Escherichia coli BarA-UvrY regulates the pks island and kills Staphylococci via the genotoxin colibactin during interspecies competition

Wound infections are often polymicrobial in nature, biofilm associated and therefore tolerant to antibiotic therapy, and associated with delayed healing. Escherichia coli and Staphylococcus aureus are among the most frequently cultured pathogens from wound infections. However, little is known about the frequency or consequence of E. coli and S. aureus polymicrobial interactions during wound infections. Here we show that E. coli kills Staphylococci, including S. aureus, both in vitro and in a mouse excisional wound model via the genotoxin, colibactin. Colibactin biosynthesis is encoded by the pks locus, which we identified in nearly 30% of human E. coli wound infection isolates. While it is not clear how colibactin is released from E. coli or how it penetrates target cells, we found that the colibactin intermediate N-myristoyl-D-Asn (NMDA) disrupts the S. aureus membrane. We also show that the BarA-UvrY two component system (TCS) senses the environment created during E. coli and S. aureus mixed species interaction, leading to upregulation of pks island genes. Further, we show that BarA-UvrY acts via the carbon storage global regulatory (Csr) system to control pks expression. Together, our data demonstrate the role of colibactin in interspecies competition and show that it is regulated by BarA-UvrY TCS during interspecies competition.

The most frequently cultured bacterial species from wound infections include Staphylococcus aureus, Pseudomonas aeruginosa, Enterococcus spp., Escherichia coli, and Klebsiella pneumoniae [17,18]. The mechanistic basis of polymicrobial interactions in wounds has been examined for S. aureus together with P. aeruginosa and Enterococcus faecalis [7,10,19]. However, despite the fact that E. coli and S. aureus are among the top five most prevalent pathogens in often polymicrobial surgical, diabetic and non-diabetic wound infections [17,18,20,21], and coexist within diabetic wound microbiomes [22][23][24], polymicrobial interaction studies between these organisms, or of E. coli within wound infections in general, are scarce. We have previously shown that E. faecalis promotes E. coli biofilm growth and virulence in vitro and in a mouse excisional wound infection model [25]. However, the mechanistic basis of interactions between S. aureus and E. coli remains largely unknown.
In this study, we show that E. coli antagonizes S. aureus in biofilms and planktonic growth. Both the E. coli pks island and the BarA-UvrY two component system (TCS) are required for killing S. aureus. The pks island encodes the biosynthetic machinery to produce colibactin, a genotoxin that causes DNA damage in eukaryotic cells and is associated with human colorectal cancer [26,27]. Here we show that E. coli colibactin kills S. aureus by causing irreparable DNA damage. E. coli also antagonizes the growth and survival of S. aureus upon co-infection in a mouse excisional wound model, and this antagonism is dependent on the pks island and the BarA-UvrY TCS. Finally, we show that the BarA-UvrY TCS regulates the expression of the pks island through the Csr system during interspecies competition. Taken together, our data demonstrate the mechanism by which E. coli colibactin acts in interspecies competition to kill S. aureus during wound infection. The colibactin pks island and BarA-UvrY TCS contributes to E. colimediated growth antagonism of S. aureus To identify the genes involved in E. coli-mediated killing of S. aureus, we screened 14,828 E. coli UTI89 transposon mutants for failure to antagonize S. aureus growth in a macrocolony assay. We validated each E. coli gene identified in the transposon screen for its inability to kill S. aureus after 24 hours of macrocolony co-culture and the transposon insertion sites for validated mutants were determined by whole genome sequencing. The majority (99 out of 108) of transposon insertions mapped to genes of the pks island and genes encoding the two-component system (TCS) BarA-UvrY (S3 Table). To confirm that these E. coli loci impacted S. aureus survival, we generated deletion mutants comprising the entire pks island, as well as for barA and uvrY, all of which had significantly higher S. aureus CFU in the mixed species biological experiments. (E) Enumeration of S. aureus after planktonic growth alone or mixed with either E. coli UTI89 or MG1655 for 24 or 48 hours. N = 6 independent experiments (A-E) Data from single species macrocolonies or planktonic culture are indicated with open bars, and data from mixed species (all inoculated at a ratio of 1 EC :1 SA ) macrocolonies are indicated with checked bars. Individual data points from each biological replicate are indicated with closed circles. (B, D, E) Statistical significance was determined by two-way ANOVA with Sidak's (A and C) and Tukey's (D) test for multiple comparisons. ���� p< 0.0001. Error bars represent SD from the mean. All statistical tests were performed on log-transformed CFU data. (See S1 Fig for paired

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus macrocolony compared to E. coli UTI89 wild type (Fig 3A). These mutants grew as well as wild type (S2 Fig), showing that attenuation of the killing phenotype was not due to growth differences. Although co-culture with E. coli Δpks did not restore S. aureus growth to single species growth levels, S. aureus CFU were similar to that in co-culture with E. coli MG1655 (Fig 1C) which does not possess the pks island [34,35] and which also failed to kill S. aureus. We therefore surmise that nutrient competition within mixed species macrocolonies results in a 1-2 log decrease in S. aureus compared to S. aureus grown alone. Notably, co-culture with either ΔbarA or ΔuvrY only partially restored S. aureus growth, suggesting that the growth antagonism is not completely abolished when the BarA-UvrY TCS is inactivated (Fig 3A). Chromosomal complementation of uvrY in the ΔuvrY mutant, restores the killing phenotype.  These data show that the pks island is necessary for S. aureus killing and suggest that the Bar-A-UvrY TCS may either directly or indirectly regulate the expression of the pks island or the activity of its gene products.

The genotoxin colibactin kills S. aureus by inducing DNA damage
The pks island encodes enzymes required for the synthesis of the genotoxin colibactin [36]. E. coli strains carrying the 54 kb pks island generate DNA adducts and induce DNA crosslinks in mammalian cells [26,[37][38][39][40]. In bacteria, DNA adducts and crosslinks can be repaired via the nucleotide excision repair (NER) pathway, facilitated by the UvrABC endonuclease complex [41]. Accordingly, pks+ E. coli strains lacking both UvrB and the ClbS colibactin resistance protein, which protects E. coli from colibactin-mediated autotoxicity, are severely impaired for growth [42]. Consistent with DNA damage, we found that S. aureus uvrAB genes were significantly upregulated in mixed species macrocolonies with E. coli (Fig 2). Accordingly, we predicted that disruption of the S. aureus NER pathway would accelerate its killing by E. coli. We thus examined CFU following macrocolony co-culture of S. aureus uvrA, uvrB and uvrC transposon mutants with E. coli at 8 hours, a timepoint prior to complete S. aureus eradication, in order to detect differences between wild type S. aureus and uvrABC mutants, and observed accelerated pks-dependent killing and significantly fewer S. aureus CFU suggesting a role for NER in the protection of S. aureus from colibactin-mediated killing (Fig 3B). To further investigate the role of colibactin in S. aureus killing, we expressed the colibactin resistance protein ClbS in S. aureus. Expression of ClbS in S. aureus cells conferred full protection from E. coli pks-mediated killing, suggesting that colibactin is responsible for S. aureus cytotoxicity ( Fig  3C). To determine the prevalence of pks island in E. coli isolated from human wound swabs, we screened 58 isolates for presence of pks by PCR flanking the clbB gene and found 17 out of the 58 clinical isolates (29.3%) were pks positive (pks+) (S4 Table). We then randomly picked ten of each pks+ and pks-E. coli clinical isolates to perform co-culture macrocolony assay with S. aureus. We found 9 of 10 pks+ clinical isolates were able to kill S. aureus, while no pks-clinical isolates killed S. aureus (S3 Fig). Finally, we show that expression of ClbS in S. aureus conferred protection against killing by a pks+ E. coli clinical isolate (Fig 3D) comparable to S. aureus CFU in co-culture with a pks-E. coli isolate. Collectively, these data support the interpretation that killing of S. aureus by colibactin-producing E. coli is mediated by DNA damage, which is consistent with the known genotoxic effects of colibactin.

N-myristoyl-D-Asn causes pore formation in S. aureus
Maturation of colibactin requires the removal of the prodrug motif, N-myristoyl-D-Asn (NMDA), by ClbP peptidase [43,44]. NMDA is the most abundant of the pks island metabolites along with its analogues that vary in acyl chain lengths (C 12 to C 16 ) [45]. These intermediates do not exhibit cytotoxic or genotoxic activity in HeLa cells; however, NMDA can modestly inhibit Bacillus subtilis growth [45]. Thus, we investigated if the production of NMDA could provide an alternative explanation for the killing of S. aureus by pks+ E. coli. We synthesized NMDA and added it at increasing concentrations to S. aureus but we observed minimal dose-dependent growth inhibition, and only at high concentrations of 600 μM (S4A Fig). Despite the absence of significant toxicity, NMDA treatment resulted in increased S. aureus membrane permeability as measured by propidium iodide uptake using flow cytometry (S4B Fig). Thus, NMDA, which is the most abundant colibactin metabolite isolated from culture supernatants [45] and is therefore likely released from E. coli along with colibactin, can compromise S. aureus membrane integrity.

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus

BarA-UvrY TCS regulates pks island genes via the Csr system during interspecies competition
In E. coli, pks island genes are upregulated when iron is limited in a Fur-dependent manner, while pks island genes are downregulated when iron is in abundance [46,47]. While gene expression profiling indicates that both E. coli and S. aureus are experiencing iron-limitation in mixed species macrocolonies, iron-supplementation experiments did not prevent E. colimediated killing of S. aureus (S5 Fig), suggesting that iron restriction is not the sole driver of pks expression in this mixed species interaction. The BarA-UvrY TCS have been recently shown to be involved in the regulation of the pks island [48]. Since both pks and barA/uvrY E. coli mutants failed to kill S. aureus, we hypothesized that the BarA-UvrY TCS is involved in the regulation of the pks island during interspecies competition. To investigate this, we first compared the expression of pks island genes (clbA and clbB) between single species macrocolonies of wild type E. coli and E. coli ΔuvrY and found out that the expression of both clbA and clbB were significantly lower in the E. coli ΔuvrY macrocolony (Fig 4A). Next, we examined the expression of pks island genes in the presence or absence of S. aureus, and found that E. coli clbA expression was significantly increased in mixed species macrocolonies compared to E. coli single species macrocolonies (Fig 4B). By contrast, expression of both clbA and clbB were significantly lower when E. coli ΔuvrY was co-cultured in macrocolonies with S. aureus, compared to wild type E. coli-S. aureus macrocolonies, suggesting that the BarA-UvrY TCS is

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus involved in regulating pks island gene expression (Fig 4C). Moreover, S. aureus-dependent induction of clbA and clbB gene expression in wild type E. coli was significantly attenuated upon macrocolony co-culture with E. coli ΔuvrY (Fig 4D), together suggesting that E. coli can use the BarA-UvrY system to sense S. aureus and induce pks island (clbA and clbB) gene expression.
The BarA-UvrY TCS regulates the expression of the Csr system, which in turn regulates a variety of metabolic and virulence genes via the global regulator CsrA [49]. CsrA is a post-transcriptional regulator that can either promote or repress gene expression [50]. Activation of the BarA-UvrY TCS, leads to the expression of the sRNAs CsrB and CsrC, which bind to CsrA and inhibit the regulatory activity of CsrA [50]. Therefore, if BarA-UvrY regulates pks transcription via CsrA, we hypothesized that increasing expression of CsrA would repress pks island gene expression and increasing expression of CsrB would lead to the upregulation of pks island genes. Consistent with these predictions, we examined 16 hour macrocolonies in which we observed a steep drop in S. aureus CFU (Fig 1A) and predicted substantial pks gene expression, and observed that overexpression of CsrA leads to the downregulation of pks island genes, suggesting that CsrA is a negative regulator of the pks island (Fig 5A). Conversely, overexpression of CsrB resulted in upregulation of pks island genes (Fig 5B). Co-culturing the E. coli overexpression strains with S. aureus in macrocolonies was consistent with the pks expression data, such that CsrA overexpression led to reduced pks gene expression and S. aureus

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus killing, and CsrB overexpression led to increased pks gene expression and enhanced killing (Fig 5C). CsrA regulates gene expression by binding to its target mRNA at GGA motifs, which can be found in the 5' untranslated region, early coding region, and the stem-loop structure of the mRNA [51][52][53]. ClbR is a transcriptional regulator of the pks locus [54]. To test whether CsrA regulates ClbR via interaction with clbR mRNA GGA motifs, we generated a E. coli strain where the GGA motifs in clbR were modified (S5 Fig), which we predicted would reduce CsrA binding efficiency to clbR mRNA, as has been reported for other CsrA mRNA substrates [49,52]. We hypothesized that this strain, E. coli clbR mut , would significantly increase S. aureus killing because CsrA repression of clbR expression would be alleviated due to reduced CsrA binding to clbR mRNA. Consistent with this hypothesis, we observed that E. coli clbR mut kills S. aureus faster than wild type E. coli, as expected as a result of pks derepression and increased colibactin synthesis (Fig 5D). Collectively, these data demonstrate that the E. coli BarA-UvrY TCS senses S. aureus and responds by inducing pks gene expression (clbA and clbB) via CsrA which acts as a negative regulator of the pks island.

E. coli antagonizes the growth of S. aureus in a mouse model of wound infection
To determine whether E. coli could similarly antagonize S. aureus growth in vivo within a mixed species wound infection, we infected excisional wounds of C57BL/6 mice with 10 6 CFU each of E. coli and S. aureus cells and monitored the bacterial burden at the wound site. At 24 hours post infection (hpi), S. aureus CFU were significantly reduced when co-infected with E. coli as compared to single species S. aureus infection (Fig 6A). Upon co-infection with S. aureus and E. coli Δpks, S. aureus CFU remained similar to S. aureus single species infected wounds (Fig 6B), whereas co-infection E. coli ΔbarA or E. coli ΔuvrY resulted in increased S. aureus survival, but not restoration to single species levels (Fig 6C), similar to our in vitro results (Fig 3A). Together, these data demonstrate that both the E. coli pks island and the Bar-A-UvrY TCS are important for the growth antagonism of S. aureus observed in mixed species infections in vivo.

Discussion
Escherichia coli and Staphylococcus aureus are both important pathogens that cause wound infections, blood infections, urinary tract infections and infective endocarditis [55][56][57]. Both E. coli and S. aureus can exhibit polymicrobial synergy with other bacterial species during infection, which is advantageous for these pathogens, but often leads to adverse disease outcomes [7,9,25]. While many studies have investigated the mechanistic basis of polymicrobial interactions between different microbial species, the molecular interactions between E. coli and S. aureus have not been reported. In this study, we found that E. coli production of colibactin is responsible for the growth antagonism toward S. aureus, resulting in significant inhibition of S. aureus in vitro and in vivo during polymicrobial wound infection. E. coli pks genes are upregulated during co-culture with S. aureus, supporting the proposed role of E. coli colibactin as an effector for niche adaptation or domination [58]. Finally, we found that the E. coli two component signal transduction system BarA-UvrY senses the polymicrobial environment leading to the upregulation of the pks island via the Csr system.
While colibactin is best studied for its genotoxicity toward eukaryotic cells, a functional role for the pks island in polymicrobial interactions has also been reported. Colibactin altered the gut microbiome composition in newborn mice when the pregnant mothers were previously colonized with pks + E. coli, with Firmicute reduction observed starting 35 days after birth [58]. Colibactin killing was also demonstrated in Vibrio cholera, with an interesting PLOS PATHOGENS E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus suggestion that the colibactin susceptibility of specific bacterial species could be related to the efficiency of endogenous DNA repair mechanisms [59]. More relevant to our work, E. coli episomally expressing the pks island spotted onto lawns of S. aureus gave rise to small zones of inhibition around the pks + E. coli colonies, although they did not confirm that colibactin was the factor producing antibiotic activity [60]. We similarly observed that this antagonism was more efficient within macrocolonies, where E. coli completely inhibits the growth of S. aureus within 24 hours as compared to planktonic cultures, where we only saw significant growth inhibition at 48 hours. These data indicate that colibactin-mediated growth inhibition of S. aureus is favored at proximity but is not a biofilm dependent phenotype.
The ClbS resistance protein, which hydrolyzes colibactin into a non-toxic compound, protects the mammalian host DNA from colibactin-mediated damage [42,61] and confers to S. aureus full protection from E. coli-mediated growth inhibition. Therefore, we conclude that mature colibactin, rather than a colibactin intermediate, is the factor inhibiting the growth of S. aureus. Consistent with this conclusion is the observation that the most efficient growth inhibition of S. aureus occurred when both E. coli and S. aureus were in close contact within a macrocolony biofilm, whereas killing was less efficient in planktonic co-culture, which would be expected for a highly unstable molecule such as colibactin [36]. Moreover, while NMDA impaired the integrity of the S. aureus membrane and has been shown to modestly inhibit B. subtilis growth [45], this colibactin intermediate is not responsible for S. aureus killing by E. coli. To date, it is not known how colibactin enters mammalian or bacterial target cells; however, the ability of NMDA to compromise the membranes of S. aureus could serve as a mechanism for colibactin entry in some circumstances.
This study, along with a contemporaneous report [48], is the first to show that pks island genes are regulated by the BarA-UvrY TCS system, extending pks regulatory inputs beyond iron limitation [46,47]. One of the direct targets of the BarA-UvrY TCS is the Csr system, which in turn regulates diverse functional pathways such as glycolysis, gluconeogenesis, and expression of virulence factors such as biofilm formation, toxin production and pilus expression [50]. BarA-UvrY activation induces the expression of csrB and csrC [62,63] which negatively regulate the activity of the CsrA transcriptional regulator [50]. Here we show that CsrA regulates the expression of the pks island via interactions with the mRNA of the pks regulator ClbR. CsrA have also been shown to bind to the mRNA of ClbQ, an enzyme involved in colibactin synthesis, suggesting that CsrA may also regulate the expression of other pks island genes by direct binding of their mRNA [48]. Together, our data support a link between the BarA-UvrY TCS, the Csr system, and colibactin synthesis. Colibactin function has largely been studied in the context of colorectal cancer [36]. Since we know that pks is regulated by both iron and the BarA-UvrY system, and since iron may not always be a limited nutrient in the gastrointestinal (GI) tract [64], these facts suggest that BarA-UvrY may be the predominant regulator of pks island expression in the GI tract. Short-chain fatty acids (SCFAs) such as acetate, propionate and butyrate are abundant in the GI tract [65]. Furthermore, these SCFAs have been demonstrated to be the stimulus of the BarA histidine kinase [66]. As such, the presence of SCFA could serve as a signal for upregulation of the pks island via the BarA-UvrY TCS in the GI tract. Consistent with a role for SCFA sensing by BarA, both E. coli and S. aureus have been reported to accumulate and increase production of these molecules when is iron is

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus limited, which is suggested by gene expression profiles of mixed species macrocolonies [67,68]. Overall, these results not only bridge the knowledge gap in understanding the polymicrobial interactions between E. coli and S. aureus, but also contribute to the understanding pks island regulation in the context of polymicrobial infections.
E. coli and S. aureus are causative agents of wound infections [69], where they can be co-isolated from polymicrobial wound infections [20,70]. In one study, E. coli and S. aureus comprised 6.8% of the observed polymicrobial infections in wounds, second only to P. aeruginosa and S. aureus co-infections, suggesting that these species may interact during infection [71]. Our report that pks+ strains of E. coli inhibit S. aureus growth in wound infections suggest several possibilities. It is possible that when E. coli and S. aureus are co-isolated from wounds, the E. coli strains do not encode or express the pks island. Alternatively, pks+ E. coli that are co-isolated with S. aureus from wound infections may retain spatial segregation within wound biofilms such that colibactin is not in close enough proximity to S. aureus to severely limit its growth. It is also possible that host-dependent factors may serve to inactivate colibactin in some individuals. More detailed epidemiological, metagenomic, and pangenomic studies of wound infection microbiota are required to understand the ecological landscape within wound infections.
In summary, in this study, we have shown that co-infection with S. aureus induces E. coli colibactin production, which in turn is inhibitory to S. aureus in vitro and in vivo, informing the microbial ecology at play during polymicrobial wound infections (Fig 7). Additionally, we report that the BarA-UvrY TCS indirectly regulates pks gene expression via the Csr system during interspecies competition. The antimicrobial spectrum of colibactin is not limited to S. aureus as previously reported [60], but extends to all the Staphylococcal species we tested. While it will be useful to have a greater understanding of the range of bacterial species that is targeted by colibactin, current knowledge raises the possibility of colibactin-related compounds as a narrow-spectrum anti-bacterial therapeutic. Its genotoxicity toward mammalian cells notwithstanding, colibactin has also been enigmatic to purify at useful yields [36], complicating its optimization as a therapeutic. Nonetheless, this work underscores the importance of the BarA-UvrY two component system in the regulation of the pks island, which could potentially be a therapeutic target to inhibit colibactin synthesis.

Ethics statement
All animal experiments were performed with approval from the Institutional Animal Care and Use Committee (IACUC) in Nanyang Technological University, School of Biological Sciences under protocol ARF-SBS/NIE-A19061.

Bacteria strains and growth conditions
All strains and plasmids used in this study are listed in Table 1. Both E. coli and S. aureus were grown in Tryptic Soy Broth (TSB; BD Bacto, USA) at 37˚C either with shaking at 200 RPM or under static conditions to late stationary phase. Overnight cultures were normalized to 1-2 x 10 8 colony forming units (CFU)/ mL by washing the cell pellets twice with phosphate buffered saline (PBS) and then normalized to optical density (OD 600nm ) of 0.4 (E. coli) and 0.5 (S. aureus) by diluting in PBS.

Generation of mutants
E. coli UTI89 mutants were generated using the positive-negative selection system as described previously [83]. Briefly, the first recombination requires amplification of the positive-negative

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus selection cassette (Kan_RelE) from the plasmid pSLC-217 via PCR. Primers contained 50 bp homology sequence that upstream or downstream to the target gene. E. coli UTI89 carrying the pKM208 plasmid were induced with 1 mM IPTG and made electro-competent. The competent cells were transformed with 1 μg of PCR product via electroporation. The electroporated cells were recovered in LB at 37˚C for 3 hr with shaking, followed by static incubation for 1 hr. The transformed cells were plated on LB agar plates supplemented with 50 μg/mL kanamycin to select for cells with the Kan_RelE selection cassette inserted into the target gene. The second recombination requires the amplification of 500 bp of DNA sequence that is upstream and downstream of the target gene and stitching them together. For the second recombination, E. coli UTI89 with the positive-negative cassette inserted into the target gene and carrying the pKM208 plasmid were induced with 1 mM IPTG and made electro-competent. The electrocompetent cells were transformed with 1 μg of the stitching fragment via electroporation. After recovery, the cells were plated on M9 agar plates supplemented with 0.2% rhamnose. The resulting knockout mutants were confirmed by colony PCR (see S5 Table for primers used in this study). To generate the E. coli clbR mut mutant strain, the first recombination was done by amplifying of the positive-negative selection cassette (Kan_RelE) using primers containing 50 bp homology sequence that is upstream or downstream of clbR. The second

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus recombination requires the amplification of the insert using primers containing 50 bp homology sequence that is upstream or downstream of clbR. This insert was ordered as a gBlock from Integrated DNA Technologies Pte. Ltd, Singapore. The gBlock sequence can be found in S7 Fig. Bases for the third GGA motif was left unmodified to avoid alteration of the ClbR protein sequence.

Generation of plasmids
To create the ClbS expression vector, plasmid pJC-2343 was linearized with primers (InFu-sion_Vector_F/InFusion_Vector_R) and ClbS was amplified with primers (InFusion_ClbS_F/ InFusion_ClbS_R) using E. coli UTI89 genomic DNA as a template. PCR was performed using Q5 High-Fidelity DNA polymerase (New England Biolabs, United States) according to the

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus manufacturer's protocol. Thereafter, PCR purification was performed using Wizard SV Gel and PCR Clean-Up System (Promega, United States) following the manufacturer's protocol. ClbS DNA was inserted into the linearized plasmid using In-Fusion HD Cloning system (Takara, Japan). The infusion product was used to transform into Stellar Competent Cells (Takara, Japan). The vector was linearized by inverse PCR with outward directed primers (SodA_RBS_F/ SodA_RBS_R) containing the SodA RBS and re-ligated using Kinase, Ligase, DpnI (KLD) mix (New England Biolabs, United States) [86]. The plasmid pJC-2343-ClbS was extracted from Stellar competent cells using the Monarch Plasmid Miniprep Kit and used to transform S. aureus USA300 LAC via phage generalized transduction. The sarA P1 promoter was amplified from NCTC 8325 with primers JCO 1141 + JCO 1142 and cloned into pJC1213 at SphI and PstI to generate pJC2343 [85]. Primers used are shown in S5 Table. Plasmids were verified by sequencing. Successful expression of ClbS in S. aureus USA300 LAC was verified by Western blot using guinea pig polyclonal antisera against ClbS and an anti-Guinea pig-HRP secondary antibody (Invitrogen, United States) for detection. To create vectors for expression of csrA and csrB, the respective genes were amplified using E. coli UTI89 genomic DNA as a template and primers containing NcoI and HindIII restriction sites. PCR was performed using Q5 High-Fidelity DNA polymerase (New England Biolabs, United States) according to the manufacturer's protocol, using primers (OEcsrA_F/ OEcsrA_R) for the csrA insert and primers (OEcsrB_F/ OEcsrB_R) for the csrB insert. PCR purification was performed using Wizard SV Gel and PCR Clean-Up System (Promega, United States) in accordance with the manufacturer's protocol. Thereafter, the vector pTrc99A and the PCR products were digested with NcoI-HF and HindIII-HF (New England Biolabs, United States) according to the manufacturer's protocol. The vector and insert were ligated with T4 DNA ligase (New England Biolabs, United States) following the manufacturer's protocol. The ligated product was used to transform Stellar Competent Cells (Takara, Japan). The plasmids pTrc99A-CsrA and pTrc99A-CsrB were extracted from Stellar competent cells using the Monarch Plasmid Miniprep Kit and used to transform electrocompetent E. coli UTI89. Plasmids were verified by sequencing. Primers used in this study are listed below in S5 Table. Generation of polyclonal antisera Recombinant protein fragments were designed, expressed, and purified using the Protein Production Platform (NTU, Singapore) as previously described [87]. The ClbS target comprised of amino acid residues 2 to 166 from NCBI RefSeq accession no. ABE07674.1 and were cloned into pNIC28-Bsa4 with an N-terminal His tag followed by a TEV protease cleavage site. Polyclonal antisera were generated commercially (SABio, Singapore) by immunization of guinea pigs with purified recombinant ClbS. Specificity of the immune sera was confirmed by the absence of signal on Western blots of whole-cell lysates from wild-type S. aureus USA300 LAC with vector control.

Macrocolony biofilm assay
E. coli and S. aureus were grown to late stationary phase and normalized as described above. Normalized cultures of E. coli and S. aureus were mixed at a 1:1 ratio for mixed species macrocolony inocula or diluted twice with PBS for single species macrocolony inocula. 5 μL of each mixture were spotted on TSB supplemented with 1.5% (w/v) agar. Macrocolonies were grown at 37˚C to the required timepoint. Thereafter, the macrocolonies were harvested using a sterile blade and resuspended in PBS. For enumeration of viable CFU of each strain, the resuspension was plated on medium to select for E. coli (MacConkey; BD BBL, USA) or S. aureus (TSB supplemented with colistin and nalidixic acid; 5 μg/mL each).

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus Planktonic co-culture assay E. coli and S. aureus were grown to late stationary phase and normalized as described above. Normalized cultures of E. coli and S. aureus were mixed at a 1:1 ratio for mixed cultures or diluted twice with PBS for single cultures. 5 μL of each mixture was inoculated in 5 mL of TSB broth and grown at 37˚C with shaking at 200 RPM. At specific timepoints, 200 μL of the culture was sampled for enumeration of viable CFU before performing serial dilution and plating on selective medium to select for E. coli and S. aureus.

RNA extraction from macrocolonies
Single species and mixed species macrocolonies were grown for 6 hours followed by RNA extraction. The macrocolony was first resuspended in TRIzol Reagent (Ambion) and physical cell lysis was performed using Lysing Matrix B (MP Biomedicals). Thereafter, nucleic acids were purified via chloroform extraction followed by isopropanol precipitation. To remove DNA, DNase treatment was performed using the TURBO DNA-free kit (Ambion, USA). Ribosomal RNA (rRNA) was depleted from the samples using the RIBO-Zero Magnetic Bacterial Kit (Epicentre). RNA was converted to cDNA using the NEBNext RNA First Strand Synthesis Module and NEBNext Ultra Directional RNA Second Strand Synthesis Module (New England Biolabs, USA). Library preparation was performed by the SCELSE sequencing facility and sequenced via Illumina Miseq2500 machine as 250 bp paired reads.

Transcriptomic analysis
RNA sequencing reads were trimmed via BBMap tools (Bushnell, 2016). The trimmed reads were mapped to S. aureus HG001 (GenBank assembly accession GCA_000013425.1) and E. coli UTI89 reference genome (GenBank assembly accession GCA_000013265.1) using BWA (version 0.7.15-r1140) [88,89]. Reads were mapped to predicted open reading frames to each reference genome using HTSeq [90]. Gene expression analyses were done in R (version 3.4.4) using Bioconductor package, edgeR [91]. Gene expression differences were considered significant if the false discovery rate (FDR) was below 0.05. Annotation of genes was done using Kyoto Encyclopedia of Genes and Genomes (KEGG). The raw and processed data for the RNA-seq can be found under the GEO accession number: GSE190571.

Generation of E. coli transposon mutant library
E. coli UTI89 were made electrocompetent, achieving a transformation efficiency of 10 7 −10 9 CFU/μg of DNA. Briefly, pre-warmed SB medium (Tryptone, 30 g/L; yeast extract, 20 g/L; MOPS, 10 g/L) were inoculated with overnight cultures at a 1:250 ratio and incubated at 37˚C with shaking at 200 RPM to mid-log phase (OD 600nm 0.8-0.9). The cultures were then chilled on ice for 15 min before washing the cell pellets 3 times in ice cold 10% glycerol. The cell pellets were resuspended in 1 mL of 10% glycerol and aliquoted into 50 μL aliquots. The aliquots were flash frozen in liquid nitrogen and stored at -80˚C. A transposon library of E. coli UTI89 was generated with the EZ-Tn5 <R6Kγori/KAN-2>Tnp Transposome Kit (Epicentre), according to the manufacturer's protocol. Following transformation, the electroporated cells were allowed to recover at 37˚C for 1 hr in SOC media (Yeast extract, 5 g/L; Tryptone, 20 g/L; 10 mM NaCl; 2.5 mM KCl; 10 mM MgCl 2 ; 10 mM MgSO 4 ; and 20 mM glucose). Finally, the electroporated cells were diluted in PBS to achieve approximately 100 CFU/plate. The diluted cells were spread on Miller's LB 1.5% (w/v) agar plates supplemented with 50 μg/mL kanamycin and incubated overnight at 37˚C.

PLOS PATHOGENS
E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus

Transposon library screen
Individual mutants of the E. coli UTI89 transposon library were inoculated in 200 μL of LB media in 96-well plates and incubated at 37˚C statically overnight. A S. aureus USA300 LAC-GFP overnight culture was normalized as described above and diluted 100-fold to a final volume of 200 μL in 96-well plates before 3 μL of each UTI89 mutant were transferred into each well. Finally, 3 μL of the mixed cultures were spotted onto TSB agar and incubated at 37˚C for 48 hours. A primary screen was conducted based on fluorescence intensity within the macrocolony, indicative of viable GFP-expressing S. aureus. Subsequently, mutants from the primary screen were validated by macrocolony biofilm assays and growth kinetic assays before whole genome sequencing was performed to identify the location of the transposon.

Solid phase synthesis of N-myristoyl-D-Asn synthesis
The synthesis of N-myristoyl-D-Asn (NMDA) was performed as previously described [45], with modifications. All the solvents and reagents were purchased from commercial suppliers and used without further purification. N2-Fmoc-N4-trityl-D-asparagine [Fmoc-D-Asn(Trt)-OH], 2-chlorotrityl chloride resin (1.0mmol/g, 100~200mesh, 1%DVB) and PyBOP were purchased from GL Biochem (Shanghai) Ltd. 1H NMR was recorded on a Bruker 400 MHz spectrometer at 298 K. All chemical shifts were quoted in ppm and coupling constants were measured in Hz. Electrospray ionization mass spectrum (ESI-MS) of NMDA was measured in negative mode on a Thermo LTQ XL system.
Pre Fmoc deprotection. To the resin (S8C Fig, 0.75mmol) pre-swollen in DCM was added 20% piperidine in DMF (10 mL) and the reaction mixture was shaken for 10 min. The solution was drained, and the resin was washed with DMF (x3), DCM (x3). This procedure was repeated twice to obtain the resin (S8D Fig). Myristic acid (S8E Fig, 3 equiv.) and PyBOP (6 equiv.) were dissolved in DMF/DCM (50/50). DIPEA (8 equiv.) was added to the mixture to activate the carboxylic acid. The solution was added to the resin (S8D Fig) and the mixture was shaken for 1 h at room temperature. Completion of the coupling reaction was checked using the Ninhydrin test. The solution was drained, and the resin was washed with DMF (3 times), DCM (3 times) successively to give the resin (S8F Fig). Cleavage of NMDA from the resin. To the resin (S8F Fig) was added the cleavage mixture TFA/H 2 O/TIS (95%/2.5%/2.5%, 5 mL) and the mixture was shaken for 3 h at room temperature. The resin was removed by filtration and the resin was washed with the cleavage mixture once (2.5 mL). To the combined filtrate was added dropwise cold diethyl ether to precipitate the crude NMDA. The precipitate was collected after centrifugation and the diethyl PLOS PATHOGENS E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus ether decanted. This solid was washed with cold diethyl ether three times (20-30 mL x3) using the centrifugation procedure. The crude product was purified by semi-preparative reversephase-HPLC. Semi-preparative RP-HPLC was preformed using a Shimadzu HPLC system equipped with a Phenomenex jupiter-C18 RP column (10 × 250 mm, 5 μm) with a flow rate of 2.5 mL per minute, eluting using a gradient of buffer B (90% acetonitrile, 10% H 2 O, 0.045% TFA) in buffer A (H 2 O, 0.045% TFA). The combined pure NMDA fractions after HPLC purification were lyophilized to afford N-myristoyl-D-asparagine in powder form.
Compound characterization. The obtained pure compound was characterized by 1H NMR (S9 Fig).

N-myristoyl-D-Asn growth inhibition assay
Overnight cultures of S. aureus were normalized to OD 600nm of 0.4 and diluted 100-fold. Thereafter, 8 μL of the diluted cultures were inoculated into 96-well plates containing 200 μL of TSB media supplemented with NMDA at 100 μM, 300 μM and 600 μM. For all tested NMDA concentrations and the vehicle control, DMSO was supplemented to 1% (v/v). The plates were incubated at 37˚C in a Tecan Infinite M200 Pro spectrophotometer. Absorbance readings at 600 nm were taken every 15 min for 12 hours.

Cell permeability assay
Cell permeability was determined using propidium iodide (PI) staining and flow cytometry. S. aureus USA300 LAC overnight cultures were sub-cultured into fresh TSB media and allowed to grow to mid-log phase at OD 600nm of 0.6. Subsequently, the cultures were treated with DMSO (1%), 100 μM palmitoleic acid and 600 μM NMDA and incubated at 37˚C for 15 min. After the treatment, the cells were washed twice with PBS and stained with PI Buffer (Abcam) for 30 min at room temperature. Flow cytometry was performed with flow cytometer Fortessa X to identify the population of S. aureus stained positive for PI after treatment with compounds. Data analysis was performed using FlowJo, version 10.

RNA extraction and real time quantitative PCR (RT-qPCR)
Macrocolonies were grown as described above. RNA from macrocolonies was extracted using the RNeasy Mini Kit (Qiagen, United States) according to the manufacturer's protocol. Genomic DNA was removed by DNase treatment (TURBO DNA-free Kit, Ambion). RNA and DNA were quantified using Qubit RNA Assay Kit and Qubit dsDNA HS Assay kits (Invitrogen, United States). RNA quality was analyzed using Agilent RNA ScreenTape (Agilent Technologies, United States). RNA samples with minimum RIN value of 7.5 and DNA contamination of not more than 10% were converted to cDNA using SuperScript III Firststrand Synthesis Supermix (Invitrogen, United States) with accordance to the manufacturer's protocol. RT-qPCR reaction mix was prepared using KAPA SYBR FAST qPCR Kit Master Mix (2X) Universal (Kapa biosystem, United States) and ran on a StepOnePlus Real-Time PCR System (Applied Biosystems, USA). Primers GyrA_F/GyrA_R were used to amplify gyrA (Housekeeping gene), primers ClbA_F/ClbA_R were used to amplify clbA, primers ClbB_F/ ClbB_R were used to amplify clbB. Primers are found in S6 Table. PLOS PATHOGENS E. coli BarA-UvrY TCS regulates colibactin production via CsrA to kill S. aureus

Mouse model of polymicrobial wound infection
Bacteria was grown as described above, normalized to 10 6 CFU/10 μL, and used to infect wounds of C57BL/6 mice (Male, 7-8 weeks old; InVivos, Singapore) as previously described [92]. Briefly, the animals were anesthetized with 3% isoflurane. Dorsal hair was shaven and fine hair was removed after the application of Nair cream (Church and Dwight Co, Charles Ewing Boulevard, USA) and shaved using a scalpel blade. The skin was disinfected with 70% ethanol and a full-thickness wound was created with a 6 mm biopsy punch (Integra Miltex, New York, USA). The wounds were inoculated with 10 μL of the respective inoculum (E. coli, 1-2 x 10 6 CFU; S. aureus 1-2 x 10 6 CFU; Mixed 1-2 x 10 6 CFU each). Thereafter, the wound site was sealed with a transparent dressing (Tegaderm 3M, St Paul Minnesota, USA). At the indicated timepoints, mice were euthanized, and the wounds were excised and homogenized in 1 mL PBS. Viable bacteria in the wound homogenates were enumerated by plating onto selective media for E. coli (MacConkey; Merck Singapore) and S. aureus (MRSASelect II Agar; Biorad USA).