Nayar Prize II Year 2 Progress Report - October 2018

Project: Microfluidic Drug-Microbiota Interaction Platform Team: Abhinav Bhushan, Genoveva Murillo, Rajendra Mehta Postdoctoral Scholar: Sonali Karnik, Students: Chengyao Wang, Rongfei Wu, Kihwan Kim, Andrea Cancino, Thao Dang, Adithya Subramanian.

In the second year of the Nayar Prize II, we have accomplished our year 2 milestones as well as formulated clear strategies for making an impact on the lives of patients. Our overall project goals are to study the role microbiota play in influencing drug metabolism. We are developing a microfluidic platform to facilitate study of interactions between the drugs, intestine, and microbiota. We have made extraordinary progress to advance rapidly towards the milestones including strengthening collaborations with Dr. Ali Keshavarzian, who leads one of the leading groups on microbiome/IBD research. Challenged by the reviewers at the end of year 1 to define how the microfluidic drug- microbiota interaction platform can help patients, we talked to several key opinion leaders and collaborated with the business school to develop strategies that will directly benefit patients. We came up with several solutions. One builds upon the concepts of a drug-microbiota interaction platform towards a patient-specific (personalized) microfluidic bioassay; in other words, we will generate an IBD patient-based intestinal platform to test microbiota reprograming that can be used to query prebiotics, probiotics, and IBD therapeutics, potentially to determine the effects of these interventions before prescribing to the IBD patients. The other plan will give individuals the opportunity to generate a personal score that will represent the drug metabolizing capability of their own microbiome. The support from the Nayar Prize was the catalyst that made this possible.

Phase II summary 1) Establish intestinal culture in the microfluidic device and characterize drug interactions. We developed a novel microfluidic device that incorporated micrometer resolution membranes which are synthesized out of extracellular matrix. As our results show, cells on the microfluidic device with biomembranes had excellent viability for extended periods of time than the cells in a single layer device. More importantly, the biomembranes induced expression of tight junction between the intestinal cells, which is an important factor in the cells mimicking in vivo phenotypes. This work was published in ACS Biomaterials Science and Engineering and the journal highlighted the science in a special feature (“More realistic and accurate organs-on-chips”). The intellectual property concerning the collagen membrane device was filed by the university and granted a provisional patent. We built upon this work in four directions.

A. Establish primary colon cells in the device (Phase II Aim 1). Development of the primary colon culture is quite challenging as the microenvironment external to the in vivo gut environment does not have all the factors to support the growth and function of the cells. We have carried out a large number of experiments to grow the primary colon on a variety of extracellular matrices – fibronectin, collagen type I, collagen type IV, and Matrigel. We have conducted these experiments in tissue culture plates as well in the microfluidic devices. Our results indicate that Matrigel with a collagen coat provided the best results in the microfluidic devices. The colon cells express tight barrier junctions and also e-cadherin, which is one of the important cellular proteins (Figure 1). These results are being finalized in Figure 1. Expression of e-cadherin collaboration with the clinical collaborators. protein in primary colon cells.

B. Characterize role drugs play in regulating bacterial metabolites PDMS top cover (Phase II, Aim 2). It has been shown that metformin boosts the capacity of gut microbiota to produce certain short-chain fatty acids (butyric acid and Top-chamber propionic acid). We have exposed the several strains of bacteria to drugs and B their selected metabolites and Membrane characterized their effects on the bacteria. Middle-chamber

C. Characterize role bacteria play in A C regulating activity of the colon drug Oxygen concentration metabolizing enzymes (Phase II, Aim 3). Objective 3. We systematically Figure 2. Oxygen transport model (not drawn to scale). A is the concentration of oxygen in the middle-chamber studied the interactions (genes and after cellular respiration. B is the oxygen concentration specific activity of phase I metabolism) in the top-chamber. C is the oxygen concentration between colon cells and bacteria. The outside the device (disclosure pending). major findings are that (1) species- specific differences exist in the regulation and activity of these enzymes, (2) regulation of the various enzymes appeared to be unique, and (3) communication between the bacteria and the colon cells is required for modulation of enzymatic activity.

The results of Aims 2 and 3 are part of a manuscript that is under peer-review, which is provided with this report.

The complexity of the gut environment includes a biphasic oxygen environment – normal oxygen for the epithelial cells and anaerobic environment for the microbiota. Creating such an environment in vitro is very challenging, and does not currently exist, especially in microfluidics. We have created a representative gas transport model and used it to create prototype microfluidic devices. The results are being finalized for a publication as well as invention disclosure to the university (Figure 2).

D. We initiated work towards (see Appendix and below)

a. IBD patient-based intestinal platform to test microbiota reprograming and b. microfluidic model of microbiota-intestine-brain to study microbiota mediated neural inflammation which is implicated in neurodegenerative disorders (Alzheimer’s).

2) Market research. We carried out interviews of several technical leaders from the pharmaceutical industry on the use of the microfluidic drug-microbiota platform in drug discovery. The overwhelming response we received was that the industry is aware of the influence of bacteria but does not have assays to test for these interactions – any device or platform, such as ours, that can measure these interactions will be hugely beneficial. This work was corroborated by a study conducted in parallel by the students from the Stuart School of Business. During our research, we discovered another possible opportunity – the evaluation of an individual’s propensity for metabolizing drugs. The reports are attached as Appendix.

The outcome of these studies is that our Nayar project has three possible market opportunities. Some of this work is included in the disclosure filed with the university. Plans to commercialize are being explored. a. drug-microbiota interaction platform, b. IBD patient-based IBD bioassay, and c. personalized score of drug metabolizing capability by microbiome.

3) Training. The project has been instrumental in exciting students about research and has helped tremendously in recruiting students.

#4 disciplines integrated (engineering, biology, clinical, business) #12 students directly supported/educated #2 Armour college PURE scholarships #1 PhD fellowship (Starr Fieldhouse), #2 travel awards #5 presentations at national & local meetings

4) Journal papers Wang C, Tanataweethum N, Karnik S, ML, Bhushan A, A Novel Microfluidic Colon with An Extracellular Matrix Membrane, ACS Biomaterials Science and Engineering, 2018, doi: 10.1021/acsbiomaterials.7b00883.

Wang C, Cancino A, Dang T, Bhushan A, Regulation of drug metabolizing enzymes by microbiota, under review.

Wang C, Dang T, Bhushan A, Tunable multiphasic oxygen environments for coculture of anaerobic and aerobic bacteria with intestinal cells, under preparation.

Presentations Chengyao Wang, Chengyao Wang, Maliha W Shaikh, Christopher Frosyth, Ali Keshavarzian, Abhinav Bhushan, “Intestinal Monolayers on a Novel Microfluidic Device with an Extracellular Matrix Membrane”, BMES Annual Meeting, 2018.

Chengyao Wang, Nida Tanataweethum, Sonali Karnik, Abhinav Bhushan, “Novel Microfluidic Colon with an Extracellular Matrix Membrane”, Rainin Foundation Innovation Symposium, 2018.

Chengyao Wang, Andrea Cancino, Abhinav Bhushan, “Role of Bacteria in Regulating Activity of Enzymes Responsible for Metabolism of Drugs”, Rainin Foundation Innovation Symposium, 2018.

Role of Bacteria and Bacterial Products in Regulating Activity of Enzymes Responsible for Metabolism of Drugs in Microfluidic Devices, MIT-Harvard Microbiome Symposium.

Chengyao Wang, Nida Tanataweethum, Genoveva Murillo, Rajendra Mehta, Abhinav Bhushan, “A novel microfluidic device with an extracellular matrix-based membrane”, Experimental Biology, 2017.

5) Collaborations Each of the collaborators has been excited by the strengths of our group, IIT, and the support from the Nayar Prize. It is needless to say that without the Nayar Prize, venturing into any of these areas would not have been possible.

Dr. Ali Keshavarzian, Rush University - Role of microbiota along the gut-brain axis in the development of Parkinson’s disease.

Dr. Gokhan Dalgin, University of Chicago - Development of mini-brains for modeling neurodegenerative disorders.

6) Grant funding

a) Dynamics of microbial interactions in Alzheimer's using a microfluidic platform, NIH R03 under review.

b) Patient-based intestinal platform to test microbiota reprograming in IBD, NIH R21. Scored, to be resubmitted in October 2018.

c) A microfluidic platform to generate functionally reproducible human stem cell derived mini- brains, NIH R21. Scored, to be resubmitted in July 2019.

d) Microfluidic platform to discover interactions between intestine, liver, and bacteria, NSF CAREER (positive feedback from Program Officer, to be resubmitted in 2019)

7) Intellectual property a) Microfluidic Device with Extracellular Matrix Support Membrane, Provisional awarded.

b) Modulation of drug metabolism by bacteria, Disclosure filed with the Office of Technology Development, Illinois Institute of Technology.

Plans for Phase III We have a clear understanding of the path to translate this research. We are on track to accomplish our original objectives as well as some of the new directions that we have identified (Figure 3). We will explore the commercialization of the “personalized score of microbiome’s drug metabolizing capability” (MyBioScore). Our office of technology development has recommended that we participate in the NSF I-Corps program. We will work with them and possibly MATTER Chicago to develop a strategy and participate for commercialization. Our objectives for Phase III are:

Objective 1. Create a biphasic oxygen environment in the microfluidic devices. Several bacteria such as VSL#3, akkermansia muciniphila, and Bifidobacterium require an anaerobic environment for growth. Although anaerobic chambers exist at the macroscale, the challenge is to culture the bacteria with the intestinal cells. We have carried out feasibility studies to model and create a microfluidic device that can simultaneously support two oxygen concentrations. In this objective, we will finish characterization of cells and bacteria in these devices and develop the devices for routine fabrication.

Objective 2. Characterize intestine-bacteria influence on drug metabolism in microfluidic devices. As the results from this year indicate, bacterial strains have tremendous effects on a wide range of cellular processes, including drug metabolizing capability. We know that cells in a microfluidic environment are more physiological than those in the tissue culture plates (we had shown this in year 1). In this objective, we will carry out the next phase of our studies to characterize the role of bacteria on drug metabolism in microfluidic devices.

Objective 3. Develop a personalized intestinal bioassay to test therapeutics. Inflammatory bowel diseases (IBD), affect millions of Americans and carry an economic burden of hundreds of billions of dollars. There are several experimental and human observational studies to support the current notion that IBD is a consequence of abnormal mucosal immune response to intestinal microbiota due to abnormal microbiota composition, increased exposure of bacterial product to mucosal immune cells [disrupted intestinal barrier integrity] and exaggerated mucosal immunity. Thus, microbiota 10/2016 2017 2018 2019 2020

Phase I: Customizable Commercialize: microfluidic platform personalized score of microbiome’s Topological biomembranes drug metabolizing capability Microfluidic devices with biomembranes

Phase II: Microfluidic intestine-on-chip Primary colon in microfluidic device

Role of drugs in regulating bacterial metabolites

Bacteria modulation of intestine activity

Phase III: Drug interactions Microfluidic devices with dual-O2 env

Intestine-microbiota drug interactions in devices

Personalized intestinal bioassay

Figure 3. Timeline for the proposed project. reprograming is a promising therapeutic approach to prevent/treat IBD. But, in spite of success of microbiota reprograming by diet/pro and prebiotics and stool transplantation in animal models of IBD, the result in patients is disappointing. This failure could be due to lack of universal response of microbiota to any given intervention and immune response to change in microbiota. For example, it is known that not all IBD patients respond to TNF antibodies or anti-IL12/23 antibody suggesting that immune pathways leading to intestinal inflammation and tissue damage is different in different patients. This limitation could explain why result of animal studies and in vitro cell line studies are not universally translatable to human studies. Thus, availability of a robust and reliable patient-based “bioassay” is a major unmet need to overcome highly individualize and complex intestinal microenvironment in order to develop new precision and personalize therapeutic intervention including microbiota reprograming in IBD. To address this critical gap, we propose to develop a discovery platform that will enable colon- microbe-immune cultures and allow query of factors that can result in favorable/unfavorable outcomes. In this objective, we will test pre-/probiotics to regulate intestinal response to inflammation or differential drug metabolism activity. We will carry out this study at Rush University. Mindful of the additional logistics involved with patient studies, we will conduct a pilot study with all the proper controls (patient +/- Crohn’s, a number of bacteria and pre-/pro-biotics). The outcome will be a personalized patient-based assay that can be queried with therapeutics that can inform the physician on the best suitable line of therapy. In later phases, we could use patient’s own microbiota in the model and be able to carry out interesting experiments by cross culturing samples from patients and controls. Examples include= - Crohn’s intestinal cells+ healthy microbiota + probiotics; Crohn’s intestinal cells + healthy microbiota+ probiotics; Healthy intestinal cells + healthy microbiota+ microbiota; Healthy intestinal cells + Crohn’s microbiota+ probiotics.

Confidential

10/15/2018 Proceedings of the National Academy of Sciences

MANUSCRIPT ABOUT // AUTHORS // REVIEWERS // LOGOUT MY HOMEPAGE SUBMISSION

Manuscript # 2018-17445 Current Revision # 0 Submission Date 2018-10-15 Manuscript Title Bacteria Regulate Activity of Enzymes Responsible for Metabolism of Drugs Manuscript Type Direct Submission Special Features and Colloquia N/A Classification Biological Sciences/Applied Biological Sciences Database deposition Corresponding Author Abhinav Bhushan (Illinois Institute of Technology) Chengyao Wang , Andrea Cancino , Rongfei Fu , Thao Dang , Qifan He , Abhinav Co-Authors Bhushan (corr-auth) The trillions of microbes that inhabit our intestines and that are present in Abstract probiotics affect health and regulate important biochemical factors. One of the important ... View full abstract The microbes that inhabit our intestines and that are present in probiotics affect Significance Statement health and regulate important biochemical factors including metabolism of drugs. Due to ... View full Significance Statement Suggested Editorial Board MemberJames Collins (Massachusetts Institute of Technology) Sangeeta Bhatia (Massachusetts Institute of Technology), Minor Coon (Victor C. Vaughan Distinguished University), Philippe J. Sansonetti (Unité de Pathogénie Suggested Editor Microbienne Moléculaire et Unité INSERM 1202, Institut Pasteur / Chaire de Microbiologie et de Maladies Infectieuses, Collège de France) Author Conflict of Interest Funding Body Archiving Mandates No Funders Please let us know whether you have posted a version of your submission to a Preprint Server preprint server: No Electronic Forms 1 of 1 forms complete - View Electronic Forms Status

Manuscript Items

1. Manuscript File PDF (239KB) 2. Parsed Citations (Parsed Citations file not available) 3. Related Manuscript File PDF (1072KB)

Manuscript Tasks Send Manuscript Correspondence Check Status

Contact Feedback Subscribe For the Press Editorial Board Terms & Conditions https://www.pnascentral.org/cgi-bin/main.plex?form_type=view_ms&j_id=1&ms_id=451782&ms_rev_no=0&ms_id_key=ftdZvUogMIm1tWmQUKvzSB… 1/1 Bacteria Regulate Activity of Enzymes Responsible for Metabolism of Drugs

Chengyao Wang, Andrea Cancino, Rongfei Fu, Thao Dang, Qifan He, Abhinav Bhushan

Department of Biomedical Engineering, Illinois Institute of Technology, Chicago, IL 60616 Abstract The trillions of microbes that inhabit our intestines and that are present in probiotics affect health and regulate important biochemical factors. One of the important factors that bacteria, both gut microbes and probiotics, can regulate is the activity of enzymes that are important for drug metabolism. Due to the challenges of an experimental system that can study intestinal-bacterial interactions, studies that can report these effects are lacking. We report a carefully designed study that systematically looks at different levels the interactions between the colon cells and different bacteria (gut microbes and probiotics). We find that paracrine signaling between the bacteria and colon cells is critical in regulating the activity of important enzymes responsible for drug metabolism. Pathway analysis revealed that the bacteria cause changes in diverse set of biological pathways, especially in those related to drug metabolism. This was confirmed by measuring specific activities of important cytochrome P450s. The major findings are that (1) species-specific differences exist in the regulation and activity of these enzymes, (2) regulation of the various enzymes appeared to be unique, and (3) communication between the bacteria and the colon cells is required for modulation of enzymatic activity. This knowledge can help a priori design of therapeutics.

Significance statement The microbes that inhabit our intestines and that are present in probiotics affect health and regulate important biochemical factors including metabolism of drugs. Due to the challenges of an experimental system that can study intestinal-bacterial interactions, studies that report these effects are lacking. This gap has limited our ability to a priori use this knowledge in drug development. We systematically studied the interactions (genes and specific activity of phase I metabolism) between colon cells and bacteria. The major findings are that (1) species-specific differences exist in the regulation and activity of these enzymes, (2) regulation of the various enzymes appeared to be unique, and (3) communication between the bacteria and the colon cells is required for modulation of enzymatic activity. Introduction The intestine is the site for processing of food, drugs, absorption of nutrients, and generation of waste; chronic intestinal ailments affect our overall physiological health. Microbiota that colonizes the intestinal microenvironment contribute to all these functions. The microbiota is perturbed by change in diet, onset of diseases, and consumption of drugs (1, 2). Almost 24% of the common drugs modulated the growth of 40 representative human gut bacterial strains (3). The microbiota themselves are causally linked to metabolic disorders (4) and to the metabolism of drugs (5). In fact, the microbiota may contribute to the metabolism of at least fifty drugs (6, 7). Recent studies have shown the effect gut microbes have on the activity of drugs such as acetaminophen and digoxin. For example, metronidazole, a drug used to treat Crohn’s disease, has antimicrobial properties which puts selective pressure on the intestinal microbiota that leads to inactivation of the drug and hence reduced efficiency (8, 9). Bacteria can inactivate a drug as in the case of Actinobacterium Eggerthella lenta which converts the cardiac drug digoxin into an inactive metabolite dihydrodigoxin (10). In another example, the drug sulfasalazine is hydrolyzed by an intestinal bacterium to 5-aminosalicylic acid, the active compound, and sulfapyridine. The parent drug, sulfasalazine, can inhibit the NF-κB inflammatory pathway whereas sulfapyridine cannot, illustrating a situation where the parent drug and its bacterial metabolites can have different mechanisms of action and presumably different targets (11). With the trillions of bacteria in the microbiome, the inter-individual diversity in the microbiome composition can have broad impact on the potential variation in PK/PD of drugs. Even probiotics, the usage of which is on the rise (4), modify the pharmacokinetics of orally administered drugs (12–14).

Cytochrome P450 enzymes (CYPs) are proteins that are used as substrates for phase I drug metabolism and account for 75% of the metabolism of drugs (15). They are a major source of variability in drug pharmacokinetics and response (15). Fifty-seven human CYPs, belonging to the CYP1, 2, and 3 families, have been identified to be responsible for the biotransformation of most foreign substances including 70–80% of all drugs in clinical use and involved in over 90% of the reactions in drug metabolism (16). It has been shown that approximately 70% of human drug metabolism and pharmacokinetics can be attributed to six major CYP450s, which include CYP1A2, 2C9, 2C19, 2D6, 2E1, 3A4 and 3A5 . The CYP3A4 in the intestine and the liver is involved in the metabolism of majority of the drugs on the market (23–27). The ability of gut bacteria to metabolize drugs is comparable to that of the liver (28). CYP450 3A4 enzyme is abundant in the human intestine (29) and its activity is markedly lower in patients with Crohn's disease (30); enterocytic CYP3A4 influences coeliac disease and cystic fibrosis in pediatric populations (31). CYP1A2, another major cytochrome in the intestine, influences the oxidative metabolism of the exogenous and endogenous molecules including caffeine (32) as well as is involved in several drug interactions (33).

CYP activity is modulated by many factors such as inflammation (34, 35), food (36), and bacteria (37–39). Blockade of pro-inflammatory cytokines and lipopolysaccharide (LPS) with aminoguanidine both extended down-regulation of CYP2A5 to the high dose range while not affecting LPS-induced depression of CYP1A and CYP2B (34, 40). The anti-inflammatory effect and modulation of cytochrome P450 activities by Artemisia annua tea infusions inhibited the activity of CYP3A4 and of CYP1A1 (36). Probiotics can also affect drug metabolism, for example, gut microbiota can influence the metabolism of nifedipine through affecting CYP3A in the intestinal mucosa (12). VSL#3, a common probiotic (41–44), which is a combination of eight strains, contains facultative or obligate, anaerobic, probiotic strains including L. acidophilus, L. plantarum, L. paracasei, L. delbrueckii subsp. bulgaricus, B. breve, B. longum, B. infantis, and Streptococcus thermophiles, modifies CYP activity (45–47). VSL#3 probiotic regulates the intestinal epithelial barrier in vivo and in vitro (42), induces mucin expression and secretion in colonic epithelial cells (48), and modulates human dendritic cell phenotype and function (49). Lactobacillus and Bifidobacterium have effects on gut disorders and drug metabolism where they can affect the metabolism of acetaldehyde and ethanol (45–47). All the Lactobacillus and Bifidobacterium strains from VSL#3 revealed a modification to oxidize ethanol and acetaldehyde which suggested a beneficial impact on high gastrointestinal acetaldehyde levels following alcohol intake (45).

As we discover more and more the causative roles of bacteria on human health (4), the need for an assay to study the role of bacteria on the potential metabolism of drugs has never been more. However, there are limited experimental tools to screen drugs for potential metabolism by the gut microbiome. The major bottleneck to understanding impact of microbes (both intestinal and probiotics) on drug metabolism is the lack of fundamental analytical approaches to quantifying the enzymes that are responsible for metabolizing the drugs. Although the interactions between several drugs and intestinal bacterial strains were recently reported (3), such studies ignore the role of intestinal cells, which we know play an important role in regulating the activity of the microbes. Thus, there remains a need for efficient approach for characterizing the role of bacteria on potential regulation of drug metabolism. In this paper, we have systematically studied the role of three different bacteria on the regulation of intestinal drug metabolizing enzymes. We have selected a probiotic (VSL#3), and two intestinal microbes E-coli Nissle 1917 and Bifidobacterium adolescentis (B. adolescentis). E-coli Nissle 1917, an anaerobic bacteria, is predominantly found in the human gut is another bacteria, has been shown to regulate drug metabolism (49–53). E. coli Nissle 1917 administration has been shown responsible for better drug absorption from the gastrointestinal tract (54) as well as for maintaining tight epithelial junction and repair of the barrier function (55, 56). B. adolescentis, is a bacterial component seen in the gut microbiota and plays a significant role in influencing the prevention or/and treatment of different gastrointestinal infections, intestinal disorders such as Crohn’s disease (57), which could be attributed to the effects of mechanisms of B. adolescentis consumption of lactate and acetate (58) . Due to the challenges of in vitro culture of intestinal cells, we used Caco2 cells because of their ability to form stable monolayers with the expression of relevant human drug transporters and CYP activity (59, 60). When grown on semi-permeable membranes, these cells can grow into 3D structures with tight barrier junctions as others and we have shown (59, 61) . When grown on semi-permeable membranes, these cells can grow into 3D structures with tight barrier junctions as others and we have shown (59, 61) . We studied more than 80 genes related to different aspects of drug metabolism under different culture of Caco2 cells and bacteria (Figure S2). We quantified not only the changes in relative mRNA but also the activity of important CYPs.

Materials and Methods

Cell Culture Human intestinal epithelial Caco2 cells were obtained from ATCC (Manassas, VA) and grown in 1X Eagle’s Modified Eagle Medium (EMEM) (VWR) containing 4.5 g/l glucose and 25 mM HEPES supplemented with 20% fetal bovine serum (FBS) (Gibco), 1% pen-strep (Gibco).

The cells were cultured at 37°C in a humidified incubator with 5% CO2. Cells were cultured in a 75cm2 cell culture flask until the cells were 80% confluent. The cells were then seeded in well plates and on Transwell inserts (Corning, NY) and grown until confluent. Bacteria Culture Three different types of bacteria strands were used which included VSL#3 (Alphasigma USA, Covington, LA), Bifidobacterium adolescentis (B. adolescentis ) (Manassas, VA), and E-Coli Nissle 1917 (Ardeypharm GmbH, Germany). All three bacteria strains were grown following the manufacturer’s instructions. The respective mediums included 1:1 mixture of Difco Lactobacilli MRS Broth (BD) and Reinforced Clostridial Medium (BD), Reinforced Clostridial Medium, and LB which contains 1% (W/V) Tryptone, 0.5% (W/V) yeast extract, and 1% (W/V) sodium hloride. The bacteria were first plated on agar plates to determine the standard curve of bacteria colonies and optical density (OD). The gradient concentrations of original bacteria suspensions (½, 1/5, 1/10, 1/25, 1/50, 1/100, 1/250, 1/500, 1/1000) were used to determine the standard curve. Twenty μl of bacteria suspension and a control of the medium were seeded at agar plate. The OD of each concentration was measured using the Spectra Max M2 (Molecular Devices) by measuring the absorbance at 600nm. After the bacteria colonies were visible, an image was taken on the microscope and the number of colonies was counted using Image J (NIH, Bethesda, MD). Once that information was obtained, a standard curve was made (Figure S1). The standard curve was used to quantify the MOI for the cocultures.

Experimental Design We carried out three sets of experiments (Figure S2). (1) Caco2 cells and bacteria were cocultured (a) in well plates and (b) in (0.4 um) Transwell inserts. In well plates, the cells and bacteria were in contact with each other. In Transwell inserts, cells were seeded on the inserts in the top chamber and bacteria were seeded at bottom where caused the interactions to be paracrine. Three different Multiplicity of Infection (MOI) of bacteria and Caco2 cells were tested - 1:1, 10:1, and 100:1. The viability of Caco2 cells seeded with the bacteria for 2, 4, 6, and 8 hours was estimated. Live/dead images were used to calculate the cell viability. (2) We also used the bacterial supernatant to test its effect on intestinal enzymatic activity. The bacteria supernatant was added to the Caco2 cell culture media in three different ratios - 1:1, 1:10, and 1:100. The conditioned media was introduced into (a) well plates and (b) bottom well of Transwell inserts. The bacteria supernatant was filtered by 0.22 um syringe filter (Corning) to clear all the bacteria effect. The viability of Caco2 cells seeded with the different ratio of bacteria supernatant for 8 hours was estimated. (3) We then used a few of the specific metabolites that the bacteria are known to produce. Three metabolites (butyrate (Sigma), propionate (Sigma), p-cresol (Sigma)) were selected. Bacterial metabolic products, such as short chain fatty acids and their relative ratios, may be causal in disturbed immune and metabolic signaling (62). Metabolites were mixed with cell culture media to result appropriate concentration - 2 mM butyrate (63, 64), 3 mM propionate (64, 65), and 1 mM p-cresol (66, 67). The conditioned metabolites were introduced into (a) tissue culture plates and (b) bottom chamber of Transwell inserts. The viability of Caco2 cells seeded with the different metabolites for 8 hours was estimated. cDNA Synthesis and Quantitative Polymerase Chain Reaction (qPCR) Caco2 cells and the bacteria co-cultures were prepared as described above. After a 4 hour incubation, the cells were washed with sterile 1X PBS (Sigma) and 1X trypsin (Gibco) was added to detach the cells from the walls. The RNA was isolated using the RNeasy Mini Kit (QIAGEN, Germantown, MD). RNA concentration of each sample was quantified, and cDNA samples prepared by using the First Stand Kit (Qiagen) and SYBR Green (Qiagen) using Thermo Cycler C1000 Touch (BIO-RAD, Hercules, CA). qPCR was performed using the RT2 Profiler PCR Arrays (QIAGEN) on the ABI ViiA7 (Thermo Fisher, Skokie, IL).

Pathway analysis Heatmap was generated using the log2 (fold change) of the different genes by FunRich (68). Data were also analyzed with FunRich, Gene Ontology (69, 69), and Reactome (70) to study biological pathways with the following analysis settings: the fold change that larger than 3 and less than 0.3 were selected. Fold change was calculated as below: first, the ΔCt value, which is the difference between Ct of target gene and of HKG genes of each treatment and control, was calculated. By subtracting the ΔCt values of experimental group from the control, ΔΔCt was calculated for each gene which was used to calculate the fold change = 2 (-ΔΔCt). Heatmap was generated from 28 genes with the p-value < 0.05. In the heatmap, log2 fold change was used to represent gene expression and the color map indicates the downregulation of gene expression (log2 fold change < 0) as blue and up regulation as red (log2 fold change > 0). The Ct values of seven important CYP 450s were normalized to same housekeeping genes (ΔCt = Ct of target gene - Ct of HKG gene). Fold change was calculated following the Qiagen protocol based on the difference between control group and experimental group and log2 fold change was plotted to show the genes that were up and down regulated. (n>=3)

CYP 450 Analysis Caco2 cells were seeded in the 48-well plate and a 48 Transwell inserts for 2-3 days, following which the bacteria was added as described above and cultured together in a 37°C,

5% CO2 incubator. P450-Glow CYP3A4 Assay with Luciferin-IPA, P450-Glo™ CYP1A2 and CYP3A4 were purchased from Promega (Madison, WI) and were used according to the manufacturer’s instructions. The luminescence of each experiment was read using the Spectra Max M2.

Data analysis Unless otherwise indicated, data are expressed as means ± SEM. Differences between experimental groups were compared using t test. Statistical analyses were done with Excel and FunRich. Differences with a p < 0.05 were considered statistically significant.

Results and Discussions

Cell viability in coculture We used VSL#3 as model bacteria to determine the time for which the co-culture could be carried out without significant loss of Caco2 cell viability. Viability of Caco2 cells co-cultured with VSL#3 bacteria in a well plate was stable at first 4 hours, which stayed at least at 97%. When the coculture was taken to 8 hours, the viability of Caco2 cells decreased to 65% (Figure 1A, D). In contrast, the coculture in Transwell insert could be taken beyond 8 hours without significant drop in viability (Figure 1B, E). Similar high viability was observed when the cells were cultured under different ratios of bacterial supernatant and cell culture media (Figure 1C, F). Based on these results, we selected the four-hour time point for the subsequent co-culture studies.

Gene expression analysis The heatmap shows that in the selected gene set, the upregulation of genes by B. adolescentis was higher than that by VSL#3 and E.coli Nissle1917 (Figure 2A, Figure S3). These genes belong to CYP450, aldehyde dehydrogenases, and flavin containing monooxygenases. Surprisingly, B. adolescentis upregulated most of the genes whereas the other two bacteria led to a downregulation of the same genes. We discovered that for VSL#3, CYP1A2, 3A4, 2C9, and 2E1 expression was highly upregulated, whereas 2C19, 2D6 and 3A5 was downregulated. E.coli Nissle 1917 upregulated CYP2C9, 2E1, 3A4, and 3A5 and downregulated CYP1A2, 2C19 and 2D6. B. adolescentis downregulated CYP 2D6 and 3A5 while upregulating the other five. If we look at just the ΔCt values (lower ΔCt value reports higher gene expression), we find that while B. adolescentis and VSL#3 upregulated several of the CYP450 genes significantly (p < 0.05), E.coli Nissle 1917 did not (Figure 2C). VSL#3 and B. adolescentis significantly upregulated the expression of CYP 1A2 and 3A4, whereas only B. adolescentis had an effect (downregulation) on the expression of CYP2D6. The data suggests that B. adolescentis and VSL#3 will potentially affect the pharmacokinetics of drugs, at least of those that rely on phase I metabolism which could be significant as these include the CYPs that contribute to almost 70% of human drug metabolism and pharmacokinetics (16).

Pathway analysis We carried out analysis to identify the major pathways that corresponded to the upregulated and downregulated genes for each of the bacteria (Figure 3). The data suggest that all three bacteria regulate drug metabolism and inflammatory signals. The analysis showed that for VSL#3 the upregulated genes contributed primarily to processes for metabolism of xenobiotics and synthesis of HETE whereas the downregulated genes to modification of inert chemicals into biologically reactive species (69) (Figure 3A). Synthesis of HETE is associated with inflammatory signals and cancers as well as with the ability to facilitate tumor cell motility (70). E.coli Nissle 1917 contributed to pathways important for xenobiotics, endogenous sterols, fatty acids, and the synthesis of epoxy, dihydroxyeicosatrienoic acids, and HETE (Figure 3B). Excessive storage of fatty acids can lead to obesity related dyslipidemia, hypertension, and hypertriglyceridemia and also alters the lipid metabolism (71). EET and DHET are involved in by formation and metabolism of CYP450s (71). The genes up- and downregulated by B. adolescentis contributed to pathways similar to E.coli Nissle 1917 (Figure 3C). The pathways affected by E.coli Nissle 1917 and B. adolescentis were primarily related to the modification of xenobiotic metabolism and inflammatory signals, but the proportion that the genes contribute to each pathway was less than VSL#3. The differences among VSL#3 and the other two bacteria may occur because VSL#3 comprises eight different strains. While it may be interesting to study the effect of each of these stains, our results are important because consumers take VSL#3 as a compound-probiotic. The knowledge can help a priori design of therapeutics. CYP 450 activity The bacteria we tested modify the activity of drug metabolizing enzymes CYP3A4 and 1A2. Three different Multiplicity of Infection (MOI) of bacteria and Caco2 cells were used - 1:1, 10:1, and 100:1 which were derived from the standard curve (Figure S1). Bringing the bacteria in contact with the cells suppressed the CYP activity for in all three cases (Figure 4). Both VSL#3 and its supernatant up-regulate CYP3A4 and 1A2 in Transwell inserts, but not in well plate (Figure 4A). We observed that for MOI =1:1, 10:1 and 100:1, the activity of CYP3A4 was more than 3-fold over the control and more than 2-fold over control for CYP1A2. However, culturing the cells with different amounts of the bacteria supernatant conditioned media recapitulated only partial activity of the two CYP enzymes (Figure 4B), suggesting that signaling between the bacteria and colon cells in an important factor in determining the activities. Paracrine signaling between VSL#3 and Caco2 cells upregulated activity of both CYP enzymes CYP3A4 and 1A2. The modification of CYP3A4 activity was higher than that of CYP1A2. VSL#3 supernatant also upregulated the activities although at a lower level than the co-culture, again, suggesting the important of bidirectional paracrine signaling. In well plate cultures, surprisingly, the supernatant did not affect CYP3A4 activity but downregulated CYP1A2 activity. E.coli Nissle 1917 modified the activity of CYP3A4 and 1A2 in the well plate as well as the Transwell inserts (Figure 4C). In plates, CYP3A4 activity was upregulated when for MOIs 1:1 and 100:1, whereas there were no changes for MOI=10:1. E.coli Nissle 1917 did not affect the activity of CYP1A2 in the well plate. In the Transwell insert, CYP3A4 activity went up and the highest fold change was 1.9 that occurred at MOI=10:1 while the fold change for MOI=1:1 and 100:1 was 1.4 and 1.5. However, the E.coli’s effect on CYP1A2 was the opposite. The highest reduction in fold change, which was 0.75 compared to control, was at MOI=10:1 (Figure 4C). E.coli Nissle 1917 supernatant did not change CYP3A4 activity but slightly decreased the activity of CYP1A2 when the ratio of supernatant and media was 1:100 (Figure 4D). By decreasing the bacterial supernatant from 1:1 to1:100, the CYP1A2 activity fold change decreased from 1.1 to 0.85 compare to control. E.coli Nissle 1917 supernatant caused the activity of CYP3A4 to increase, while decreasing that of CYP1A2 (Figure 4D). When the ratio of E.coli Nissle 1917 supernatant to cell culture media was 1:10, the upregulation on CYP3A4 activity was the highest. The other two treatments, ratio of 1:10 and 1:100, just showed upregulation in the activity of CYP3A4. Unlike CYP3A4, the activity of CYP1A2 was downregulated under all three ratios. B. adolescentis slightly lowered CYP3A4 activity at all MOIs in cell culture plate (Figure 4E). The CYP1A2 activity was also slightly downregulated. In the Transwell insert, CYP3A4 activity was increased only at MOIs 10:1 and 100:1. When the MOI decreased to 1:1, there was no influence on 3A4 activity any more. In contrast, CYP1A2 activity was slightly upregulated at MOI of 10:1. There was no influence on activity when the MOIs were 1:1 and 100:1. The effect of B. adolescentis supernatant on the activity of both CYPs was similar to that in the tissue culture plate – higher ratio of the supernatant corresponded to lower activity. In Transwell inserts, B. adolescentis supernatant slightly increased the CYP3A4 activity (Figure 4F). There was no significant change in the activity of CYP1A2 at 1:1; however, when the ratio decreased to 1:10 and 1:100, the fold change value of CYP1A2 activity was higher. Our data suggests that the bacteria have significant effect on CYP activity when the cells and bacteria are not in contact. Bacterial communication with the colon cells appears to be important in modulating the CYP activity. Although bacterial supernatant can modify CYP activity as well, the levels are not the same as those by the bacteria. In contrast, tests with individual bacterial metabolites (2mM butyrate, 3mM propionate and 2mM p-cresol) indicated no significant influence on CYP3A4 activity and a downregulation of CYP1A2 activity (Figure 4G), although at levels lower than the other cases. The results indicate that the individual metabolites do not significantly regulate CYP activity as compared to the co-culture and the supernatant.

Conclusions Despite advances in our understanding a bacteria’s (gut microbe and probiotics) contribution to human health, one factor of the host-microbe interactions that remains understudied is the influence of bacteria on drugs. These interactions could have an important role in explaining the variation in human responses to drugs or the variable response to therapy. However, there is a serious lack of available data that one could use to determine how bacteria would potentially regulate a drug’s metabolism. As a first step to fill that gap, we have established and thoroughly characterized a simple co-coculture between colon cells and different types of bacteria. We have systematically dissected the role of colon-bacteria interactions versus paracrine signaling through co-cultures of human colon cells with three different strains of gut bacteria. We have used this setup to study alterations in general phase I enzymes and more importantly, measured the activity of human CYP450. Our findings indicate that the bacteria and their supernatant both modify the activity of drug metabolizing enzymes. We found species specific differences in the regulation and activity of the phase I enzymes uniquely. Our results indicate that communication between the bacteria and the colon cells is required for modulation of metabolic activity. The regulation of the various CYP450’s appeared to be different, which suggested that studies are needed to uncover the regulatory mechanisms.

Supplementary Figures are included in the supplementary material.

Abbreviations TV: the colon cells co-cultured with VSL#3 in Transwell inserts, MOI=1:10 TN: the colon cells co-cultured with E.coli Nislle 1917 in Transwell inserts, MOI=1:10 TB: the colon cells co-cultured with B. adolescentis in Transwell inserts, MOI=1:10 TC: the colon cells cultured in Transwell inserts, without bacteria. MOI: Multiplicity of Infection P.MOI: the Multiplicity of Infection bacteria and cells tissue culture plate, ranging from 1:1 to 100:1 T.MOI: the Multiplicity of Infection bacteria and cells Transwell inserts, ranging from 1:1 to 100:1 P.CM: the bacteria supernatant conditioned media in tissue culture plate with different ratio range from 1:1 to 1:100 P.CM: the bacteria supernatant conditioned media in Transwell inserts with different ratio range from 1:1 to 1:100.

Acknowledgments This project was partly funded by the Nayar Prize II and student scholarships from the Armor College of Engineering. We acknowledge technical help from Dr. Hong lab.

Author contributions CW designed the study, conducted experiments, and wrote the manuscript. AC conducted experiment and wrote the manuscript. RF, TD, and QH conducted experiment. AB designed the study and wrote the manuscript.

References 1. Flowers SA, Evans SJ, Ward KM, McInnis MG, Ellingrod VL (2017) Interaction Between Atypical Antipsychotics and the Gut Microbiome in a Bipolar Disease Cohort. Pharmacotherapy 37 (3):261–267. 2. Forslund K, et al. (2015) Disentangling type 2 diabetes and metformin treatment signatures in the human gut microbiota. Nature 528 (7581):262–266. 3. Maier L, et al. (2018) Extensive impact of non-antibiotic drugs on human gut bacteria. Nature 555 (7698):623–628. 4. Scheithauer TPM, Dallinga-Thie GM, de Vos WM, Nieuwdorp M, van Raalte DH (2016) Causality of small and large intestinal microbiota in weight regulation and insulin resistance. Mol Metab 5 (9):759–770. 5. Fuller AT (1937) IS p-aminobenzenesulphonamide the active agent in prontosil therapy? The Lancet 229 (5917):194–198. 6. Li H, He J, Jia W (2016) The influence of gut microbiota on drug metabolism and toxicity. Expert Opin Drug Metab Toxicol 12 (1):31–40. 7. Spanogiannopoulos P, Bess EN, Carmody RN, Turnbaugh PJ (2016) The microbial pharmacists within us: a metagenomic view of xenobiotic metabolism. Nat Rev Microbiol 14 (5):273–287. 8. Yoo DH, et al. (2014) Gut microbiota-mediated drug interactions between lovastatin and antibiotics. Drug Metab Dispos 42 (9):1508–13. 9. Arndt H, Palitzsch KD, Grisham MB, Granger DN (1994) Metronidazole inhibits leukocyte-endothelial cell adhesion in rat mesenteric venules. Gastroenterology 106 (5):1271–6. 10. Saha JR, Butler J VP, Neu HC, Lindenbaum J (1983) Digoxin-inactivating bacteria: identification in human gut flora. Science 220 (4594):325–7. 11. Wahl C, Liptay S, Adler G, Schmid RM (1998) Sulfasalazine: a potent and specific inhibitor of nuclear factor kappa B. J Clin Invest 101 (5):1163–74. 12. Kato R, et al. (2007) Effect of Lactobacillus casei on the absorption of nifedipine from rat small intestine. Drug Metab Pharmacokinet 22 (2):96–102. 13. Lee HJ, Zhang H, Orlovich DA, Fawcett JP (2012) The influence of probiotic treatment on sulfasalazine metabolism in rat. Xenobiotica 42 (8):791–7. 14. Stojancevic M, Bojic G, Salami HA, Mikov M (2014) The Influence of Intestinal Tract and Probiotics on the Fate of Orally Administered Drugs. Curr Issues Mol Biol 16:55–68. 15. Pikuleva IA, Waterman MR (2013) Cytochromes P450: Roles in Diseases. J Biol Chem 288 (24):17091–17098. 16. Lin JH (2007) Pharmacokinetic and pharmacodynamic variability: a daunting challenge in drug therapy. Curr Drug Metab 8 (2):109–136. 17. Meyer UA, Skoda RC, Zanger UM (1990) The genetic polymorphism of debrisoquine/sparteine metabolism-molecular mechanisms. Pharmacol Ther 46 (2):297–308. 18. von Moltke LL, et al. (1998) Multiple human cytochromes contribute to biotransformation of dextromethorphan in-vitro: role of CYP2C9, CYP2C19, CYP2D6, and CYP3A. J Pharm Pharmacol 50 (9):997–1004. 19. Uttamsingh V, Lu C, Miwa G, Gan L-S (2005) Relative contributions of the five major human cytochromes P450, 1A2, 2C9, 2C19, 2D6, and 3A4, to the hepatic metabolism of the proteasome inhibitor bortezomib. Drug Metab Dispos Biol Fate Chem 33 (11):1723–1728. 20. Turpault S, et al. (2009) Pharmacokinetic assessment of a five-probe cocktail for CYPs 1A2, 2C9, 2C19, 2D6 and 3A. Br J Clin Pharmacol 68 (6):928–935. 21. Yenilmez ED, Tamam L, Karaytug O, Tuli A (2018) Characterization CYP1A2, CYP2C9, CYP2C19 and CYP2D6 Polymorphisms Using HRMA in Psychiatry Patients with Schizophrenia and Bipolar Disease for Personalized Medicine. Comb Chem High Throughput Screen 21 (5):374–380. 22. Paine MF, et al. (2006) THE HUMAN INTESTINAL CYTOCHROME P450 “PIE.” Drug Metab Dispos Biol Fate Chem 34 (5):880–886. 23. Labroo RB, Paine MF, Thummel KE, Kharasch ED (1997) Fentanyl Metabolism by Human Hepatic and Intestinal Cytochrome P450 3A4: Implications for Interindividual Variability in Disposition, Efficacy, and Drug Interactions. Drug Metab Dispos 25 (9):1072– 1080. 24. Fang J, Bourin M, Baker GB (1999) Metabolism of risperidone to 9-hydroxyrisperidone by human cytochromes P450 2D6 and 3A4. Naunyn Schmiedebergs Arch Pharmacol 359 (2):147–151. 25. Khan KK, He YQ, Correia MA, Halpert JR (2002) Differential Oxidation of Mifepristone by Cytochromes P450 3A4 and 3A5: Selective Inactivation of P450 3A4. Drug Metab Dispos 30 (9):985–990. 26. Zhou S-F (2008) Drugs behave as substrates, inhibitors and inducers of human cytochrome P450 3A4. Curr Drug Metab 9 (4):310–322. 27. Yamazaki H, Nakamoto M, Shimizu M, Murayama N, Niwa T (2010) Potential impact of cytochrome P450 3A5 in human liver on drug interactions with triazoles. Br J Clin Pharmacol 69 (6):593–597. 28. Kim DH (2015) Gut Microbiota-Mediated Drug-Antibiotic Interactions. Drug Metab Dispos 43 (10):1581–9. 29. Lin JH, Chiba M, Baillie TA (1999) Is the Role of the Small Intestine in First-Pass Metabolism Overemphasized? Pharmacol Rev 51 (2):135–158. 30. Wilson A, Tirona RG, Kim RB (2017) CYP3A4 Activity is Markedly Lower in Patients with Crohn’s Disease. Inflamm Bowel Dis 23 (5):804–813. 31. Johnson TN, Tanner MS, Taylor CJ, Tucker GT (2001) Enterocytic CYP3A4 in a paediatric population: developmental changes and the effect of coeliac disease and cystic fibrosis. Br J Clin Pharmacol 51 (5):451–460. 32. Härtter S, et al. (2003) Effects of caffeine intake on the pharmacokinetics of melatonin, a probe drug for CYP1A2 activity. Br J Clin Pharmacol 56 (6):679–682. 33. Chen H, et al. (2014) Bioactive components of Glycyrrhiza uralensis mediate drug functions and properties through regulation of CYP450 enzymes. Mol Med Rep 10 (3):1355– 1362. 34. De-Oliveira ACAX, Poça KS, Totino PRR, Paumgartten FJR (2015) Modulation of Cytochrome P450 2A5 Activity by Lipopolysaccharide: Low-Dose Effects and Non-Monotonic Dose-Response Relationship. PLoS ONE 10 (1). doi:10.1371/journal.pone.0117842. 35. De-Oliveira ACA, et al. (2010) Up- and down-modulation of liver cytochrome P450 activities and associated events in two murine malaria models. Malar J 9 (1):81. 36. Melillo de Magalhaes P, et al. (2012) Anti-inflammatory effect and modulation of cytochrome P450 activities by Artemisia annua tea infusions in human intestinal Caco-2 cells. Food Chem 134:864–71. 37. Claus SP, et al. (2011) Colonization-Induced Host-Gut Microbial Metabolic Interaction. mBio 2 (2):e00271-10. 38. Bezirtzoglou EEV (2012) Intestinal cytochromes P450 regulating the intestinal microbiota and its probiotic profile. Microb Ecol Health Dis 23 (1):18370. 39. Boer J, et al. (2016) The Role of UGT, CYP and Gut Microbiota in the Metabolism of Epacadostat (EPA) in Humans. Drug Metab Dispos:dmd.116.070680. 40. Li-Masters T, Morgan ET (2001) Effects of Bacterial Lipopolysaccharide on Phenobarbital-Induced CYP2B Expression in Mice. Drug Metab Dispos 29 (3):252–257. 41. Selwyn FP, Cheng SL, Klaassen CD, Cui JY (2016) Regulation of Hepatic Drug-Metabolizing Enzymes in Germ-Free Mice by Conventionalization and Probiotics. Drug Metab Dispos Biol Fate Chem 44 (2):262–274. 42. Dai C, Zhao D-H, Jiang M (2012) VSL#3 probiotics regulate the intestinal epithelial barrier in vivo and in vitro via the p38 and ERK signaling pathways. Int J Mol Med 29 (2):202–208. 43. Cho SS, Finocchiaro T (2009) Handbook of Prebiotics and Probiotics Ingredients: Health Benefits and Food Applications (CRC Press). 44. Gionchetti P, et al. (2002) Probiotics — Role in inflammatory bowel disease. Dig Liver Dis 34:S58–S62. 45. Nosova T, Jousimies-Somer H, Jokelainen K, Heine R, Salaspuro M (2000) ACETALDEHYDE PRODUCTION AND METABOLISM BY HUMAN INDIGENOUS AND PROBIOTIC LACTOBACILLUS AND BIFIDOBACTERIUM STRAINS. Alcohol Alcohol 35 (6):561–568. 46. Presti I, et al. (2015) Evaluation of the probiotic properties of new Lactobacillus and Bifidobacterium strains and their in vitro effect. Appl Microbiol Biotechnol 99 (13):5613–5626. 47. Paturi G (2007) Probiotic characteristics of Lactobacillus acidophilus and Lactobacillus paracasei and their effects on immune response and gene expression in mice. Available at: https://researchdirect.westernsydney.edu.au/islandora/object/uws%3A2504/ [Accessed October 12, 2018]. 48. Caballero-Franco C, Keller K, De Simone C, Chadee K (2007) The VSL#3 probiotic formula induces mucin gene expression and secretion in colonic epithelial cells. Am J Physiol-Gastrointest Liver Physiol 292 (1):G315–G322. 49. Lammers KM, et al. (2002) Effect of probiotic strains on interleukin 8 production by HT29/19A cells. Am J Gastroenterol 97 (5):1182–1186. 50. Wassenaar TM (2016) Insights from 100 Years of Research with Probiotic E. Coli. Eur J Microbiol Immunol 6 (3):147–161. 51. Schultz M, et al. (2004) Preventive Effects of Escherichia coli Strain Nissle 1917 on Acute and Chronic Intestinal Inflammation in Two Different Murine Models of Colitis. Clin Diagn Lab Immunol 11 (2):372–378. 52. Maltby R, Leatham-Jensen MP, Gibson T, Cohen PS, Conway T (2013) Nutritional basis for colonization resistance by human commensal Escherichia coli strains HS and Nissle 1917 against E. coli O157:H7 in the mouse intestine. PloS One 8 (1):e53957. 53. Schultz M (2008) Clinical use of E. coli Nissle 1917 in inflammatory bowel disease. Inflamm Bowel Dis 14 (7):1012–1018. 54. Matuskova Z, et al. (2014) Administration of a Probiotic Can Change Drug Pharmacokinetics: Effect of E. coli Nissle 1917 on Amidarone Absorption in Rats. PLOS ONE 9 (2):e87150. 55. Ukena SN, et al. (2007) Probiotic Escherichia coli Nissle 1917 Inhibits Leaky Gut by Enhancing Mucosal Integrity. PLOS ONE 2 (12):e1308. 56. Zyrek AA, et al. (2007) Molecular mechanisms underlying the probiotic effects of Escherichia coli Nissle 1917 involve ZO-2 and PKCζ redistribution resulting in tight junction and epithelial barrier repair. Cell Microbiol 9 (3):804–816. 57. Belenguer A, et al. (2006) Two Routes of Metabolic Cross-Feeding between Bifidobacterium adolescentis and Butyrate-Producing Anaerobes from the Human Gut. Appl Env Microbiol 72 (5):3593–3599. 58. Kim HJ, Huh D, Hamilton G, Ingber DE (2012) Human gut-on-a-chip inhabited by microbial flora that experiences intestinal peristalsis-like motions and flow. Lab Chip 12 (12):2165–2174. 59. Zhan T, et al. (2018) Metabolic Susceptibility of 2-Chlorothioxanthone and Its Toxic Effects on mRNA and Protein Expression and Activities of Human CYP1A2 and CYP3A4 Enzymes. Environ Sci Technol. doi:10.1021/acs.est.8b04643. 60. Wang C, Tanataweethum N, Karnik S, Bhushan A (2018) Novel Microfluidic Colon with an Extracellular Matrix Membrane. ACS Biomater Sci Eng 4 (4):1377–1385. 61. Zhao L, et al. (2018) Gut bacteria selectively promoted by dietary fibers alleviate type 2 diabetes. Science 359 (6380):1151–1156. 62. Barcenilla A, et al. (2000) Phylogenetic Relationships of Butyrate-Producing Bacteria from the Human Gut. Appl Env Microbiol 66 (4):1654–1661. 63. Gibson SA, Sewell GW (1992) Stimulation of Reductive Dechlorination of Tetrachloroethene in Anaerobic Aquifer Microcosms by Addition of Short-Chain Organic Acids or Alcohols. Appl Env Microbiol 58 (4):1392–1393. 64. Güven D, et al. (2005) Propionate Oxidation by and Methanol Inhibition of Anaerobic Ammonium-Oxidizing Bacteria. Appl Env Microbiol 71 (2):1066–1071. 65. Shields MS, Montgomery SO, Chapman PJ, Cuskey SM, Pritchard PH (1989) Novel Pathway of Toluene Catabolism in the Trichloroethylene-Degrading Bacterium G4. Appl Env Microbiol 55 (6):1624–1629. 66. Beller HR, Spormann AM, Sharma PK, Cole JR, Reinhard M (1996) Isolation and characterization of a novel toluene-degrading, sulfate-reducing bacterium. Appl Env Microbiol 62 (4):1188–1196. 67. Pathan M, et al. (2015) FunRich: An open access standalone functional enrichment and interaction network analysis tool. PROTEOMICS 15 (15):2597–2601. 68. Ashburner M, et al. (2000) Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 25 (1):25–29. 69. Joshi-Tope G, et al. (2005) Reactome: a knowledgebase of biological pathways. Nucleic Acids Res 33 (suppl_1):D428–D432. 70. Patterson AD, Gonzalez FJ, Idle JR (2010) Xenobiotic Metabolism: A View through the Metabolometer. Chem Res Toxicol 23 (5):851–860. 71. Powell WS, Rokach J (2015) Biosynthesis, biological effects, and receptors of hydroxyeicosatetraenoic acids (HETEs) and oxoeicosatetraenoic acids (oxo-ETEs) derived from arachidonic acid. Biochim Biophys Acta 1851 (4):340–355. 72. Hanif A, Edin ML, Zeldin DC, Morisseau C, Nayeem MA (2016) Effect of Soluble Epoxide Hydrolase on the Modulation of Coronary Reactive Hyperemia: Role of Oxylipins and PPARγ. PLOS ONE 11 (9):e0162147.

List of figures and tables

Figure 1. Viability of colon cells co-cultured with VSL#3 bacteria in (A) well plate and (B) Transwell inserts, in (C) different proportion of VSL#3 bacteria supernatant and cell culture media mixture in tissue culture plate for 4 hr. Live/dead stain of colon cells co-cultured with VSL#3 for 4 hr and 8 hr in (D) well plate and (E) Transwell inserts; (F) Live/dead stain of colon cells cocultured with VSL#3 supernatant conditioned media for 4 hr, in ratio of 1:1 and 1:10. S: VSL3 supernatant, M: cell culture media; Green live, Red dead.

Figure 2. qPCR analysis of the colon cells co-cultured with VSL#3, E.coli Nissle 1917 and B. adolescentis bacteria in Transwell inserts. (A) Heatmap generated from 28 genes (p-value < 0.05). The gene expression is plotted as log2 fold change. (B) Log2 fold change plot of seven CYPs. Fold change was calculated based on the control group. (n>=3). (C) Ct values of seven primary CYP450s genes normalized to same housekeeping gene. Values represent mean ± SEM (Student’s t test; *p < 0.05). TV: colon cells co-cultured with VSL#3 in Transwell inserts, MOI=1:10; TN: colon cells co-cultured with E.coli Nissle 1917 in Transwell inserts, MOI=1:10; TV: colon cells co-cultured with B. adolescentis in Transwell inserts, MOI=1:10; TC: colon cells cultured in Transwell inserts, without bacteria.

Figure 3. (A) – (C) Biological pathway analysis of colon cells co-cultured with VSL#3, E.coli Nissle 1917 and B. adolescentis in Transwell inserts for upregulated (fold change >3) and downregulated genes (fold change <0.3). All the pathways that are affected by more than 10%. HETE: (16-20)-hydroxyeicosatetraenoic acids; AAD: Aflatoxin activation and detoxification; EET: epoxy; DHET: dihydroxyeicosatrienoic acids; Phase I-FC: Phase I - Functionalization of compounds; PPARA: PPARA activates gene expression; LT: Leukotrienes and EX: Eoxins.

Figure 4. Enzymatic activity of CYP3A4 and CYP1A2 in colon cells co-cultured with (A) VSL3; (C) E.coli Nissle 1917; (E) B. adolescentis; and (B, D, F) their respective supernatant conditioned media in tissue culture plate and Transwell inserts. (G) Enzymatic activity of CYP3A4 and CYP1A2 activity of colon cells when cultured with specific bacterial metabolites (butyrate, propionate and p-cresol). P.MOI: the Multiplicity of Infection bacteria and cells tissue culture plate, ranging from 1:1 to 100:1; T.MOI: the Multiplicity of Infection bacteria and cells Transwell inserts, ranging from 1:1 to 100:1; P.CM: bacterial supernatant conditioned media in well plate with different ratios ranging from 1:1 to 1:100; P.CM: bacterial supernatant conditioned media in Transwell inserts in different ratios ranging from 1:1 to 1:100.

Figure S1. Bacteria standard curve of OD and density (CFU/mL)

Figure S2. Schematic diagram of co-culture between colon cells and bacteria in well plate and Transwell inserts.

Figure S3. qPCR analysis of the colon cells co-cultured with VSL#3, E.coli Nissle 1917 and B. adolescentis bacteria in Transwell inserts. Heatmap generated from 84 genes.

Figure S5. Biological pathways analysis of colon cells co-cultured with VSL#3 bacteria in Transwell inserts based on genes fold change >3 and <0.3.

Figure S6. Biological pathways analysis of colon cells co-cultured with E.coli Nissle 1917 bacteria in Transwell inserts based on genes fold change >3 and <0.3.

Table 1. Up- and downregulated genes influenced by VSL#3, E.coli Nissle 1917, and B. adolescentis. Fold change >3 was considered as upregulated and < 0.3 as downregulated.

A B C

100 100 100

75 75 75

50 50 50 % Viability % Viability 25 % Viability 25 25

0 0 0 Ctrl 2 4 8 Ctrl 2 4 8 0 0.5 0.1 0.04 0.02 0.01

Coculture time (h) Coculture time (h) proportion of VSL#3 bacteria supernatant

D E F

4 hrs 8 hrs 4 hrs 8 hrs S:M=1:1 S:M=1:10

Log2 (Fold Change) (Fold Log2 -1 TB -2 -3

C 20 18 16 14 * * * 12 * TV * 10 TN

ΔCt 8 6 TB 4 TC 2 0

FMO oxidises nucleophiles 28.6

Fatty acids 16.7 28.6

Synthesis of EET and DHET 16.7 28.6

Xenobiotics 33.3 28.6

Biological Pathways Biological Endogenous sterols 33.3 14.3

100 0 100

TN (%) C UP DOWN Synthesis of LT and EX 22.2 7.9 Eicosanoids 22.2 10.5 Synthesis of HETE 22.2 13.2 Fatty acids 22.2 15.8 Xenobiotics 44.4 18.4

Biological Pathways Biological RA biosynthesis pathway 33.3 18.4 Endogenous sterols 21.1

100 0 100 TB (%) Figure 3. (A) – (C) Biological pathway analysis of colon cells co-cultured with VSL#3, E.coli Nissle 1917 and B. adolescentis in Transwell inserts for upregulated (fold change >3) and downregulated genes (fold change <0.3). All the pathways that are affected by more than 10%. HETE: (16-20)-hydroxyeicosatetraenoic acids; AAD: Aflatoxin activation and detoxification; EET: epoxy; DHET: dihydroxyeicosatrienoic acids; Phase I-FC: Phase I - Functionalization of compounds; PPARA: PPARA activates gene expression; LT: Leukotrienes and EX: Eoxins. A B VSL#3 3A4 VSL#3 supernatant 3A4 4 1A2 4 1A2 3 3 2 2 1

Fold Fold change 1 0 Fold change 0 Figure 4. Enzymatic activity of CYP3A4 and CYP1A2 in colon cells co-cultured with (A) VSL3; (C) E.coli Nissle 1917; (E) B. adolescentis; and (B, D, F) their respective E.coli Nissle 1917 supernatant conditioned media in tissue C 3A4 D E.coli Nissle 1917 supernatant 3A4 culture plate and Transwell inserts. (G) 4 1A2 4 1A2 Enzymatic activity of CYP3A4 and CYP1A2 3 3 activity of colon cells when cultured with 2 specific bacterial metabolites (butyrate, 2 1 propionate and p-cresol). P.MOI: the Fold Fold change

Fold Fold change 1 0 Multiplicity of Infection bacteria and cells 0 tissue culture plate, ranging from 1:1 to 100:1; T.MOI: the Multiplicity of Infection bacteria and cells Transwell inserts, ranging from 1:1 to 100:1; P.CM: bacterial supernatant B. adolescentis conditioned media in well plate with different E B. Adolescentis supernatant 3A4 F 3A4 ratios ranging from 1:1 to 1:100; P.CM: 4 4 bacterial supernatant conditioned media in 3 1A2 1A2 3 Transwell inserts in different ratios ranging 2 2 from 1:1 to 1:100. 1 Fold Fold change 1 0 Fold change 0

Metabolites G 4

2 3a4 Fold Fold change 1 1a2

0 caco2+2mM caco2+3mM caco2+2mM butyrate propionate p-cresol TV TN TB

CYP11A1, CYP11B2, CYP17A1, CYP19A1, CYP1A1, CYP1B1, CYP21A2, CYP26A1, CYP26C1, CYP2A13, CYP2C8, CYP2C9, CYP2F1, CYP3A4, CYP3A43, CYP4A11, CYP1A1, CYP1B1, CYP4A22, CYP4B1, CYP4F2, CYP1A2, CYP3A4, CYP4F2, CYP2A13, CYP4F2, ADH4, Upregulated (fold change > 3) CYP4F8, CYP7A1, CYP7B1, AADAC FMO1, FMO2 CYP8B1, ADH1A, ADH4, ADH7, AADAC, GZMB, UCHL1, ALDH1A2, ALDH1A3, ALDH3A1, ALDH8A1, FMO1, FMO2, FMO3, FMO4, MAOB, PTGS2

CYP2B6, CYP2D6, CYP3A5, CYP11B1, CYP2C18, CYP11B1, CYP2C19, CYP4F11, CYP4F12, ADH1C, Downregulated (fold change < 1/3) CYP4F11, ADH1B, GZMA, CYP4F11, ADH1B ADH6, HSD17B10, PTGS1 MAOA, PTGS1

Table 1. Up- and downregulated genes influenced by VSL#3, E.coli Nissle 1917, and B. adolescentis. Fold change >3 was considered as upregulated and < 0.3 as downregulated. E.coli Nissle 1917 B. adolescentis VSL#3 3.5 0.2 2.5 3 2 2.5 0.15 2 1.5 0.1 1.5 1 1 y = 126.7x - 0.1084 y = 3.8371x + 0.0413

CFU/ml (*106) 0.05 CFU/ml (*106 ) y = 1.7638x - 0.5133 R² = 0.9754 R² = 0.9549 0.5 0.5 CFU/ml (*106 ) R² = 0.9842 0 0 0 0 0.01 0.02 0.03 0 0.01 0.02 0.03 0.04 0 0.5 1 1.5 2 OD OD OD1 Linear (OD1)

Figure S1. Bacteria standard curve of OD and density (CFU/mL) Cell culture media Cell culture media

Colon cells Colon cells Cell culture media

Conditioned Cell culture media media

Colon cells Colon cells Conditioned media

Cell culture media Microbiome

Colon cells Colon cells Microbiome

Figure S2. Schematic diagram of co-culture between colon cells and bacteria in well plate and Transwell inserts. Figure S3. qPCR analysis of the colon cells co-cultured with VSL#3, E.coli Nissle 1917 and B. adolescentis bacteria in Transwell inserts. Heatmap generated from 84 genes. CYP450 10.0 8.0 6.0

4.0 TV 2.0 TN 0.0 TB -2.0 -4.0 CYP2F1 CYP4F2 CYP4F3 CYP4F8 CYP2S1 CYP2E1 CYP2C8 CYP2C9 CYP1B1 CYP2B6 CYP4B1 CYP7B1 CYP8B1 CYP2R1 CYP1A1 CYP1A2 CYP3A4 CYP3A5 CYP3A7 CYP7A1 CYP2D6 CYP2W1 CYP4F11 CYP4F12 CYP26C1 CYP2C18 CYP2C19 CYP11B1 CYP11B2 CYP26B1 CYP27B1 CYP11A1 CYP17A1 CYP19A1 CYP21A2 CYP24A1 CYP26A1 CYP27A1 CYP2A13 CYP3A43 CYP4A11 CYP4A22

ADH Esterases ALDH FMO 10 10 10 10 8 8 8 8 6 6 6 6 4 4 4 TV 4 2 TV TV TV 2 2 TN 2 0 TN 0 TN TN 0 0 -2 TB TB -2 TB TB -2 -4 -4 -2 -4 -4 ADH4 ADH5 ADH6 ADH7 DHRS2 ADH1C ADH1B ADH1A ALDH2 ALDH3B2 HSD17B10 ALDH1A1 ALDH1A3 ALDH3A2 ALDH5A1 ALDH7A1 ALDH9A1 FMO1 FMO2 FMO3 FMO4 FMO5

Figure S4. Log2 Fold Change plot of the all 84 genes in the kit. Fold change was calculated based on the control group and separated to 6 different enzymes group. CYP450: Cytochrome P450s; ADH:Alcohol Dehydrogenases; ALDH: Aldehyde Dehydrogenases; FMO: Flavin Containing Monooxygenases. COX reactions 16.7 Defective CYP11B1 causes Adrenal hyperplasia 4… 16.7 UP DOWN Defective MAOA causes Brunner syndrome… 16.7 Enzymatic degradation of Dopamine by… 16.7 Enzymatic degradation of dopamine by COMT 16.7 Norepinephrine Neurotransmitter Release Cycle 16.7 Biogenic amines are oxidatively deaminated to… 16.7 Metabolism of serotonin 16.7 Interleukin-4 and Interleukin-13 signaling 16.7 Synthesis of Prostaglandins (PG) and… 16.7 Endogenous sterols 16.7 Ethanol oxidation 16.7 Glucocorticoid biosynthesis 16.7

Eicosanoids 16.7 16.7 Fatty acids 16.7 16.7 Aromatic amines can be N-hydroxylated or N-… 16.7

Biological Biological Pathways Methylation 16.7 Synthesis of Leukotrienes (LT) and Eoxins (EX) 16.7 16.7 PPARA activates gene expression 16.7 Phase I - Functionalization of compounds 16.7 Miscellaneous substrates 16.7 33.3 Biosynthesis of maresin-like SPMs 33.3 Biosynthesis of protectins 33.3 Synthesis of epoxy (EET) and… 33.3 Aflatoxin activation and detoxification 33.3 Synthesis of (16-20)-hydroxyeicosatetraenoic… 50.0 Xenobiotics 16.7 50.0 100 80 60 40 20 0 20 40 60 80 100 TV (%)

Figure S5. Biological pathways analysis of colon cells co-cultured with VSL#3 bacteria in Transwell inserts based on genes fold change >3 and <0.3. Defective CYP27A1 causes Cerebrotendinous… 16.7 Defective CYP11B1 causes Adrenal hyperplasia… 16.7 UP DOWN Synthesis of bile acids and bile salts via 27-… 16.7 Synthesis of bile acids and bile salts via 24-… 16.7 Synthesis of bile acids and bile salts via 7alpha-… 16.7 Glucocorticoid biosynthesis 16.7 Synthesis of (16-20)-hydroxyeicosatetraenoic… 16.7 42.9 FMO oxidises nucleophiles 28.6 Fatty acids 16.7 28.6 Synthesis of epoxy (EET) and… 16.7 28.6 Xenobiotics 33.3 28.6 Eicosanoids 16.7 14.3 Miscellaneous substrates 16.7 14.3 CYP2E1 reactions 16.7 14.3

Biological Biological Pathways Defective CYP1B1 causes Glaucoma 14.3 Biosynthesis of protectins 14.3 Synthesis of Leukotrienes (LT) and Eoxins (EX) 16.7 14.3 Endogenous sterols 33.3 14.3 RA biosynthesis pathway 14.3 Aflatoxin activation and detoxification 14.3 Ethanol oxidation 16.7 14.3 PPARA activates gene expression 14.3 100 80 60 40 20 0 20 40 60 80 100

TN (%)

Figure S6. Biological pathways analysis of colon cells co-cultured with E.coli Nissle 1917 bacteria in Transwell inserts based on genes fold change >3 and <0.3. 11.1 rRNA processing in the mitochondrion 11.1 UP DOWN 11.1 tRNA modification in the mitochondrion 11.1 11.1 Biosynthesis of DPAn-3 SPMs 2.6 2.6 Biosynthesis of electrophilic ω-3 PUFA oxo-derivatives 2.6 2.6 Synthesis of 15-eicosatetraenoic acid derivatives 2.6 2.6 NOTCH2 intracellular domain regulates transcription 2.6 2.6 Defective CYP7B1 causes Spastic paraplegia 5A, autosomal… 2.6 2.6 Defective CYP21A2 causes Adrenal hyperplasia 3 (AH3) 2.6 2.6 Biosynthesis of protectins 2.6 2.6 Defective CYP17A1 causes Adrenal hyperplasia 5 (AH5) 2.6 2.6 Defective CYP11B2 causes Corticosterone methyloxidase 1… 2.6 2.6 Biogenic amines are oxidatively deaminated to aldehydes by… 2.6 2.6 Estrogen biosynthesis 2.6 2.6 Pregnenolone biosynthesis 2.6 2.6 Abacavir metabolism 2.6 2.6 Androgen biosynthesis 2.6 5.3

Biological Biological Pathways Vitamins 5.3 5.3 Synthesis of Prostaglandins (PG) and Thromboxanes (TX) 11.1 5.3 11.1 5.3 Phase I - Functionalization of compounds 5.3 5.3 FMO oxidises nucleophiles 7.9 11.1 7.9 Ethanol oxidation 22.2 7.9 7.9 Synthesis of bile acids and bile salts via 27-hydroxycholesterol 7.9 7.9 Glucocorticoid biosynthesis 7.9 22.2 10.5 Synthesis of epoxy (EET) and dihydroxyeicosatrienoic acids… 10.5 22.2 13.2 Synthesis of Leukotrienes (LT) and Eoxins (EX) 11.1 13.2 22.2 15.8 Synthesis of (16-20)-hydroxyeicosatetraenoic acids (HETE) 15.8 44.4 18.4 Xenobiotics 33.3 18.4 11.1 18.4 Endogenous sterols 21.1 100 80 60 40 20 0 20 40 60 80 100

TB (%)

Supplementary Figure S5. Biological pathways analysis of colon cells co-cultured with B. adolescentis bacteria in Transwell inserts based on genes fold change >3 and <0.3