Multiplexed volatile-releasing probes and uses thereof

Abstract

Compositions are described herein for the detection of the activity of a plurality of enzymes. The compositions comprise a plurality of molecular probes. Each molecular probe comprises a recognition domain / substrate structured to interact with a specific enzyme, a reporter molecule, and a linking group forming a covalent bond between the recognition domain and reporter molecule. For each of the plurality of molecular probes, upon interaction of the recognition domain with specific enzyme, the covalent bond is destroyed, rendering reporter molecule detectable by a chemical detection device. Methods of detecting enzymatic activity of a plurality of enzymes are also described herein. Such methods comprise reacting a composition, as described herein, with a plurality of enzymes, and identifying detectable reporter molecules with a chemical detection device.

Description

MULTIPLEXED VOLATILE-RELEASING PROBES AND USES THEREOFCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 550,957, filed February 7, 2024.STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OF DEVELOPMENT

[0002] This invention was made with Government support under contract R00EB028311 awarded by the National Institutes of Health (NIH). The Government has certain rights in this invention.FIELD OF THE INVENTION

[0003] The present disclosure generally relates to compositions and methods for multiplexed sensing of gastrointestinal (GI) enzyme activities for diagnosis of GI disorders in a patient. More specifically, the present disclosure generally relates to compositions and methods for detection of gastrointestinal (GI) disorders in a patient with the use of an orally-administered cocktail or combination of a plurality of molecular probes (also referred to as nanosensors or sensors), where each molecular probe is designed to degrade or decompose in the GI tract to generate a unique volatile organic compound (VOC) which can be detected in breath and quantified and correlated with the presence of one or more GI disorders in a patient.BACKGROUND

[0004] Endoscopic imaging and biopsy are the gold standard for the detection and clinical assessment of many gastrointestinal (GI) diseases including GI cancers, ulcers, inflammatory bowel disease, and disaccharidase deficiencies. However, patient discomfort, potential complications with anesthesia and bowel perforation, cost, and length of the procedure deter individuals from recommended testing. Furthermore, the invasivenature of endoscopy makes it an unsuitable tool for routine monitoring. Non-invasive diagnostics provide a more tenable path to early GI disease detection and effective disease management by reducing the logistical barriers to testing. Capsule endoscopy and blood and stool tests are non-invasive diagnostic tools currently used. However, the former, which involves swallowing a miniaturized camera to capture images of the GI tract, has a high miss rate for intestinal lesions, and the latter relies on blood or fecal biomarkers with poor sensitivity for early-stage disease or poor disease specificity. Therefore, early and accurate diagnosis of GI disease via non-invasive testing remains a significant challenge.

[0005] Of the many GI diseases, advances in early detection of colorectal cancer (CRC) is especially important as CRC is currently the second leading cause of cancer death in the United States. Current screening relies on tests for fecal biomarkers such as hemoglobin from intestinal bleeding and DNA epigenetic markers from cancerous lesions, which have good sensitivity for advanced- stage CRC but poor sensitivity for early-stage CRC. For that reason, direct visualization of the colon via colonoscopy remains the gold standard for early detection. However, the invasiveness of colonoscopy, patient discomfort during bowel prep, cost, and potential complications with anesthesia and bowel perforation deter individuals from completing recommended diagnostic testing. Currently, there is a 67% adherence rate for screening, and nearly 25% of cases are diagnosed when the cancer has spread to distant sites resulting in a greater than 70% drop in 5-year survival rate. In particular, adherence rates are disproportionately low for racial and ethnic minorities, who are more likely to complete screening when provided a non-invasive test option. Thus, there is an outstanding need for non-invasive diagnostic tests for sensitive detection of early-stage CRC.

[0006] Another focus in the treatment of GI diseases has been the development of diagnostic tools for the clinical management of inflammatory bowel disease (IBD), where early detection of flare-ups is needed to facilitate timely treatment. Furthermore, with increasing drug options, diagnostic tools are needed for early and frequent assessment of treatment response, and drug and dosing optimization to achieve shorter times to remission. Currently, clinical assessment of disease activity and treatment response relies on highly subjective symptom-based clinical scores (e.g. Crohn’s Disease Activity Index, CD Al) that correlate poorly with direct measures of disease activity acquired via endoscopy andhistopathology. Quantitative methods to monitor disease activity also exist, but rely on blood / fecal biomarkers of inflammation that are not IBD-specific (e.g. C-reactive protein, calprotectin, lactoferrin). For these reasons, direct visualization of the intestinal mucosa via endoscopy remains the gold standard for assessing disease activity and treatment response in IBD. However, the invasiveness of endoscopy, patient discomfort, cost, and technical challenges with reaching remote intestinal segments hinder its use as a routine monitoring tool for treatment response. Taken together, there remains a dire need for non- invasive methods that can be used for frequent and quantitative measures of disease activity.

[0007] Biomarker testing has potential utility for all stages of GI disease management - from risk assessment to initial diagnosis to prognosis and treatment monitoring. However, a critical limitation of current blood and stool tests is their reliance on single biomarkers, which indicate disease with poor specificity. Quantification of multiple biomarkers (i.e. a disease signature) can significantly improve the ability to discriminate between different diseases and levels of disease severity. However, biomarker signatures do not yet exist for many GI diseases, and conventional biomarker discovery is an arduous process that yields few clinically-validated biomarkers. Biomarker discovery involves -omic analysis of clinical samples, and its poor success rate can be attributed to poor standardization of sample collection or analysis methods, a trade-off between analytical throughput and accuracy, and increased likelihood of false “voo-doo” correlations due to overfitting models to datasets with a small ratio of subjects to tested variables.SUMMARY OF THE DISLCOSURE

[0008] Various aspects of the present disclosure are directed to ingestible chemical compounds that act as molecular probes to sense disease-associated enzyme (e.g., glycosidase and protease) activities in a patient. Activity -based signatures incorporating one or multiple enzymes, as described herein, is a novel and effective mechanism for detection of disease or query health status. Molecular probes according to various aspects of the disclosure have various diagnostic applications including disease detection, disease classification, disease prognosis, as a means to monitor treatment response and use as clinical endpoint.

[0009] In some instances, molecular probes according to various aspects of the disclosure are particularly useful for the detection of GI disorders in a patient such as, but not limited to, GI cancers (e.g., gastric cancer, colorectal cancer), inflammatory bowel disease (IBD), irritable bowel syndrome (IBS), disaccharidase deficiencies, ulcers, GI infections, gastroesophageal reflux disease (GERD), and Celiac disease. In some instances, molecular probes according to various aspects of the disclosure are particularly useful for the detection of non-GI disorders in a patient such as, but not limited to, cancers, infections, thrombosis and coagulopathies, autoimmune diseases, respiratory diseases, inflammation, injury, fibrosis, and wound healing.

[0010] Various non-limiting aspects of the present disclosure are as follows.

[0011] In some instances a first aspect of the disclosure can be described as a composition for the detection of the activity of a plurality of enzymes, the composition comprising a plurality of molecular probes, each molecular probe comprising a recognition domain / substrate structured to interact with a specific enzyme; a reporter molecule; and a linking group forming a covalent bond between the recognition domain and reporter molecule, wherein for each of the plurality of molecular probes, upon interaction of the recognition domain with specific enzyme, the covalent bond is destroyed, rendering reporter molecule detectable by a chemical detection device. In some instances, at least one of the plurality of molecular probes, of a composition according to the first aspect, comprises more than one recognition domain. In some instances at least one of the plurality of molecular probes, of a composition according to the first aspect, comprises more than one reporter molecule. In some instances, at least one of the plurality of molecular probes, of a composition according to the first aspect, comprises more than one recognition domain and more than one reporter molecule.

[0012] In some instances, a second aspect of the disclosure can be described as a composition according to the first aspect, wherein the reporter molecule is a volatile organic compound, which is optionally labeled with a radioactive / non-radioactive isotope, and is volatile after destruction of the linking group.

[0013] In some instances, a third aspect of the disclosure can be described as a composition according to the first or second aspect, wherein the enzyme is a glycosidase and the recognition domain is a carbohydrate substrate (a mono- / di- / poly-saccharide).

[0014] In some instances, a fourth aspect of the disclosure can be described as a composition according to any one of the first through third aspects, wherein the linking group is an O-, N-, or S- glycosidic bond.

[0015] In some instances, a fifth aspect of the disclosure can be described as a composition according to any one of the first through fourth aspects, wherein the linking group comprises a self-immolative group.

[0016] In some instances, a sixth aspect of the disclosure can be described as a composition according to any one of the first through fifth aspects, wherein the recognition domain / substrate is covalently bound to a biomolecule (for example, peptide / proteins, lipids, carbohydrates, and nucleic acids), a polymer (for example, polyethylene glycol and dextran), or a nanoparticle scaffold nanoparticle scaffold (for example, iron oxide nanoparticles, gold nanoparticles, and porous silicon nanoparticles).

[0017] In some instances, a seventh aspect of the disclosure can be described as a composition according to any one of the first through sixth aspects, wherein the molecular probes are encapsulated in nanoparticles or microparticles.

[0018] In some instances, an eighth aspect of the disclosure can be described as a method of detecting enzymatic activity of a plurality of enzymes, the method comprising reacting a composition according to any one of the first through seventh aspects with the plurality of enzymes, and identifying detectable reporter molecules with a chemical detection device.

[0019] In some instances, a ninth aspect of the disclosure can be described as a method according to the eighth aspect, wherein reacting the composition with the plurality of enzymes is performed in vivo.

[0020] In some instances, a tenth aspect of the disclosure can be described as a method according to the eighth or ninth aspect, wherein a breath sample is collected from a subject administered the composition to quantify the detectable reporter molecules.

[0021] In some instances, an eleventh aspect of the disclosure can be described as a method according to any one of the eighth through tenth aspects, wherein the composition is administered to a subject by an oral route, inhalation, an intravenous route, a subcutaneous route, an intramuscular route, an intraperitoneal injection route, an ocular route, asublingual route, a topical route, an aural route, a rectal route, or via an implanted device or an applied device (for example, a microneedle patch, an osmotic pump).

[0022] In some instances, a twelfth aspect of the disclosure can be described as a method according to the eighth aspect, wherein reacting the composition with the plurality of enzymes is performed ex vivo (for example, with enzymes in tissue or blood samples collected from a human or animal subject, or with enzymes in a water sample, a plant sample, or other environmental sample) and a reaction solution or headspace is analyzed for the detectable reporter molecules.

[0023] In some instances, a thirteenth aspect of the disclosure can be described as a method according to the eighth aspect, wherein reacting the composition with the plurality of enzymes is performed in vitro (for example, with recombinant / purified enzymes or enzymes in a mammalian tissue culture, a microbial culture, a water sample, a plant, an environmental sample, or a bioreactor) and a reaction solution or headspace is analyzed for the detectable reporter molecules.

[0024] In some instances, a fourteenth aspect of the disclosure can be described as a method according to any one of the eighth through thirteenth aspects, wherein the chemical detection device is a mass spectrometer, an infrared spectrometer, an ion mobility spectrometer, an electronic nose, a breathalyzer device, a colorimetric VOC sensor array, a microfluidic device, a human or animal (for example, canine) nose, an engineered microbial sensor, or any combination thereof.

[0025] In some instances, a fifteenth aspect of the disclosure can be described as a method according to any one of the eighth through fourteenth aspects, wherein the identity and abundance of the detectable reporter molecules is indicative of a disease or health status.

[0026] In some instances, a sixteenth aspect of the disclosure can be described as a method according to any one of the eighth through fifteenth aspects, wherein the identity and abundance of the detectable reporter molecules is used to monitor a disease progression.

[0027] In some instances, a seventeenth aspect of the disclosure can be described as a method according to any one of the eighth through sixteenth aspects, wherein the identity and abundance of the detectable reporter molecules is used for a disease prognosis.

[0028] In some instances, an eighteenth aspect of the disclosure can be described as a method according to any one of the eighth through seventeenth aspects, wherein theidentity and abundance of the detectable reporter molecules is used to monitor a treatment response and, optionally, used as a clinical endpoint.

[0029] In some instances, a nineteenth aspect of the disclosure can be described as a as a method according to any one of the eighth through eighteenth aspects wherein the identity and abundance of the detectable reporter molecules is used to monitor microbial contamination in the environment (for example, in water sources, natural vegetation) or food sources (for example, crops and livestock).BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order that the present disclosure may be readily understood, aspects of the vial adapter are illustrated by way of examples in the accompanying drawings, in which like parts are referred to with like reference numerals throughout.

[0031] FIG. 1 is a schematic illustration of an ingestible molecular probe according to various aspects of the disclosure before and after cleavage by an enzyme (for example, a glycosidase).

[0032] FIG. 2 is a schematic illustration of exemplary ingestible molecular probe according to various aspects of the disclosure for use as glycosidase probes.

[0033] FIG. 3 is a schematic illustration showing various steps for determining the presence of one or more GI disorders in a patient using ingestible molecular probes according to various aspect of the disclosure.

[0034] FIG. 4 is a schematic illustration of an ingestible molecular probe according to various aspects of the disclosure before and after cleavage by a protease.

[0035] FIG. 5A is a schematic illustration showing the use of ingestible molecular probes in in vitro assays to assess probe activity. As described in Example 1, the probes are reacted with a glycosidase, and the reaction headspace is analyzed using a mass spectrometer to quantify volatilized reporters generated from probe cleavage.

[0036] FIG. 5B is a graphical display showing signal counts over time of volatile reporter molecules tested in Example 1 and other intestinal glycosidases.

[0037] FIG. 5C is a graphical display showing signal counts of volatile reporter molecules tested in Example 1 after reaction for 1 hour with their respective glycosidase targets versus other intestinal glycosidases.

[0038] FIGS. 6A-C are graphical displays showing signal counts of volatile reporter molecules tested in Example 2 for [3-D-glucuronidase (GUS) activity (FIGS. 6A and B) or a-D-glucosidase activity (FIG. 6C).

[0039] FIG. 7A is a schematic illustration showing the experimental timeline for mouse breath study of Example 3. ADGlc-vl24 probes were delivered into fasted mice via oral gavage, and 5 breath samples were collected from 0-1, 1-2, 2-3, 3-4, and 4-5 hours.

[0040] FIG. 7B is a graphical display showing reporter signals in breath samples measured using mass spectrometry in Example 3. Each line represents breath signal over time for one mouse (n = 5). The greatest breath signal coincides with the time during which the probe is in the cecum and large intestine, the portions of the GI tract containing most of the gut microbiome.

[0041] FIG. 7C is a graphical display comparing summed breath signals from 0-3 hours after probe administration in healthy control mice and antibiotic-treated mice in Example 3. Antibiotic-treated mice have depleted microbiome. Therefore, this comparison was used to determine the contribution of the microbiome to intestinal a-D-glucosidase activity (n = 4-5). **Indicates p < 0.01.

[0042] FIG. 8A is a schematic illustration showing the experimental timeline for a second mouse breath study in Example 3. Here, to deplete the microbiome, one of the experimental groups was treated with an antibiotic cocktail daily for 4 days before the breath study. Molecular probes were administered via oral gavage into mice, and 3 breath samples were collected per mouse from 0-1, 1-2, and 2-3 hours after probe administration.

[0043] FIG. 8B is a graphical display comparing breath signal with and without antibiotic treatment to determine the contribution of the microbiome to intestinal P-D-galactosidase activity (n = 3-5) for the second mouse breath study in Example 3.

[0044] FIG. 9A is a schematic illustration of a GUS sensor synthesized via conjugation of glucuronic acid (GlcA) to a VOC reporter molecule (deuterated ethanol, m / z 52) in Example 4.

[0045] FIG. 9B is a graphical display showing quantification of the VOC reporter molecule via mass spectrometry in Example 4, which confirmed that deuterated ethanol is released upon sensor cleavage by active GUS (n = 3).

[0046] FIG. 9C is an IVIS image of GUS activity (Example 4) in the mouse large intestine. Signal was generated by oral delivery of a fluorogenic probe for GUS activity and appeared 2-5 hours after oral gavage of the probe, the time during which the probe traffics through the large intestine.

[0047] FIG. 9D is a graphical display of breath signal data collected hourly after oral delivery of molecular probes in healthy mice in Example 4. Elevated breath signal was observed from 2-4 h after molecular probe (“nanosensor”) delivery. Elevated breath signal was suppressed by oral administration of a GUS inhibitor, demonstrating that breath signal is specifically driven by intestinal GUS activity.

[0048] FIG. 10a is schematic illustration of a general structure of a molecular probe according to various aspects of the disclosure, specifically a peptide-VOC conjugate with a self-immolative (SI) linking group, and protease-triggered VOC release.

[0049] FIG. 10b is a schematic illustration of a molecular probe according to various aspects of the disclosure, specifically peptide-VOC conjugate as in FIG. 10a, designed to sense FAP activity and the workflow for assessing function. The conjugate was reacted with FAP in a VOA vial, and the reaction solution and headspace were sampled and analyzed via mass spectrometry to identify the cleavage products.

[0050] FIG. 10c is a graphical display of a MALDI-MS analysis of a reaction solution (Example 9) confirming cleavage of the amide bond between the peptide and SI linker.

[0051] FIG. lOd is a graphical display of a PTR-MS analysis of a reaction headspace confirming (Example 9) volatilization of the liberated VOC reporter with the addition of FAP.

[0052] FIG. lOe is a graphical display of an FAP concentration-dependent VOC signal (Example 9) produced by the conjugate (n = 3).

[0053] FIG. lOf is a graphical display of a VOC signal specifically generated by FAP activity (Example 9) over that of other proteases that can be active in the intestine (n = 3).

[0054] FIG. 1 la is a schematic illustration of four molecular probes used in Example 10, each having reporter molecules with a distinct molecular mass. The 4-plex was reacted with either a single protease, as illustrated in FIG. 1 lb, or with a combination of proteases, as illustrated in FIG. 11c, and the signal from released VOC mass barcodes was measured via mass spectrometry. At a given acquisition time, a single sample is being analyzed, andthe signal across the four mass channels represent the abundance of each reporter in that sample.

[0055] FIG. 12 provides (Left) a schematic illustration of a sucrase sensor synthesized via conjugation of a-D-glucopyranoside (aDGlc) to a VOC reporter molecule (guaiacol, m / z 125) and (Right) a graphical display confirming sucrase-triggered reporter release via an in vitro cleavage study in which the VOC glycoside was reacted with sucrase, and vaporized reporter in the headspace was quantified using mass spectrometry from 0-30 min after sucrase addition (n = 3).DETAILED DESCRIPTION

[0056] The following description of the embodiments is merely exemplary in nature and is not intended to limit the subject matter of the present disclosure, their application, or uses.

[0057] It is noted that, as used in this specification and the appended claims, the singular forms “a,” “an,” and “the,” include plural references unless expressly and unequivocally limited to one referent. As used herein, the term “include” and its grammatical variants are intended to be non-limiting, such that recitation of items in a list is not to the exclusion of other like items that can be substituted or added to the listed items. For example, as used in this specification and the following claims, the terms “comprise” (as well as forms, derivatives, or variations thereof, such as “comprising” and “comprises”), “include” (as well as forms, derivatives, or variations thereof, such as “including” and “includes”) and “has” (as well as forms, derivatives, or variations thereof, such as “having” and “have”) are inclusive (i.e., open-ended) and do not exclude additional elements or steps. Accordingly, these terms are intended to not only cover the recited element(s) or step(s), but may also include other elements or steps not expressly recited. Furthermore, as used herein, the use of the terms “a” or “an” when used in conjunction with an element may mean “one,” but it is also consistent with the meaning of “one or more,” “at least one,” and “one or more than one.” Therefore, an element preceded by “a” or “an” does not, without more constraints, preclude the existence of additional identical elements.

[0058] Various aspects of the present disclosure are directed to ingestible chemical compounds that act as molecular probes to sense disease-associated enzyme (e g.glycosidase and protease) activities in a patient. Activity-based signatures incorporating one or multiple enzymes, as described herein, is a novel and effective mechanism for detection of disease or query health status. Molecular probes according to various aspects of the disclosure have various diagnostic applications including disease detection, disease classification, disease prognosis, as a means to monitor treatment response and use as clinical endpoint. Ingestible chemical compounds according to various aspects of the disclosure can be administered to a subject in various ways including, but not limited to, orally, via inhalation, intravenously, subcutaneously, intramuscularly, intraperitoneally, ocularly, sublingually, topically, aurally, rectally, or via an implanted or applied device (for example, a microneedle patch or an osmotic pump).

[0059] In some instances, molecular probes according to various aspects of the disclosure are particularly useful for the detection of GI disorders in a patient, such as, but not limited to, GI cancers (e.g., gastric cancer, colorectal cancer), inflammatory bowel disease (IBD), irritable bowel syndrome (IBS), disaccharidase deficiencies, ulcers, GI infections, gastroesophageal reflux disease (GERD), and Celiac disease.

[0060] In some instances, molecular probes according to various aspects of the disclosure are particularly useful for the detection of non-GI disorders in a patient, such as, but not limited to, cancers, infections, thrombosis and coagulopathies, autoimmune diseases, respiratory diseases, inflammation, injury, fibrosis, and wound healing.

[0061] In some instances, molecular probes according to various aspects of the disclosure are particularly useful for the detection (both identity and abundance) of microbial contamination in an environment (for example, in water sources and natural vegetation) or food sources (for example, crops and livestock).

[0062] Various forms of chemical detection devices may be used for detection of molecular according to the disclosure including, but not limited to, mass spectrometers, infrared spectrometers, ion mobility spectrometers, electronic noses, breathalyzer devices, colorimetric VOC sensor arrays, microfluidic devices, human or animal (for example, canine) noses, engineered microbial sensors, or any combination thereof.

[0063] Various aspects of the present disclosure are directed to ingestible chemical compounds that act as molecular probes to sense disease-associated enzyme (e.g. glycosidase and protease) activities in the GI tract and produce an amplified, mass-encodeddisease signature for non-invasive detection in breath. Breath tests are non-invasive, rapid, and can be completed with ease at great frequency and in a variety of settings for monitoring applications. Herein, we describe diagnostic platforms to induce exhalation of biomarker signatures that reflect disease activity in the GI tract. For this, we leverage intestinal enzyme activities. The GI tract houses a diverse repertoire of enzymes from the host and microbiome that play significant roles in human health and disease. Proteases and glycosidases catalyze the breakdown of proteins and carbohydrates for digestion, drug metabolism, host immunity, and tissue repair and remodeling. At the molecular scale, dysregulated enzyme activities contribute to GI disease pathology and occur early in disease development, in advance of changes in tissue morphology or clinical symptoms. Multiplexed sensing of intestinal enzyme activities can enable sensitive and specific detection of early-stage GI disease. Activity-based signatures incorporating one or both of proteases and glycosidases, as described herein, is a novel and effective mechanism for GI disease detection.

[0064] Glycosidases are enzymes that hydrolyze glycosidic bonds to break down carbohydrates (i.e. sugars) into smaller subunits. The human GI tract contains thousands of distinct human and microbial glycosidases that are important for digestion, drug metabolism, host immunity, and tissue repair and remodeling. Aberrant intestinal glycosidase activity is a hallmark of many GI diseases. Therefore, functional readouts of intestinal glycosidase activity have potential utility as diagnostic biomarkers of GI disease.

[0065] Various aspects of the disclosure are directed to ingestible molecular probes that traffic through the GI tracts and release volatile organic compounds (VOCs) upon degradation by specific glycosidases. Various aspects of the disclosure are also directed to ingestible molecular probes that traffic through the GI tracts and release volatile organic compounds (VOCs) upon degradation by specific proteases. The VOC reporters (or “volatile reporter molecules”) are subsequently eliminated from the body in breath and can be quantified via various chemical detection devices and methodologies such as, for example, mass spectrometry. This strategy enables GI disease detection via breath testing - a non-invasive alternative to endoscopy and biopsy.

[0066] Ingestible molecular probes according to various aspects of the disclosure can include three components for the detection of glycosidase activity. A first component is arecognition domain or recognition molecule (for example, a carbohydrate or peptide) that has a chemical structure configured to bind to an active site of a specific enzyme such as a glycosidase or protease. Various recognition domains such as, for example, a carbohydrate substrate such as a mono- / di- / poly-saccharide, a peptide substrate, amino acids, lipids, nucleic acids, phosphate groups, sulfate groups, ester groups to sense activity of aminopeptidases, lipases, nucleases, phosphatases, sulfatases, or esterases can be used. A second component is a reporter molecule that is covalently bound to the recognition domain. A third component is a linking group bonding the recognition domain and the reporter molecule. When the molecular probe is intact (i.e., the recognition domain is bound to the reporter molecule via the linking group), the reporter molecule is not volatile and is undetectable. When the linking group of the molecular probe is destroyed, however, the reporter molecule is converted to a volatile reporter molecule (or VOC). The now volatile reporter molecule exhibits a various properties such as a characteristic molecular mass. As the volatile reporter molecules are volatile and have characteristic molecular masses, they can be detected in various analytical detection devices such as, for example, but not limited to, mass spectrometers, infrared spectrometers, ion mobility spectrometers, electronic noses, breathalyzer devices, colorimetric VOC sensor arrays, microfluidic devices, human or animal (for example, canine) noses, engineered microbial sensors, or any combination thereof. Like naturally-occurring volatile compounds produced in the body, volatile reporter molecules will be exhaled after diffusing from the intestinal lumen into blood circulation, followed by pulmonary gas exchange. Concentrations of individual volatile reporter molecules in breath can be quantified (via, for example, mass spectrometry) and used to build a classifier of disease using machine learning. Altogether, catalytic processing of volatile-releasing probes by intestinal enzymes will produce an amplified, mass-encoded disease signature for non-invasive detection in breath.

[0067] In some instances, a molecular probe according to various aspects of the disclosure includes a single recognition domain, a single reporter molecule, and a single linking group. In some instances, a molecular probe according to various aspects of the disclosure can include a plurality of recognition domains such as, for example 2-10 distinct recognition domains. In some instances, a molecular probe according to various aspects of the disclosure can include a plurality of reporter molecules and corresponding linkinggroups such as, for example 2-10 distinct reporter molecules. In some instances, a molecular probe according to various aspects of the disclosure can include a plurality of recognition domains and a plurality of reporter molecules.

[0068] In some instances, the linking group is made of a single atom. In some instances the single atom is an oxygen, nitrogen or sulfur atom. Upon destruction of the single atom linking group, the single atom forms part of the volatile reporter molecule such as a hydroxyl (-OH) group, an amine (-NH2) group or a thiol (-SH) group. In some instances, the linking group may include a self-immolative group. Representative molecular probe structures with an exemplary self-immolative group are provided in Formulae (I) and (II) as follows:S M clative link r Thlchfe volatile (II)

[0069] In exemplary Formulae (I) and (II), both contain a peptide as the recognition domain. Formula (I) contains a reporter molecule with a sulfur atom bound to the self- immolative group, which results in the formation of a volatile reporter molecule comprising a thiol group after destruction of the self-immolative group. Formula (I) contains a reporter molecule with an oxygen atom bound to the self-immolative group, which results in the formation of a volatile reporter molecule comprising a hydroxyl group after destruction of the self-immolative group.

[0070] In some instances, the linking group is a glycosidase-cleavable covalent bond. FIG. l is a schematic illustration showing an exemplary ingestible molecular probe before andafter cleavage by an active intestinal enzyme, specifically a glycosidase. In some instances, the linking group is a protease-cleavable covalent bond.

[0071] There are thousands of distinct glycosidases in the GI tract. Exemplary ingestible molecular probes designed to sense a-D-glucosidase, P-D-galactosidase, and -D- glucuronidase activity in the intestine are illustrated in FIG. 2. In the small intestine, the brush border disaccharidases - sucrase-isomaltase and lactase - are responsible for majority of a-D-glucosidase and P-D-galactosidase activity. In disaccharidase deficiencies, the absence of sucrase-isomaltase and lactase results in the inability to digest sucrose or lactose, respectively, leading to non-specific GI symptoms such as diarrhea, bloating, excess gas production, and abdominal pain and possible malnutrition. B-D-glucuronidase (GUS) is a microbial glycosidase that reactivates formerly metabolized drugs. Drug reactivation can cause GI toxicity. These are just a few examples that illustrate the role of glycosidases in GI health and disease.

[0072] During GI disease, intestinal glycosidase activities are altered. Thus, breath readouts for intestinal glycosidase activity have potential utility as biomarkers for GI disease. In some instances, a single molecular probe according to various aspects of the disclosure can be logically synthesized and structured to detect the activity of a particular glycosidase and can be orally administered for production of single biomarkers. In some instances, multiple molecular probes can be logically synthesized and structured such that each can detect the activity of a different glycosidase and can be co-administered in a mixture for production of a biomarker signature. Biomarkers can be used for disease detection, for treatment prognosis, to monitor treatment response, and used as clinical endpoints for treatment / clinical trials. For example, disaccharidase deficiencies are currently treated via disaccharidase replacement therapies (e.g. pre-prandial ingestion of lactase pills for those who are lactose intolerant). Glycosidase probes can be used to detect disaccharidase deficiencies as well as monitor the restoration of disaccharidase activity via breath testing.

[0073] There are over 600 distinct human proteases. Proteases catalyze proteolysis, breaking down proteins into smaller polypeptides or single amino acids, and spurring the formation of new protein products. Proteases do this by cleaving the peptide bonds withinproteins via hydrolysis. Proteases are involved in numerous biological pathways, including digestion of ingested proteins, protein catabolism and cell signaling.

[0074] FIG. 3 is a schematic illustration showing various steps for determining the presence of one or more GI disorders in a patient using ingestible molecular probes according to various aspect of the disclosure. In FIG. 3, the recognition domain (or “enzyme substrate”) can be, for example, a carbohydrate when the molecular probe is for use to detect glycosidase activity. Alternatively, the recognition domain can be, for example, a peptide when the molecular probe is for use to detect protease activity.

[0075] In some instances, molecular probes according to various aspects of the disclosure can be used to sense glycosidase and / or protease activities in other tissues in locations of the body other than the GI tract via alternative administration routes (e.g. intravascular, intraperitoneal, subcutaneous injections). When the molecular probes are released in tissues and cleaved, the resulting volatile reporter molecules will diffuse into blood circulation and will be exhaled after pulmonary gas exchange. Thus, the molecular probes disclosed herein can also be used to generate breath biomarkers for non-GI diseases.

[0076] In some instances, molecular probes according to various aspects of the disclosure can be mixed with biological samples (e.g. pure glycosidases and / or proteases, bacterial / mammalian cell cultures, cell / tissue homogenates and supernatants) to assay for enzymatic activity. Probe cleavage can be monitored by measuring the concentration of volatile reporters molecules in, for example the sample headspace using a mass spectrometer. Potential applications include microbial pathogen identification, in-line monitoring for cell manufacturing processes, and for tissue engineering.

[0077] Alternative molecular probes according to various aspects of the disclosure are also envisioned herein. In some instances, the recognition domain / substrate of molecular probes according to the disclosure may be covalently bound to a biomolecule (for example, a peptide / protein, lipid, carbohydrate, or nucleic acid), a polymer (for example, polyethylene glycol, dextran), or a nanoparticle scaffold (for example, iron oxide, gold, porous silicon nanoparticles). For example, in some instances, molecular probes may comprise a plurality of reporter molecules bound to a biocompatible polymeric nanoparticle (or “nanocarrier”) by peptide linking groups (FIG. 4). When reporter molecules are bound to the nanocarrier, they exist in a non-volatile pro-form that cannot be detected. Upon cleavage of the peptidelinking groups by target proteases, the reporter molecules are released and recover their characteristic mass and volatility (i.e., are converted to volatile reporter molecules) for mass-based detection in vapor form.

[0078] Molecular probe engineering as an alternative to biomarker discovery. Diverse biological species (e.g. proteins, nucleic acids, metabolites, circulating tumor cells) have been identified as biomarkers of disease. However, reliance on these naturally-occurring biomarkers for disease identification is limiting due to fundamental biological and technical challenges such as low concentrations, noise from biological background, instability, and difficulties in their isolation from complex biological matrices (e.g. blood, stool). Moreover, biomarker type dictates the tools and methods used for biomarker quantification. Thus, there are limited degrees of freedom for optimizing biomarker diagnostic performance. Therefore, rather than rely on biomarker discovery to identify naturally-occurring biomarkers of GI disorders, the present invention allows for logical engineering of ingestible molecular probes de novo by leveraging known disease biology and newly-engineered enzyme-responsive molecules. Using that approach, we are no longer at the mercy of what Nature provides, but have greater control over factors contributing to biomarker signal-to-noise ratio such as (1) choice of volatile reporter molecule(s), (2) kinetics of biomarker(s) production via molecular probe design, and (3) probe dosing regimen(s). Furthermore, we can optimize for specificity by engineering multiplexed biomarkers. That approach allows one to fully leverage the benefits of breath testing in clinical diagnostics, as breath biomarker discovery has had limited success to date.

[0079] Plug-and-play chemistry to induce the exhalation of diverse VOCs in breath. A handful of prior art clinical breath tests for GI diseases rely on metabolic processing of ingested VOC precursors (small molecules or macromolecules) into volatile biomarkers for exhalation. These include the 13C-urea breath test for H. pylori infection, the 13C- methacetin breath test for liver function, and the small intestinal bacterial overgrowth (SIBO) breath test. These breath tests induce the exhalation of a limited range of VOCs in breath - 13C isotope-labeled carbon dioxide, hydrogen, and methane. In contrast to the VOC precursors used in existing breath tests, ingestible molecular probes according to various aspects of the disclosure are highly modular, and the volatile reporter moleculesformed therefrom are exchangeable. In accordance with various aspects of the disclosure, hydroxyl-, thiol- and amine-containing VOCs can be used as volatile reporter molecules and can be covalently bound to a recognition domain via a cleavable bond. We have already identified hundreds of VOCs with generally regarded as safe (GRAS) FDA designation containing such functional groups that can be incorporated into molecular probes according to various aspects of the disclosure. GRAS compounds are highly benign molecules, many of which are already used as food flavorings and should have minimal regulatory barriers to translation.

[0080] VOC-barcoded molecular probes for multiplexed glycosidase sensing & breath biomarkers. Existing prior art breath tests rely on single biomarkers, which provide limited disease specificity. Furthermore, while thousands of distinct volatile organic compounds (VOCs) are naturally exhaled in breath, identification of diseasespecific VOCs have been unsuccessful due to limitations with breath biomarker discovery methods. Using plug-and-play chemistry to generate a library of specific recognition domains (i.e., specific activity for certain glycosidases, proteases, disease-specific enzymes, etc.) covalently bound to correspondingly distinct reporter molecules (which are converted to volatile reporter molecules exhibiting, for example, distinguishable molecular masses) as described above, various and diverse arrays or panels of molecular probes can be generated to simultaneously query multiple types of enzymatic activity. Oral administration of the molecular probe arrays or panels result in mass-encoded, or barcoded, signatures with improved disease specificity.

[0081] In the context of CRC detection in a patient, multiplexed sensing of intestinal enzyme activities will be invaluable to generate a CRC-specific breath biomarker signature. To build multiplexing capabilities, molecular probes for different enzymes can be engineered, as described elsewhere herein, to release volatile reporter molecules of distinct molecular masses.

[0082] Multiplexed breath volatiles as disease-specific “breath fingerprints”. Intense efforts are currently underway by industry and academic researchers to identify diseasespecific “breath fingerprints”, which are comprised of multiple breath VOCs with a pattern in their relative abundance. The driving force behind this search is the need for greater disease specificity, beyond what single biomarkers can offer. Alkanes, for example, are asingle class of naturally-occurring breath volatiles that are quantified in breath tests for heart allograft rejection, and have poor diagnostic performance (78.6% sensitivity and 62.4% specificity in clinical studies). Alkanes are produced from peroxidation of lipid membranes during inflammation. As post-surgical inflammation can stem from numerous sources other than immune response against the allograft, specificity of this breath test is poor. Likewise, exhaled nitric oxide, is used to diagnose asthma and can be confounded by airway inflammation from smoking and allergies. To date, no “breath fingerprints” have been identified and clinically validated. However, we propose that “breath fingerprints” in the form of multiplexed volatile reporter molecules can be engineered by combining nanomedicine approaches with machine learning. In essence, the delivery and controlled release of volatile reporter cargos triggered by orthogonal enzymatic activities yields multidimensional datasets that can be used in supervised machine learning to develop disease classification algorithms. The ubiquitous role of dysregulated enzyme activities across a broad range of diseases combined with the ability of volatiles to traffic from different tissues to breath, suggests that this approach can be used to engineer multiplexed biomarkers for numerous other diseases beyond GI diseases.

[0083] Ultrasensitive detection via enzyme-catalyzed signal amplification. Breath biomarker signal is amplified via enzyme-driven reporter release. In this process, enzyme activity is not consumed, and a single enzyme can cleave thousands of molecular probes per hour, thus, driving signal amplification for sensitive tumor detection. Sensitivity is further enhanced through breath sampling methods that concentrate resulting volatile reporter molecules in large volumes of breath collected over time, which can be collected onto, for example, sorbent tubes for downstream analysis via mass spectrometry.

[0084] Example 1

[0085] In this example, glycosidase activity sensing was assessed in vitro using assays in which probes were incubated with recombinant or purified glycosidases in volatile organic analysis (VOA) vials (FIG. 5A). Analysis of the reaction headspace via mass spectrometry was used to quantify reporter release and volatilization. We confirmed that molecular probes - aDGlc-vl24, pDGal-vl40, and GlcA-vl52 (see FIG. 2) - react with their respective glycosidase targets, triggering the release of volatile reporter molecules into the reaction headspace (FIGS. 5B-C). When probes were incubated with other commonintestinal glycosidases, minimal reporter signal was observed (FIGS. 5B-C). Altogether, this data demonstrates that probes are highly glycosidase-specific. For optimal signal-to- noise ratio (SNR) in breath, glycosidase specificity is important to minimize noise from off-target probe degradation. Notably, probe reactions with glycosidases were also highly stereospecific. The aDGlc-vl24 probe only reacted with a-D-glucosidase and not with |3- D-glucosidase (FIG. 5B). A final important observation is that probes were not degraded by a-amylase (FIGS. 5B-C), a highly abundant intestinal glycosidase that has broad substrate specificity due to its role in carbohydrate digestion. These results suggest that non-specific degradation of probes during transit through the GI tract is unlikely.

[0086] Example 2

[0087] Using the same in vitro assays as in Example 1, we also confirmed that the glycosidase molecular probe structure is highly modular, allowing for easy exchange of the volatile reporters (FIG. 6A-B). Two probes for P-D-glucuronidase (GUS) were synthesized - GlcA-vl24 containing a guaiacol (molecular mass = 124) volatile reporter molecule and GlcA-vl52 containing a 4-ethylguaiacol (molecular mass =152 volatile reporter molecule (see FIG. 2). Both reporter molecules were covalently bound to the GUS recognition domain ( -D-glucuronic acid) via a glycosidic bond and were released into the reaction headspace when molecular probes were reacted with GUS (FIGS. 6A-B). The same guaiacol reporter in the GlcA-vl24 probe was used in the aDGlc-124 probe for a-D- glucosidase activity, demonstrating that probes for different glycosidase activities can be easily synthesized by simple exchange of the carbohydrate substrate (FIGS. 6A and C).

[0088] Example 3

[0089] The plug-and-play chemistry described herein is powerful not only for the ease of molecular probe design, but also for multiplexed sensing of intestinal glycosidase activities. Volatile reporter molecules of distinct mass can be used to barcode an array of carbohydrate substrates, and ingestion of the resulting probe panel can be used to produce a breath biomarker signature for GI disease. To assess the ability of molecular probes to induce exhaled readouts for intestinal glycosidase activity, aDGlc-vl24 probes were administered to healthy mice via oral gavage, and reporter signal in breath samples was measured using mass spectrometry (FIG. 7A). Breath was collected over 5 h in ~1 h increments after probe administration to monitor breath signal kinetics as the probeprogressed through different segments of the GI tract. Breath signal in probe-treated mice was consistently elevated above that of mice administered a vehicle control (FIG. 7B). Immediately after oral gavage, breath signal rose rapidly until the greatest signal was observed when the probes were present in the cecum and large intestine (FIG. 7B). These are the segments of the GI tract that house most of the microbiome. To determine if most of the probe metabolism was due to microbial a-D-glucosidase, a subsequent experiment was completed where molecular probes were administered into healthy control mice and mice pre-treated with antibiotics to deplete the microbiome. Without the microbiome, little to no volatile reporter molecules were measured in breath samples (FIG. 7C). These results indicate that the majority of molecular probe is metabolized by microbial a-D-glucosidase activity.

[0090] Similarly, administration of pDGal-v!40 probes into healthy mice resulted in elevated breath signal over that of mice administered vehicle controls, and breath signal was significantly reduced with microbiome depletion (FIG. 8). Thus, microbial P-D- galactosidase also contributes significantly to molecular probe metabolism.

[0091] Example 4

[0092] In this example, we determined that oral delivery of a sensor can be used to induce exhalation of reporters for intestinal enzyme activity. Our initial target was P-glucuronidase (GUS), a glycosidase secreted by the microbiome and overexpressed in cancer cells that contributes to tumorigenesis and tumor invasion in CRC. Microbial GUS reactivates carcinogenic compounds locally in the intestine via removal of glucuronic acid from glucuronide metabolites, and, thus, creates a source of genotoxicity for tumorigenesis. While mammalian GUS is a lysosomal enzyme, it is found to be overexpressed and secreted by tumor cells. GUS activity in CRC cell lines correlate with tumor invasiveness and is, therefore, thought to degrade ECM glycosaminoglycans for tumor invasion. Taken together, GUS is one of several promising enzyme targets we aim to harness for a CRC breath biomarker signature.

[0093] A GUS molecular probe (or “nanosensor”) was synthesized via conjugation of glucuronic acid (the GUS recognition domain) to deuterated ethanol (the volatile reporter molecule, a bio-orthogonal volatile that is not naturally found in breath) (FIG. 9A). We confirmed in vitro that these molecular probes release deuterated ethanol upon cleavage byactive GUS (FIG. 9B). Like humans, healthy mice possess baseline GUS activity in the large intestine, where the bulk of the microbiome resides (FIG. 9C). Similarly, mechanisms for VOC elimination in breath are conserved across humans and rodents, and clinical breath tests (e.g. for Helicobacter pylori infection in the stomach, gastric emptying, and intestinal sucrase deficiency) have conserved function in humans and rodent models of disease. Therefore, molecular probe delivery into healthy mice was used to assess breath signal from GUS-triggered volatile reporter molecule release. Based on imaging studies, the molecular probe travels through the large intestine 2-5 hours after oral delivery. During this window of time, normalized breath signal (ratio of breath signal in sensor-dosed mice to that of PBS-dosed mice) was elevated (FIG. 9D). Furthermore, elevated breath signal was suppressed via oral delivery of a small molecule GUS inhibitor (FIG. 9D). Taken together, these results demonstrate that the elevated breath signals were specifically driven by intestinal GUS activity. Thus, we have established the feasibility of engineering breath readouts for intestinal enzyme activity via orally-delivered sensors.

[0094] Example 5 - Building a molecular probe (“nanosensor”) panel for multiplexed sensing of IBP protease activity.

[0095] A significant feature of the present invention is multiplexing capability, where each nanosensor in a panel is designed to sense a different target protease and release volatile reporter molecules exhibiting distinct masses. Thus, nanosensors can be simultaneously delivered in ingestible formulations, and all reporters represented in the nanosensor panel can be measured in a single breath sample using mass spectrometry or other breath analysis devices. In past studies, disease signatures comprised of 14-20 proteases have been sufficient to detect disease with >90% sensitivity and specificity. Therefore, a 15-20-plex panel of molecular probes according to various aspects of the disclosure will be generated for detection of IBD flare-ups. This strategy should produce breath biomarker signatures that offer greater specificity for IBD disease activity than existing single biomarkers. Protease targets will be nominated based on literature search and analysis of transcriptional datasets. Exemplary protease targets are provided in Table 1 below.Table 1.

[0096] Proteases with altered activity in both human IBD and mouse models of IBD will be prioritized. Due to the interchangeability of the peptide linker in the nanosensor, protease specificity is easily tuned. To identify peptides cleaved by IBD proteases, a library of fluorogenic peptide substrates will be screened against recombinant proteases in 384- well plate format. Fluorescence from peptide cleavage will be measured over time using a plate reader. Hierarchical cluster analysis of cleavage rates will be used to eliminate redundant substrates and downselect for a peptide panel that is cleaved by orthogonal IBD proteases. The peptides will then be barcoded with volatile reporters of distinct mass. Volatile reporters will be covalently bound to peptides via self-immolative linkers, as described elsewhere herein. Self-immolative linkers can adjoin the C-terminus of peptide recognition domains to diverse functional groups frequently found in VOCs (e.g. hydroxyl, thiol, and amine groups). In this format, we have shown that peptide cleavage triggers the linker to self-immolate (i.e. self-destruct), thereby converting reporter molecules to corresponding volatile reporter molecules without any linker remnants that could reduce their volatility. Barcoded peptides will be purified via HPLC, and the product mass will beconfirmed with LC-MS. Protease-triggered volatile reporter molecule release will be confirmed by reacting barcoded peptides with their respective proteases and quantifying volatilized reporters in the reaction headspace using a proton transfer reaction-mass spectrometer (PTR-MS). Nanosensors will then be synthesized via surface conjugation of barcoded peptides onto nanocarriers. To minimize barriers to in-human testing, nanosensor components will be restricted to compounds safe for human consumption. This includes the use of VOCs with Generally Regarded As Safe (GRAS) FDA designation as reporter molecules. GRAS VOCs are used as food flavorings and are highly benign. To date, we have identified more than 200 GRAS VOCs that are compatible for use in accordance with various aspects of the disclosure. Other components include peptides and polyethylene glycol nanocarriers, which are components in diagnostic nanoparticles that have undergone successful phase I clinical trials for safety.

[0097] Example 6 - Formulation of molecular probes (nanosensors”) for oral delivery and characterize nanosensor activity in the GI tract.

[0098] An important consideration for in vivo molecular probe function is delivery of intact probes to the intestine. In some instances, encapsulation of the molecular probe(s) may be necessary for testing intestinal enzyme activity due to potential instability of linking groups in the acidic environment of the stomach. Similar to drugs that require protection from the acidic and proteolytic gastric environment, nanosensors will be encapsulated in Ph- responsive microparticle formulations using microspraying methods. Ph-responsive formulations will enable controlled release of nanosensors in the small intestine (which is more alkaline than the stomach in both humans and mice). In some instances according to this example, microparticle formulations will incorporate Eudragit®, a copolymer of methacrylic acid and ethylacrylate, which is insoluble at gastric Ph but soluble at intestinal Ph and therefore useful to protect drugs that are unstable in gastric fluid or to prevent off- target effects in the gastric mucosa. Spray-drying methods have been previously used to synthesize Ph-responsive microparticles incorporating Eudragit®. Such methods will be investigated and pharmacokinetic and biodistribution studies completed to determine the transit time and controlled release of probe formulations through the small and large intestine. Prior studies have shown that a number of breath tests have conserved functionin humans, mice and other rodent models due to overlapping disease biology and VOC elimination pathways. Therefore, we plan to validate our approach in mice.

[0099] Additionally, since intestinal neutrophil elastase (NE) activity is elevated during flare-ups in human and mouse IBD, previously-developed NE nanosensors will be used as an IBD-relevant model cargo to validate our delivery strategies. Stability of microformulated NE nanosensors will be evaluated in vitro in simulated gastric fluid, which models the acidic gastric Ph and contains pepsin, the most abundant gastric protease. After incubation with gastric fluids, volatile reporter molecules in the reaction headspace will be measured using PTR-MS. We expect nanosensors to remain encapsulated at low Ph, preventing non-specific nanosensor degradation by pepsin. Similarly, microformulations will be tested with simulated intestinal fluid containing NE to confirm nanosensor release at higher Ph and subsequent reporter release via NE cleavage. Optimal formulations will be moved forward into animal studies. Fluorescently-labeled formulations will be delivered orally in mice to characterize their pharmacokinetics through the GI tract. Trafficking time to the small and large intestine will be used to determine the window of time during which breath samples should be collected. Breath studies will then be completed in healthy control mice and DSS-induced colitis mouse models, the latter of which have elevated NE activity in the large intestine. After oral gavage of the microformulation, breath samples will be collected and analyzed using previously established methods. Mice will be placed in a chamber for hour-long increments and exhaled volatiles will be concentrated onto carbon beds in sorbent tubes by continuous pulling of the chamber headspace through sorbent tubes using a vacuum source. That will increase detection sensitivity and is a method used for human breath analysis. Thermal desorption will be used to release VOCs from sorbent tubes into the PTR-MS for analysis. Oral delivery of an NE inhibitor will be used to determine if breath signals are driven specifically by NE activity. Peak breath signal between 2-5 hours after oral gavage are expected, the window of time during which microformulations will traffic through the large intestine.

[0100] Example 7 - Development of a classifier for IBD flare-ups via multiplexed breath-based protease sensing.

[0101] IL-10 knockout mice are extensively-used IBD animal models that spontaneously develop colitis by 8 weeks of age. This model recapitulates many of the immunoregulatory and microbiome aspects of human IBD and also responds to anti-TNFa treatment. For these reasons, we will use IL-10 knockout mice, and develop IL-10 knockout models, for validation of our breath biomarker approach. To develop a protease-based classifier for IBD flare-ups, we will use a training and validation cohort approach. The nanosensor panel will be administered into a training cohort of wildtype mice and IL- 10 knockout mice after the onset of inflammation. Breath will be collected from 2-3, 3-4 and 4-5 hours after molecular probe (or “nanosensor”) administration, the anticipated window of time during which nanosensors transit the large intestine. Proton-transfer reaction-mass spectrometry (PTR-MS) will be used to measure reporter concentrations in breath samples. Using established machine learning algorithms (i.e. support vector machine, random forest, and regularized logistic regression), a classifier with maximum sensitivity and specificity for IBD flare-up will be built using the training cohort data. Classifier accuracy (as defined by area under the receiver operating characteristic curve (AUROC)) will be determined using a dataset generated in an independent validation cohort, which will include mice with non- IBD enteritis from infection to test specificity of the classifier for IBD. It is expected that a minimum sample size of 15 mice per experimental group per cohort will be needed to generate a classifier with >90% detection accuracy. Multiple timepoints during disease progression will be tested to identify the earliest timepoint in the disease model at which the detection sensitivity and specificity of engineered breath biomarkers exceed that of clinical blood and stool biomarkers. Histological grading of intestinal tissues using an established grading system will be completed to determine disease severity corresponding to the earliest point of detection. It is predicted that the classifier will be specific for IBD colitis over other sources of inflammation (such as infection) and will provide an indicator of IBD flare-up before severe disease (lesions affecting the majority of the colon, transmural inflammation with extensive mononuclear and neutrophil infiltration, crypt abscesses).

[0102] Example 8 - Combining multiplexed breath biomarkers with machine learning to establish classifiers of GI disease.

[0103] In this example, breath biomarker signatures will be established to non-invasively detect and monitor GI disease. To do so, inflammatory bowel disease (IBD) will again be used as our model GI disease. IBD is a debilitating, idiopathic disorder and is characterized by alternating periods of remission and relapse of chronic inflammation in the GI tract. Disease flare-ups are unpredictable, can last from days to months, and, if not treated in a timely manner, causes cumulative tissue damage over time that requires surgical intervention. Therefore, detection of subclinical inflammation before symptoms occur is needed for timely initiation of treatment with immunosuppressive therapies. Furthermore, invasive endoscopy is currently used to visualize intestinal healing to assess treatment response and clinical endpoints. Thus, IBD is a GI disease that could significantly benefit from non-invasive breath tests for early detection, treatment monitoring, and to provide quantitative endpoints for treatment. Machine learning can be used with our multiplexed breath biomarkers to generate a GI disease classifiers, and a probe panel will be developed for multiplexed sensing of IBD-relevant intestinal enzymes and tested in an IBD mouse model.

[0104] While the exact etiology of IBD is largely unknown and generally attributed to genetic factors, immune dysfunction and microbiome alterations, enzyme-mediated processes contribute significantly to IBD pathology. Enhanced tryptic protease activities in the colon, for example, initiate protease-activated receptor-2 (PAR-2) signaling pathways that lead to increased intestinal permeability and subsequent immune activation. Metalloproteases and serine proteases secreted locally by infiltrating immune cells, stromal cells, and epithelial cells degrade proteins in the extracellular matrix, driving intestinal tissue damage and remodeling. In addition to host enzymes, proteases and glycosidases from commensal microbes and enteric pathogens contribute to inflammation by breaking down the protective colonic mucosal layer and increasing intestinal permeability via disassembly of tight junction proteins. Differential activity of these and other enzymes during remission, subclinical inflammation, relapse, and treatment can be harnessed to generate multiplexed breath biomarkers to predict flare-up occurrences and to monitor treatment response. Exemplary enzyme targets and their respective molecular probe recognition domains (or “substrates”) are provided in Table 2 Additional targets will be identified via literature search and analysis of publicly-available transcriptional datasets.Table 2.

[0105] To generate data to train an IBD classifier for early detection of subclinical inflammation, probes will be administered via oral gavage into a dextran sulfate sodium (DSS)-induced colitis model, and breath samples will be collected from mice using previously established methodologies. Volatile reporter molecule(s) quantification in breath samples will be completed using mass spectrometry. The resulting dataset will consist of measured abundance for each reporter present in the probe panel. Using training cohort data, established machine learning algorithms (i.e. support vector machine, random forest, and regularized logistic regression) will be used to build IBD classifiers with maximum sensitivity and specificity for subclinical inflammation. Classifier accuracy (as defined by area under the receiver operating characteristic curve (AUROC)), sensitivity,and specificity will be determined using a dataset generated in an independent validation cohort. A classifier to monitor treatment response and endpoint will be similarly established by administering the probe panel into DSS colitis models that are treated with immunosuppressive anti-TNF antibodies and small molecule immunomodulators (e.g. methotrexate). Intestinal inflammation in DSS colitis mouse models will be characterized using histopathology to identify a timepoint for subclinical inflammation. However, the acute nature of the model means it may not provide a good model for subclinical inflammation. In this instance, IL- 10 knockout mice will be used. IL- 10 knockout mice are genetically-engineered models that spontaneously develop discontinuous and transmural patches of inflammation in the small intestine similar to what is observed in human IBD by 8 weeks of age and is an extensively-used model to study the immunoregulatory and microbiome aspects of IBD. IL-10 knockout mice also exhibit therapeutic response to anti- TNF antibody treatment similar to that of human patients. Thus, the more gradual onset of inflammation in this model and its response to common IBD treatments makes it a suitable alternative to generate IBD classifiers for early detection and treatment response.

[0106] Example 9 - VOC conjugation to peptides for protease sensing

[0107] Single biomarkers have poor disease specificity. Therefore, multiplexed sensing of intestinal enzyme activities is needed to generate a CRC-specific breath biomarker signature. To build multiplexing capabilities, nanosensors for different enzymes will be engineered to release volatile reporters of distinct mass. The enzyme sensing mechanism depends on reporter attachment to enzyme substrates (i.e. peptides, glycosides) via enzyme-cleavable bonds. Therefore, this aim will focus on establishing modular “plug- and-play” chemistries for attachment of volatile reporters to enzyme substrates. Additionally, non-specific breakdown of nanosensors during transit through the GI tract needs to be minimized to avoid high background signal. Therefore, this aim will also focus on developing oral formulations so that nanosensors can overcome delivery barriers such as the acidic and proteolytic gastric environment.

[0108] Example 9.1 - VOC conjugation to peptides for protease sensing

[0109] VOCs (volatile reporter molecules) that are covalently bound to the cleavage site of peptide substrates are released upon peptide cleavage. Liberated VOCs undergo phase transition into a gas to produce a signal quantifiable by mass spectrometry. In this example,linking group chemistries, specifically the use of self-immolative linkers, are established to incorporate diverse VOCs into peptide-VOC conjugates for multiplexing. Self- immolative linkers can adjoin the C-terminus of peptide substrates to functional groups frequently found in VOCs (e.g., hydroxyl and thiol groups), in addition to amine groups. In this format, peptide cleavage will trigger the linker to self-immolate (i.e. self-destruct), thereby releasing the VOC reporter without any linker remnants that could reduce its volatility (FIG. 10a). The feasibility of this approach has been confirmed by synthesizing a conjugate containing a peptide substrate for fibroblast activation protein a (FAP) and methyl salicylate (a safe-to-ingest, hydroxyl-containing VOC with a mass-to-charge (m / z) ratio of 153) (FIG. 10b). FAP is an ECM-remodeling protease with elevated expression in cancer-associated fibroblasts and helps with tumor cell invasion in CRC and other cancers. Elevated FAP expression has been associated with worse clinical outcomes, and the protease has garnered widespread attention as a promising pan-cancer biomarker and therapeutic target. Thus, FAP is a highly relevant protease target for CRC detection. Characterization of VOC reporter release from peptide-VOC conjugates is possible via in vitro cleavage assays, where conjugates are reacted with proteases in volatile organic analysis (VOA) vials, and gastight syringes are used to pierce the rubber septa caps for headspace sampling (FIG. 10b). For the FAP-sensing conjugate, reaction products confirmed cleavage of the amide bond between the FAP peptide substrate and the self- immolative linker (FIG. lOJc) and vaporization of the released methyl salicylate reporter (FIG. lOd). In vitro cleavage studies further confirmed FAP concentration-dependent reporter signal (FIG. lOe) and specific sensing of FAP activity over that of other proteases (FIG. 1 Of). In this example, we will build upon this preliminary work and systematically test conjugation of different VOC reporters to our model peptide substrate for FAP. By holding the substrate constant and exchanging the reporter, the modularity of this approach will be tested for VOC reporters with different functional groups and overall structure (aliphatic versus aromatic). Since VOCs are distanced from the cleavage site by the self- immolative linker, we expect that VOC chemical structure will have a lesser effect on Michaelis-Menten cleavage kinetics than if they were directly conjugated to the peptide. The reaction rate constant (kcat) and Kmwill be determined for each conjugate for comparison of catalytic efficiency. Exemplary VOC reporter molecules with the GenerallyRegarded as Safe (GRAS) FDA designation have been identified for testing (Table 3). GRAS compounds are commonly used as food flavorings and are, therefore, safe for ingestion and have well-characterized toxicity profiles. For future translation in humans, we expect patients to fast 12 hours before testing to minimize background signal from food volatiles, which is common practice for clinical breath tests.Table 3.

[0110] Example 9.2 - Multiplexing via VOC mass barcodes[0U1] Past studies have shown that 14 and 19-plex nanosensor panels provide sufficiently broad coverage of cancer-associated protease activity for early detection of lung and prostate cancers in mouse models with 100% sensitivity and 95% specificity. Therefore,using self-immolative linker chemistry, a 15-20-plex nanosensor panel will be created for CRC detection. In preliminary multiplexing studies, a 4-plex was synthesized by coupling 4 different peptide substrates for Proteases A, B, C, and D to 4 VOC reporters of distinct mass (FIG. I la). In in vitro cleavage assays, the 4-plex was either reacted with a single protease (FIG. 1 lb) or a combination of proteases (FIG. 11c). These studies show that the activity of multiple proteases could be monitored simultaneously via measurement of mass- encoded reporter signals in the reaction headspace. These preliminary studies confirm multiplexed sensing in vitro, which is a crucial milestone towards a CRC breath biomarker signature.

[0112] To achieve a larger panel of 15-20 nanosensors, 15-20 VOC reporters with optimal physicochemical properties for trafficking from the GI tract to the breath will be identified. Through review of publications from the Flavor Extract Manufacturers Association (FEMA), more than 200 GRAS compounds with suitable functional groups for coupling to self-immolative linkers have been identified and rank-ordered on their vapor pressure, logP (i.e. hydrophobicity), and airwater partition coefficients. The top 20 VOC reporter candidates are shown in Table 3. These will be assessed empirically via dosing of compound mixtures in mice via oral gavage and subsequent measurement of exhaled reporters at multiple timepoints from 5 min to Ih after dosing. VOCs traffic from their tissue of origin into breath in minutes in both humans and mice. In both species, VOCs in the GI lumen diffuse across the epithelium to enter systemic circulation and are exhaled following pulmonary gas exchange. Due to conserved VOC elimination pathways, breath tests have comparable function in humans and mouse models of disease. Therefore, mice are a suitable models to assess our approach. VOC pharmacokinetics are expected to differ based on their physicochemical properties. The magnitude of peak breath signal and time- to-peak signal will be used to identify VOCs that are readily eliminated from the body in breath versus those that are sequestered in tissues and fat. Breath samples will be collected and analyzed using methods we have previously established in which mice are placed in sealed chambers for a fixed time and the chamber headspace is subsequently displaced into evacuated glass vials for analysis via mass spectrometry. For increased detection sensitivity, breath volatiles will be concentrated onto carbon beds in sorbent tubes by continuous pulling of the chamber headspace through sorbent tubes using a vacuum source.The latter method was used in past studies (FIG. 9d). Using these methods, mouse breath can readily be analyzed to identify optimal VOC mass barcodes for multiplexing.

[0113] Example 9.3- VOC conjugation to carbohydrate substrates for glycosidase sensing

[0114] Combining protease and glycosidase activities can create highly specific biomarker signatures. This approach is currently used in the clinical laboratories, where protease and glycosidase activities are measured in vitro from patient-derived microbial cultures for species-level pathogen identification. In the context of CRC, the intestine contains both classes of enzymes and both contribute to tumorigenesis and tumor invasion. As described herein, nanosensors can be logically synthesized to leverage the activity of both types of enzymes for breath biomarkers. Thus, there is an opportunity to harness both intestinal proteases and glycosidases to generate a highly specific breath biomarker signature for CRC detection.

[0115] This synthesis of molecular probes containing sugar (i.e., carbohydrate) recognition domains for detection glycosidase activity (which may be referred to as “VOC glycosides”) for requires two main steps: (1) a nucleophilic substitution reaction and (2) a deprotection reaction. Briefly, reactions will start with a sugar containing protected hydroxyl groups and a halide leaving group at the anomeric carbon. Introduction of a nucleophilic VOC will trigger substitution of the halide group with the VOC. In this reaction, both hydroxyl- and thiol-containing VOCs will be used as nucleophiles to synthesize VOC glycosides. After substitution, protecting groups will be removed and glycosides will be purified using HPLC. Enantiomer products will be separated using chiral column chromatography, and stereochemistry of the collected fractions will be determined using NMR. Purity and percent yield of the desired intermediate(s) and product will be determined at each step.

[0116] As described in previous examples, VOC glycosides have been synthesized for glycosidase sensing. In the first example, a VOC reporter (deuterated ethanol, m / z 52.1) was conjugated to the P-glucuronidase substrate (glucuronic acid) (FIG. 9a). In a second example, a different VOC reporter (guaiacol, m / z 125.2) was conjugated to a sucrase substrate (a-D-glucopyranoside). In vitro cleavage assays showed that reaction of each VOC glucoside with their respective glycosidase resulted in VOC reporter release and vaporization (FIG. 12). More importantly, oral delivery of the former resulted in a breath signal for intestinal P-glucuronidase activity (FIG. 9d). Taken together, these results showthe feasibility of harnessing Intestinal glycosidase activity to generate breath biomarkers for CRC. In this aim, in vivo testing of VOC glycosides for sucrase sensing will further be explored. Sucrase-isomaltase, the glycosidase responsible for most intestinal sucrase activity, is normally restricted to the small intestine. However, sucrase-isomaltase expression is observed in the large intestine in pre-cancerous polyps and metastatic colon adenocarcinomas and is thought to contribute to cancer cell metabolism by providing glucose for glycolysis. Therefore, it is a potentially relevant target for CRC detection. The VOC glycoside will be orally delivered into healthy mice to determine if sucrase- isomaltase in the small intestine will cleave the glycoside and trigger the exhalation of reporters.

[0117] Example 9.4- Formulating nanosensors for oral delivery

[0118] In certain examples below, completed nanosensor panels will be tested in Apcmml+mice, the most widely-used CRC mouse model for early tumor development. In contrast to human CRC, Apcmm' mice develop tumors throughout the small intestine rather than in the large intestine. Therefore, we will develop nanosensor formulations that are actively sensing in the small and large intestine. For optimal si nal-to-noise ratio (SNR) of engineered CRC breath biomarkers, two things must occur: (1) signal must be maximized by on-target delivery of VOC-barcoded substrates to the intestine and (2) noise must be minimized by preventing non-specific substrate degradation.

[0119] To address the first point, nanosensors will be synthesized via conjugation of VOC- barcoded substrates to nanocarriers or polymers to modulate substrate biodistribution. The small intestine is a major site of small molecule and peptide absorption, containing a large surface area of enterocytes that mediate uptake. To maximize interaction of substrates with enzymes in the intestinal mucosa, VOC-barcoded substrates will be conjugated onto nanocarriers that avoid intestinal absorption and are retained in the intestinal lumen for transit through the small and large intestine. Several of these have been reported in the literature and include iron oxide nanoparticles, dextran-coated nanoparticles, and cellulose- based nanoparticles. Although long-term mucoadhesion is not desired, mucoadhesive polymers such as chitosans and carboxymethylcellulose will be investigated to determine if mucoadhesion increases breath signal. To identify the optimal nanocarrier, substrates will be conjugated to several types of nanocarriers and will be evaluated for uptake in vitrovia well-established transwell assays with Caco-2 cell monolayers and in vivo in biodistribution studies.

[0120] To minimize noise, non-specific breakdown of nanosensors in the GI tract must be prevented. Interestingly, glycosides can remain stable in transit till they encounter their respective glycosidases in the intestine. A fluorogenic glycoside for 0 -glucuronidase was administered via oral gavage in mice and fluorescence signal was only observed in the large intestine, where the glycosidase is produced by the microbiome (FIG. 9c). Elevated breath signal was also only observed during the window of time when the VOC glycoside for 0-glucuronidase was in the large intestine (FIG. 9d). However, this observation may be specific to these glycosides. Therefore, nanosensor stability will be characterized in well- established assays using simulated gastric and intestinal fluids. Simulated gastric and intestinal fluids recapitulate the PH and digestive enzymatic activity of their respective compartments. The former contains pepsin, while the latter contains porcine pancreatin. Nanosensors will be mixed with these fluids in vitro and reporter release in the headspace will be measured using mass spectrometry.

[0121] In general, microencapsulation of nanosensors may be necessary due to potential degradation in the stomach, and to a lesser extent, in the mouth. PH-responsive formulations are attractive as this strategy is already used in the pharmaceutical industry to deliver drugs to the intestine. Eudragit® is a commercially-available copolymer of methacrylic acid and ethylacrylate that is used in enteric release coatings for drugs that are unstable in gastric fluid. It is insoluble at low, gastric PH but soluble at higher, intestinal PH. Spray-drying methods will be used to synthesize PH-responsive Eudragit® microparticles to encapsulate nanosensors for delivery to the intestine. Eudragit formulations retain their function in the mouse GI tract. To test this approach, fluorogenic peptide substrates for gastric and intestinal digestive enzymes (pepsin and trypsin, respectively), will be used as model cargos in Eudragit microparticles. Mice will be euthanized at hourly timepoints after oral delivery and the GI tract will be imaged using an IVIS imager. A fluorescent signal in the small intestine and no signal in the stomach is expected due to site-specific substrate release. Alternatively, microemulsion strategies will be tested, which have been used for drug delivery to the intestine.

[0122] Example 9.5 Establishing optimal nanosensor dosing and characterize breath signal kinetics

[0123] Due to rapid VOC elimination pathways, VOC reporters that are released in the GI lumen are exhaled within minutes. Therefore, the timepoints for breath sampling are dictated by transit time of the nanosensors through the small and large intestine. The pharmacokinetics of microencapsulated nanosensors will be determined using fluorescently-labeled formulations. 1, 2, 3, 4, and 5 hours after oral delivery, mice will be euthanized, and the GI tract will be imaged using an IVIS imager. In a separate cohort, breath samples will be collected at the same timepoints after oral delivery of nanosensors. These studies will be first completed using previously-developed nanosensors for neutrophil elastase activity and mouse models of colitis, which have elevated neutrophil elastase activity in the intestine. A range of doses (10, 100, and 1000 nmol equiv. peptide) will be administered to identify the lowest dose needed to generate breath signals that are able to discriminate colitis models from healthy controls with AUROC > 90% and MTD will be determined via dose escalation studies. These studies will set a benchmark for dosing and sample timepoints.

[0124] Using the proposed linker chemistries, VOC barcoding of enzymatic substrates for multiplexed sensing of proteases and glycosidases will be achieved. Nanosensor synthesis via conjugation of barcoded substrates onto nanoparticles synthesized with dietary fibers, such as cellulose, will help with substrate retention in the GI tract to maximize interaction with intestinal enzymes. Our preliminary data indicates that glycosidase nanosensors can be stable during transit through the GI tract. However, we anticipate peptides in protease nanosensors will need protection from non-specific degradation by promiscuous digestive proteases such as pepsin, trypsin, and chymotrypsin. An alternative or supplemental strategy to microencapsulation is to implement a pepsin decoy before delivery of the nanosensors. When decoy proteins (e.g. milk proteins) are administered orally before ingestion of a therapeutic protein, the decoys will compete away pepsin activity to reduce degradation of the therapeutic protein.

[0125] Example 10- Synthesizing a nanosensor library to sense CRC-associated enzymes

[0126] Intestinal enzymes from the host and microbiome are involved in fundamental processes for tumor growth, angiogenesis, invasion and metastases. Transcriptomic, metagenomic, and activity-based analysis of samples from CRC patient samples and CRC mouse models can be used to identify a common or similar enzyme signature with classification power for CRC. From there, a CRC-specific nanosensor panel capable of sensing and releasing distinct volatile barcodes for each protease or glycosidase in the enzyme signature will be prepared.

[0127] Example 10.1- Identifying intestinal enzymes with classification power for human CRC

[0128] Host enzymes. Preliminary protease targets have been identified from published literature (Table 4). To identify additional protease targets and glycosidase targets, human RNA-Seq data for colorectal tumor samples and matched normal pairs from the Cancer Genome Atlas (TCGA) will be analyzed. DESeq2 differential expression library in the R statistical environment will be used to complete differential expression analysis, which will be cross-referenced with a list of human extracellular protease and glycosidase genes obtained from UniProt. FPKM values from colorectal tumor samples and matched normal samples will be used to complete AUROC analysis using GraphPad Prism software. 20 proteases and 20 glycosidases with highest importance values for classification and greatest FPKM fold change over normal samples will be validated in clinical samples in Example 10.3.Table 4.

[0129] Microbial enzymes. Carcinogenesis in the colon is associated with dysbiosis, or an imbalance in the composition of the microbiome and altered bacterial metabolic activities. CRC microbiomes are enriched in oral pathobionts that translocate to the colon, including Fusobacterium micleatum, Parvimonas micra, Peptostreptococcus stomatis and others. Additional pathobionts, such as enterotoxigenic Bacteroides fragilis (ETBF) and certain Escherichia coli can produce the genotoxin colibactin and promote pro-oncogenic phenotypes. The presence of ‘keystone pathobionts’ in CRC, suggests that metabolic signatures of these organisms can be detected and used diagnostically. In particular, the microbiome is a major source of intestinal glycosidases due to its extensive carbohydrate utilization. Therefore, that altered intestinal glycosidases activities in CRC microbiomes may be useful to discriminate CRC from healthy state and other GI diseases. In preliminary studies, a large-scale metagenomic analysis of CRC cohorts was completed to determine the classification power of microbial glycosidases and to determine the extent of nanosensor multiplexing required to identify CRC patients. Specifically, HUMANN3 software was used to align reads from gut microbiome samples from CRC and healthy patients from four cohorts. Genes encoding for glycosidases were then annotated using the Carbohydrate- Active enZYmes (CAZY) Database. This data was used to train a random forest machine learning algorithm to classify patients based on metagenomic abundance of glycosidase genes. This analysis showed that as few as ~20 glycosidase genes can achievegood discrimination between CRC and healthy patients with an AUROC (area under the receiver operating characteristic curve) of up to 0.8. For reference, an AUROC value of 1.0 indicates perfect classification. This analysis is consistent with results from past studies using 14- and 19-plex protease sensing for detection of other cancers and justifies development of a panel of 15-20 nanosensors. The glycosidase genes were rank-ordered by importance score for classification and those with consensus across multiple cohorts are promising candidates for CRC detection (Table 5). In this example, metagenomic datasets from inflammatory bowel disease and irritable bowel syndrome cohorts will be assessed to identify an enzyme signature that can distinguish CRC from benign GI diseases. We will also subset patients by cancer stage, to ensure that we include enzymes relevant for early- stage CRC. Through this analysis, a final list of 10-15 glycosidase targets for which to build nanosensors will be produced.

[0130] Example 10.2 — Identifying intestinal enzymes with classification power for CRC in preclinical models

[0131] Experimental validation will be shown in a commonly-used CRC mouse model, which carries a mutation in one copy of tumor suppressor, adenomatous polyposis coli (Ape) (Apcmm'+f This mutation predisposes the mice to intestinal tumor development. Apdmamice are commonly used as a model for human colorectal cancer because the biallelic loss of Ape is an initiating event in CRC pathophysiology (detected in -80-90% of some CRC patient cohorts). To validate our approach in preclinical models, we will needto determine intestinal enzyme targets that are relevant to the Apcinmmice and that overlap with human intestinal enzyme targets. Therefore, equivalent analysis described in Example10.1 will be completed on generated Apcmm!+mouse RNA-Seq datasets as well as published metagenomic datasets to identify mouse CRC enzyme targets and for comparison to human CRC enzyme targets. Further characterization will be completed by harvesting the small and large intestine from Apma,+mice during early tumor development and using immunohistochemistry to stain for proteases and glycosidases. Alternatively, the enzyme profdes of an inflammation-induced model of CRC, namely the azoxymethane (AOM)- dextran sulfate sodium (DSS) model, can be evaluated. Application of the AOM mutagen followed by repeated week-long doses of DSS, which results in colonic inflammation, causes robust tumorigenesis. In the case that we do not find overlap in the enzymes observed in the Apcmins+or AOM-DSS mice, germ-free mice can be ‘humanized’ by gavaging them with stool samples from healthy and CRC patients prior to inducing tumorigenesis with AOM-DSS.

[0132] Example 10.3 — Validating CRC enzyme targets

[0133] In this example, the activity of the enzyme targets identified in Examples 10.1 and10.2 will be assessed in stool and microbial cultures to confirm differential activity in CRC. Due to multiple levels of control in gene expression, abundance of mRNA transcripts and metagenomic abundance do not necessarily equate to enzyme activity. Therefore, assays have been established to measure enzyme activity in stool, which contains both host and microbiome enzymes. Clinical stool samples can be homogenized in buffers and centrifuged to produce supernatants, which can be assayed for enzyme activity. Supernatants can be incubated with an array of commercially-available fluorogenic substrates in 384-well plate format to quantify relative levels of enzyme activity using a platereader. Measurements are normalized to protein concentration. Stool assays have been used for other GI diseases such as Clostridium difficile infection (CDI) and showed the ability to resolve differences in enzyme activities in healthy and disease cohorts. The same assays will be used to measure enzyme activity in CRC and healthy stool samples.

[0134] In preliminary studies, we have also shown that the same assay can be used to assess enzyme activity in culture supernatants from intestinal microbes including keystone pathobionts in CRC. Intestinal microbe isolates were grown in monocultures in standardbacterial growth media. Culture supernatants from CRC pathobionts - E. faecalis B. fragilis - produced distinct cleavage signatures and cleaved select glycoside substrates more rapidly than other microbes. These results support microbial glycosidases as a class of enzymes for CRC classification and show that CRC -associated microbes exhibit unique glycosidic profiles. To build on this work, a library of 50 glycosidase substrates for enzyme targets identified in Example 10.1 will be assembled and screened against culture supernatants to validate microbial enzyme targets.

[0135] Example 10.4 — Synthesizing a CRC nanosensor panel

[0136] After identification of 20-30 total protease and glycosidase targets, the enzymes will be purchased and used in cleavage assays with a library of -100 commercially- available fluorogenic peptide / gly coside substrates in 384-well plate format. The purpose of these cleavage assays is to identify substrates that can be modified with VOC reporters for protease / glycosidase sensing. We expect that multiple substrates will be cleaved by the same enzyme and that some enzymes will have overlapping substrate specificities. To minimize redundant enzymes and substrates in our final nanosensor panel, hierarchical cluster analysis will be used to identify similarly-cleaved substrates to downselect for substrates that have orthogonal enzyme susceptibility. That will ensure that the final nanosensor panel has broad enough coverage of CRC enzyme activities to create a disease signature that can discriminate CRC from other GI diseases. Based on past studies, we expect that these studies will result in at least 15-20 substrates that we can then barcode with VOC reporters and formulate for oral delivery using chemistry from previous examples.

[0137] Microbial gene expression is environment-dependent, and, therefore, enzyme activities in culture may not reflect enzyme activities in the GI tract. Therefore, activity in stool may be a more accurate representation, which we will be assaying. Furthermore, it is possible to grow up cultures from stool using media that has been reported to maintain gut microbiome diversity, namely Gifu Anaerobic Media (GIFU) and Gut Microbiome Media (GMM), which may more accurately represent enzyme expression in the context of complex microbial communities compared to monocultures. We will perform metagenomic sequencing of the input stool sample and final culture to ensure that diversity was maintained. Murine models of CRC may not comprise organisms with the sameglycosidases as seen in human CRC patients’ microbiomes. As an alternative, we can utilize germ-free mice colonized with human CRC patient’ and healthy control patients’ microbiomes to test for enzymatic activity, with and without induction of tumorigenesis via the AOM-DSS model. Alternative models include the FabplCre; Ape 151ox / |) mouse, which was developed more recently and produce tumors in the large intestine.

[0138] Example 11 - Development of a classifier for early detection of CRC via breath analysis of intestinal enzyme signatures

[0139] Current non-invasive testing for CRC relies on fecal biomarkers, which have been shown to have low sensitivity for stage I CRC (65-79%), during which the cancer is still localized to the intestinal mucosa. This sensitivity serves as a benchmark for our approach. For sensitive detection of intestinal tumors, we are leveraging the catalytic activity of intestinal enzymes to generate amplified breath signals. In other examples described herein (Fig. 9), we demonstrated the feasibility of generating breath signals for baseline intestinal activity via nanosensor delivery in healthy mice. In this Example, nanosensor panels prepared as described herein will be evaluated in mouse models of CRC.

[0140] Example 11.1 — Developing classifier for early detection o f intestinal tumors via breath analysis

[0141] CRC nanosensor panels prepared as described herein will be evaluated in the Apcm,n / +mice, a well-characterized CRC model that develops tumors by 90 days, usually in the small intestine. The development of pre-cancerous tumors (adenomas) in Apcmm / +mice provides an opportunity to test ingestible nanosensors, according to various aspects of the disclosure, for early detection. To generate data to train a CRC classifier, nanosensors will be administered via oral gavage into Apcmm+mice of C57BL / 6 background and wild-type controls, and breath samples will be collected using methods discussed in Example 9.2. Reporter quantification in breath samples will be completed using mass spectrometry. The resulting dataset will consist of measured abundance for each reporter present in the nanosensor panel. Using established machine learning algorithms (i.e. support vector machine, random forest, and regularized logistic regression), a CRC classifier with maximum sensitivity and specificity will be built using the training cohort data. Classifier accuracy (as defined by area under the receiver operating characteristic curve (AUROC)), sensitivity, and specificity will be determined using adataset generated in an independent validation cohort. For these studies, breath datasets will be generated at three different timepoints during tumor development (60, 75, and 90 days of age), with separate cohorts that are euthanized at each timepoint for characterization of the average number of tumors, tumor size, and histological evaluation to assess the degree of dysplasia in adenomas and presence of invasive carcinoma. These data will allow us to determine the earliest timepoint during tumor development that engineered breath biomarkers can discriminate CRC mouse models from healthy controls.

[0142] Example 11.2 — Developing classifier for detection of intestinal tumors in microbiome-modified CRC models

[0143] Due to differences in the human and mice microbiome, nanosensor testing in Apcmin / +mice may not recapitulate breath signals generated by enzyme activity from the human microbiome. Apcmm / +mice have been used to examine the transcriptional responses of single cells to colonization of human CRC keystone pathobionts, including enteropathogenic B. fragilis (ETBF), and have shown that ETBF promotes oncogenic reprogramming and accelerates tumorigenesis. To determine the effect of a keystone pathobiont on CRC detection, ETBF-modified Apcmm +mice, unmodified Apcmm+mice, and healthy controls will be used to develop a new classifier for CRC. Due to ETBF-driven acceleration of CRC pathophysiology in mice, robust discrimination of ETBF -modified Apcm,n / +from healthy controls via breath analysis is expected.

[0144] Example 11.3 — Evaluating specificity of classifier for CRC over other benign GI diseases

[0145] Aberrant intestinal activity is a common feature across several GI diseases, including inflammatory bowel disease (IBD), irritable bowel syndrome (IBS), and disaccharidase deficiencies. However, it is expected that differences in enzyme activity signatures will be able to discriminate CRC from benign GI diseases. To test this hypothesis, nanosensors will be tested in mouse models of CRC and IBD. Dextran sulfate sodium (DSS)-induced colitis mouse models are commonly used for IBD research, and intestinal inflammation is established via oral delivery of DSS in autoclaved drinking water for up to 7 days. For these studies, nanosensors will be delivered via oral gavage into both disease models and breath will again be collected and analyzed using mass spectrometry.The classifier developed in Example 11.1 will be applied to this dataset to determine the classification accuracy, sensitivity, and disease specificity.

[0146] In this Example 11, a classifier that discriminates CRC from benign conditions early on in tumor development with AUROC > 0.85 is expected. Success of this platform technology will provide a path to noninvasive, breath-based detection for CRC and a broad range of GI diseases and eliminate the logistical barriers to early disease detection.

[0147] While certain implementations have been described in terms of what may be considered to be specific aspects, the present disclosure is not limited to the disclosed aspects. Additional modifications and improvements to the aforementioned vial adapter may be apparent to those skilled in the art. Moreover, the many features and advantages of the disclosure are apparent from the detailed specification, and thus, it is intended by the appended claims to cover all such features and advantages of the present disclosure which fall within the spirit and scope of the disclosure.

Claims

CLAIMSWhat is claimed is:

1. A composition for the detection of the activity of a plurality of enzymes, the composition comprising: a plurality of molecular probes, each molecular probe comprising: a recognition domain / substrate structured to interact with a specific enzyme; a reporter molecule; and a linking group forming a covalent bond between the recognition domain and reporter molecule, wherein for each of the plurality of molecular probes, upon interaction of the recognition domain with a specific enzyme, the covalent bond is destroyed, rendering reporter molecule detectable by a chemical detection device.

2. The composition of claim 1, wherein the reporter molecule is a volatile organic compound, which is optionally labeled with a radioactive / non-radioactive isotope, and is volatile after destruction of the linking group.

3. The composition of claim 1, wherein the enzyme is a glycosidase and the recognition domain is a carbohydrate substrate (a mono- / di- / poly-saccharide).

4. The composition of claim 1, wherein the linking group is an O-glycosidic bond, N- glycosidic bond, or S-glycosidic bond.

5. The composition of claim 1, wherein the linking group comprises a self-immolative group.

6. The composition of claim 1, wherein the recognition domain / substrate is covalently bound to a biomolecule (for example, peptides / proteins, lipids, carbohydrates, and nucleic acids), a polymer (for example, polyethylene glycol and dextran), or a nanoparticle scaffold(for example, iron oxide nanoparticles, gold nanoparticles, and porous silicon nanoparticles).

7. The composition of claim 1, wherein the molecular probes are encapsulated in nanoparticles or microparticles.

8. The composition of claim 1, wherein at least one of the plurality of molecular probes comprises more than one recognition domain.

9. The composition of claim 1, wherein at least one of the plurality of molecular probes comprises more than one reporter molecule.

10. The composition of claim 1, wherein at least one of the plurality of molecular probes comprises more than one recognition domain and more than one reporter molecule.

11. A method of detecting enzymatic activity of a plurality of enzymes, the method comprising: reacting a composition according to any one of claims 1 to 10 with the plurality of enzymes; and identifying detectable reporter molecules with a chemical detection device.

12. The method of claim 11, wherein reacting the composition with the plurality of enzymes is performed in vivo.

13. The method of claim 11 or 12, wherein a breath sample is collected from a subject administered the composition to quantify the detectable reporter molecules.

14. The method of claims 11 to 13, wherein the composition is administered to a subject by an oral route, inhalation, an intravenous route, a subcutaneous route, an intramuscular route, an intraperitoneal injection route, an ocular route, a sublingual route, a topical route,an aural route, a rectal route, or via an implanted device or an applied device (for example, a microneedle patch, an osmotic pump).

15. The method of claim 11, wherein reacting the composition with the plurality of enzymes is performed ex vivo (for example, with enzymes in tissue or blood samples collected from a human or animal subject or with enzymes in a water sample, a plant sample, or another environmental sample) and a reaction solution or headspace is analyzed for the detectable reporter molecules.

16. The method of claim 11, wherein reacting the composition with the plurality of enzymes is performed in vitro (for example, with recombinant / purified enzymes or enzymes in a mammalian tissue culture, a microbial culture, a water sample, a plant, an environmental sample, or a bioreactor) and a reaction solution or headspace is analyzed for the detectable reporter molecules.

17. The method of any one of claims 11 to 16, wherein the chemical detection device is a mass spectrometer, an infrared spectrometer, an ion mobility spectrometer, an electronic nose, a breathalyzer device, a colorimetric VOC sensor array, a microfluidic device, a human or animal (for example, canine) nose, an engineered microbial sensor, or any combination thereof.

18. The method of any one of claims 11 to 16, wherein the identity and abundance of the detectable reporter molecules is indicative of a disease or health status.

19. The method of any one of claims 11 to 16, wherein the identity and abundance of the detectable reporter molecules is used to monitor a disease progression.

20. The method of any one of claims 11 to 16, wherein the identity and abundance of the detectable reporter molecules is used for a disease prognosis.21 . The method of any one of claims 11 to 16, wherein the identity and abundance of the detectable reporter molecules is used to monitor a treatment response and, optionally, used as a clinical endpoint.

22. The method of any one of claims 11 to 16, wherein the identity and abundance of the detectable reporter molecules is used to monitor microbial contamination in the environment (for example, in water sources and natural vegetation) or food sources (for example, crops and livestock).

Images