biological samples into two groups, a diseased group and a disease-free group. The usefulness of the test is then evaluated based on the positive predictive value, negative predictive value, specificity, and sensitivity. Diagnostic performance is also evaluated using a receiver operating characteristic (ROC) curve to test which biomarker or combination of multiple biomarkers is a statistically superior diagnostic test for the disease or condition. Individual biomarkers that are successfully validated are examined for subsequent incorporation to form a panel of composite markers. The composite markers are constructed by weighted multivariate logistic regression or other classification algorithms.

[0038] Validation Phase Candidate biomarkers retained in the pre-validation phase are then validated through independent validation using independent blinded biological samples from a large number of subjects. The purpose of this phase is to evaluate the diagnostic accuracy of the selected biomarkers.

[0039] In one embodiment, the biomarker discovery method is applied to biological samples obtained from subjects with cancer. In some embodiments, biological samples from at least 20, at least 40, at least 60, at least 80, or at least 100 subjects are analyzed in each group (i.e., with cancer or without cancer).

[0040] Both targeted and / or non-targeted approaches, together with the machine learning methods described herein, provide new diagnostic methods for identifying possible risks and / or early stage detection of various diseases.In one embodiment, the present disclosure provides a method for identifying biomarkers based on the convergence of targeted and non-targeted approaches combined with machine learning methods.The biomarkers identified by the methods of the present disclosure are useful in diagnostic methods, methods of prognostic evaluation, monitoring of treatment results, identifying subjects who may respond to a particular treatment, drug screening, etc.

[0041] In one embodiment, the disclosure provides a method for identifying glycosylated peptide fragments as potential biomarkers, the method comprising: fragmenting glycosylated proteins with one or more proteases in each of a plurality of biological samples isolated from a subject to generate glycosylated peptide fragments; Quantifying the glycosylated peptide fragments by liquid chromatography and mass spectrometry (LC-MS), providing a quantitative result; analyzing the quantitative results together with the classification of interest using machine learning methods to select glycosylated peptide fragments that are useful for predicting the classification; and determining the identity of the glycosylated peptide fragment.

[0042] In another embodiment, the disclosure provides a method as described herein, wherein the subjects include subjects having a disease or condition and subjects not having a disease or condition. In a further embodiment, the subjects include subjects receiving treatment for a disease and subjects with a disease but not receiving treatment for the disease.

[0043] The method of the present disclosure is applicable to any disease or condition that can be detected by analyzing glycosylated peptide fragments from a biological sample of a subject. In one embodiment, the disease is cancer. In another embodiment, the cancer is selected from breast cancer, cervical cancer, or ovarian cancer. In another embodiment, the disease is an autoimmune disease. In another embodiment, the autoimmune disease is HIV, primary sclerosing cholangitis, primary biliary cirrhosis, or psoriasis.

[0044] In another embodiment, the disclosure provides a method as described herein, wherein the glycosylated protein is one or more of alpha-1-acid glycoprotein, alpha-1-antitrypsin, alpha-1B-glycoprotein, alpha-2-HS-glycoprotein, alpha-2-macroglobulin, antithrombin-III, apolipoprotein B-100, apolipoprotein D, apolipoprotein F, beta-2-glycoprotein 1, ceruloplasmin, fetuin, fibrinogen, immunoglobulin (Ig) A, IgG, IgM, haptoglobin, hemopexin, histidine-rich glycoprotein, kininogen-1, serotransferrin, transferrin, vitronectin, and zinc-alpha-2-glycoprotein. In another embodiment, the glycosylated protein is one or more of alpha-1-acid glycoprotein, immunoglobulin (Ig) A, IgG, or IgM.

[0045] In another embodiment, the disclosure provides a method as described herein, wherein the glycosylated peptide fragment is N-glycosylated or O-glycosylated.

[0046] In another embodiment, the disclosure provides a method as described herein, wherein the glycosylated peptide fragments have an average length of about 5 to about 50 amino acid residues. In some embodiments, the glycosylated peptide fragments have an average length of about 5 to about 45, or about 5 to about 40, or about 5 to about 35, or about 5 to about 30, or about 5 to about 25, or about 5 to about 20, or about 5 to about 15, or about 5 to about 10, or about 10 to about 50, or about 10 to about 45, or about 10 to about 40 or about 10 to about 35, or about 10 to about 30, or about 10 to about 25, or about 10 to about 20, or about 10 to about 15, or about 15 to about 45, or about 15 to about 40, or about 15 to about 35, or about 15 to about 30, or about 15 to about 25, or about 15 to about 20 amino acid residues. In one embodiment, the glycosylated peptide fragments have an average length of about 15 amino acid residues. In another embodiment, the glycosylated peptide fragments have an average length of about 10 amino acid residues. In another embodiment, the glycosylated peptide fragments have an average length of about 5 amino acid residues.

[0047] In another embodiment, the present disclosure provides a method as described herein, wherein the one or more proteases include any protease used for protein fragmentation. In one embodiment, the protease is a serine protease, a threonine protease, a cysteine ​​protease, an aspartic acid protease, a glutamic acid protease, a metalloprotease, an asparagine peptide lyase, or a combination thereof. Some representative examples of proteases include, but are not limited to, trypsin, chymotrypsin, endoproteinase, Asp-, Arg-C, Glu-C, Lys-C, pepsin, thermolysin, elastase, papain, proteinase K, subtilisin, clostripain, and carboxypeptidase. In another embodiment, the present disclosure provides a method as described herein, wherein the one or more proteases include at least two proteases.

[0048] The disclosed method has several additional applications. For example, one or more biomarkers can help distinguish between pre-disease and disease states, or between disease and normal states. Other non-disease specific health conditions can also be determined. For example, changes in biomarkers can be analyzed at various time points. In subjects with disease, disease progression can be monitored, in subjects undergoing treatment, the effectiveness of treatment can be monitored, and in subjects after treatment, the possibility of recurrence can be monitored. The level of a particular amount of biomarkers can also select a course of treatment for a disease. For example, a biological sample can be provided from a subject undergoing a treatment plan for a disease. Such treatment regimens include, but are not limited to, exercise regimens, dietary supplements, weight loss, surgical intervention, device implantation, and treatment with therapeutic or prophylactic agents used in subjects diagnosed or identified with a disease or condition.

[0049] Furthermore, changes in the glycopeptide ratios of multiple glycoproteins may be associated with a particular disease state or the absence of disease. For example, the presence of multiple specific glycopeptides in a biological sample may indicate the absence of disease, while the presence of multiple other specific glycopeptides in a biological sample may indicate the presence of disease. Thus, different glycopeptide profiles or panels of glycopeptide biomarkers may correlate with different states of disease.

[0050] Example 2 shows the quantification of changes in IgG1, IgG0 and IgG2 glycopeptides in plasma samples from breast cancer patients versus controls. Figure 2 shows that levels of glycopeptides A1 and A2 were elevated compared to controls, whereas levels of glycopeptides A8, A9 and A10 were decreased compared to controls in all stages of breast cancer studied in this experiment, indicating that glycopeptides A1, A2, A8, A9 and A10 are promising biomarkers for breast cancer.

[0051] Example 3 shows the quantitative results of changes in IgG, IgM and IgA glycopeptides in plasma samples from patients with PSC and those without PSC. Figure 3 shows that glycopeptide A was elevated in plasma samples from patients with PBC and PSC compared to healthy donors, whereas glycopeptides H, I and J were decreased in plasma samples from patients with PBC and PSC compared to healthy donors. Thus, glycopeptides A, H, I and J are promising biomarkers for PBC and PSC. Furthermore, Figures 4 and 5 show the individual and combined analysis results of the discriminant analysis, respectively, suggesting an accuracy of 88% for predicting disease status in the combined discriminant analysis.

[0052] In some embodiments, the disclosure provides methods in which the number of biomarkers detected and analyzed is one, or more than one, such as 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, 30 or more. Thus, the disclosure also provides panels of biomarkers useful for diagnosing a disease or condition.

[0053] mass spectrometry In one embodiment, the disclosure provides a method as described herein, which includes quantifying glycosylated peptide fragments by using a mass spectrometer. In one embodiment, the method uses a technique called "multiple reaction monitoring (MRM)". Often, this technique is combined with liquid chromatography (LC / MRM-MS) to quantify hundreds of glycosylated peptide fragments (and their parent proteins) in a single LC / MRM-MS analysis. The advanced mass spectrometry techniques of the disclosure provide detectors with efficient ion sources, higher resolution, faster separation, and higher dynamic range, allowing a wide range of untargeted measurements that retain the advantages of targeted measurements.

[0054] The mass spectrometry method of the present disclosure is applicable to several glycosylated proteins at once, for example, at least 50 or more, or at least 60 or more, or at least 70 or more, or at least 80 or more, or at least 90 or more, or at least 100 or more, or at least 110 or more, or at least 120 or more glycosylated proteins can be analyzed at once using a mass spectrometer.

[0055] In one embodiment, the mass spectrometry method of the present disclosure uses a QQQ or qTOF mass spectrometer. In another embodiment, the mass spectrometry method of the present disclosure provides data with a resolution of 5,000 or more, or 10,000 or more, or 25,000 or more, or 50,000 or more, or 100,000 or more, and a high mass accuracy of 10 ppm or more, or 5 ppm or more, or 2 ppm or more, or 1 ppm or more, or 0.5 ppm or more, 0.2 ppm or more, or 0.1 ppm or more.

[0056] Biological samples The present disclosure provides a method based on quantification of glycosylated peptide fragments from a biological sample. In some embodiments, the biological sample is one or more clinical samples collected in the past, thus reducing the resources and time that must be devoted to the identification of new biomarkers. In some embodiments, the biological sample is from one or more previous studies that have occurred over a span of 1-50 years or more. In some embodiments, the studies are accompanied by various other clinical parameters and known information such as the subject's age, height, weight, ethnicity, medical history, etc. Such additional information helps to associate the subject with a disease or condition. In some embodiments, the biological sample is one or more clinical samples collected prospectively from the subject.

[0057] In one embodiment, the present disclosure provides a method as described herein, wherein the biological sample isolated from the subject is one or more of saliva, tears, sputum, sweat, mucus, feces, gastric fluid, peritoneal fluid, amniotic fluid, cyst fluid, peritoneal fluid, spinal fluid, urine, synovial fluid, whole blood, serum, plasma, pancreatic juice, breast milk, lung lavage fluid, and bone marrow. In another embodiment, the biological sample isolated from the subject is body tissue, saliva, tears, sputum, spinal fluid, urine, synovial fluid, whole blood, serum, or plasma. In another embodiment, the biological sample isolated from the subject is whole blood, serum, or plasma. In some embodiments, the subject is a mammal. In other embodiments, the subject is a human.

[0058] disease The method of the present disclosure is applicable to any disease or condition that can be detected by analyzing glycosylated peptide fragments from a biological sample of a subject. In some embodiments, the disease or condition is cancer. In other embodiments, the cancer is acute lymphocytic leukemia (ALL), acute myeloid leukemia (AML), adrenal cortical carcinoma, anal cancer, bladder cancer, blood cancer, bone cancer, brain cancer, breast cancer, cancer of the female reproductive system, cancer of the male reproductive system, central nervous system lymphoma, cervical cancer, childhood rhabdomyosarcoma, childhood sarcoma, chronic lymphocytic leukemia (CLL), chronic myeloid leukemia (CML), colon and rectal cancer, colon cancer, endometrial cancer, endometrial sarcoma, esophageal cancer, eye cancer, gallbladder cancer, stomach cancer, gastrointestinal cancer, hairy cell leukemia, head and neck cancer, hepatocellular carcinoma, Hodgkin's disease, hypopharyngeal cancer, Kaposi's sarcoma, kidney cancer, laryngeal cancer, leukemia, liver cancer , lung cancer, malignant fibrous histiocytoma, malignant thymoma, melanoma, mesothelioma, multiple myeloma, myeloma, nasal cavity and paranasal sinus cancer, nasopharyngeal cancer, nervous system cancer, neuroblastoma, non-Hodgkin's lymphoma, oral cavity cancer, oropharyngeal cancer, osteosarcoma, ovarian cancer, pancreatic cancer, parathyroid cancer, penile cancer, pharyngeal cancer, pituitary tumors, plasma cell neoplasms, primary CNS lymphoma, prostate cancer, rectal cancer, respiratory system, retinoblastoma, salivary gland cancer, skin cancer, small intestine cancer, soft tissue sarcoma, gastric cancer, testicular cancer, thyroid cancer, urinary tract cancer, uterine sarcoma, vaginal cancer, vasculature, Waldenstrom's macroglobulinemia, Wilms' tumor, etc. In another embodiment, the cancer is breast cancer, cervical cancer, or ovarian cancer.

[0059] In another embodiment, the disease is an autoimmune disease. In another embodiment, the autoimmune disease is acute disseminated encephalomyelitis, Addison's disease, agammaglobulinemia, age-related macular degeneration, alopecia areata, amyotrophic lateral sclerosis, ankylosing spondylitis, antiphospholipid syndrome, antisynthetase syndrome, atopic allergy, atopic dermatitis, autoimmune anemia, autoimmune cardiomyopathy, autoimmune enteropathy, autoimmune hemolytic anemia, autoimmune hepatitis, autoimmune inner ear disease, autoimmune lymphoproliferative syndrome, autoimmune peripheral neuropathy, autoimmune pancreatitis, autoimmune polyendocrine syndrome, autoimmune progesterone dermatitis, autoimmune hemoglobinopathy ... plaque-reduced purpura, autoimmune eczema, autoimmune uveitis, Baro's disease / Baro's concentric sclerosis, Behçet's disease, Berger's disease, Bickerstaff encephalitis, Blau syndrome, bullous pemphigoid, cancer, Castleman's disease, celiac disease, Chagas disease, chronic inflammatory demyelinating polyneuropathy, chronic relapsing ifocal osteomyelitis, chronic obstructive pulmonary disease, Churg-Strauss syndrome, cicatricial pemphigoid, Cogan's syndrome, cold agglutinin disease, complement component 2 deficiency, contact dermatitis, cranial arteritis, CREST syndrome, Crohn's disease, Cushing's syndrome, cutaneous leukocytoclastic vasculitis, Dego's disease skin disease, dermatitis herpetiformis, dermatomyositis, type 1 diabetes mellitus, diffuse cutaneous systemic sclerosis, Dressler syndrome, drug-induced lupus, discoid lupus erythematosus, eczema, endometriosis, tendonitis-associated arthritis, eosinophilic fasciitis, eosinophilic gastroenteritis, epidermolysis bullosa acquisita, erythema nodosum, erythroblastosis fetalis, essential mixed cryoglobulinemia, Evan's syndrome, progressive fibrous dysplasia, fibrosing alveolitis, gastritis, gastrointestinal pemphigoid, glomerulonephritis, Goodpasture's syndrome, Graves' disease, Guillain-Barre syndrome, Hashimoto's myopathy, Hashimoto's Enoch-Schönlein purpura, HIV, pemphigoid gestationis, suppuration Hidradenitis ulcerans, Hughes-Stobin syndrome, hypogammaglobulinemia, idiopathic inflammatory demyelinating disease, idiopathic pulmonary fibrosis, idiopathic thrombocytopenic purpura, IgA nephropathy, inclusion body myositis, chronic inflammatory demyelinating polyneuropathy, interstitial cystitis, juvenile idiopathic arthritis, Kawasaki disease, Lambert-Eaton myasthenic syndrome, leukocytoclastic vasculitis, lichen planus, lichen sclerosus, linear IgA disease, lupus erythematosus, Majeed syndrome, Meniere's disease, microscopic polyangiitis, mixed connective tissue disease, morphea, Mucca-Habermann disease, multiple sclerosis, myasthenia gravis, myositis, narcolepsy, neuromyelitis optica,Neuromyotonia, Ocular cicatricial pemphigoid, Opsoclonus myoclonus syndrome, Ord's thyroiditis, Palindromic rheumatism, Streptococcus-associated pediatric autoimmune neuropsychiatric disorders, Paraneoplastic cerebellar degeneration, Paraneoplastic paroxysmal nocturnal hemoglobinuria, Parry-Romberg syndrome, Parsonage-Turner syndrome, Parsplanitis, Pemphigus vulgaris, Pernicious anemia, Venous encephalomyelitis, POEMS syndrome, Polyarteritis nodosa, Polymyalgia rheumatoid arthritis, Polymyositis, Primary biliary cirrhosis, Primary sclerosing cholangitis, Progressive inflammatory neuropathy, Psoriasis, Psoriatic arthritis, Pyoderma gangrenosum, Pure red cell aplasia, Rasmussen syndrome In another embodiment, the autoimmune disease is HIV, primary sclerosing cholangitis, primary biliary cirrhosis, or psoriasis. ...

[0060] Machine Learning Biological samples are obtained from thousands of subjects. These subjects are used for digitization for deep mining and validation of previously undiscovered markers. In some embodiments, the biological samples are tumor or blood samples. These are digitized using LC / MS instruments to generate vast amounts of data, which are then subjected to deep machine learning analysis to discover new targets for various diseases. In some embodiments, the disease is cancer or an autoimmune disease.

[0061] In one embodiment, the disclosure provides a method for identifying glycosylated peptide fragments as potential biomarkers, the method comprising: fragmenting glycosylated proteins with one or more proteases in each of a plurality of biological samples isolated from a subject to generate glycosylated peptide fragments; Quantifying the glycosylated peptide fragments by liquid chromatography and mass spectrometry (LC-MS), providing a quantitative result; analyzing the quantitative results together with the classification of interest using machine learning methods to select glycosylated peptide fragments that are useful for predicting the classification; and determining the identity of the glycosylated peptide fragments, where the machine learning approach is deep learning, neural network, linear discriminant analysis, quadratic discriminant analysis, support vector machine, random forest, nearest neighbor, or a combination thereof. In some embodiments, the machine learning approach is deep learning, neural network, or a combination thereof. The analysis further comprises genomic data, proteomics, metabolism, lipidomics data, or a combination thereof. Figure 1 is a schematic diagram showing the integration of glycomics, LC / MS, and machine learning, further combined with proteomics, genomics, lipidomics, and metabolism to identify biomarkers for various diseases. EXAMPLES

[0062] Example 1 Common methods for biomarker discovery In a targeted approach, glycoproteins of interest are first identified in biological samples, and then LC / MS is used to analyze the site of modification, the nature of the modification, the identity of the modification, and the relative abundance of each modification for peptide fragment identity and peptide fragment quantification. This approach uses a triple quadrupole (QQQ) mass spectrometer to quantify glycosylated peptide fragments and analyze their relationship to the classification of interest.

[0063] In the non-targeted approach, the glycosylation patterns of all peptide fragments (known and unknown) are analyzed to obtain information on the changes in glycosylation patterns of different subjects. Specifically, the up-regulation or down-regulation of glycoproteins is monitored in relation to the classification of subjects. For example, glycoprotein fragments are monitored for subjects with disease or pathology versus subjects without disease or pathology. In this approach, a quadrupole time-of-flight (qTOF) mass spectrometer is used to analyze glycosylated peptide fragments.

[0064] Example 2 Quantification of IgG glycopeptides as potential biomarkers for breast cancer Plasma samples from breast cancer patients at various stages of cancer and age-matched controls were analyzed for IgG1, IgG0, and IgG2 glycopeptides and their ratio changes were compared. Specifically, 20 samples from Tis stage, 50 samples from EC1 stage, 25 samples from EC2 stage, 25 samples from EC3 stage, 9 samples from EC4 stage, and 73 age-matched control samples were subjected to MRM quantitative analysis on a QQQ mass spectrometer. As can be seen from the quantitative results in Figure 2, when the levels of specific IgG1 glycopeptides were evaluated compared to the controls, the levels of specific IgG1 glycopeptides were decreased compared to the controls in all stages of breast cancer studied in this experiment. For example, when IgG1 glycopeptides named A1-A11 were monitored, the levels of glycopeptides A1 and A2 were elevated compared to the controls, whereas the levels of glycopeptides A8, A9, and A10 were decreased compared to the controls in all stages of breast cancer studied in this experiment. Therefore, glycopeptides A1, A2, A8, A9 and A10 are promising biomarkers for breast cancer.

[0065] Example 3 Quantification of IgG glycopeptides as potential biomarkers for PSC and PBC Plasma samples from patients with primary sclerosing cholangitis (PSC), patients with primary biliary cirrhosis (PBC), and plasma samples from healthy donors were analyzed for IgG1 and IgG2 glycopeptides to compare changes in glycopeptide ratios. Specifically, 100 PBC plasma samples, 76 PSC plasma samples, and plasma samples from 49 healthy donors were subjected to MRM quantitative analysis on a QQQ mass spectrometer. As can be seen from the quantitative results in Figure 3, certain IgG1 glycopeptides were elevated compared to healthy donors, whereas certain IgG1 glycopeptides were decreased compared to controls in plasma samples from patients with PBC and PSC. For example, glycopeptide A was elevated compared to healthy donors, whereas glycopeptides H, I, and J were decreased compared to healthy donors in plasma samples from patients with PBC and PSC. Thus, glycopeptides A, H, I, and J are promising biomarkers for PBC and PSC.

[0066] A similar analysis was performed on IgA and IgM glycoproteins in plasma samples from patients with PBC and in plasma samples from patients with PSC. The results of the discriminant analysis are shown in Figure 4. It shows that the accuracy with which predictions can be made based on the individual data for IgG, IgM, and IgA is 59%, 69%, and 74%, respectively. However, when the results for all IgG, IgM, and IgA are combined, the discriminant analysis gives an accuracy of about 88%, as shown in Figure 5.

Claims

1. 1. A method for identifying glycosylated peptide fragments as potential biomarkers, comprising: fragmenting glycosylated proteins with one or more proteases in each of a plurality of biological samples isolated from a subject to generate glycosylated peptide fragments; quantitating the glycosylated peptide fragments by liquid chromatography and mass spectrometry (LC-MS), providing a quantitative result; analyzing said quantification results together with a classification of said subject with a machine learning method to select glycosylated peptide fragments useful for predicting said classification; determining the identity of the glycosylated peptide fragment.

2. 10. The method of claim 1, wherein the subjects include subjects with a disease or condition and subjects without the disease or condition.

3. 3. The method of claim 1 or 2, wherein the subjects include subjects undergoing treatment for a disease and subjects having the disease but not undergoing treatment.

4. The method of claim 2 or 3, wherein the disease is cancer or an autoimmune disease.

5. The method of claim 4, wherein the disease is a cancer selected from breast cancer, cervical cancer, or ovarian cancer.

6. 6. The method of claim 5, wherein the disease is an autoimmune disease selected from HIV, primary sclerosing cholangitis, primary biliary cirrhosis, or psoriasis.

7. The method of any one of claims 1 to 6, wherein the glycosylated peptide fragment is N-glycosylated.

8. The method of any one of claims 1 to 7, wherein the glycosylated peptide fragment is O-glycosylated.

9. 9. The method of any one of claims 1 to 8, wherein the glycosylated protein is one or more of alpha-1-acid glycoprotein, alpha-1-antitrypsin, alpha-1B-glycoprotein, alpha-2-HS-glycoprotein, alpha-2-macroglobulin, antithrombin III, apolipoprotein B-100, apolipoprotein D, apolipoprotein F, beta-2-glycoprotein 1, ceruloplasmin, fetuin, fibrinogen, immunoglobulin (Ig) A, IgG, IgM, haptoglobin, hemopexin, histidine-rich glycoprotein, kininogen-1, serotransferrin, transferrin, vitronectin, zinc-alpha-2-glycoprotein.

10. 10. The method of claim 9, wherein the glycosylated protein is one or more of alpha-1-acid glycoprotein, immunoglobulin (Ig) A, IgG or IgM.

11. The method according to any one of claims 1 to 10, wherein the glycosylated peptide fragments have an average length of 5 to 50 amino acid residues.

12. The method of any one of claims 1 to 11, wherein the one or more proteases comprises at least two proteases.

13. The method of any one of claims 1 to 12, wherein the mass spectrometry comprises multiple reaction monitoring mass spectrometry (MRM-MS).

14. The method according to any one of claims 1 to 13, wherein the biological sample is body tissue, saliva, tears, sputum, spinal fluid, urine, synovial fluid, whole blood, serum or plasma.

15. 15. The method of claim 14, wherein the biological sample is whole blood, serum or plasma.

16. The method of any one of claims 1 to 15, wherein the subject is a mammal.

17. 17. The method of claim 16, wherein the subject is a human.

18. 18. The method of any one of claims 1 to 17, wherein the machine learning approach is deep learning, neural networks, linear discriminant analysis, quadratic discriminant analysis, support vector machines, random forests, nearest neighbors or a combination thereof.

19. The method of any one of claims 1 to 18, wherein the machine learning approach is deep learning, neural networks, or a combination thereof.

20. The method of any one of claims 1 to 19, wherein the analysis further comprises genomic data, proteomic, metabolic, lipidomic data, or a combination thereof.

Images