The molecular mechanisms that allow pathogenic bacteria to infect animals have been intensively studied. On the other hand, the molecular mechanisms by which bacteria acquire virulence functions are not fully understood. In the present study, we experimentally evaluated the evolution of a non-pathogenic strain of Escherichia coli in a silkworm infection model and obtained pathogenic mutant strains. As one cause of the high virulence properties of E. coli mutants, we identified amino acid substitutions in LptD (G580S) and LptE (T95I) constituting the lipopolysaccharide (LPS) transporter, which translocates LPS from the inner to the outer membrane and is essential for E. coli growth. The growth of the LptD and LptE mutants obtained in this study was indistinguishable from that of the parent strain. The LptD and LptE mutants exhibited increased secretion of outer membrane vesicles containing LPS and resistance against various antibiotics, antimicrobial peptides, and host complement. In vivo cross-linking studies revealed that the conformation of the LptD-LptE complex was altered in the LptD and LptE mutants. Furthermore, several clinical isolates of E. coli carried amino acid substitutions of LptD and LptE that conferred resistance against antimicrobial substances. This study demonstrated an experimental evolution of bacterial virulence properties in an animal infection model and identified functional alterations of the growth-essential LPS transporter that led to high bacterial virulence by conferring resistance against antimicrobial substances. These findings suggest that non-pathogenic bacteria can gain virulence traits by changing the functions of essential genes, and provide new insight to bacterial evolution in a host environment.
Author summary: Pathogenic bacteria developed their virulence properties by changing the functions of various genes after the emergence of the host animals on earth. The types of gene function alterations that confer bacterial virulence properties, however, have remained unclear. We utilized a silkworm infection model to perform an experimental evolution of bacterial virulence activity. From a non-pathogenic strain of Escherichia coli, we obtained a mutant strain that exhibited 500-fold higher virulence than the original strain and identified mutations of the lipopolysaccharide (LPS) transporter, which translocates LPS onto the bacterial surface, as one cause of the high virulence. The mutations changed the structure of the LPS transporter, increased the secretion of outer membrane vesicles, and enabled bacterial survival in the presence of host antimicrobial substances. This mechanism to gain high virulence occurs naturally, as several E. coli clinical isolates carried mutations of the LPS transporter that confer resistance against antimicrobial substances. Our study unveiled a novel mechanism by which bacteria increase their virulence through modifying their gene function.
Uncovering the difference between pathogenic bacteria and non-pathogenic bacteria is important toward understanding the mechanisms of bacterial virulence and developing novel medicines against bacterial infectious diseases. Although previous studies using gene-knockout methods or comparative genomics have revealed many bacterial genes required for virulence including toxin genes encoded on mobile genetic elements, how pathogenic bacteria acquire virulence traits, especially the gene mutations that confer bacterial virulence, during evolution has remained unclear. Experimental evolution is a novel, recently developed system to uncover the molecular mechanisms by which cells acquire various functions, such as temperature resistance [[
LPS are glycolipids constituting the outer leaflet of Gram-negative bacteria that defend bacteria against various extracellular stresses [[
In the present study, we found that amino acid substitutions of the LPS transporter increased the amount of OMVs and conferred bacterial resistance against host-derived antimicrobial substances. Furthermore, we found amino acid substitutions of the LPS transporter in clinical isolates of E. coli that contributed to resistance against antimicrobial substances. The present study is the first to demonstrate that amino acid substitutions of the growth-essential LPS transporter increase bacterial virulence properties.
We repeatedly treated a laboratory strain of E. coli with a mutagen and subsequently infected silkworms to obtain E. coli mutants with increased killing activity against the silkworms (Fig 1A). As we performed more rounds of mutagenesis and infection, the median lethal dose (LD
Graph: Fig 1 Isolation of E. coli mutants with high virulence by experimental evolution.(A) A schematic representation of experimental evolution utilizing the silkworm infection model. E. coli strain treated with a mutagen was injected into silkworm. After the silkworm died, E. coli strains were isolated from the dead silkworm hemolymph. The mutagenesis and infection were repeated for 21 cycles. (B) An E. coli single colony isolated by the experimental evolution was cultured overnight, serially diluted, and injected into silkworms. Silkworm survival was determined at 48 h after the injection and the LD50 was determined by logistic regression. The horizontal axis represents the number of times the mutagenesis and infection experiments were performed. Each colony isolated from a dead silkworm that appeared different from the others was examined for its killing activity. The three symbols marked by green dotted lines represent the HV1, HV10, and HV11 strains. (C)E. coli strains isolated in the experimental evolution were subjected to whole genome sequencing and the number of amino acid substitutions is presented. The horizontal axis and strains are the same as in B. (D) Overnight cultured bacterial cells of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were serially diluted and injected into silkworms. Silkworm survival was determined at 48 h after the injection and the LD50 was determined by logistic regression. Data shown are the mean ± standard errors from three independent experiments. The asterisk represents a p value less than 0.05 (Student's t test). (E)E. coli strains of LptD WT, LptD G580S, LptE WT, and LptE T95I were aerobically cultured in LB broth at 37˚C. The vertical axis represents the OD600 of bacterial culture, and the horizontal axis represents the culture time.
Three mutant strains, named HV1, HV10, and HV11, which were obtained by a single round of mutagenesis and infection, had a lower LD
LptD and LptE are essential for E. coli growth. We examined whether the LptD G580S mutant and LptE T95I mutant impair growth. The growth of the LptD G580S mutant and LptE T95I mutant was indistinguishable from that of their respective parent strains in nutrient broth (Fig 1E). The results suggest that LptD G580S and LptE T95I do not affect the growth capability of E. coli.
Based on the finding that the LptD G580S and LptE T95I mutants exhibited increased virulence against silkworms, we examined whether these mutants are resistant to antimicrobial peptides that constitute the innate immune system of silkworms. In silkworm hemolymph in which antimicrobial peptides were induced, the viable cell numbers of the LptD G580S and LptE T95I mutants were more than 30-fold that of the respective parent strain (Fig 2A). In contrast, there was no difference in the viable cell number between the mutants and parent strains in silkworm hemolymph in which antimicrobial peptides were not induced, or in phosphate-buffered saline (PBS) (Fig 2A). These findings suggest that the LptD G580S and LptE T95I mutants are resistant to silkworm antimicrobial peptides.
Graph: Fig 2 The LptD and LptE mutants are resistant to various antimicrobial substances.(A) The LptD WT, LptD G580S, LptE WT, and LptE T95I strains were incubated at 37˚C for 30 min in PBS, 1.7% or 3.3% silkworm hemolymph in which antimicrobial peptides were induced (AMP-induced Hemolymph), or 3.3% silkworm hemolymph in which antimicrobial peptides was not induced (Normal Hemolymph). After incubation, the number of live bacterial cells was determined. Data shown are the mean ± standard errors from two independent experiments performed in triplicate. The asterisk represents a p value less than 0.05 (Student's t test). (B) The LptD WT, LptD G580S, LptE WT, and LptE T95I strains were incubated at 37˚C for 45 min in PBS, porcine serum (Serum), or heat-treated porcine serum (Inactivated Serum). After incubation, the number of live bacterial cells was determined. Data shown are the mean ± standard errors from two independent experiments performed in triplicate. The asterisk represents a p value less than 0.05 (Student's t test). (C) Overnight cultures of LptD WT, LptD G580S, LptE WT, and LptE T95I strains were 5-fold serially diluted, spotted onto LB agar plates supplemented without or with vancomycin, levofloxacin, tetracycline, chloramphenicol, ampicillin, colistin, streptomycin, or cholic acid, and incubated at 37˚C. (D) The LptD WT, LptD G580S, LptE WT, and LptE T95I strains were aerobically cultured at 37˚C in LB broth, and n-hexane was added to the culture at 1 h after the incubation. The OD600 of the bacterial cultures were measured every 1 h.
To determine whether the LptD G580S and LptE T95I mutants are resistant to the mammalian immune system, we examined bacterial resistance against swine serum complement. The number of viable cells of the LptD G580S and LptE T95I mutants was more than 60-fold that of the respective parent strain in swine serum (Fig 2B). No difference in the viable cell numbers was observed in heat-treated swine serum in which complements were inactivated, or in PBS (Fig 2B). These results suggest that the LptD G580S and LptE T95I mutants were resistant to mammalian complement.
LptD and LptE mutations sensitize E. coli to vancomycin and cholic acid by abolishing the outer membrane barrier function [[
To clarify whether the LptD G580S and LptE T95I mutants have increased virulence in a mammalian infection model, we examined their bacterial colonization efficiency in mouse intestine. The number of colonies recovered from mice feces did not differ between the LptD G580S and LptE T95I mutants and their respective parent strains (S1A Fig). The competition assay did not detect the difference of the colonization efficiency between the LptD G580S and LptE T95I mutants and their respective parent strains (S1B Fig). Therefore, the LptD G580S and LptE T95I mutants do not increase colonization ability in mouse intestine.
To clarify how the LptD G580S and LptE T95I mutants increase resistance against extracellular antimicrobial substances, we examined the amount of LPS in bacterial cells and in the OMVs. The amount of LPS in the bacterial cells did not differ between the LptD G580S and LptE T95I mutants and their respective parent strains (Fig 3A). In contrast, the amount of LPS in the OMV fraction was increased in the LptD G580S and LptE T95I mutants compared with that in their respective parent strains (Fig 3B). In addition, the outer membrane proteins OmpC, OmpA, and OmpX in the OMV fraction were increased in the LptD G580S and LptE T95I mutants compared with their respective parent strains (Fig 3B, S2 Table). To confirm the presence of these molecules in the OMVs, we further fractionated the OMV fraction by density gradient centrifugation. LPS, OmpC, and OmpA were detected in the same intermediate fractions (fr. 7–9) of the density gradient centrifugation, indicating that these molecules are present in the OMVs (Fig 3C). Measurement of the diameter by dynamic light-scattering assay revealed that the amount of OMVs with a diameter less than 100 nm was increased in the LptD G580S and LptE T95I mutants (S2 Fig). Taken together, these results suggest that OMVs are increased in the LptD G580S and LptE T95I mutants.
Graph: Fig 3 The LptD and LptE mutants increase production of OMVs.(A) Overnight cultured bacterial cells of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were lysed, electrophoresed in SDS-polyacrylamide gels, and subjected to Western blot analysis using an anti-LPS antibody. (B) Culture supernatants of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were ultracentrifuged and the precipitates were electrophoresed in SDS-polyacrylamide gels. The gels were stained by Coomassie Brilliant Blue (upper panel). Bands for OmpC, OmpA, and OmpX were identified by peptide mass fingerprinting analysis (S2 Table). The same samples were also subjected to Western blot analysis using an anti-LPS antibody (lower panel). (C) Culture supernatants of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were ultracentrifuged and the precipitates were further fractionated by density gradient centrifugation. The fractionated samples were electrophoresed in SDS-polyacrylamide gels and stained with Coomassie Brilliant Blue (upper panel). Also, the fractionated samples were subjected to Western blot analysis using an anti-LPS antibody (lower panel).
OMVs adsorb various antimicrobial substances and thus contribute to bacterial resistance against antimicrobial substances [[
Graph: Fig 4 The LptD and LptE mutants extrude foreign chemicals together with OMVs.(A) The LptD WT, LptD G580S, LptE WT, and LptE T95I strains were transformed with plasmids to produce isoprenoids (zeaxanthin or β-carotene). The culture supernatants were ultracentrifuged for collection of OMVs. Images are the precipitates of ultracentrifugation. (B) The amounts of zeaxanthin in OMV fractions from the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were measured. Data shown are means ± standard errors from three independent experiments. The asterisk represents a p value less than 0.05 (Student's t test). (C) The amounts of zeaxanthin in the culture supernatants from the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were measured. Data shown are means ± standard errors from three independent experiments. The asterisk represents a p value less than 0.05 (Student's t test). (D) The amount of zeaxanthin in bacterial cells of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains was measured. Data shown are means ± standard errors from three independent experiments. (E) The amount of β-carotene in OMV fractions from the LptD WT, LptD G580S, LptE WT, and LptE T95I strains was measured. Data shown are means ± standard errors from three independent experiments. The asterisk represents a p value less than 0.05 (Student's t test). (F) The amounts of β-carotene in the culture supernatants from the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were measured. Data shown are means ± standard errors from three independent experiments. The asterisk represents a p value less than 0.05 (Student's t test). (G) The amount of β-carotene in bacterial cells of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains was measured. Data shown are means ± standard errors from three independent experiments.
To understand the effects of LptD G580S and LptE T95I mutations on the molecular function of LptD and LptE, we first examined the amount of LptD and LptE by Western blot analysis. Immature LptD is changed to mature LptD by disulfide bond formation [[
Graph: Fig 5 In vivo cross-linking analysis of the LptD-LptE complex in the LptD and LptE mutants.(A) Bacterial cells of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains at a stationary phase or logarithmic growth phase were lysed and subjected to Western blot analysis using an anti-LptD antibody or anti-LptE antibody. (B) The LptD WT, LptD G580S, LptE WT, and LptE T95I strains carrying the lptE amber mutations (T86am, F90am, R124am, and R150am) were cultured in the presence of pBPA. Each strain expressing pBPA-substituted LptE was irradiated with UV light or not irradiated, and subjected to Western blot analysis using anti-LptE antibody. Lower graphs indicate the relative band intensity of the LptD-LptE complex in the LptD or LptE mutants compared with that in the parent strain. Data shown are means ± standard errors from three independent experiments. The asterisk represents a p value less than 0.05 (Student's t test). (C) UV-irradiated samples of the LptE WT and LptE T95I strains expressing LptE T86am were mixed and electrophoresed (right lane). The gel was subjected to Western blot analysis using anti-LptE antibody.
We then hypothesized that the molecular functions of LptD and LptE are affected by the LptD G580S and LptE T95I mutations. LptD and LptE form a complex on the outer membrane and constitute a final exit tunnel for LPS. We examined whether the LptD-LptE complex structure was changed in the LptD G580S and LptE T95I mutants. LptD and LptE interact with each other at several sites, which can be detected by in vivo cross-linking [[
To clarify the altered molecular functions of LptD and LptE in the LptD G580S and LptE T95I mutants, we searched for suppressor mutations against the cholic acid sensitivity of the LptD G580S and LptE T95I mutants. We isolated 47 strains resistant to cholic acid from the LptD G580S and LptE T95I mutants and determined the nucleotide sequences of lptD and lptE genes. In 10 of the 47 strains, there were amino acid substitutions in LptD and LptE, or nucleotide substitutions in the Shine-Dalgarno sequence upstream of lptE (Fig 6A). To examine whether these mutations are responsible for resistance against cholic acid, we constructed LptD G580S and LptE T95I mutants carrying these mutations by phage transduction or single-strand DNA (ssDNA) mutagenesis. The newly constructed LptD G580S and LptE T95I mutants carrying these mutations were resistant to cholic acid compared with the parental LptD G580S and LptE T95I mutants (Fig 6B), indicating that the amino acid substitutions in LptD and LptE or the nucleotide substitution in the Shine-Dalgarno sequence of lptE suppress the cholic acid sensitivity of the LptD G580S and LptE T95I mutants. Expression analysis of LptD and LptE revealed that the LptD G580S mutant carrying -6G>A lptE and the LptE T95I mutant carrying -9G>A lptE decreased the expression of both LptD and LptE (Fig 6C). The LptD G580S and LptE T95I mutants carrying the amino acid substitutions of LptD or LptE decreased the expression of LptD (Fig 6C). These results suggest that the suppressor mutations decrease the LptD expression to restore the cholic acid resistance. The suppressor mutants not only restored cholic acid resistance, but also lost vancomycin resistance (Fig 6B). Furthermore, these suppressor mutants decreased the production of OMVs and exhibited decreased killing activity against silkworms as compared with the parental LptD G580S and LptE T95I mutants (Fig 6D and 6E). These suppressor mutants showed indistinguishable growth from the parental LptD G580S and LptE T95I mutants in nutrient broth (S4 Fig). Therefore, the decreased expression of LptD caused by the suppressor mutations cancels the vancomycin resistance, the increased production of OMVs, and the increased killing activity against silkworms in the LptD G580S and LptE T95I mutants without affecting growth capability. Because decreased expression of LptD canceled the phenotypes of LptD G580S and LptE T95I, LptD G580S and LptE T95I are not considered to be loss-of-function mutations, but rather gain-of-function mutations.
Graph: Fig 6 Genetic analysis of suppressors against the LptD and LptE mutations.(A) LptD and LptE mutations in the mutants resistant to cholic acid that originated from the LptD G580S and LptE T95I mutants are presented. Mutations colored in black, cyan, or red were found in the mutants originating from the LptD G580S mutant, the LptE T95I mutant, or both mutants, respectively. (B) The LptD G580S and LptE T95I mutants carrying the mutations listed in A were constructed by phage transduction or ssDNA mutagenesis. Five-fold serial dilutions of the bacterial overnight culture were spotted onto LB plates supplemented with vancomycin or cholic acid and incubated overnight. (C) The LptD WT, LptD G580S, LptE WT, LptE T95I strains, and the suppressor mutants in B were subjected to Western blot analysis using an anti-LptD antibody or anti-LptE antibody. (D) OMV fractions were prepared from the LptD WT, LptD G580S, LptE WT, LptE T95I strains, and the suppressor mutants in B. The OMV fractions were electrophoresed in SDS-polyacrylamide gels and stained with Coomassie Brilliant Blue. (E) Silkworm killing activity of the LptD WT, LptD G580S, LptE WT, LptE T95I strains, and the suppressor mutants in B were examined. Silkworms (n = 10) were injected with bacterial cells (1.5 x 108 CFU) and survival was monitored. The vertical axis represents the survival of silkworms and horizontal axis represents time after bacterial injection. The log-rank test p value was less than 0.05 between the LptD G580S mutant and other strains (upper graph), or between the LptE T95I mutant and other strains (lower graph).
To explore the possibility that amino acid substitutions of LptD and LptE occur in clinical E. coli isolates, we examined the LptD and LptE amino acid sequences of E. coli strains whose genome information was available in public databases. We found several amino acid substitutions in LptD and LptE and classified the mutations into several patterns (Fig 7A). We constructed E. coli mutants carrying the amino acid substitutions from an E. coli laboratory strain and examined their sensitivity to vancomycin and cholic acid. The E. coli mutants carrying the O7-type, O157-type, O55-type, and O127-type mutations showed better growth than the parent strain in the presence of vancomycin (Fig 7B). In contrast, there was no growth difference between mutants and the parent strain in the presence of cholic acid (Fig 7B). The O7-type, O157-type, and O55-type mutations contained the common amino acid substitutions of LptE that are assumed to contribute to vancomycin resistance (Fig 7A); therefore, we used the O55-type mutant as a representative strain for further analysis. We then examined whether the O55-type and O127-type mutants are resistant to antimicrobial peptides and thus have increased virulence in a silkworm model. The number of viable cells of the O55-type and O127-type mutants was greater than that of the parent strain in the silkworm hemolymph containing antimicrobial peptides (Fig 7C), but the killing ability against silkworms was not increased (Fig 7D). These results suggest that amino acid substitutions of LptD and LptE found in clinical E. coli isolates contribute to resistance against extracellular antimicrobial substances, but the resistance level is not sufficient to confer high killing ability against silkworms.
Graph: Fig 7 E. coli clinical isolates carry LptD and LptE mutations conferring resistance against antimicrobial substances.(A) Amino acid substitutions of LptD and LptE that are present in 61 E. coli strains (KEGG database) are shown. "Representative strain" is one of the strains carrying the mutations. "n" indicates the number of strains carrying identical mutations. (B)E. coli strains carrying the LptD and LptE mutations listed in A were constructed by ssDNA mutagenesis from a laboratory strain of E. coli (BW25113, wild-type). The bacterial strains were cultured overnight and 5-fold serial dilutions were spotted onto LB plates supplemented with vancomycin or cholic acid. (C)E. coli strain (BW25113) and E. coli strains carrying O55-type or O127-type mutations were incubated in 1.7% silkworm hemolymph in which AMP was induced (AMP-induced Hemolymph) or in PBS, and the number of viable bacterial cells was determined. Data shown are the mean ± standard errors from two independent experiments performed in duplicate. The asterisk represents a p value less than 0.05 (Student's t test). (D) Silkworm killing activity of E. coli strain (BW25113) and E. coli strains carrying O55-type or O127-type mutations was examined. Silkworms (n = 10) were injected with bacterial cells (4.1 x 108 CFU) and survival was monitored. The log-rank test p value was larger than 0.05 between three strains.
Next, we tried to identify the substitution(s) responsible for resistance to antimicrobial substances. Because the O7-type, O157-type, and O55-type mutations contained common amino acid substitutions in LptE that are assumed to contribute to vancomycin resistance, we examined the vancomycin resistance of lptE mutants carrying each amino acid substitution from the common mutations. The LptE A83S/A84Q mutant exhibited better growth than the parent strain in the presence of vancomycin (S5 Fig), indicating that either or both A83S and A84Q are responsible for the vancomycin resistance. The O127-type mutant had slightly better growth than the O6-type mutant in the presence of vancomycin (Fig 7B), suggesting that LptD N316K mutation, the difference between the two types of mutations, contributes to vancomycin resistance.
To gain more insights into the LptD and LptE amino acid substitutions conferring high virulence properties, we searched for other LptD and LptE mutations causing high killing activity of E. coli against silkworms by performing a single-round mutagenesis and infection experiment (Fig 1A). We identified that LptD G348D, LptD S350N, and LptE E139K have high killing activity in E. coli against silkworms (S6A Fig). These amino acid substitutions increased E. coli resistance to vancomycin (S6B Fig), sensitized E. coli against cholic acid (S6B Fig), and increased the production of OMVs (S6C Fig). Therefore, LptD G348D, LptD S350N, and LptE E139K increase E. coli virulence in a similar manner as LptD G580S and LptE T95I. Next, we mapped the LptD and LptE mutations in the LptD-LptE complex structure. LptD G348 and LptD S350 were present on loop 4 of LptD (Fig 8A). LptE T95 interacted with loop 4 of LptD [[
Graph: Fig 8 Location of the LptD and LptE mutations in the LptD-LptE complex structure.(A) The structural location of the LptD and LptE mutations that increase E. coli virulence is presented. The LptD-LptE complex structure is from S. flexneri (4rhb). The left image is from the exterior side, and the right image is from the horizontal side. In the right image, LptD 635–784 was removed for clarity. Loop 4 of LptD is shown in cyan, the other parts of LptD are shown in moss green, and LptE is shown in purple. The mutations obtained in the experimental evolution are in red, and the mutations present in clinical isolates are in orange. (B) Model for structural alteration and LPS translocation in the LptD and LptE mutants. In the wild-type strain, loop 4 plugs the LPS tunnel and assists in the translocation of LPS into the outer membrane (left). In the LptD and LptE mutants, the structure of loop 4 is changed, thereby attenuating its plug function and allowing LPS to be secreted into the extracellular milieu (right).
In the present study, we performed an experimental evolution of pathogenic bacteria using an animal infection model and identified mutations of the growth-essential LPS transporter as causing high virulence properties of E. coli. Furthermore, this study revealed that amino acid substitutions of the LPS transporter were present in E. coli clinical isolates and contributed to resistance against antimicrobial substance. This is the first report to demonstrate that mutations in growth-essential genes can increase the bacterial virulence potential.
We found that LptD and LptE mutants were resistant to several antibiotics, host antimicrobial substances, and an organic solvent. The LptD and LptE mutants increased OMV production and increased efflux activity of isoprenoids together with OMVs. In addition, the suppressor mutants of the LptD and LptE mutants exhibited decreased OMV production, decreased resistance against vancomycin, and attenuated killing activity against silkworms. OMVs adsorb various antimicrobial molecules at the outside of bacteria and confer bacterial resistance to antimicrobial molecules [[
LptD G580S, LptD G348D, LptD S350N, LptE T95I, and LptE E139K induced the high production of OMVs. Among these mutations, LptD G348, LptD S350N, and LptE T95I locate around loop 4 of LptD (Fig 8A). In addition, in the LptD G580S mutant, the interaction between loop 4 of LptD and LptE T86 was decreased compared with that in the parent strain (Fig 5B). In the LptE T95I mutant, the interaction between loop 4 of LptD and LptE T86 was altered compared with that in the parent strain (Fig 5C). These results suggest that most of the LptD and LptE mutants isolated in this study had an altered LptD loop structure. Loop 4 of LptD acts as a plug for the LPS tunnel at the exterior side and assists the translocation of LPS to outer membrane [[
The LptD G580S and LptE T95I mutants were resistant to several antibiotics, host antimicrobial factors, and the organic solvent n-hexane, but were sensitive to cholic acid. The LptD G348D, LptD S350N, and LptE E139K mutants were also resistant to vancomycin, but sensitive to cholic acid. The antimicrobial substances to which the LptD and LptE mutants were resistant were hydrophobic molecules with a low molecular weight (e.g., tetracycline, chloramphenicol, levofloxacin, ampicillin, and n-hexane) or cationic molecules with a high molecular weight (e.g., vancomycin, antimicrobial peptide, complement, and colistin). In contrast, the substance to which the LptD and LptE mutants were sensitive, cholic acid, is an anionic molecule with a low molecular weight. These characteristics of the molecules to which the LptD and LptE mutants were resistant or sensitive can be attributed to the nature of the OMVs; (
Several E. coli clinical isolates had LptD N316K and LptE A83S/K84Q mutations, which confer resistance against vancomycin and antimicrobial peptides, but do not cause sensitivity to cholic acid. The LptD N316K and LptE A83S/K84Q mutations locate in the LPS tunnel, which is distal from loop 4 (Fig 8A). This may imply that LptD N316K and LptE A83S/K84Q do not alter the structure of loop 4 and do not permit the influx of cholic acid. In the natural habitat, E. coli exists in the mammalian colon and is exposed to bile acids containing cholic acid. Therefore, it is assumed that E. coli clinical isolates accumulate gene mutations that confer resistance against antimicrobial molecules without increasing their sensitivity to cholic acid. On the other hand, during our evolutional experiments, E. coli strains were exposed to silkworm hemolymph that did not contain cholic acid, so the high virulence mutants would easily lose their resistance against cholic acid and develop resistance against antimicrobial peptides. In this regard, we revealed that the high virulence mutants obtained by the experimental evolution did not have increased colonization capability in mouse intestine (S1 Fig). It is possible that the sensitivity of these mutants to cholic acid cancels the advantage of their resistance to host antimicrobial peptides or complement. Further studies are needed to verify whether the high virulence of the mutants against silkworms also results in greater virulence in mammals.
We also found E. coli clinical isolates carrying amino acid substitutions in the subunits of the LPS transporter other than LptD and LptE. LptA, LptB, LptC, LptF, and LptG, which locate in the periplasm or inner membrane, were highly conserved among E. coli strains and contained fewer amino acid substitutions than LptD and LptE (S7 Fig). This finding may indicate that LptD and LptE, which locate on the outer membrane, are exposed to more variable selection pressure than the other subunits and accumulate more mutations. In relation to this idea, the amino acid sequences of LptD and LptE as well as the LptD-LptE complex structure are different among E. coli, Shigella flexneri, Salmonella typhimurium, Pseudomonas aeruginosa, Yersinia pestis, and Klebsiella pneumoniae [[
In this study, we obtained a 500-fold high virulence bacterium from a non-pathogenic bacterium by experimental evolution using an animal infection model. The method is a powerful tool for observing the process in which non-pathogenic bacterium accumulates gene mutations to increase virulence properties. Future studies to identify gene mutations that cause high virulence properties, as well as to investigate the interactions among the gene mutations will advance our understanding of how bacteria acquire virulence functions.
This study was carried out in strict accordance with the recommendation in the Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions under the jurisdiction of the Ministry of Education, Culture, Sports, Science and Technology, 2006. All mouse protocols followed the Regulations for Animal Care and Use of Okayama University and were approved by the Animal Use Committee at Okayama University (approval number: OKU-2019593).
E. coli strains were aerobically cultured in Lysogeny broth (LB) medium at 37˚C. E. coli strains carrying a miniTn10 marker or chloramphenicol-resistant gene from pKD3 were cultured in the presence of 50 μg/ml kanamycin or 25 μg/ml chloramphenicol. E. coli strains transformed with pKD46 or pEVOL-pBpF were cultured in the presence of 100 μg/ml ampicillin or 12.5 μg/ml chloramphenicol. The bacterial strains or plasmids used in this study are listed in S3 Table.
Overnight culture of the E. coli KP7600 strain (W3110 type-A, F
Fifth instar silkworms (Fu/Yo × Tsukuba/Ne) were raised from fertilized eggs (Ehime Sansyu, Ehime, Japan) at 27˚C in our laboratory [[
Genomic DNA of KP7600, HV1, HV10, and HV11 strains was extracted using the PureLink Genomic DNA Mini Kit (Invitrogen). Genomic DNA libraries of these strains were prepared using the Ion Xpress Plus Fragment Library Kit (Life Technologies). Sequencing was performed using an Ion PGM sequencer (Life Technologies). At least 100 million base sequences of 100-base single end reads were generated per sample. Genomic DNA of other mutant strains isolated after the second and further rounds of mutagenesis and infection was extracted using the QIAamp DNA Blood Mini Kit (Qiagen). The genomic DNA libraries of these mutant strains were prepared using the Nextera XT DNA Library Prep Kit (Illumina, San Diego, CA, USA) and sequenced using MiSeq (Illumina). At least 100 million base sequences of 300-base paired end reads were generated per sample. The data were analyzed using the CLC Genomics Workbench software. The reads were mapped to a reference genome of the E. coli W3110 strain and the single nucleotide polymorphism causing amino acid substitutions were identified using the basic variant detection program in the CLC software. The sequencing reads were deposited in DDBJ (KP7600, HV1, HV10, HV11 strains, DRA008387; 16 strains after the second or more round of mutagenesis, DRA005482 [round 2, SAMD00071302; round 3, SAMD00071303; round 4, SAMD00071304; round 5, SAMD00071305 and SAMD00071306; round 6, SAMD00071307; round 7, SAMD00071308-SAMD00071311; round 11, SAMD00071312 and SAMD00071313; round 16, SAMD00071314 and SAMD00071315; round 21, SAMD00071316 and SAMD00071317]).
We examined E. coli resistance against silkworm antimicrobial peptides according to our previous method with minor modifications [[
To examine E. coli resistance against serum complement, a slightly modified protocol from our previous method was used [[
To examine E. coli resistance against antibiotics and cholic acid, autoclaved LB agar was mixed with antibiotics or cholic acid and poured into a disposable square dish (Eiken Chemical Co., ltd., Tokyo, Japan). E. coli overnight cultures were 5-fold serially diluted with LB broth in a 96-well plate. The serially diluted bacterial solution was spotted onto the LB agar plates supplemented with antibiotic or cholic acid using a 12-channel pipette. The plates were incubated at 37˚C.
To examine E. coli resistance against n-hexane, we used a previously described method with minor modifications [[
Overnight cultures of the LptD G580S and LptE T95I mutants were inoculated into LB broth supplemented with or without ethylmethane sulfonate (final concentration 0.2%). The overnight culture was inoculated into fresh LB broth and aerobically cultured overnight. The overnight culture was spread on LB agar plates supplemented with cholic acid (final concentration 6%) and incubated overnight. A colony was streaked on LB plates to isolate single colonies. The single colony was inoculated into LB broth and cultured overnight. The culture was used as the stock of bacterial strains and extraction of genome DNA. The lptD and lptE genes were amplified by polymerase chain reaction (PCR) from the genome DNA using primers (lptD-F, lptD-R, lptE-F, and lptE-R) (S4 Table).
The miniTn10 marker located near the lptD gene in JD20181 (yabP:miniTn10) [[
The miniTn10 marker (yabP:miniTn10) in the LptD G580S mutant carrying the LptD R429H, LptD G489D, or LptD E509K mutations was transferred to KP7600 by transduction, resulting in the LptD G580S suppressor mutants carrying the LptD R429H, LptD G489D, or LptD E509K mutations. The miniTn10 marker (cobC:miniTn10) in the LptE T95I mutant carrying the lptE -9G>A mutation was transferred to KP7600 by transduction, resulting in the LptE T95I suppressor mutant carrying the lptE -9G>A mutation.
For the LptD G580S mutant carrying the LptE W21R mutation, a chloramphenicol-resistant marker was introduced to the genomic region near to the lptE gene by Red-recombinase mediated targeting using targeting primers (S4 Table) [[
Because the LptD G580S mutants carrying the LptD G445D or lptE -6G>A mutations showed resistance against phage P1 vir, we introduced these mutations to the LptD G580S mutant by ssDNA mutagenesis [[
ssDNA mutagenesis was performed according to the previous methods [[
The mouse colonization experiment was performed according to the previous method [[
To perform competition assay, chloramphenicol resistance marker was transferred to LptD WT, LptD G580, LptE WT, LptE T95I strains by transduction (S3 Table). Mice were administered 0.2 ml of bacterial solution (10
E. coli overnight culture (1 ml) was inoculated into 100 ml of LB broth and aerobically cultured at 37˚C for 24 h. The bacterial culture was centrifuged at 16,200 g for 10 min, and the supernatant was filtered with a 0.45-μm polyvinylidene difluoride (PVDF) membrane (Millex-HV, Millipore). The sample was ultracentrifuged at 235,000 g for 3 h (45Ti rotor, Beckman). The ultracentrifuged precipitate was suspended in HEPES-NaCl (10 mM HEPES-NaOH [pH 6.8], 146 mM NaCl). The sample was mixed with 3x sodium dodecyl sulfate (SDS) sample buffer, boiled for 5 min, and electrophoresed on a 15% SDS polyacrylamide gel. The gel was stained with Coomassie Brilliant Blue. To identify the protein, the band was excised, digested with trypsin, and subjected to matrix-assisted laser desorption ionization time-of-flight mass spectrometry.
To measure OMV diameter, the OMV fraction was filtered with 0.2-μm PVDF membranes and analyzed using a dynamic light-scattering assay (Delsa Nano C, Beckman). Further purification of the OMVs was performed according to the previously method [[
The E. coli strains were transformed with pAC-ZEAXipi [[
According to previous reports [[
E. coli overnight cultures (50 μl) were inoculated into 5 ml of LB broth and aerobically cultured for 3 h (logarithmic growth phase) or 25 h (stationary phase) at 37˚C. The bacterial culture was centrifuged at 21,500 g for 2 min, and the precipitated cells were suspended in PBS and mixed with 3x SDS sample buffer. To detect mature LptD, 3x SDS sample buffer without β-mercaptoethanol was used. The bacterial cell concentration in the sample was adjusted to OD
In vivo cross-linking experiments were performed according to the previously described method with minor modifications [[
S1 Fig. Colonization ability of the LptD and LptE mutants in mouse intestine.
(A) ICR mice (n = 6) were orally administered the LptD WT, LptD G580S, LptE WT, or LptE T95I strains. The number of bacterial colonies recovered from the mouse feces was counted on days 1, 3, and 6 after the bacterial administration. Dotted lines indicate the detection limit in the assay (20 CFU/g feces). ND, not detected. (B) The LptD WT, LptD G580S, LptE WT, or LptE T95I strains were labeled with a cassette conferring resistance to chloramphenicol. The chloramphenicol-resistant strains were mixed with the chloramphenicol-sensitive LptD WT or LptE WT strains at the ratio of 1:1 and were administered to ICR mice (n = 6–7). The number of bacterial colonies recovered from the mouse feces was counted on day 1 after the bacterial administration. The competitive index was calculated by dividing the number of chloramphenicol-resistant colonies by the number of chloramphenicol-sensitive colonies. Gray circle represents the value was more than 10 or less than 0.1, because either of the chloramphenicol-resistant or -sensitive colony was not detected.
(TIF)
S2 Fig. Diameter measurement of OMVs from the LptD and LptE mutants.
OMV fractions of the LptD WT, LptD G580S, LptE WT, and LptE T95I strains were subjected to dynamic light-scattering analysis. The horizontal axis represents the particle diameter, and the vertical axis represents the relative distribution of the particles to the total particles.
(TIF)
S3 Fig. In vivo cross-linking analysis of the LptD-LptE complex using an anti-LptD antibody.
The LptD WT, LptD G580S, LptE WT, and LptE T95I strains expressing pBPA-substituted LptE were irradiated with UV light or not irradiated, and subjected to Western blot analysis using an anti-LptD antibody.
(TIF)
S4 Fig. Growth curves of the LptD and LptE suppressor mutants.
E. coli strains of LptD WT, LptD G580S, LptE WT, LptE T95I, and the suppressor mutants were aerobically cultured in LB broth at 37˚C. The vertical axis represents the OD
(TIF)
S5 Fig. Identification of LptE mutations conferring vancomycin resistance from O55-type mutations.
E. coli strains carrying O55-type mutations were constructed by ssDNA mutagenesis. The strains were cultured overnight and 5-fold serial dilutions were spotted onto LB plates supplemented with vancomycin. The left panel indicates the amino acid substitutions carried by the mutants.
(TIF)
S6 Fig. Phenotypic characterization of the LptD G348D, LptD S350N, LptE E139K mutants.
(A) The LptD WT, LptD G348D, LptD S350N, LptE WT, and LptE E139K strains were cultured overnight and serial dilutions of bacterial cells were then injected into silkworms. Silkworm survival was counted at 48 h after the injection. The LD
(TIF)
S7 Fig. Number of amino acid substitutions in the LPS transporter subunits in various E. coli strains.
Genome data of 65 E. coli strains (KEGG database) were examined to count the number of amino acid substitutions in LptA, LptB, LptC, LptD, LptE, LptF, and LptG. Horizontal axis represents the number of amino acid substitutions, and the vertical axis represents the number of strains.
(TIF)
S1 Table. Amino acid substitutions identified in high virulence mutants.
(DOCX)
S2 Table. Identification of proteins increased in the LptD and LptE mutants.
The protein band stained with Coomassie Brilliant Blue was excised and digested in-gel with trypsin. The sample was subjected to Matrix-assisted laser desorption ionization time-of-flight mass spectrometry analysis (Microflex LRF 20, Bruker Daltonics). Database searching was performed using the Mascot search program (
(DOCX)
S3 Table. List of bacterial strains and plasmids used.
(DOCX)
S4 Table. Primers used in this study.
(DOCX)
We thank H Maki and M Akiyama for providing the P1 phage. We also thank the National BioResource Project (NBRP)-E. coli (NIG, Japan) for providing the KP7600, miniTn10 library, and BW25113. In addition, we thank S Hori for insightful comments on this study and M Urata for technical assistance.
By Chikara Kaito; Hirono Yoshikai; Ai Wakamatsu; Atsushi Miyashita; Yasuhiko Matsumoto; Tomoko Fujiyuki; Masaru Kato; Yoshitoshi Ogura; Tetsuya Hayashi; Takao Isogai and Kazuhisa Sekimizu
Reported by Author; Author; Author; Author; Author; Author; Author; Author; Author; Author; Author