The colorectal adenocarcinoma cell lines (CRC cells) HT-29, SW480, HCT116, and Caco-2 (www.atcc.org) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Lonza) supplemented with 10% fetal calf serum (FCS), 20 mM HEPES, 2 mM L-glutamine, and 1× non-essential amino acids (Gibco) at 37°C/5% CO₂. For gene-expression analysis, cells were serum starved to 1% FCS for 24 h before co-culturing with bacterial cells. These culturing conditions were used unless stated otherwise.

The following Streptococcus bovis strains were used: S. gallolyticus subsp. gallolyticus strains UCN34 (Rusniok et al., 2010), NTB1 (Boleij et al., 2011), 1293, 1294 (Tripodi et al., 2005), and NTB12; Streptococcus infantarius NCTC8133 (Biarc et al., 2004) (NZ_LR594042.1); Streptococcus lutetiensis NTB2; S. gallolyticus subsp. macedonicus strains CIP105685T, 19AS, and ACA-DC205; and S. gallolyticus subsp. pasteurianus strains 992 and NTB7. A phylogenetic tree of these S. bovis strains was published in Taddese et al., 2020 (Taddese et al., 2020). Other bacterial strains were E. faecalis 19433 (www.atcc.org), Enterobacter cloacae NTB9, Staphylococcus lugdunensis NTB8, Salmonella typhimurium NTB6, Escherichia coli Nissle 1917, and E. coli NTB5 (Boleij et al., 2011). All strains were grown on Columbia blood agar or in brain heart infusion (BHI) broth (Difco) supplemented with 1% glucose at 37°C/5% CO₂. E. coli was grown at 200 rounds per minute (rpm).

For MTT assays, optimal cell seeding in 96-well plates (Greiner Bio-One, Austria) was defined at 5,000 cells/well for Caco-2, 1,500 cells/well for HCT116, 10,000 cells/well for HT29, and 6,000 cells/well for SW480 to allow for 72 h growth observation. Cells were incubated overnight to attach before addition of bacterial secretomes in a fourfold dilution in DMEM supplemented with 10% FCS. Secretomes were collected as follows. After culturing in BHI broth, bacteria were centrifuged at 4,700 rpm for 20 min and filter sterilized using 0.2-µm filters (Sigma-Aldrich, USA). Molecules larger than 10 kDa were concentrated using Amicon ultra-15 centrifugal filters (Merck Millipore, Merck, USA). Concentrated secretomes were frozen at −80°C until further use. MTT-assays were performed as described previously, and MTT assay was performed after 24, 48, and 72 h. Metabolic activity of CRC cells measured by MTT assay was used as a measure for cell growth. All experiments were performed in quadruplicate. Area under the curve was calculated for each condition (strain) and compared to cells incubated with cell culture media containing BHI broth as control using independent sample t-test in GraphPad Prism 9.

HT-29 cells were grown to confluence in T75 flasks and subsequently incubated with or without S. gallolyticus [multiplicity of infection (MOI) of 20] in two duplicate experiments. After 2 h, the culture medium was refreshed, and incubation was prolonged for another 2 h. This 4-h incubation period was chosen for microarray analysis because induction of the COX-2 marker for chronic inflammation and CRC progression (Ellmerich et al., 2000) could be determined under these conditions. Next, HT-29 monolayers were washed three times with prewarmed PBS after which RNA was isolated according to the Qiagen protocol (Qiagen RNeasy kit) with on-column DNA digestion. RNA quantity was measured with Nano-drop (Thermo Scientific, USA), and RNA quality was checked with the Bio-analyzer 2000 (Agilent Technologies, USA). Samples with RNA integrity scores ≥9.0 were approved for microarray analysis.

Gene expression profiling was performed using Affymetrix GeneChip Human Gene 1.0 ST arrays, representing all known human genes (Affymetrix Inc., Santa Clara, CA, USA). The Affymetrix GeneChip Whole Transcript Sense Target Labeling Assay was used to generate amplified and biotinylated sense-strands DNA targets from the entire expressed genome (1.0 µg of total RNA). The manufacturer’s manual was followed for the hybridization, washing, and scanning steps (version 4, P/N 701880 Rev. 4). Arrays were hybridized by rotating them at 60 rpm in the Affymetrix GeneChip hybridization oven at 45°C for 17 h. After hybridization, the arrays were washed in the Affymetrix GeneChip Fluidics station FS 450. Arrays were scanned using the Affymetrix GeneChip scanner 3000 7G system.

The Affymetrix CEL files were first imported into Affymetrix Expression Console version 1.1 where control probes were extracted using the default robust multichip averaging (RMA) algorithm in order to perform quality analysis checks. The Area Under the Curve (AUC) of the Receiver Operator Characteristic was calculated using the positive and negative control probes. All arrays had an AUC score above the empirically defined threshold of 0.85 indicating a good separation of the positive controls from the negative controls.

Subsequently the CEL files were imported into Partek^® (Partek^® Genomic Suite software, version 6.4 Copyright^© 2008 Partek Inc., St. Louis, MO, USA) where only core exons were extracted and normalized using the RMA algorithm with GC background correction. Core transcript summaries were calculated using the mean intensities of the corresponding probesets. The correspondence of the replicate samples was confirmed using principle component analysis (PCA) and Pearson correlation analysis. After grouping the samples by cell type, an ANOVA was performed on the log2 intensities, and contrast p-values were calculated for all pair-wise comparisons between the sample groups. Genes with a fold change of at least 1.5 and a p-value below 0.05 were selected for further evaluation. Raw data files are deposited in the Gene Expression Omnibus (GEO) database, available through: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=jtcvzsakgwaiibo&acc=GSE29295.

For determination of CYP1A1 and ALDH1A3 gene expression in time-series experiments, CRC cell lines were incubated with S. gallolyticus (MOI, 20) for the indicated time periods. To determine concentration dependence and additive effects of 1 and 10 nM 3-methylcholanthrene (3MC; Sigma) on CYP gene expression, cells were incubated for 4 h with the indicated bacterial MOIs. To determine AhR-dependent CYP gene expression, CRC cells were incubated in the presence of the AhR-antagonist CH-223191 (3 µM; Calbiochem) for 1 h and subsequently incubated with S. gallolyticus (MOI, 20) and/or 1 nM 3MC. These relatively low concentrations of 3MC (1 and 10 nM) were used to avoid saturation of CYP1A1 induction by 3MC, which allowed to determine the additive effect of S. gallolyticus on CYP1A1 expression under these conditions. To determine the effect of bacteria-associated factors on CYP gene expression, S. gallolyticus, E. faecalis, or E. coli were cocultured with Caco-2 cells for 6 h. Then, the supernatants were spun for 10 min at 4,000 g and subsequently passed through a 0.2-µm filter to remove bacteria. Next, fresh serum-starved Caco-2 cells (24 h) were incubated with the secretomes for the indicated time periods. Caco-2 and HT-29 cells were incubated with the bacterial strains at a MOI of 20 for 2 and 4 h to determine the differences between CYP1A1 induction for the different S. bovis subspecies in comparison with the Gram-positive bacterium E. faecalis and the Gram-negative bacterium E. coli. The above-described interventions were all performed in at least two replicate experiments. CRC cells were disrupted in RLT-lysis buffer, and RNA was extracted according to the Qiagen protocol (RNeasy Mini-Kit). Next, Iscript Reverse Transcriptase PCR (Bio-Rad) was performed to synthesize 1 µg of cDNA under the following conditions: 5 min at 25°C, 30 min at 42°C, and 5 min at 85°C. Expression of the genes listed in Supplementary Table S1 was compared to the expression of the household gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (4310884E gene expression assay; Applied Biosystems) using the following quantitative real-time PCR (qPCR) protocol: 2 min at 50°C, 10 min at 95°C, and 40 cycles of 15 s at 95°C and 60 s at 60°C (7900 HT, Applied Biosystems). All genes measured with custom primers (Biolegio) were analyzed with SYBR green Mastermix (Applied Biosystems), and all gene-expression assays were analyzed with Universal Mastermix (Applied Biosystems). The relative quantity (RQ) values were calculated via the ΔΔCt method (Pfaffl, 2001) using SDS 2.2.1 software. The log2 values of the ΔΔCt values were plotted on a linear scale, whereby the 0 level represents no expression change. For all data with two grouping variables, two-way ANOVAs were performed, and for data with one grouping variable, one-way ANOVA or Student’s t-test were performed in GraphPad Prim 5. Differences were considered significant below p-value of 0.05.

Expression of CYP1A1 was evaluated at the protein level in Caco-2 cells, as these cells are best suited to measure CYP1A1 at the protein level (Meunier et al., 1995). However, it should be noted that also in this case, CYP1A1 levels are below the level of detection without addition of specific stimulating agents (Rosenberg and Leff, 1993). The experiments of this study were further challenged by the fact that only viable bacterium strongly stimulates CYP1A1, whereas prolonged coincubations with viable bacteria resulted in the death of Caco-2 cells. To overcome these challenges, confluent Caco-2 monolayers were incubated with 1 µM 3MC or cocultured with S. gallolyticus at a MOI of 50. The use of this high 3MC concentration and high MOI allowed a rapid and relatively strong induction of CYP1A1, while cellular damage by bacterial proliferation at ≤6 h time points was minimized. After 6 h of incubation, Caco-2 cells were refreshed with medium containing 10 µg/ml chloramphenicol to block further bacterial growth for prolonged incubations. After 6, 9, and 12 h of incubation, cells were dislocated in PBS-EDTA and disrupted in lysis buffer (50 mM HEPES, 0.5 M NaCl, 1.5 mM MgCl₂, 1 mM EDTA, 10% glycerol, and 1% Trition-X100) (Anwar-Mohamed and El-Kadi, 2009). As control of induction and to test several different CYP1A antibodies (Supplementary Information) Caco-2 cells were incubated with 1 μM 3MC for 20 h or non-treated and subsequently lysed. Lysates were vortexed every 10 min during incubation on ice for 1 h. Next, samples were spun at 12,000 g for 10 min and stored at −20°C until further analysis. Fifty micrograms of protein in sodium dodecyl sulfate (SDS) sample buffer (50 mM Tris–HCl, pH 6.0, 2% SDS, 5% β-mercaptoethanol, and 10% glycerol) was incubated for 5 min at 95°C prior to 12.5% glycine SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (Towbin et al., 1979). Proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Amersham) by Western blotting. Next, membranes were blocked for 1 h in 5% bovine serum albumin (BSA) in Tris-buffered saline supplemented with Tween 20 (0.1%) (TBS-T) and incubated with monoclonal mouse anti-CYP1A1 antibodies (Santa Cruz) diluted 1:500 in TBS-T, monoclonal mouse anti-β-actin antibody (Sigma A5441) diluted 1:25,000 in TBS-T, monoclonal mouse anti-CYP1A1/1A2 antibodies (K06; collection of Dr. W. Peters) diluted 1:800 in TBS-T, or polyclonal rabbit anti-CYP1A1 antibody (Human Biologics Inc.) diluted 1:1,000 in 5% BSA, or polyclonal rabbit anti CYP1A1/1A2 antibody (Human Biologics Inc.) diluted 1:800 in 5% BSA. Bound antibodies were visualized with the ECL detection system (Amersham) using antimouse immunoglobulin G (IgG) horseradish peroxidase (HRP) conjugates diluted 1:50,000 or antirabbit IgG HRP conjugates diluted 1:50,000 in 5% milk in TBS-T (Jackson ImmunoResearch).

After incubation for 6, 9, and 12 h, Caco-2 cells were washed with ice-cold PBS and lysed in 0.1M potassium phosphate buffer (0.1M dipotassium phosphate, 0.1M EDTA, and 20% glycerol, pH 7.4). Note that viable bacteria were removed after the 6-h time point to avoid premature cell death by excessive bacterial proliferation as described in the previous section. Cells were disrupted by mechanical shearing on ice (crude cell lysates) and stored at −80°C until use. Protein concentrations were determined with the Bradford protein assay following the manufacturers’ description (BioRad). Crude cell lysates were diluted to 100 µg/ml protein in assay buffer (0.1M dipotassium phosphate and 0.1M EDTA, pH 7.4), and 4 µM ethoxyresorufin (Molecular Probes) was added. Samples were prewarmed at 37°C before addition of 10 µM reduced nicotinamide adenine dinucleotide phosphate (NADPH) (Sigma). The conversion of ethoxyresorufin (non-fluorescent) into resorufin (fluorescent) is catalyzed by CYP1A1 and requires the presence of NADPH. The reaction was started by the addition of NADPH and resorufin production in the samples was measured during a 10-min time period on a Shimadzu RF-5000 spectrofluorometer at an excitation of 550 nm and emission of 580 nm. Known concentrations of resorufin (Molecular Probes) were used to create a calibration curve to calculate the amount of resorufin formed from ethoxyresorufin, catalyzed by the CYP1A1 enzyme (Yu et al., 2001). To compare individual samples, the amount of resorufin produced per minute was related to the protein concentration in the sample (resorufin/min·mg). The amount of resorufin produced per minute per milligram is a measure of the CYP1A1 activity in the sample. Two-way ANOVA statistics was used to determine significant changes.

The additional effects of bacterial products on DNA damage by 0.1 µM 3MC were measured using the COMET assay (Trevigen). Secretomes of S. gallolyticus, E. faecalis, and E. coli were incubated with Caco-2 cells in quadruplicate. After priming Caco-2 cells for 6 h with these secretomes, 0.1 µM 3MC was added, and incubation was prolonged for 18 h. This relatively low concentration of 3MC was used to allow assessment of a possible additive effect of bacterial stimulation on 3MC toxicity. Higher concentrations could mask such an additive effect due to saturation of CYP1A1 induction by 3MC itself. H₂O₂ (100 µM) was used as positive control for DNA damage. After incubation of Caco-2 cells with the supernatants, 3MC or H₂O₂, cells were washed with warm PBS, harvested by trypsin treatment and spun at 200 g for 5 min. Cells were counted, and 8,000 cells per condition were washed with Mg- and Ca-free ice-cold PBS. Next, cells were dissolved in low-melting agarose and mounted on COMET slides. After 30 min at 4°C in the dark to solidify the agarose, the immobilized cells were disrupted in Trevigen lysis solution for 60 min at 4°C. After lyses of the cells in the solidified agarose, DNA was unwound by alkaline treatment (pH >13) for 30 min. Subsequent electrophoresis for 30 min at 25 V was employed to yield migration of DNA in the agarose. In this assay, the migration distance of DNA is a measure of the amount of DNA damage under a certain condition. After electrophoresis, COMET slides were washed in 0.4M Tris–HCl (pH 7.5) and fixed for 10 min in absolute alcohol. DNA damage (observed as COMET-like structures) was visualized by SYBRgreen staining and fluorescence microscopy (Leica). A total of 100 COMETs per condition were independently scored by two researchers that were uninformed about the experimental conditions, following the methodology of Collins et al. (Collins, 2004). DNA damage was calculated as reference to the positive H₂O₂ control. One-way ANOVA in GraphPad Prism 4.00 was performed to determine significance of the results (p < 0.05).

The mouse strains used in this study were C57bl6 purchased from Jackson Laboratory (Bar Harbor, ME, USA) or obtained as littermates of in-house breeding. We administered kanamycin (1 g/L) and doxycyclin (100 mg/L) for 5 days in drinking water, followed by oral inoculation with the S. gallolyticus strain UCN34 (~5 × 10⁷ and 5 × 10⁸ bacteria in PBS) or PBS alone (sham control) in mice at 4 weeks of age. Simultaneously with antibiotics pretreatment, mice were orally gavaged with the AhR-inhibitor CH223191 (10 µg/g/day) dissolved in corn oil (vehicle) or with vehicle alone, continued until sacrifice at day 7 postinoculation of S. gallolyticus. Mice were randomly placed in groups, making sure that there was no cage effects for mice experiments. Sham mice were housed separately from S. gallolyticus colonized mice. We quantified fecal bacterial colonization as colony-forming units (CFUs) per gram stool on BHI agar supplemented with 5 µg/ml doxycyclin. Single colonies were boiled in 50 μl dH2O, and S. gallolyticus colonization was confirmed with SodA PCR (SodA d1- CCTTATGCATATGATGCTCTTGAGCC, SodA d2-AGATAGTAAGCGTGTTCCCAAACGTC, 488 bp product; DNA was denatured at 95°C for 10 min followed by 40 cycles at 94°C for 30 s, 56°C for 35 s, and 72°C for 72 s, and subsequent elongation for 7 min at 72°C) (Poyart et al., 2002). At experimental time points, we harvested one piece each of the cecum in Trizol reagent for RNA analysis. qPCR was performed on RNA extracted from Trizol using TaqMan gene expression assays for CYP1A1 (Mm00487218_m1), IL4 (Mm00445259_m1), AhR (Mm00478932_m1), and PTGS2 (Mm00478374_m1) relative to 18s (4318839) and standard operation conditions on 7500 fast system (Applied Biosystems). The colons were Swiss rolled, paraffin embedded, and subsequently sectioned at 4 μm for H&E/periodic acid–Schiff (PAS) staining or IHC ki67 analysis. Histology was reviewed by an expert GI pathologist. All mice were kept in specific pathogen-free (SPF) conditions prior to S. gallolyticus colonization in the Johns Hopkins University (JHU) animal facility. The mouse protocols were approved by the Johns Hopkins University Animal Care and Use Committee in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care International.

To measure serum response to S. gallolyticus, 96-well plates were coated with 50 µg/ml collagen IV from rat tail in 0.02M acetic acid for 1 h. Plates were washed twice in PBS. Next, plates were coated with 1 × 10⁹ S. gallolyticus UCN34 per well o/n in PBS at 37°C. Plates were fixed in 4% paraformaldehyde for 1 h at room temperature and stored at 4°C until further use. Wells were blocked with 3% skim milk in PBS–0.5% Tween-20 (PBS-T). Serum was diluted in 3% skim milk in PBS-T 1:50 and incubated for 1.5 h at room temperature with gentle shaking. A serum-free control was used as reference together with non-infected mouse serum. Plates were washed four times in PBS-T for 5 min and subsequently incubated with antimouse-HRP in 3% skim milk in PBS-T 1:5,000 for 1 h at room temperature. Plates were washed four times for 5 min in PBS-T and developed with 100 µl TMB substrate in 5–10 min. The reaction was stop with 50 µl 2N H₂SO₄. The OD was measured at 450 nm in a BioRad plate reader with reference filter at 655 nm within 30 min after stopping the reaction.

To unravel the effects of S. gallolyticus on gene expression of CRC cells, a transcriptome analysis was performed 4 h after coculturing of S. gallolyticus UCN34 with HT29 CRC cells. Microarray analysis showed a total of 44 significantly differentially expressed genes upon S. gallolyticus exposure (Table 1), which was a surprisingly low number in comparison to about 150 differentially expressed genes upon coincubation with other non-pathogenic bacteria under the same conditions (GEO-GSE29295). Strikingly, as much as nine of these S. gallolyticus-regulated genes appeared to belong to oxidoreductase pathways (Table 1 in bold), including CYP1A1, ADH1A, and ALDH1A3 (Supplementary Figure S1). For validation, 18 (out of 44) genes with normalized expression levels >6 on the used Affymetrix GeneChip were selected. The corresponding log-2 values for microarray and qPCR of 10 of these genes showed similar levels by both methods as predicted by Pearson correlation (r = 0.94; p < 0.001) (Figures 1A, B). In line with microarray results, CYP1A1 (log 2 RQ, 1.94) and ALDH1A3 (log 2 RQ, 1.02) were also found to be the most significant differentially expressed genes by qPCR (Figure 1 and Supplementary Figure S2). Taken together, these data imply that S. gallolyticus induces the expression of CYP1A1 and ALDH1A3. Interestingly, the corresponding enzyme activities of these genes are involved in biotransformation of drugs and food components and are best known for their clearing function in the liver (Nebert and Dalton, 2006). As the expression of these enzymes is elevated during CRC progression (McKay et al., 1993; Ding and Kaminsky, 2003; Huang et al., 2009), and thereby could be related to colon carcinogenesis, we decided to further focus on these genes.

The temporal induction of CYP1A1 and ALDH1A3 expression was examined in the cell lines HT29, Caco-2, SW480, and HCT116 after exposure to S. gallolyticus. CYP1A1 was consistently upregulated (p < 0.001) in all cell lines upon 2, 4, and 6 h of coculturing with S. gallolyticus (Figure 2A). The induction of ALDH1A3 was less evident among the different cell lines, although ALDH1A3 was significantly increased to a log2 RQ between 1.1 for Caco-2 and 6.9 for HT-29 after 4 and 6 h (p < 0.001) (Figure 2B). Next, CYP1A1 induction was evaluated at the protein and functional level. To this purpose, Western blot analysis was performed on protein extracts from Caco-2 cells harvested at different time points after S. gallolyticus coincubation. Our initial experiments, and the results of others (Meunier et al., 1995), indicated that Caco-2 cells were best suited to detect CYP1A1 at the protein level, and CYP1A1 is only detectably present after stimulation with inducing agents, such as 3-methylcholanthrene (3MC), with maximum protein expression levels after 18–20 h (Rosenberg and Leff, 1993). Because such long incubation times are not feasible with live S. gallolyticus, Caco-2 cells were stimulated with S. gallolyticus at a high MOI of 50 for 6 h to obtain a short but strong induction, after which bacteria were removed for prolonged incubations of 9 and 12 h (see Materials and Methods). 3MC was used in the same time course as a control for CYP1A1 induction in Caco-2 cells (Aboutabl et al., 2009), and human liver microsomes expressing CYP1A1 at high level were loaded as positive control on the Western blot. S. gallolyticus induced cellular CYP1A1 levels (~58 kDa) after 6 and 9 h compared to untreated Caco-2 cells (Figure 2C and Supplementary Figure S3), which was reduced to background levels after 12 h. 3MC continued to increase CYP1A1 levels up to 12 h of stimulation. To validate these findings, CYP1A1 enzyme activity was determined in Caco-2 by the 7-ethoxyresorufin-O-deethylase (7-EROD) assay that measures conversion of ethoxyresorufin into resorufin by functional CYP1A1 enzyme. Resorufin production was found to be significantly increased by S. gallolyticus after 6 h incubation (p < 0.05), while 3MC increased CYP1A1 activity at all three time points (p < 0.001) (Figure 2D). Thus, these activity data corroborate the real-time and Western blot data that show increased expression of CYP1A1 upon incubation with viable S. gallolyticus cells.

Polycyclic aromatic hydrocarbon (PAH) substrates, such as 3MC, of CYP1 enzymes interact with the intracellular AhR to induce CYP1 expression. After the subsequent binding of this AhR-PAH complex to several cofactors, it translocates to the nucleus to activate a xenobiotic response element (XRE) that is present in the promoter regions of the CYP1A1, CYP1A2, and CYP1B1 genes (McMillan and Bradfield, 2007) (Figure 3A). To examine whether S. gallolyticus also targets AhR, the induction of CYP1A1, CYP1A2, and CYP1B1 was examined in HT29 cells in combination with 3MC. These experiments showed that S. gallolyticus significantly increased CYP1A1, CYP1A2, and CYP1B1 and had an additive effect on 0.001 µM 3MC-mediated CYP1A1 induction (Figure 3B; p < 0.05). S. gallolyticus also had a slight additive effect on 3MC stimulation for both CYP1A2 and CYP1B1 (Figure 3B), although this was not significant. Furthermore, S. gallolyticus induced CYP1 expression in a concentration-dependent manner, similar to that of CYP1A1 (Figure 3C; p < 0.001). An AhR antagonist that prevents the binding of PAHs to this receptor (Kim et al., 2006) was added to cocultures to evaluate whether S. gallolyticus also depends on AhR. Preincubation of HT-29 cells with this AhR inhibitor clearly blocked the upregulation of CYP1A1 (Figure 3D), CYP1A2, and CYP1B1 (Supplementary Figure S4) by both 3MC and S. gallolyticus. Taken together, these results show that CYP1 induction by S. gallolyticus is mediated by an AhR-dependent mechanism.

To investigate whether the induction of CYP1A1 is a specific feature of S. gallolyticus, the expression of this gene was investigated upon coincubation of HT-29 cells with other S. bovis and other intestinal bacterial strains. Besides S. gallolyticus UCN34, the other S. bovis group strains S. macedonicus CIP105685T, S. infantarius NCTC8133, and S. gallolyticus NTB1 and 1293 increased CYP1A1 gene expression (Supplementary Figure S5A). From the other intestinal bacteria, E. faecalis and E. cloacae were strong inducers of CYP1A1 gene expression in a concentration-dependent manner (p < 0.01) (Supplementary Figure S5B). Intermediate activity was observed for S. lugdunensis, E. coli Nissle 1917, and S. typhimurium. Contrarily, coincubation of HT-29 cells with E. coli NTB5 did not result in an induction of CYP1A1 and might even reduce the expression of this gene (Supplementary Figure S5C). Moreover, the additive effect of S. gallolyticus in combination with 0.001 µM 3MC (Figure 3B) was not observed for E. coli NTB5 (Supplementary Figures S5C, D). Thus, these data show that CYP1 induction is not specific for S. gallolyticus and might confer a more general mechanism among a subset of intestinal bacteria.

A remarkable finding was the fact that the intracellular AhR is stimulated by S. gallolyticus cells, which lack the ability to invade CRC cells (Boleij et al., 2011). This suggests that these bacteria release a factor themselves, convert and activate a medium component, or provoke the expression of a diffusible host factor that re-enters CRC cells to induce CYP1A1 expression. To confirm this hypothesis, secretome from S. gallolyticus (SGS) was filtered and added to fresh Caco-2 cells for 4, 8, and 12 h and to HT29-cells for 12 h. As shown in Figure 4A and Supplementary Figure S5E, SGS induced CYP1A1 expression in Caco-2 cells and HT29-cells (log2 RQ 3.0; p < 0.01). Interestingly, secretomes from E. faecalis only marginally induced CYP1A1 expression after 12 h (log2 RQ 1.5; NS), whereas viable E. faecalis cells strongly induced expression of this gene (Supplementary Figure S5A). In line with the previous results, secretomes from E. coli NTB5 had no effect on CYP1A1 expression. Secretomes from other S. bovis-group bacteria, such as S. gallolyticus subsp. pasteurianus, subsp. macedonicus, and subsp. gallolyticus, S. equinus, or S. lutetiensus, were, however, unable to induce CYP1A1 gene expression (Supplementary Figure S5E). These results confirm the idea that a S. gallolyticus-associated factor present in the secretome of strain UCN34 induces CYP1 gene expression through the AhR; however, why secretomes of other strong inducers of CYP1 are unable to stimulate AhR needs to be resolved.

Next, we tested whether CYP1 upregulation by S. gallolyticus may contribute to the DNA-damaging effect of 3MC. To test this hypothesis, Caco-2 cells were first incubated in SGS for 6 h to prime CYP1 expression, after which the level of 3MC-induced DNA damage was compared to that in untreated cells by the COMET assay, which quantifies the level of DNA damage on the cellular level. As shown in Figures 4B–D, the DNA damaging effect of 0.1 µM 3MC was significantly increased by preincubation of Caco-2 cells in SGS (p < 0.05). In contrast, priming of these cells with secretomes from E. faecalis or E. coli did not increase 3MC genotoxicity. In fact, secretome from E. faecalis had even a surprising inhibitory effect on 3MC-induced DNA damage (p < 0.05). It is presently unknown how this relates to the induction of CYP1 expression by viable E. faecalis cells. Together, these results show that secreted or otherwise released S. gallolyticus-associated factors can prime CRC cells towards an increased susceptibility to 3MC-induced DNA damage under the applied experimental conditions.

It was previously shown that the related bacterium S. infantarius NCTC 8133, now reclassified as S. equinus, induced COX-2 expression in colon epithelial cells in vitro (Ellmerich et al., 2000). Moreover, it has been described that COX2 expression can, similarly to CYP1A1, also be activated via AhR-ligand binding to the XRE promotor (Degner et al., 2007). Therefore increased AhR activity and CYP1A1 might also relate to COX-2 upregulation. S. gallolyticus UCN34 showed a low, but significant, induction of COX-2 expression to a log2 RQ of 1.6 in HT-29 cells after 2 and 4 h (p = 0.017) of coculturing (Figure 5A). However, at protein level, PGE2 release was not significantly increased in HT29 cells compared to control cells after 24 h exposure to SGS (74.2 vs. 45.0 pg/ml, p > 0.05, Figure 5B). Furthermore, it has been described that AhR activation may lead to cell proliferation. Previously, it was shown that S. gallolyticus increases cell proliferation of several colon epithelial cell lines, dependent on the bacterial strain (Kumar et al., 2018). We show here that at least SGS of strain UCN34 and NTB12 are unable to increase cell proliferation in HT29, SW480, HCT116, and Caco-2 cells. NTB12 even decreased cell proliferation significantly in HCT116 cells (Figure 5C). Similarly, S. lutetiensis NTB2 consistently decreased cell proliferation in Caco-2, SW480, and HCT116 cells. Only S. gallolyticus subsp. pasteurianus NTB7 and 992 were significantly increasing growth of HCT116 cells (p < 0.001). Hence, a clear correlation between secretomes, AhR-activation, cell proliferation, and PGE2 release is not observed from these results.

To verify that CYP1A1 activation via AhR in in vitro cell models is also observed in vivo, wild-type C57bl6 mice were colonized with S. gallolyticus UCN34 after 5 days of antibiotic treatment with kanamycin/doxycyclin. Mice were either pretreated with vehicle or with daily gavage of 10 mg/kg AhR inhibitor CH223191. After 1 week of colonization, a clear induction of CYP1A1 mRNA was detected in the cecum of mice colonized with S. gallolyticus, which was significantly reduced in mice fed AhR inhibitor (Figure 6A). Mice colonization was measured in fecal pellets with levels of 1-3*10⁹ CFU/g stool at 3 days postcolonization waning to levels of 1–10 × 10⁷ CFU/g stool at 1 week (Figure 6B and Supplementary Figure S6). As levels of S. gallolyticus started to drop at 1 week, we tested how long colonization would persist and whether effective immune response would eliminate S. gallolyticus. We observed colonization up to 21–30 days after which S. gallolyticus in stool was below the detection level. Simultaneous with reduced colonization levels, IgGs detecting immobilized S. galloltyicus in an ELISA assay were present in serum of mice colonized with S. galloltyicus (Figure 6C). To confirm that a B-cell response was initiated, IL4 mRNA was measured and detected at 1 week with increasing levels 4 weeks postcolonization (Figure 6D). CYP1A1 was no longer increased at 4 weeks postinoculation when colonization of S. gallolyticus disappears. In addition, in vivo, we did not observe any increase in prostaglandin expression at mRNA and protein level (data not shown) or differences in cell proliferation in proximal or distal colon (Figure 6E and Supplementary Figure S7). The cecum showed some immune cell infiltration and reactive epithelial changes with S. gallolyticus colonization (Supplementary Figure S8), but this was not significantly different from sham mice, and no differences were observed between mice treated with vehicle or AhR inhibitor. No inflammation or aberrant growth was observed in the cecum or colon. Together, these data show that also in vivo CYP1 is induced by S. gallolyticus and that S. gallolyticus is effectively recognized and eliminated by the mouse intestinal immune system.