In Escherichia coli, the biosynthetic pathways of several small iron-scavenging molecules known as siderophores (enterobactin, salmochelins and yersiniabactin) and of a genotoxin (colibactin) are known to require a 4′-phosphopantetheinyl transferase (PPTase). Only two PPTases have been clearly identified: EntD and ClbA. The gene coding for EntD is part of the core genome of E. coli, whereas ClbA is encoded on the pks pathogenicity island which codes for colibactin. Interestingly, the pks island is physically associated with the high pathogenicity island (HPI) in a subset of highly virulent E. coli strains. The HPI carries the gene cluster required for yersiniabactin synthesis except for a gene coding its cognate PPTase. Here we investigated a potential interplay between the synthesis pathways leading to the production of siderophores and colibactin, through a functional interchangeability between EntD and ClbA. We demonstrated that ClbA could contribute to siderophores synthesis. Inactivation of both entD and clbA abolished the virulence of extra-intestinal pathogenic E. coli (ExPEC) in a mouse sepsis model, and the presence of either functional EntD or ClbA was required for the survival of ExPEC in vivo. This is the first report demonstrating a connection between multiple phosphopantetheinyl-requiring pathways leading to the biosynthesis of functionally distinct secondary metabolites in a given microorganism. Therefore, we hypothesize that the strict association of the pks island with HPI has been selected in highly virulent E. coli because ClbA is a promiscuous PPTase that can contribute to the synthesis of both the genotoxin and siderophores. The data highlight the complex regulatory interaction of various virulence features with different functions. The identification of key points of these networks is not only essential to the understanding of ExPEC virulence but also an attractive and promising target for the development of anti-virulence therapy strategies.
The synthesis of numerous molecules involved in the virulence potential and fitness of pathogenic bacteria requires a particular enzyme family, i.e. phosphopantetheinyl transferases (PPTases). To date, the synthesis of a given bioactive metabolite was thought to require a specific PPTase. As PPTases are being investigated as promising targets for antibacterial development, we addressed the question of a possible functional interchangeability between PPTases in Escherichia coli. PPTases are known to be involved in the synthesis of low-molecular weight iron chelators (siderophores), and of a genotoxin named colibactin. Here we demonstrated interplay between the synthesis pathways leading to the production of siderophores and of colibactin. We showed that inactivation of both PPTases abolished the virulence of extra-intestinal pathogenic E. coli (ExPEC) in a mouse sepsis model. To our knowledge, this is the first demonstration of interplay between multiple PPTases-requiring pathways leading to the biosynthesis of functionally distinctive virulence factors, in a given microorganism. The extensive substrate specificity of PPTase ClbA could account for the co-selection and co-evolution of genomic islands encoding colibactin and yersiniabactin siderophore.
Citation: Martin P, Marcq I, Magistro G, Penary M, Garcie C, Payros D, et al. (2013) Interplay between Siderophores and Colibactin Genotoxin Biosynthetic Pathways in Escherichia coli. PLoS Pathog 9(7): e1003437. https://doi.org/10.1371/journal.ppat.1003437
Editor: Xavier Nassif, Faculté de Médecine Paris Descartes, site Necker, France
Received: February 19, 2013; Accepted: May 2, 2013; Published: July 11, 2013
Copyright: © 2013 Martin et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was supported by grants from FEDER (MYCA), from the Centre de Physiopathologie de Toulouse Purpan (CPTP, Sepsis: the critical role of iron) and from the French National Research Agency (ANR). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Escherichia coli is a normal resident of the lower-gut of humans and animals. Although usually a commensal, E. coli can be also a pathogen, associated with diarrheal disease and extra-intestinal infections , . The majority of E. coli strains can be assigned to one of five main phylogenetic groups: A, B1, B2, D and E . Strains of the distinct phylogenetic groups differ in their phenotypic and genotypic characteristics –. Extra-intestinal pathogenic E. coli (ExPEC), which display enhanced ability to cause infection outside the intestinal tract, carry specific genetic determinants or virulence factors that are clustered on different pathogenicity islands . These virulence factors associated with extra-intestinal infections are nonrandomly distributed, and strains of the E. coli phylogenetic group B2 harbor the greatest frequency and diversity of virulence traits , .
As iron bioavailability is limited in the host, ExPEC are known to synthesize up to four types of siderophores involved in iron uptake: enterobactin, salmochelins, yersiniabactin and aerobactin , . The biosynthesis of the first three requires a 4′-phosphopantetheinyl transferase (PPTase). These enzymes activate polyketide synthases (PKSs) and nonribosomal peptide synthetases (NRPSs) by catalyzing the transfer of a phosphopantetheinyl (P-pant) moiety from coenzyme A to conserved serine residues on PKSs and NRPSs , . In organisms containing multiple P-pant-requiring pathways, each pathway generally involves a dedicated cognate PPTase . In E. coli, the EntD PPTase is involved in the synthesis of enterobactin  and salmochelins, which are glycosylated forms of enterobactin . The IroA locus responsible for salmochelins production is located either on a chromosomal pathogenicity island or on a transmissible plasmid . Contrary to enterobactin, salmochelins are able to evade the mammalian innate immune response protein lipocalin 2 (siderocalin) and are therefore more potent virulence factors . The other siderophore necessitating a PPTase is yersiniabactin. This siderophore is encoded by the high-pathogenicity island (HPI) that was acquired through horizontal transfer . The HPI core region was detected in more than 70% of ExPEC isolated from blood cultures, urine samples and cerebrospinal fluid . While yersiniabactin production in Yersinia requires the YbtD PPTase encoded outside the HPI , no gene homologous to ybtD has been identified in the genome of E. coli strains producing yersiniabactin. The PPTase committed to the synthesis of yersiniabactin in E. coli remains unknown.
We have shown that a number of E. coli strains from phylogenetic group B2 display also the pks island, which codes for the production of colibactin, a polyketide-non ribosomal peptide genotoxin . Colibactin is known to induce DNA double-strand breaks, cell cycle arrest in G2-phase and megalocytosis in infected eukaryotic cells . E. coli strains harboring the pks island can induce DNA damage in enterocytes in vivo and trigger genomic instability in mammalian cells . In a rodent model of colon inflammation, colibactin was demonstrated to potentiate the development of colon cancer . Surprisingly, colibactin is also required for the colonic anti-inflammatory properties of the probiotic E. coli strain Nissle 1917 . The synthesis of colibactin requires a PPTase encoded by the clbA gene located on the pks island . Epidemiological studies revealed that the majority (73.1%) of the colibactin-positive E. coli strains was clinical ExPEC and that the pks island was significantly associated with a highly virulent subset of ExPEC isolates . Strikingly, an analysis of the prevalence of the colibactin island among Enterobacteriaceae revealed that the pks island was constantly associated with the yersiniabactin gene cluster .
In this work we investigated a potential interplay between the biosynthetic pathways leading to the production of siderophores and of the colibactin genotoxin, through a possible functional interchangeability between PPTases in E. coli. We demonstrated that ClbA can contribute to the synthesis of siderophores both in vitro and in vivo. We proved in a mouse model of sepsis that the presence of either functional EntD or ClbA is required to maintain full virulence of ExPEC. This evidenced the interconnection between pathways leading to the synthesis of distinct secondary metabolites, via the PPTase ClbA. Therefore, the strict association of the pks island with HPI could have been selected in highly virulent E. coli isolates because ClbA can contribute to the synthesis of both the genotoxin and yersiniabactin.
The pks island does not code for the biosynthesis of a siderophore in vitro
Because colibactin and siderophores belong to the same family of chemical compounds, we investigated first whether the pks island could not only allow the production of a genotoxin, but also of a siderophore. The entE gene, that encodes the ligase component of synthase multienzyme complex necessary for the enterobactin biosynthesis, was inactivated in the enterobactin producer E. coli strain MG1655. The resulting MG1655 entE mutant strain was shown not to produce any siderophore, as detected on CAS plate (Fig. 1A). The wild type (WT) and entE derivative of strain MG1655 were transformed with the bacterial artificial chromosome (BAC) harboring the entire pks island (BAC pks+). Both strains MG1655+BAC pks+ and MG1655 entE+BAC pks+ were shown to produce the genotoxin, as evidenced by the induction of double-strand breaks in eukaryotic cells (data not shown). The production of siderophore was qualitatively investigated in the resulting strains by plating on CAS plates (Fig. 1A). A yellow halo was not observed surrounding the bacterial colonies of strain MG1655 entE+BAC pks+. This showed that the pks island did not code for the biosynthesis of a siderophore.
Chrome azurol S (CAS) plates upon which the E. coli strain MG1655 and derivatives have been grown overnight. A. Wild type and entE derivatives of strain MG1655. B. Wild type and entD derivatives of strain MG1655. BAC pks+ is a bacterial artificial chromosome (BAC) harboring the entire pks island. BAC pksΔclbA is a BAC harboring the entire pks island where the clbA gene was deleted. A yellow halo is produced around siderophore secreting bacteria.
ClbA, the PPTase encoded on the pks island, can support the enterobactin siderophore synthesis in vitro
In order to test whether the ClbA PPTase was functionally capable of participating to the biosynthesis of enterobactin, the entD gene was disrupted in E. coli strain MG1655. The resulting MG1655 entD mutant strain was subsequently transformed with BAC pks+ and with the BAC harboring the entire pks island where the clbA gene was deleted (BAC pksΔclbA). The production of siderophore was investigated by plating the resulting strains on CAS medium (Fig. 1B). This revealed that disruption of the entD gene in strain MG1655 resulted in the abrogation of the production of enterobactin (Fig. 1B). The introduction of the intact pks island in strain MG1655 entD restored the production of yellow pigmentation surrounding the colonies. This was not observed upon the introduction of the pks island disrupted for the clbA gene (Fig. 1B). Introduction of a functional plasmidic clbA gene in strain MG1655 entD+BAC pksΔclbA and in strain MG1655 entD restored the production of enterobactin (data not shown).
These data evidenced that the ClbA PPTase can contribute to the enterobactin siderophore synthesis in vitro.
Both EntD and ClbA can support the yersiniabactin siderophore synthesis in vitro
Yersiniabactin is a siderophore the biosynthesis of which requires the PPTase YbtD in Yersinia pestis . Although numerous E. coli strains were shown to produce yersiniabactin, an in silico analysis of the genome of all the E. coli strains available to date did not reveal any gene homologous to the ybtD gene.
In order to test whether the ClbA PPTase was functionally proficient to participate to the biosynthesis of yersiniabactin, we analyzed the enterobactin and yersiniabactin producer E. coli strain SE15. The entD gene was disrupted in E. coli strain SE15. The resulting SE15 entD mutant was subsequently transformed with plasmids carrying wild type entD gene or clbA gene. The production of total siderophores was qualitatively (Fig. 2A) and quantitatively (Fig. 2B) investigated using the CAS assay. This revealed that disruption of the entD gene resulted in the abrogation of the production of siderophores in strain SE15 (Fig. 2A and 2B). As expected, complementation with entD gene restored the production of siderophores. Remarkably, complementation with clbA gene also resulted in the synthesis of siderophores (Fig. 2A and 2B). The synthesis of yersiniabactin was specifically quantified in the different SE15 derivatives (Fig. 2C). This revealed that in the entD mutant, the yersiniabactin biosynthesis was abolished. The introduction of entD or clbA genes in SE15 entD mutant strain resulted in the restoration of yersiniabactin production.
Siderophore production by the enterobactin and yersiniabactin siderophores producer Escherichia coli strain SE15 and derivatives. A. Chrome azurol S (CAS) plate upon which E. coli strain SE15 and derivatives have been streaked for overnight growth. B. Quantification of total siderophore production in supernatants of E. coli strain SE15 and derivatives determined by the CAS assay. The data are the means and standard deviations of 5 independent experiments. C. Quantification of the yersiniabactin siderophore production in E. coli strain SE15 and derivatives. E. coli strains HB101, MG1655 and DH5α were used as negative controls (K12). RLU: relative light units. ***: P<0.001, **: P<0.01, *: P<0.05.
These data showed that in E. coli strain SE15, EntD is the PPTase dedicated to the synthesis of yersiniabactin. Moreover, the EntD function can be substituted by ClbA. This suggests that both EntD and ClbA are involved in the synthesis of yersiniabactin in E. coli strains producing endogenously EntD and ClbA.
Colibactin synthesis cannot be sustained by EntD in vitro
As our data demonstrated that ClbA could complement EntD for the synthesis of enterobactin and yersiniabactin, we investigated whether EntD could rescue a clbA mutant for the production of colibactin. The entD gene was disrupted alone or in combination with the clbA gene in the colibactin producing E. coli strain M1/5. The M1/5 entD clbA double mutant was transformed with multicopy plasmids harboring wild type entD or clbA genes. The production of colibactin was quantified in the resulting strains through the quantification of megalocytic cells (Fig. 3A) and phosphorylation of H2AX histone (Fig. 3B) which correlate with DNA double strand breaks resulting from the genotoxic effect of colibactin , .
Colibactin production by Escherichia coli strain M1/5 and derivatives determined by megalocytosis (A) and by quantification of DNA double strand breaks (B) in infected HeLa cells. A. Live E. coli wild type strain M1/5, mutants and complemented derivatives were added directly onto HeLa cells [multiplicity of infection (MOI) = 100], cocultivated for 4 h, then washed as described in Nougayrède et al. . The cells were incubated for 72 h with gentamicin before protein staining with methylene blue. The quantification of staining was measured at OD 660 nm. **: P<0.01, *: P<0.05, ns: not significant. B. Quantification of DNA double strand breaks through the quantification of phosphorylated H2AX (γ-H2AX) using In Cell Western method . HeLa cells were infected 4 h with strain M1/5 and derivatives [MOI = 50 to 6] fixed, and examined 8 h post infection for quantification of γ-H2AX.
HeLa cells were infected with the different strains for 4 hours, fixed and stained with methylene blue in order to quantify the megalocytosis effect, as previously described . This revealed that the megalocytosis effect observed with the M1/5 entD mutant strain was similar to the effect measured with the wild type M1/5 strain (Fig. 3A). Inactivation of the clbA gene in the M1/5 entD mutant abrogated the colibactin effect (Fig. 3A). Transformation of the M1/5 entD clbA mutant strain with plasmids carrying the functional wild type clbA gene resulted in the restoration of the megalocytosis. A partial complementation of the double mutation was observed with plasmid p-clbA (1) whereas the double mutant was fully complemented with p-clbA (2). The different copy number of the plasmids can account for the quantitative differences observed below. A complementation was not observed when the wild type entD gene was expressed from a multicopy plasmid in the double mutant (Fig. 3A).
Genotoxicity of colibactin  was also examined in HeLa cells using H2AX assay based on indirect DNA double strand break detection using In Cell Western (ICW) with infrared fluorescence for H2AX phosphorylation (γ-H2AX) quantification . HeLa cells were infected with strains M1/5, M1/5 entD, M1/5 entD clbA or M1/5 entD clbA complemented with entD. Following the quantification of the γ-H2AX (green) and the DNA (red) signals (Fig. 3B), respectively, the fold induction of γ-H2AX per cell was calculated. This revealed a genotoxic dose–response depending on the multiplicity of infection (MOI, Fig. 3B). No difference of γ-H2AX per cell was observed between WT and entD mutant strains. Infection of HeLa cells with mutant M1/5 entD clbA did not induce phosphorylation of H2AX. Moreover, the introduction of the functional entD gene did not result in the generation of DNA double strand breaks in strain M1/5 entD clbA (Fig. 3B).
Altogether, these data evidenced that EntD does not contribute to the colibactin synthesis, even when highly expressed on a multicopy plasmid.
Colibactin synthesis can be sustained by exogenous PPTases in vitro
We then investigated whether other PPTases, originated from other bacterial species, could rescue a clbA mutant for the production of colibactin. The clbA gene was disrupted in E. coli strain M1/5. The M1/5 clbA mutant was transformed with plasmids harboring wild type ybtD gene that encodes the YbtD PPTase in Yersinia pestis, pptT gene the PptT PPTase in Mycobacterium tuberculosis, sfp gene the Sfp PPTase in Bacillus subtilis, and clbA gene. PptT is involved in biosynthesis of the mycobactin siderophore  and is essential for mycobacterial viability . Sfp is required for production of the peptide antibiotic surfactin . The production of colibactin was quantified in the resulting strains through the quantification of megalocytic cells (Fig. 4A) and phosphorylation of H2AX histone (Fig. 4B). This revealed that both the megalocytosis and the H2AX phosphorylation were restored in the clbA mutant upon introduction of ybtD, pptT and sfp genes.
Colibactin production by Escherichia coli strain M1/5 and derivatives determined by megalocytosis (A) and by quantification of DNA double strand breaks (B), as in Fig. 3. ***: P<0.001, **: P<0.01, ns: not significant. E. coli strain SE15, which is devoid of colibactin locus, was used as a negative control. The ybtD gene encodes the YbtD PPTase in Yersinia pestis, the pptT gene the PptT PPTase in Mycobacterium tuberculosis, and the sfp gene the Sfp PPTase in Bacillus subtilis.
These data evidenced that ClbA can be xeno-complemented for the colibactin synthesis.
ClbA is more promiscuous in its substrate specificity than EntD
In order to confirm that EntD and ClbA have narrow and broad substrate-specificity, respectively, we investigated whether EntD and ClbA had the capacity to activate the carrier protein involved in a reporter biosynthetic pathway. When activated by a PPTase, the single-module non-ribosomal peptide synthetase BpsA from Streptomyces lavendulae synthesizes a colored product (indigoidine), from a single substrate (L-glutamine) . Plasmid p-BpsA that encodes BspA was transformed into strain MG1655 entD. The resulting MG1655 ΔentD+p-BspA strain was subsequently transformed with plasmids carrying ybtD, pptT, sfp, clbA, or entD genes. In addition, E. coli strain MG1655 BAC pks+ and MG1655 BAC pksΔclbA were transformed with p-BpsA. The resulting strains that carry both the NRPS and a functional PPTase were grown in auto-induction medium, as previously described . A blue coloration was detectable in cultures after overnight incubation for all strains but strain MG1655 ΔentD+p-BspA+p-entD (Fig. 5A). A quantification of the indigoidine production was determined for all the strains (Fig. 5B). This confirmed that contrary to EntD, the PPTases YbtD, PptT, Sfp and ClbA were able to participate to the synthesis of the blue pigment.
The synthesis of the single-module non-ribosomal peptide synthetase BpsA from Streptomyces lavendulae resulting in the production of indigoidine was qualitatively (A) and quantitatively (B) assessed as previously described , . E. coli strain MG1655 ΔentD+p-BpsA was transformed with the plasmids p-ybtD, p-pptT, p-sfp, p-clbA (1), p-entD and pUC19 (left). E. coli strain MG1655 BAC pks+ and MG1655 BAC pksΔclbA were transformed with p-BpsA (right). A. All the strains were grown overnight at 18°C in an auto-inducer medium. The cultures were left at room temperature so that the blue pigmentation (indigoidine) is produced. B. Quantification of indigoidine in the different strains. The data are the means and standard deviations of 3 independent experiments. ***: P<0.001, **: P<0.01, ns: not significant.
This strengthens the fact that ClbA is more promiscuous in its substrate specificity than EntD in E. coli.
Both EntD and ClbA must be inactivated to abolish virulence of ExPEC in a mouse model of sepsis
In order to address the consequences, on the virulence of E. coli, of the cross talk between the synthesis pathways of colibactin and siderophores demonstrated in vitro, we investigated E. coli strain SP15, an extra-intestinal pathogenic E. coli strain (ExPEC) of serotype O18:K1:H7 isolated from neonatal meningitis, in a mouse model of sepsis. E. coli strain SP15 produces colibactin and four different siderophores (aerobactin, yersiniabactin, enterobactin and salmochelin). The entD or clbA genes were disrupted individually and in combination. The strains were injected individually into the mice footpad; and the mice survival was monitored (Fig. 6A). This revealed that all the strains but SP15 entD clbA induced 70% mortality within 40 hours after injection. In contrast, virulence of strain SP15 entD clbA was completely attenuated in this mouse model of sepsis (Fig. 6A).
Mice underwent footpad injection with 108 CFU of E. coli SP15 wild type strain or derivatives. A. The percentage of mice survival was monitored. 10 to 25 mice were used per group. B. 18 h post infection 4 to 10 mice per group were sacrificed. Bacteria were quantified in spleen and blood collected from each animal. For statistical analysis, two-factor ANOVA and Bonferroni's multiple comparison test was performed. ***: P<0.001, **: P<0.01.
The bacterial dissemination in the mice was analyzed (Fig. 6B). Mice were sacrificed 18 hours post injection with PBS, WT strain, single or double mutants. Spleens and blood samples were collected, and bacteria were quantified by plating on selective medium (Fig. 6B). We observed that in both spleen and blood of infected animals the bacterial loads were similar with all the strains, but strain SP15 entD clbA. No bacteria were recovered from spleen or blood of mice injected with the double mutant SP15 entD clbA (Fig. 6B).
This demonstrated that both EntD and ClbA must be inactivated to abolish virulence of ExPEC in a mouse model of sepsis.
The presence of either functional EntD or ClbA is required to maintain full virulence of ExPEC in a mouse model of sepsis
In order to investigate the relative importance of EntD and ClbA in the virulence of E. coli, the SP15 entD clbA mutant strain was transformed with plasmids harboring clbA or entD functional genes. The resulting complemented strains were injected in mice (Fig. 7). This showed that complementation of strain SP15 entD clbA with either clbA or entD totally restored the virulence of the strain (Fig. 7A). A slight but statistically significant delay in survival kinetics was observed when strain SP15 entD clbA complemented with the clbA gene was used for the injections (Fig. 7A). The quantification of bacteria in spleen and blood of the infected animals was determined (Fig. 7B). This revealed that complementation with clbA or entD allowed the survival of strain SP15 entD clbA in vivo, in a statistically significant manner at least in blood (Fig. 7B).
Mice underwent footpad injection with 108 CFU of E. coli SP15 entD clbA strain and complemented derivatives. A. The percentage of mice survival was monitored. 10 to 25 mice were used per group. B. 18 h post infection 3 to 5 mice per group were sacrificed. Bacteria were quantified in spleen and blood collected from each animal. For statistical analysis, two-factor ANOVA and Bonferroni's multiple comparison test was performed. ***: P<0.001, ns: not significant.
This evidenced that the presence of either functional EntD or ClbA is required to maintain full virulence of ExPEC in a mouse model of sepsis.
Our work demonstrates the interplay between the biosynthetic pathways of a genotoxin and multiple siderophores. We have shown that ClbA, encoded by the pks island, is a promiscuous PPTase which promotes the synthesis of colibactin, yersiniabactin, enterobactin and consequently salmochelins. Although we demonstrated that ClbA could substitute for an entD mutation, the reciprocity was not observed. EntD seems to be specific for the synthesis of siderophores, which is consistent with other published reports . In contrast, YbtD, the PPTase involved in yersiniabactin production in Yersinia was shown to substitute for a clbA mutation and allowed the production of colibactin. Attempts to relate conserved motifs of the group II subfamily of PPTases  with substrate specificity did not allow us to understand the functional promiscuity evidenced among certain PPTases, since type II PPTases usually have very remote primary sequences. Unfortunately, it is not possible to compare either the 3D structure of these PPTases because only the structure of Sfp is available . Type II PPTases are predicted to have a similar folding and very similar secondary structures . However it is difficult to draw conclusions on the folding of proteins and to correlate it with substrate specificity. Only the comparison of 3D high-resolution structures would provide information about the structure/function relationship of PPTases. Our work provides novel evidence that make PPTases promising targets for antibacterial development , because these enzymes are crucial for the biosynthesis of a multitude of a pathogen's collection of essential metabolites and virulence factors .
Iron is an essential element for survival of E. coli. Therefore, E. coli strains have evolved a strategy for iron acquisition which uses multiple siderophores with high-affinity for ferric iron. These include enterobactin, salmochelins, aerobactin and yersiniabactin . Each siderophore has specific affinity for iron and may be differentially regulated to provide different advantages, potentially allowing extra-intestinal pathogenic E. coli (ExPEC) to adapt to different environmental conditions or to overcome host innate immunity , , . In our model of sepsis, the ExPEC mutant that produced only aerobactin as a siderophore (strain SP15 entD clbA) was completely attenuated. This suggests that aerobactin plays a minor role in the iron uptake in this sepsis model; but the importance of each siderophore can be host and strain dependent . Interestingly, either ClbA or EntD were able to restore the virulence of strain SP15 entD clbA. However, we have shown that colibactin synthesis cannot be sustained by EntD. This suggests that not colibactin, but the siderophore systems (alone or in combination) are critical during the first step of the infection in this mouse model of sepsis. Indeed, the bacterial loads in both spleen and blood were similar in animal infected with SP15 entD clbA mutant complemented either with ClbA or EntD. Analysis of bacteria present in the popliteal lymph node confirmed this analysis (data not shown). Since the carriage of the pks island is correlated with successful long-term gut colonization in humans , colibactin could be important for the commensal lifestyle of ExPEC. Moreover, our unpublished data suggest that the genotoxin colibactin could also play a role in natural sepsis since lymphocytes are susceptible to the genotoxin.
Phylogroup B2, which includes the majority of ExPEC isolates, is considered to represent the evolutionary eldest lineage within the species . Interestingly, the pks island found in B2 isolates is highly conserved, and is physically associated to a highly conserved High-Pathogenicity Island. This might even point towards a recent emergence of a distinct subgroup within phylogroup B2. In fact, epidemiological knowledge allows defining specific clonal lineages with high ExPEC virulence potential . We believe that the most virulent and also the best colonizer of human gut resulted from a step-by-step acquisition and selection of different mobile elements. We propose here a scenario with the sequential integration of at least two pathogenicity islands and the cross talk via two PPTases (Fig. 8). At first, all E. coli strains produce at least one siderophore i.e. enterobactin. The entD gene and the other genes of the enterobactin system are part of the core genome and have been identified in all the E. coli strains isolated so far . In contrast, the HPI encoding the yersiniabactin siderophore system devoid of any PPTase gene was acquired by horizontal gene transfer. Almost all E. coli HPIs appear to result from a single ancestor, which entered the E. coli species rather recently . All strains of the phylogenetic group B2 and almost all of group D carry the HPI, whereas strains of groups A and B1 were found to be only occasionally HPI positive (Fig. 8, ). The spread of the HPI must have occurred in a dramatically fast fashion, which may indicate a strong selective pressure. We have shown in this study that EntD is actually the PPTase that mediates the synthesis of a functional yersiniabactin. E. coli strains that contain the HPI were demonstrated to be more virulent than isolates that lack the island . Moreover, yersiniabactin is frequently associated with urinary tract infections , . The pks island is known to be confined to the phylogenetic group B2. Besides, the pks island is highly represented within an especially highly virulent subset of B2 strains that exhibit extremely elevated virulence scores and an increased likelihood of causing bacteremia . It has been previously demonstrated that all the E. coli strains that acquired the pks island encoding the colibactin through horizontal transfer, also displayed the HPI locus, with an integration site in tRNA asnW gene and asnT gene, respectively (Fig. 8; ). The pks island appears to be highly conserved (or even identical) in terms of nucleotide sequence in different E. coli isolates . This may be a hint to a more recent acquisition of the pks island, compared to the HPI, which displays about 1–2% sequence divergence among the E. coli isolates (Schubert, unpublished data). We hypothesize that the association of the pks island with HPI has been selected in the highly virulent E. coli isolates because ClbA can contribute to the synthesis of both the genotoxin and yersiniabactin (and also enterobactin and consequently salmochelins). This deadly association is not confined in E. coli. Similar events also occurred in other pathogenic Enterobacteriacae since the pks island was also detected in Klebsiella pneumoniae, Enterobacter aerogenes, and Citrobacter koseri isolates where the island is also physically associated on the chromosome with the HPI locus .
Left. Phylogenetic relationships amongst the E. coli reference strains (ECOR, , ). The phylogeny was based on MLST of back-bone genes , . Ent locus positive strains are indicated in blue, HPI island positive strains are indicated in green , iro locus positive strains are indicated in orange and pks island positive strains are indicated in pink . The presence of the iro locus was determined only in B2 strains. Right. The archetypal chromosome of phylogroup B2 E. coli strains. The loci encoding enterobactin (ent), yersiniabactin (HPI), salmochelin (iro) and colibactin (pks) are represented. The arrows originating from PPTases EntD and ClbA and pointing towards other loci illustrate the capacity of the PPTase to contribute to the synthesis of metabolites from other loci.
Materials and Methods
Bacterial strains, mutagenesis procedures and growth conditions
Bacterial strains used in this study are listed in Table 1. E. coli SE15 (O150:H5) is a human commensal bacterium isolated from feces of a healthy adult and classified into E. coli phylogenetic group B2 . Strain SE15 is devoid of the pks island. E. coli M1/5 is a human commensal bacterium isolated from feces of a healthy adult and classified into E. coli phylogenetic group B2. Strain M1/5 harbors of the pks island. Strain SP15 is an extra-intestinal pathogenic E. coli strain (ExPEC) of serotype O18:K1:H7 isolated from neonatal meningitis. Strain SP15 harbors the pks island. The repertoire of siderophores the E. coli strains possess is indicated in Table 1. Gene inactivations were engineered by using the lambda Red recombinase method  using primers listed in Table 2. For complementation, the clbA gene was cloned into plasmid pASK75, a cloning vector that harbors a pBR322 origin of replication and therefore is low copy number plasmid (p-clbA (1), table 1) or PCR-Script, a cloning vector that harbors a ColE1 origin of replication and therefore is high copy number plasmid (p-clbA (2), table 1). For complementation, the entD gene was cloned into PCR-Script (p-entD, table 1).
Before injection to mice, all E. coli strains were grown overnight in LB broth supplemented with antibiotics if required, at 37°C with shaking. These cultures were diluted 1∶100 in LB broth with antibiotics when necessary and grown for 3 h at 37°C with shaking. Bacterial cells were resuspended in sterile PBS to the appropriate concentration (2×109 CFU/mL). All the strains were shown to display similar growth kinetics in vitro in LB broth (data not shown).
Detection and quantification of total siderophores
Chrome azurol S (CAS) assay was used to detect siderophores produced by E. coli. The CAS solution was prepared according to Schwyn and Neilands . E. coli strains were grown on CAS agar plates and incubated at 37°C overnight in the dark. The colonies with orange zones were siderophore-producing strains .
To quantify siderophore synthesis, 500 µL of CAS indicator solution containing 4 mM sulfosalicylic acid was mixed with the same volume of supernatant. The reaction mixtures were incubated for 60 min at room temperature to allow complex formation, and the siderophore-dependent color change was determined at OD630 nm. For quantification, the iron chelating agent 8 hydroxyquinoline (8HQ, sigma-aldrich) was used as the standard.
Quantification of yersiniabactin
The expression of the fyuA gene encoding the yersiniabactin receptor (FyuA) is known to be up-regulated in the presence of extracellular yersiniabactin . Thus, yersiniabactin-dependent up-regulation of fyuA expression can be monitored by means of a fyuA-reporter fusion in the indicator strain .
Bacterial strains were cultivated in NBD medium, i.e. Nutrient Broth (NB) medium supplemented with 200 µM α,α'-dipyridyl (Sigma), for 24 h at 37°C. Bacteria were pelleted by centrifugation and the supernatant was added to the indicator strain WR1542 carrying plasmid pACYC5.3L (kind gift of W. Rabsch, Wernigerode). The plasmid encodes all genes necessary for yersiniabactin uptake; i.e. irp6, irp7, irp8, fyuA and ybtA. Additionally, the fyuA promoter region fused to the luciferase reporter gene is included on pACYC5.3L. After further 24 h of incubation at 37°C the indicator strain was pelleted and resuspended in bacterial lysis buffer (100 mM potassium phosphate buffer [pH 7.8], 2 mM EDTA, 1% [wt/vol] Triton X-100, 5 mg/ml bovine serum albumin, 1 mM dithiothreitol, 5 mg/ml lysozyme). Complete lysis was performed by incubation at room temperature for 20 min and repeated mixing. The samples were centrifuged and supernatants of lysates were analyzed by addition of luciferase reagent (20 mM Tricine-HCl (pH 7.8), 1.07 mM (MgCO3)4 Mg(OH)2, 100 µM EDTA, 470 µM D(-) luciferin, 33.3 mM dithiothreitol, 270 µM Li3 coenzyme A, 530 µM Mg-ATP). Luciferase activities were determined in triplicates using the multimode reader Berthold Tristar LB 941. Values were corrected by relating luciferase activity to the OD600 of bacterial cultures grown 24 h in NBD medium. K12 E. coli strain DH5α served as negative control. The experiments were repeated at least three times.
Detection and quantification of indigoidine
After overnight cultures in LB broth supplemented with the appropriate antibiotics, bacteria were diluted 1∶10 in M9 minimal medium supplemented with 100 mM L-glutamine and 1 mM IPTG, and cultivated 16 h at 18–20°C under shaking , . Bacteria were then collected by centrifugation at 900× g for 5 min. At this speed, bacterial cells were pelleted while indigoidine still remained in the supernatant . Indigoidine production was quantified by measuring the absorbance of blue-colored supernatant (OD612 nm). The bacterial pellet was resuspended in PBS, and biomass was quantified by measuring the absorbance (OD450 nm). Finally, the indigoidine production was normalized with the ratio Indigoidine/Biomass (e.g. ratio OD612 nm/OD450 nm).
Determination of the megalocytosis induced by colibactin
HeLa cells were maintained by serial passage in DMEM supplemented with 10% FCS, non-essential amino acids and 50 µg/mL gentamicin. HeLa cells were dispensed in 96-well cell culture plate (5×103 cells/well). For bacterial infections, overnight LB broth cultures of E. coli were diluted in interaction medium (DMEM, 5% FCS, 25 mM HEPES) and cell cultures (∼70% confluent) were infected with a multiplicity of infection (number of bacteria per HeLa cell at the onset of infection) of 3 to 400. Four hours post-inoculation, cells were washed 3 times with HBSS and incubated in cell culture medium 72 h with 200 µg/mL gentamicin before protein staining with methylene blue (1% w/v in Tris-HCl 0.01M). The methylene blue was extracted with HCl 0.1N. The quantification of staining was measured at OD660 nm.
Determination of the genotoxic effect induced by colibactin
The In Cell Western procedure was performed as described previously . Briefly, HeLa cells were dispensed in 96-well cell culture plate (1.5×105 cells/200 µL/wells). Twenty four hours later, cells were infected with E. coli strains for 4 h. Eight hours post-infection the cells were directly fixed in the plate with 4% paraformaldehyde. Paraformaldehyde was neutralized, and cells were permeabilized as previously described . Cells were blocked with MAXblock Blocking medium (Active Motif, Belgium) supplemented with phosphatase inhibitor PHOSTOP (Roche), followed by overnight incubation with rabbit monoclonal anti γ-H2AX (Cell Signaling) (1∶200). An infrared fluorescent secondary antibody absorbing at 800 nm (IRDyeTM 800CW, Rockland) was then applied (dilution 1∶500). For DNA labeling, RedDot2 (Biotium) was used (dilution 1∶500) together with the secondary antibody. The DNA and the γ-H2AX were simultaneously visualized using an Odyssey Infrared Imaging Scanner (Li-Cor ScienceTec, Les Ulis, France) with the 680 nm fluorophore (red color) and the 800 nm fluorophore (green dye). Relative fluorescence units from the scanning allowed a quantitative analysis. Relative fluorescent units for γ-H2AX per cell (as determined by γ-H2AX divided by DNA content) were divided by vehicle controls to determine percent change in phosphorylation of H2AX levels relative to control. All experiments were carried out in triplicate.
Mouse sepsis model
Animal experimentations were carried out in accordance with the European directive for the protection of animals used for scientific purposes. The protocols were validated by the local ethics committee on animal experiment “Comité d'éthique Midi Pyrénées pour l'expérimentation animale” which is affiliated to “Comité National de Réflexion Ethique sur l'Expérimentation Animale” and linked to the french ministry of research (Referenced protocols: PX-ANI-A2-94, 95, 96, 99, 100, and 101). Nine week old female C57BL/6J mice (JANVIER) were injected into the footpad with 108 ExPEC WT, clbA mutants, entD mutant, entD clbA mutant and entD clbA mutant complemented with clbA (entD clbA+p-clbA(2)) and entD (entD clbA+p-entD), together with intraperitoneal injection of 100 µL of carbenicillin (1.6 mg/mL) or PBS. When required, mice were sacrificed by lethal anaesthesia (rompun/ketamine in 0.9% NaCl) 18 h post injection. The abdominal cavity of anesthetized mouse was opened. The widest part of the posterior vena cava was localized and sectioned. Blood was collected by aspiration from the abdominal cavity. Spleens were surgically removed. Bacteria located in spleen cells were isolated from the mechanical dissociation of the splenic tissue using Precellys tissue homogenizer. Bacteria were quantified by plating of serial dilutions of blood and dissociated spleen on appropriate selective MacConkey agar. The antibiotics used to supplement the medium correspond to the resistance displayed by the different strains and are indicated in Table 1.
Statistical analyses were conducted using GraphPad Prism 5.0d. The mean ± standard deviation (SD) is shown in figures, and P values were calculated using a one-way or two-way ANOVA followed by a Bonferroni post-test unless otherwise stated. For bacterial quantification, CFU by mg of spleen or mL of blood were log transformed for the analysis. A P value of less than 0.05 was considered statistically significant and is denoted by *. P<0.01 is denoted by ** and P<0.001 by ***.
We thank Dr. Damien Dubois for the in silico analysis of conserved domains of PPTases and for critical reading of the manuscript. We thank Claude Watrin for excellent technical help.
Conceived and designed the experiments: PM IM JPN EO. Performed the experiments: PM IM GM MP CG DP MB MO. Analyzed the data: PM GM MO MA JPN SS EO. Contributed reagents/materials/analysis tools: MA CC. Wrote the paper: PM EO.
- 1. Kaper JB, Nataro JP, Mobley HL (2004) Pathogenic Escherichia coli. Nat Rev Microbiol 2: 123–140.
- 2. Russo TA, Johnson JR (2000) Proposal for a new inclusive designation for extraintestinal pathogenic isolates of Escherichia coli: ExPEC. J Infect Dis 181: 1753–1754.
- 3. Chaudhuri RR, Henderson IR (2012) The evolution of the Escherichia coli phylogeny. Infect Genet Evol 12: 214–226.
- 4. Gordon DM, Cowling A (2003) The distribution and genetic structure of Escherichia coli in Australian vertebrates: host and geographic effects. Microbiology 149: 3575–3586.
- 5. Escobar-Paramo P, Grenet K, Le Menac'h A, Rode L, Salgado E, et al. (2004) Large-scale population structure of human commensal Escherichia coli isolates. Appl Environ Microbiol 70: 5698–5700.
- 6. Escobar-Paramo P, Le Menac'h A, Le Gall T, Amorin C, Gouriou S, et al. (2006) Identification of forces shaping the commensal Escherichia coli genetic structure by comparing animal and human isolates. Environ Microbiol 8: 1975–1984.
- 7. Hacker J, Blum-Oehler G, Muhldorfer I, Tschape H (1997) Pathogenicity islands of virulent bacteria: structure, function and impact on microbial evolution. Mol Microbiol 23: 1089–1097.
- 8. Johnson JR, Delavari P, Kuskowski M, Stell AL (2001) Phylogenetic distribution of extraintestinal virulence-associated traits in Escherichia coli. J Infect Dis 183: 78–88.
- 9. Bingen E, Picard B, Brahimi N, Mathy S, Desjardins P, et al. (1998) Phylogenetic analysis of Escherichia coli strains causing neonatal meningitis suggests horizontal gene transfer from a predominant pool of highly virulent B2 group strains. J Infect Dis 177: 642–650.
- 10. Gao Q, Wang X, Xu H, Xu Y, Ling J, et al. (2012) Roles of iron acquisition systems in virulence of extraintestinal pathogenic Escherichia coli: salmochelin and aerobactin contribute more to virulence than heme in a chicken infection model. BMC Microbiol 12: 143.
- 11. Garenaux A, Caza M, Dozois CM (2011) The Ins and Outs of siderophore mediated iron uptake by extra-intestinal pathogenic Escherichia coli. Vet Microbiol 153: 89–98.
- 12. Copp JN, Neilan BA (2006) The phosphopantetheinyl transferase superfamily: phylogenetic analysis and functional implications in cyanobacteria. Appl Environ Microbiol 72: 2298–2305.
- 13. Lambalot RH, Gehring AM, Flugel RS, Zuber P, LaCelle M, et al. (1996) A new enzyme superfamily - the phosphopantetheinyl transferases. Chem Biol 3: 923–936.
- 14. Gehring AM, Bradley KA, Walsh CT (1997) Enterobactin biosynthesis in Escherichia coli: isochorismate lyase (EntB) is a bifunctional enzyme that is phosphopantetheinylated by EntD and then acylated by EntE using ATP and 2,3-dihydroxybenzoate. Biochemistry 36: 8495–8503.
- 15. Fischbach MA, Lin H, Liu DR, Walsh CT (2005) In vitro characterization of IroB, a pathogen-associated C-glycosyltransferase. Proc Natl Acad Sci U S A 102: 571–576.
- 16. Sorsa LJ, Dufke S, Heesemann J, Schubert S (2003) Characterization of an iroBCDEN gene cluster on a transmissible plasmid of uropathogenic Escherichia coli: evidence for horizontal transfer of a chromosomal virulence factor. Infect Immun 71: 3285–3293.
- 17. Goetz DH, Holmes MA, Borregaard N, Bluhm ME, Raymond KN, et al. (2002) The neutrophil lipocalin NGAL is a bacteriostatic agent that interferes with siderophore-mediated iron acquisition. Mol Cell 10: 1033–1043.
- 18. Schubert S, Picard B, Gouriou S, Heesemann J, Denamur E (2002) Yersinia high-pathogenicity island contributes to virulence in Escherichia coli causing extraintestinal infections. Infect Immun 70: 5335–5337.
- 19. Schubert S, Rakin A, Heesemann J (2004) The Yersinia high-pathogenicity island (HPI): evolutionary and functional aspects. Int J Med Microbiol 294: 83–94.
- 20. Bobrov AG, Geoffroy VA, Perry RD (2002) Yersiniabactin production requires the thioesterase domain of HMWP2 and YbtD, a putative phosphopantetheinylate transferase. Infect Immun 70: 4204–4214.
- 21. Nougayrede JP, Homburg S, Taieb F, Boury M, Brzuszkiewicz E, et al. (2006) Escherichia coli induces DNA double-strand breaks in eukaryotic cells. Science 313: 848–851.
- 22. Cuevas-Ramos G, Petit CR, Marcq I, Boury M, Oswald E, et al. (2010) Escherichia coli induces DNA damage in vivo and triggers genomic instability in mammalian cells. Proc Natl Acad Sci U S A 107: 11537–11542.
- 23. Arthur JC, Perez-Chanona E, Muhlbauer M, Tomkovich S, Uronis JM, et al. (2012) Intestinal inflammation targets cancer-inducing activity of the microbiota. Science 338: 120–123.
- 24. Olier M, Marcq I, Salvador-Cartier C, Secher T, Dobrindt U, et al. (2012) Genotoxicity of Escherichia coli Nissle 1917 strain cannot be dissociated from its probiotic activity. Gut Microbes 3: 501–509.
- 25. Johnson JR, Johnston B, Kuskowski MA, Nougayrede JP, Oswald E (2008) Molecular epidemiology and phylogenetic distribution of the Escherichia coli pks genomic island. J Clin Microbiol 46: 3906–3911.
- 26. Putze J, Hennequin C, Nougayrede JP, Zhang W, Homburg S, et al. (2009) Genetic structure and distribution of the colibactin genomic island among members of the family Enterobacteriaceae. Infect Immun 77: 4696–4703.
- 27. Audebert M, Riu A, Jacques C, Hillenweck A, Jamin EL, et al. (2010) Use of the gamma H2AX assay for assessing the genotoxicity of polycyclic aromatic hydrocarbons in human cell lines. Toxicology Letters 199: 182–192.
- 28. Cole ST, Brosch R, Parkhill J, Garnier T, Churcher C, et al. (1998) Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 393: 537–544.
- 29. Chalut C, Botella L, de Sousa-D'Auria C, Houssin C, Guilhot C (2006) The nonredundant roles of two 4′-phosphopantetheinyl transferases in vital processes of Mycobacteria. Proc Natl Acad Sci U S A 103: 8511–8516.
- 30. Quadri LE, Weinreb PH, Lei M, Nakano MM, Zuber P, et al. (1998) Characterization of Sfp, a Bacillus subtilis phosphopantetheinyl transferase for peptidyl carrier protein domains in peptide synthetases. Biochemistry 37: 1585–1595.
- 31. Muller M, Auslander S, Auslander D, Kemmer C, Fussenegger M (2012) A novel reporter system for bacterial and mammalian cells based on the non-ribosomal peptide indigoidine. Metab Eng 14: 325–335.
- 32. Owen JG, Robins KJ, Parachin NS, Ackerley DF (2012) A functional screen for recovery of 4′-phosphopantetheinyl transferase and associated natural product biosynthesis genes from metagenome libraries. Environ Microbiol 14: 1198–1209.
- 33. Reuter K, Mofid MR, Marahiel MA, Ficner R (1999) Crystal structure of the surfactin synthetase-activating enzyme sfp: a prototype of the 4′-phosphopantetheinyl transferase superfamily. EMBO J 18: 6823–6831.
- 34. Yasgar A, Foley TL, Jadhav A, Inglese J, Burkart MD, et al. (2010) A strategy to discover inhibitors of Bacillus subtilis surfactin-type phosphopantetheinyl transferase. Mol Biosyst 6: 365–375.
- 35. Duckworth BP, Aldrich CC (2010) Development of a high-throughput fluorescence polarization assay for the discovery of phosphopantetheinyl transferase inhibitors. Anal Biochem 403: 13–19.
- 36. Fischbach MA, Lin H, Zhou L, Yu Y, Abergel RJ, et al. (2006) The pathogen-associated iroA gene cluster mediates bacterial evasion of lipocalin 2. Proc Natl Acad Sci U S A 103: 16502–16507.
- 37. Dozois CM, Daigle F, Curtiss R 3rd (2003) Identification of pathogen-specific and conserved genes expressed in vivo by an avian pathogenic Escherichia coli strain. Proc Natl Acad Sci U S A 100: 247–252.
- 38. Caza M, Lepine F, Dozois CM (2011) Secretion, but not overall synthesis, of catecholate siderophores contributes to virulence of extraintestinal pathogenic Escherichia coli. Mol Microbiol 80: 266–282.
- 39. Nowrouzian FL, Oswald E (2012) Escherichia coli strains with the capacity for long-term persistence in the bowel microbiota carry the potentially genotoxic pks island. Microb Pathog 53: 180–182.
- 40. Tenaillon O, Skurnik D, Picard B, Denamur E (2010) The population genetics of commensal Escherichia coli. Nat Rev Microbiol 8: 207–217.
- 41. Kohler CD, Dobrindt U (2011) What defines extraintestinal pathogenic Escherichia coli? Int J Med Microbiol 301: 642–647.
- 42. Dobrindt U, Hochhut B, Hentschel U, Hacker J (2004) Genomic islands in pathogenic and environmental microorganisms. Nat Rev Microbiol 2: 414–424.
- 43. Schubert S, Darlu P, Clermont O, Wieser A, Magistro G, et al. (2009) Role of intraspecies recombination in the spread of pathogenicity islands within the Escherichia coli species. PLoS Pathog 5: e1000257.
- 44. Henderson JP, Crowley JR, Pinkner JS, Walker JN, Tsukayama P, et al. (2009) Quantitative metabolomics reveals an epigenetic blueprint for iron acquisition in uropathogenic Escherichia coli. PLoS Pathog 5: e1000305.
- 45. Johnson JR, Kuskowski MA, Gajewski A, Soto S, Horcajada JP, et al. (2005) Extended virulence genotypes and phylogenetic background of Escherichia coli isolates from patients with cystitis, pyelonephritis, or prostatitis. J Infect Dis 191: 46–50.
- 46. Toh H, Oshima K, Toyoda A, Ogura Y, Ooka T, et al. (2010) Complete genome sequence of the wild-type commensal Escherichia coli strain SE15, belonging to phylogenetic group B2. J Bacteriol 192: 1165–1166.
- 47. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A 97: 6640–6645.
- 48. Schwyn B, Neilands JB (1987) Universal chemical assay for the detection and determination of siderophores. Anal Biochem 160: 47–56.
- 49. Pelludat C, Rakin A, Jacobi CA, Schubert S, Heesemann J (1998) The yersiniabactin biosynthetic gene cluster of Yersinia enterocolitica: organization and siderophore-dependent regulation. J Bacteriol 180: 538–546.
- 50. Schubert S, Cuenca S, Fischer D, Heesemann J (2000) High-pathogenicity island of Yersinia pestis in enterobacteriaceae isolated from blood cultures and urine samples: prevalence and functional expression. J Infect Dis 182: 1268–1271.
- 51. Yu D, Xu F, Valiente J, Wang S, Zhan J (2012) An indigoidine biosynthetic gene cluster from Streptomyces chromofuscus ATCC 49982 contains an unusual IndB homologue. J Ind Microbiol Biotechnol
- 52. Ochman H, Selander RK (1984) Standard reference strains of Escherichia coli from natural populations. J Bacteriol 157: 690–693.
- 53. Johnson JR, Oswald E, O'Bryan TT, Kuskowski MA, Spanjaard L (2002) Phylogenetic distribution of virulence-associated genes among Escherichia coli isolates associated with neonatal bacterial meningitis in the Netherlands. J Infect Dis 185: 774–784.