The Polyphosphate Kinase of Escherichia coli Is Required for Full Production of the Genotoxin Colibactin

Colibactin-producing E. coli induces DNA damage in eukaryotic cells and promotes tumor formation in mouse models of intestinal inflammation. Recent studies have provided strong evidence supporting the causative role of colibactin in human colorectal cancer (CRC) progression.

Given the role of colibactin in bacterial virulence and tumorigenesis, it is important to understand the regulation of its production, to provide clues for the development of anticolibactin strategies. It was recently reported that ClbR is an (auto)transcriptional activator of the clbB gene (17). In addition, the two master regulators of bacterial iron homeostasis Fur (ferric uptake regulator) and the small regulatory noncoding RNA RyhB regulate the transcription of clbA (18,19). In vivo studies showed that the expression of pks genes was upregulated in human urine (20) and enriched in intestinal inflammation and CRC development (21)(22)(23).
In this work, we used a random mutagenesis strategy to find regulators involved in colibactin production. We determined that a mutant of the gene ppk encoding polyphosphate kinase (PPK) has a lower clbB promoter (PclbB) activity than the wild type (WT). PPK catalyzes the reversible conversion of the terminal (g ) phosphate of ATP to long chains of inorganic polyphosphate (polyP; ca. 750 residues), which has been found to be involved in bacterial virulence and stress responses (24). In this work, we found that PPK played a positive role for PclbB activity and colibactin production. As mesalamine (also known as 5-aminosalicylic acid) is an inhibitor of PPK enzymatic activity (25), we tested and confirmed that this commonly prescribed drug is capable of inhibiting PclbB activity and colibactin biosynthesis.

RESULTS
Identification of PPK as an enhancer of PclbB activity. On the assembly line of colibactin, ClbB is the enzyme that accepts the prodrug motif C 14 AsnOH and constructs the amide bond cleavable by ClbP for releasing active colibactin (3). In order to investigate the regulation of colibactin production, we constructed a transcriptional fusion expressing luxCDABE (lux) under the control of PclbB, resulting in plasmid pMT3a (Table 1; see also Fig. S1a in the supplemental material). This plasmid was transformed into E. coli strain SP15, isolated from a patient with neonatal meningitis. Thus, the expression level of luminescence of SP15(pMT3a) reflects PclbB activity. Relative luminescence units (RLUs) and optical density at 600 nm (OD 600 ) of SP15(pMT3a) were monitored for 8 h in Dulbecco's modified Eagle's medium (DMEM)-HEPES at 37°C (Fig. 1a). According to RLUs normalized to OD 600 (RLU/OD 600 ), the peak of PclbB activity was just after 4 h. Therefore, we decided to set the measurement time point at 4 h for the following experiments. First, we tested the intrinsic luminescence variability of the WT strain SP15(pMT3a) by measuring RLU/OD 600 for 300 isolates (Fig. 1b). Relative to (versus) the median value of RLU/OD 600 of the 300 isolates, the values of RLU/OD 600 of individual isolates were increased between 220% and 130%; And, relative to the median value of OD 600 , the OD 600 s of individual isolates were found to be between 220% and 130%.
Next, a transposon (Tn) mutant library of SP15(pMT3a) was constructed by using the EZ-Tn5 ,KAN-2.Tnp Transposome. Under the same condition as described above, RLU/OD 600 values of 823 Tn mutants were measured at 4 h. Seventeen mutants showing growth retardation compared with the WT (increase of OD 600 of less than 220%) were excluded; 41 mutants showed an increase of RLU/OD 600 of less than 220% (Fig. 1c). By sequencing, 40 mutants were identified to have the Tn inserted into the lux operon, and 1 mutant (named P1D10) was identified to have the Tn inserted into the gene ppk (1,754 bp after the start codon) (Fig. S2). The attenuated PclbB activity in P1D10 was also observed by time course monitoring (Fig. 1d).
To confirm that the attenuated PclbB activity was due to the inactivation of ppk in P1D10, we constructed an isogenic ppk deletion mutant of SP15 (SP15 Dppk). SP15 Dppk carrying the PclbB-lux reporter fusion pMT3 (Table 1) had significantly lower PclbB activity than the WT ( Fig. 2a and b). Additionally, we deleted ppk in the previously described reporter strain Nissle 1917 (EcN) carrying a transcriptional fusion of PclbB and the lux operon on the chromosome (EcN clbB::lux) (17,26), resulting in strain EcN clbB::lux Dppk (Fig. S1b). PclbB activity was also significantly lower in EcN clbB::lux Dppk than in EcN clbB::lux ( Fig. 2c and d). These results consistently suggest that PPK is an enhancer of PclbB activity.
PPK is required for full production of colibactin. To determine whether PPK is associated with the biosynthesis of colibactin, we performed DNA interstrand crosslinking (ICL) assays in which bacteria were in direct contact with DNA. The ICL amount is directly correlated with the production of active colibactin (27). After incubation with bacteria, exposed DNA was purified and migrated under the alkaline-denaturing conditions. DNA with ICL is nondenaturable and displays delayed migration compared to that of unaffected denatured single-stranded DNA. Our results showed that SP15 Dppk caused less ICL than the WT, and this ability was restored in the complemented strain, SP15 Dppk carrying a plasmid, pGEN-ppk, expressing ppk (SP15 Dppk-c) (  and b). This ppk deletion-associated phenotype was also observed in other strains, including the probiotic strain EcN, the colitogenic strain NC101, and the uropathogenic strain UTI89 (Fig. 3c and d). These results indicate that PPK is required for colibactin biosynthesis in different genetic contexts. Since colibactin cannot be directly quantified yet, we quantified the production of the prodrug motif C 14 AsnOH, which is correlated with colibactin production and maturation ( Fig. 4a) (3). The result showed that the production level of C 14 AsnOH of SP15 Dppk was about 10 times less than that of the WT (Fig. 4b), and this level was restored to the WT level in SP15 Dppk-c. Taken together, these findings indicate that PPK is required for full PclbB activity, thereby enhancing colibactin biosynthesis.
PPK is required for full genotoxicity of colibactin-producing E. coli. As ICLs induce the DNA damage response in the host cells, we quantified g H2AX, which is a sensitive marker for colibactin-induced DNA damage by in-cell Western (ICW) assay (2). After a 4-h transient infection and 4 h of growth, HeLa cells grown on a 96-well plate were fixed, and g H2AX was stained by immunofluorescence. The fluorescent signal of g H2AX is pseudocolored in green, and the fluorescent signal of DNA is pseudocolored in red (Fig. 5a). The genotoxic index was determined by quantification of the signal of g H2AX relative to DNA content and normalized to the control (Fig. 5b)  showed that the genotoxicity of SP15 Dppk was significantly lower than that of the SP15 WT and was restored in SP15 Dppk-c. This indicates that PPK is required for the full genotoxicity of SP15.
Colibactin-producing E. coli induces megalocytosis in cultured eukaryotic cells, characterized by a progressive enlargement of the cell body and nucleus and a reduced cell number (1). To corroborate the previous result, we investigated the role of PPK in megalocytosis. Fewer giant cells were observed with infection by SP15 Dppk than for cells infected by the WT (Fig. 6a). Through the quantification of stained methylene blue on infected HeLa cells relative to noninfected cells, our results indicate that the mutation of ppk significantly reduced the ability of SP15 to induce megalocytosis, which was restored in SP15 Dppk-c (Fig. 6b). This indicates that PPK is required for colibactin-producing E. coli to induce DNA damage and subsequent megalocytosis of host cells.
Mesalamine reduces PclbB activity and represses colibactin production. One well-known PPK enzymatic activity inhibitor is the anti-inflammatory drug mesalamine, commonly prescribed for IBD and proposed for CRC prevention (25). We thus investigated whether mesalamine has an effect on colibactin synthesis similar to what we observed in Dppk mutants. First, we tested the effect of mesalamine on PclbB activity in two genetic backgrounds, SP15 and EcN. Luminescence emitted by the bacteria was monitored in DMEM-HEPES at 37°C with or without the presence of mesalamine (2 or 4 mM). The results showed that mesalamine reduced PclbB activity in a dose-dependent manner in both SP15 ( Fig. 7a and b) and EcN ( Fig. 7c and d), while it did not cause growth retardation of bacteria (Fig. S3). This indicates that mesalamine has an inhibitory effect on PclbB activity.
To test whether mesalamine treatment has an impact on the production of colibactin, we quantified the production of the prodrug motif C 14 AsnOH of the bacteria with or without mesalamine. We observed that a dose of 8 mM mesalamine decreased  (Fig. S4). This result indicates that mesalamine inhibits the biosynthesis of colibactin.
We investigated the effect of mesalamine on ICL formation induced by various colibactin-producing E. coli strains, including SP15 ( Fig. 9a and b), EcN ( Fig. 9c and d), NC101 ( Fig. 9e and f), and UTI89 ( Fig. 9g and h). We observed that a dose of 15 mM mesalamine significantly reduced ICL formation in all the strains tested, while the bacterial CFU were not reduced and even showed a slight increase with the treatment with mesalamine (Fig. S5). To confirm these results, we also evaluated the genotoxicity in eukaryotic cells induced by colibactin-producing E. coli with or without mesalamine treatment. By using ICW assays, we observed that the genotoxicity induced by SP15  was reduced in a dose-dependent manner with the treatment of mesalamine (Fig. 10), while the viability of HeLa cells was not affected (Fig. S6a), and the bacterial CFU were not reduced, by mesalamine (Fig. S6b). Taken together, these results demonstrated that treatment with mesalamine inhibited the production of colibactin, thereby protecting eukaryotic cells from the genotoxicity of colibactin-producing E. coli.
We then investigated the impact of mesalamine treatment in a Dppk mutant on PclbB activity, colibactin production, and genotoxicity. We observed that mesalamine decreased PclbB activity ( Fig. S7a and b), the production level of C 14 AsnOH (Fig. S7d), and genotoxicity ( Fig. S7f and g), without reducing bacterial viability (Fig. S7c, e, and  h). These results indicate that mesalamine inhibits PclbB activity and colibactin production independently from its inhibition effect on PPK enzymatic activity (25).

DISCUSSION
The data implicating colibactin in virulence and colorectal tumorigenesis have motivated extensive structural and pharmacological studies of colibactin (5,(28)(29)(30)(31)(32)(33) and other metabolites of the pks pathway (2,34,35). However, very limited data are available about the regulation of the production of this important genotoxin. In this study, we developed a high-throughput screening of regulators involved in colibactin biosyn-  The cell viability relative to that of NI controls was determined by quantification of methylene blue staining. The methylene blue was extracted and quantified by the measurement of OD 660 . Bars represent means 6 SEMs (n = 4 independent experimental replicates). The significance of the difference between each strain and the WT was determined using the Kruskal-Wallis test followed by the two-stage step-up method of Benjamini, Krieger, and Yekutieli; P values are shown. thesis based on the construction of a library of random mutants of a colibactin-producing E. coli strain harboring a PclbB-reporter fusion. We reasoned that because ClbB is essential for the production of colibactin, regulators of PclbB activity should impact the production level of colibactin. Of 823 mutants screened, 1 mutant had lower PclbB activity than the WT. This mutant had the transposon inserted in the gene ppk, encoding the polyphosphate kinase (PPK). We then constructed isogenic mutants of ppk in different colibactin-producing E. coli strains to test PclbB activity and the production of colibactin as well. Consistently, the deletion of ppk reduced PclbB activity and caused a lower production level of colibactin.
This work highlights the role of PPK in PclbB activity, which is correlated with the production of colibactin. A recent study has shown that ClbR is the transcriptional activator of clbB (17). In this work, we discovered the first regulator of clbB transcriptional activity outside of the pks island. PPK is essential for the production of long-chain polyP (36). E. coli mutants lacking ppk were described to be defective in virulence and responses to multiple stresses (i.e., nutrient starvation, oxidants, acidic challenge, osmotic shock, and heat shock) (24,37,38). Additionally, the ppk deletion mutant of meningitis E. coli strain E44 showed less ability than the WT to cross the blood-brain barrier (BBB) (37). The ppk deletion mutant of the uropathogenic strain UTI89 was a. b.

SP15 pMT3 SP15 pMT3
EcN clbB::lux EcN clbB::lux  shown to have defects in biofilm formation, resistance to oxidation, and formation of antibiotic-resistant persister cells (25,39). PPK is distributed across a wide spectrum of bacterial pathogens and absent in mammalian cells, and it has been therefore proposed as a new target for developing antibacterial agents that specifically target pathogens without affecting the host and its beneficial bacteria (40). In this work, we observed the deletion of ppk reduced the genotoxicity of colibactin-producing E. coli, including the meningitic strain SP15, the probiotic strain EcN, the colitogenic strain NC101, and the uropathogenic strain UTI89. Future research should clarify whether this is the case in vivo. Our finding reinforces the idea to take PPK as a target of antibacterial drugs and provided a new path for developing an anticolibactin strategy.
Several studies have focused on finding inhibitors of PPK (25,(41)(42)(43). One of the iden-  tified PPK inhibitors, mesalamine (also known as mesalazine or 5-aminosalicylic acid), has been validated by treating different bacteria ranging from clinically isolated uropathogenic E. coli and P. aeruginosa strains to human gastrointestinal luminal samples (25). Mesalamine is a drug commonly used to treat IBD patients, and rare side effects have been reported (44)(45)(46). Mesalamine exerts its anti-inflammatory effects locally on the colorectal mucosa, and the efficacy is dependent on achieving high intraluminal concentrations (47,48). In patients conventionally treated with mesalamine, stool concentrations of mesalamine are on the median order of 30 mM, ranging from 10 to 100 mM; these concentrations correspond to luminal concentrations of mesalamine 100 times greater than the concentrations in the colonic mucosa (49). Mesalamine has been shown to have chemopreventive effects on CRC and has been proposed as a first-line treatment that should be given daily in high doses and long term to reduce the possibility of recurrence and risk of CRC (45,50,51). The effects of mesalamine on the host have been intensely researched (51)(52)(53)(54)(55)(56), while few studies have investigated the effects on bacteria. Mesalamine has been shown to affect bacterial gene expression (49) and to alter gut microbiota (57)(58)(59). Interestingly, a recent report showed that mesalamine downregulated the transcription of the pks gene (60), but it did not show which pks gene was downregulated. This study also showed that mesalamine (9.8 mM and 13 mM) inhibited DNA breakage in colonic epithelial Caco-2 cells induced by colibactin-producing E. coli (60). In our study, we first identified PPK as an enhancer of colibactin production, which led us to test the PPK inhibitor mesalamine. We tested not only the inhibitory effects of mesalamine on the genotoxicity of colibactin-producing E. coli in eukaryotic cells but also directly the amount of colibactin-correlated metabolite C 14 AsnOH and the formation of ICL. We also tested a wider range of colibactin-producing E. coli strains and demonstrated that the effect of mesalamine on colibactin production is universal. Among the strains tested, one strain should especially get our attention: the probiotic strain EcN, which is the active component of microbial drug Mutaflor (61). EcN has been widely used in the treatment of IBD and has proven to be as effective as the gold standard mesalamine for the maintenance of remission in ulcerative colitis patients (61). It has been suggested that a combination of mesalamine and EcN might exert additive or synergistic therapeutic efficacy, and mesalamine has no effect on the viability of EcN in vivo (62). Here, our data suggest that in vitro mesalamine has a suppressive effect on the genotoxicity of EcN without altering the viability of EcN. Future research should clarify whether this is the case in vivo.
In this study, we also investigated whether mesalamine treatment inhibits the biosynthesis of colibactin in a Dppk mutant. Interestingly, an additional inhibition effect on colibactin production was observed in the Dppk mutant treated with mesalamine, indicating that mesalamine is capable of inhibiting colibactin production independently from its inhibitory effect on PPK enzymatic activity (25). Future research is needed to clarify this new mechanism.
In summary, this study showed that PPK played a role in the transcriptional activity of clbB and was required for the genotoxicity of colibactin-producing E. coli. This provided us a new perspective on the regulatory network of colibactin production and brought us a novel clue for anticolibactin strategy development. By using the PPK inhibitor mesalamine, we confirmed the role of PPK in colibactin production and also identified mesalamine as an effective drug for inhibiting pks 1 E. coli genotoxicity to eukaryotic cells. Further studies are necessary to test the synergistic activity of mesalamine and EcN in vivo and to determine if treatment of IBD with both mesalamine and EcN protects patients against CRC.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. The bacterial strains and plasmids used in this study are listed in Table 1. Gene mutagenesis was performed by using the l red mutagenesis method with the primers listed in Table 2 and confirmed by PCR. For genetic manipulations, all E. coli strains were grown routinely in lysogenic broth (LB) medium. When appropriate, antibiotics were added at the following concentrations: 50 mg/ml for kanamycin, 50 mg/ml for carbenicillin, and 25 mg/ml for chloramphenicol.
Chemicals and reagents. Unless otherwise indicated, chemicals were from Sigma-Aldrich or Fisher. The stock solution of mesalamine (400 mM) was extemporaneously prepared in dimethyl sulfoxide (DMSO), and dilutions were made immediately before each experiment.
Plasmid construction. The plasmids used in this study are listed in Table 1. For the construction of clbB promoter (PclbB) reporter fusions, the promoter sequence of clbB (from bp 2473 to 117 relative to the initiation start codon of clbB), 490 bp, named MT3, containing PclbB was amplified from the genome of SP15 and cloned into the reporter plasmid pCM17 preceding the luxCDABE operon (Fig. S1). The primers used are listed in Table 2. The result plasmid, pMT3, was verified by sequencing. After pMT3 is introduced into the target bacteria, the luxCDABE operon encodes a luciferase (LuxA and LuxB) and the enzymes that produce its substrate (LuxC, LuxD, and LuxE) under the control of PclbB, so bacteria that have PclbB activated and express the cluster emit 490-nm luminescence spontaneously. The promoterreporter fusion pMT3 contains the kanamycin resistance (Kan r ) cassette. For compatibility with the transposon containing the Kan r cassette, the Kan r cassette of pMT3 was disrupted by inserting an ampicillin resistance (Amp r ) gene at the restriction site BssHII, which resulted in pMT3a. The plasmids were verified by sequencing. For complementation, the coding sequence of gene ppk plus its putative promoter region was amplified (the primers used are listed in Table 2) and cloned into pGEN-MCS using HindIII and BamHI restriction sites. All restriction enzymes were purchased from New England BioLabs (NEB) and used based on the supplier's recommendations.
Construction of Tn mutant library and identification of Tn insertion sites of selected mutants. The transposon (Tn) mutant library of E. coli strain SP15 containing pMT3a was prepared using the EZ-Tn5 ,KAN-2.Tnp Transposome kit (Lucigen). Mutants were stored at -80°C with 20% (vol/vol) glycerol as a cryoprotectant. To identify Tn insertion sites of selected mutants, DNA fragments spanning the Tn insertion junction were amplified by arbitrarily primed PCR (AP-PCR) for sequence analysis (63), and then the resulting sequence was mapped to the bacterial genome and plasmids.
Luminescence measurement. For monitoring clbB promoter (PclbB) activity in SP15 (carrying pMT3a or pMT3) and the mutants, each strain was inoculated into 150 ml of LB and grown at 37°C without shaking. A total of 5 ml of overnight culture was inoculated into 100 ml of Dulbecco's modified Eagle's medium (DMEM)-HEPES (Gibco) in a black 96-well plate (Greiner Bio-One), and then the bacteria were grown without shaking at 37°C. The luminescence emission (relative light units [RLU]; 2,000-ms aperture per sample) and the optical density at 600 nm (OD 600 ) were measured at 4 h by a luminometer (Tecan Spark multimode reader). To have the time course PclbB activity, the bacteria were grown without shaking at 37°C in the luminometer, and RLU and OD 600 were measured every 0.5 h. The area under the curve (AUC) of RLU/OD 600 , which quantifies the cumulative luminescence, was calculated with GraphPad Prism (version 8.0) software.
To monitor PclbB activity in EcN clbB::lux (Fig. S1b) (17,26) and the derivatives, each strain was inoculated into 3 ml of LB and grown at 37°C with shaking at 240 rpm overnight. A total of 500 ml of overnight culture was inoculated into 9.5 ml of DMEM-HEPES and then grown at 37°C with shaking at 240 rpm for 8h. Ten-microliter subcultures were inoculated into 100 ml of DMEM-HEPES in a black 96-well plate. Bacteria were grown without shaking at 37°C in the luminometer, and RLU and OD 600 were measured at 0.5 h. The AUC was determined as previously described. To detect the effect of mesalamine on PclbB activity in EcN and SP15, the same protocol was used; mesalamine was added in 100 ml of DMEM-HEPES in the black 96-well plate inoculated with 10-ml subcultures. C 14 AsnOH (colibactin cleavage product) quantification. Each E. coli strain was inoculated in triplicate into 3 ml of LB and grown at 37°C with shaking at 240 rpm overnight. A total of 500 ml of overnight culture was inoculated into 9.5 ml of DMEM-HEPES and then grown at 37°C with shaking at 240 rpm to an optical density at OD 600 of 0.4 to ;0.6. Then 500 ml of subculture was inoculated into 9.5 ml of DMEM-HEPES and grown under the same condition for 8 h. Bacterial cells were pelleted by centrifugation at 5,000 Â g for 10 min, and the supernatants were filtered through a 0.22-mm-pore-size polyvinylidene difluoride (PVDF) filter (Millipore). The supernatants were stored at -80°C until N-myristoyl-D-asparagine (C 14 AsnOH) extraction. With the same protocol for lipid extraction as previously described (34), 5 ml of internal standard (IS) mixture (deuterium-labeled compounds) (400 ng/ml) and 0.3 ml of cold methanol (MeOH) was added to each 1-ml supernatant sample. An Oasis HLB 96-well plate was conditioned with 500 ml of MeOH and 500 ml of 10% MeOH/H 2 O. The samples were loaded in this conditioned plate and then washed with 500 ml of 10% MeOH/H 2 O and dried under aspiration. Lipids were eluted with 750 ml of MeOH, evaporated twice under N 2 , and then suspended in 10 ml of methanol. The quantification of C 14 AsnOH was performed by the MetaToul Lipidomics Facility (Inserm UMR1048, Toulouse,

Deletion of ppk
MT44_ppk-mut-R GTTATTCAGATTGTTCGAGTGATTTGATGTAGTCGTAAAT CGCCAACTGCGCATATGAATATCCTCCTTAGTTC MT54_pGEN-ppk-HindIII-F CCGAAGCTTGTACATCGGTGCATTTCGTC Amplification of ppk plus its putative promoter region MT55_pGEN-ppk-BamHI-R CGCGGATCCAGGGTTATTCAGATTGTTCGAG France), using an in-house quantification assay by high-performance liquid chromatography/tandem mass spectrometry analysis. Genotoxicity assay. HeLa cells (1.5 Â 10 5 /200 ml/well) were grown in DMEM GlutaMAX supplemented with 10% fetal calf serum (FCS) and 1% nonessential amino acids (NEAA), in 96-well culture plates, at 37°C in a 5% CO 2 incubator for 24 h. Each E. coli strain was inoculated into 3 ml of LB and grown at 37°C with shaking at 240 rpm overnight. A total of 500 ml of overnight culture was inoculated into 9.5 ml of DMEM-HEPES and then grown at 37°C with shaking at 240 rpm to an OD 600 of 0.4 to ;0.6. Then HeLa cells were infected at a multiplicity of infection (MOI) of 100, 50, 25, or 12.5 with each strain with or without mesalamine. At 4 h postinfection, the cells were washed 3 times with Hanks' balanced salt solution (HBSS) and incubated at 37°C in DMEM GlutaMAX supplemented with FCS and NEAA for 3 h with 200mg/ml of gentamicin. The in-cell Western (ICW) procedure was performed as previously described (2). Briefly, after cells were fixed, permeabilized, and blocked, they were incubated overnight at 4°C with rabbit monoclonal anti-g H2AX antibody 9718 (Cell Signaling Technology; 1:200). An infrared fluorescent secondary antibody absorbing at 800 nm (IRDye 800CW, 1:500; Rockland Immunochemicals) was then applied. DNA was counterstained with RedDot2 (Biotium; 1:500). DNA and g H2AX were visualized simultaneously using an Odyssey infrared imaging scanner (LI-COR Biosciences) at 680 nm and 800 nm. Relative fluorescent units for g H2AX per well (as determined by the 800-nm signal divided by the 700-nm signal) were divided by untreated controls to determine the genotoxic index.
DNA cross-linking assay. The assay was performed as previously described (27). Briefly, linearized DNA was obtained by digesting plasmid pUC19 with BamHI (NEB). Each E. coli strain was inoculated into 3 ml of LB and grown at 37°C with shaking at 240 rpm overnight. A total of 500 ml of overnight culture was inoculated into 9.5 ml of DMEM-HEPES and then grown at 37°C with shaking at 240 rpm to an OD 600 of 0.4 to ;0.6. For bacterium-DNA interactions, 1.5 Â 10 6 bacteria were inoculated into 100 ml of DMEM-HEPES with or without mesalamine for 4 h at 37°C without shaking. Following centrifugation for 10 min at 5,000 Â g, bacteria were pelleted and resuspended in sterile Milli-Q H 2 O. Then, 500 ng of linearized DNA was added into the bacterial suspension and incubated for 40 min at 37°C without shaking. The bacteria were then pelleted by centrifugation for 5 min at 5,000 Â g, and the DNA was extracted from the supernatant by purification using a PCR purification kit (Qiagen) according to the manufacturer's recommendations.
A denaturing agarose gel was prepared by dissolving 1.0 g of agarose in 100 ml of a 100 mM NaCl and 2 mM EDTA solution (pH 8.0). The gel was then soaked (2 h) in an alkaline running buffer solution (40 mM NaOH and 1 mM EDTA [pH ;12.0]). A total of 100 ng of each DNA sample was loaded onto the agarose gel. The gel was run for 45 min at 1 V/cm and then 2 h at 2 V/cm. The gel was then neutralized for a total of 45 min in a 100 mM Tris (pH 7.4) buffer solution containing 150 mM NaCl. The gel was stained with GelRed for 20 min and revealed with UV exposure using the ChemiDoc imaging system (Bio-Rad).
Megalocytosis assay. Quantification of the colibactin-associated genotoxic effect by megalocytosis assay was performed as previously described (1). Briefly, HeLa cells (5 Â 10 3 /well) were grown in DMEM GlutaMAX (Gibco) supplemented with 10% (vol/vol) FCS (Eurobio) and 1% (vol/vol) NEAA (Invitrogen), in 96-well culture plates, at 37°C in a 5% CO 2 incubator for 24 h. Each E. coli strain was inoculated into 3 ml of LB and grown at 37°C with shaking at 240 rpm overnight. A total of 500 ml of overnight culture was inoculated into 9.5 ml of DMEM-HEPES and then grown at 37°C with shaking at 240 rpm to an OD 600 of 0.4 to ;0.6. Then HeLa cells were infected at MOIs of 100 and 50 with each strain in 100 ml of DMEM-HEPES. At 4 h postinfection, the cells were washed 3 times with HBSS (Gibco) and incubated in DMEM GlutaMAX supplemented with FCS, NEAA, and 200 mg/ml of gentamicin for 72 h before fixation (4% formaldehyde) and protein staining with methylene blue (1% [wt/vol] in 0.01 M Tris-HCl). The methylene blue was extracted with 0.1 M HCl. Staining was quantified by measurement of the OD 660 .
Statistical analyses. The mean and the standard error of the mean (SEM) are shown in the figures, unless otherwise stated. P values were calculated in GraphPad Prism 8.0 by the Mann-Whitney test or Kruskal-Wallis test followed by the two-stage step-up method of Benjamini, Krieger, and Yekutieli. P values of ,0.05 were considered statistically significant.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. and Marion Garofalo for helping read fluorescence in ICW assay. We also thank the lipidomic facility of MetaToul, Toulouse, France.