ABSTRACT
Antimicrobial peptides play an important role in host defense against Vibrio cholerae. Generally, the V. cholerae O1 classical biotype is polymyxin B (PB) sensitive and El Tor is relatively resistant. Detection of classical biotype traits like the production of classical cholera toxin and PB sensitivity in El Tor strains has been reported in recent years, including in the devastating Yemen cholera outbreak during 2016-2018. To investigate the factor(s) responsible for the shift in the trend of sensitivity to PB, we studied the two-component system encoded by carRS, regulating the lipid A modification of El Tor vibrios, and found that only carR contains a single nucleotide polymorphism (SNP) in recently emerged PB-sensitive strains. We designated the two alleles present in PB-resistant and -sensitive strains carRr and carRs alleles, respectively, and replaced the carRs allele of a sensitive strain with the carRr allele, using an allelic-exchange approach. The sensitive strain then became resistant. The PB-resistant strain N16961 was made susceptible to PB in a similar fashion. Our in silico CarR protein models suggested that the D89N substitution in the more stable CarRs protein brings the two structural domains of CarR closer, constricting the DNA binding cleft. This probably reduces the expression of the carR-regulated almEFG operon, inducing PB susceptibility. Expression of almEFG in PB-sensitive strains was found to be downregulated under natural culturing conditions. In addition, the expression of carR and almEG decreased in all strains with increased concentrations of extracellular Ca2+ but increased with a rise in pH. The downregulation of almEFG in CarRs strains confirmed that the G265A mutation is responsible for the emergence of PB-sensitive El Tor strains.
INTRODUCTION
Cholera, a life-threatening disease, is still present in developing countries, where people do not have a good sanitary system, good hygiene practices, and a safe drinking water supply (1). The Haitian outbreak during 2010 provided direct evidence of these health-related factors (2). Recently, Yemen, a country in Western Asia, experienced its worst cholera outbreak, which affected more than 1.1 million people. The high infection rate (69%) has placed the disease at the top of the public health agenda (3). The Gram-negative motile bacterium Vibrio cholerae O1 is the main cause of the disease and was responsible for the seven pandemics in the recorded history of cholera. The first six pandemics were presumed to be caused by the V. cholerae O1 classical biotype, and the current (seventh) pandemic is due to the El Tor biotype (1, 4, 5). Generally, the biotypes share very close affiliation in the taxonomy of vibrios. V. cholerae O1 biotyping is based on standard biochemical tests, and polymyxin B (PB) susceptibility is one of them. The classical biotype is sensitive to the antibiotic, whereas El Tor is resistant (6). The conventional approach has become anecdotal due to the recent appearance of PB-sensitive clinical El Tor strains in Kolkata and other parts of India (7, 8). According to recent reports, the strains that caused the Kenyan and Yemen cholera outbreaks also have the PB-sensitive signature (9, 10).
Antimicrobial peptides act strongly against microorganisms, especially Gram-negative bacteria (11, 12). In V. cholerae, the outer membrane protein OmpU confers resistance to PB by interfering with the alternative sigma factor sigma E (13). In addition, the other gene products, MsbB/LpxN, involved in lipid A acylation, and AlmE, AlmF, and AlmG, which are active in glycine and diglycine modification, were also proven to be important for V. cholerae PB resistance (14, 15). Apart from cell surface modifications, V. cholerae RND (resistance-nodulation-division) family efflux systems, particularly a VexB (VC0164)-mediated pump, may contribute significantly to PB resistance (16).
The cationic antimicrobial peptide (CAMP) PB exerts its action on the bacterial outer membrane by binding to the negatively charged lipid A molecules of lipopolysaccharide existing on the surfaces of Gram-negative bacteria. After binding, CAMPs dissipate membrane integrity by forming pores, leading to outflow of cellular components (11). However, some bacteria modify their lipid A molecules to resist the detrimental effect of CAMPs; for example, El Tor vibrios modify their lipid moiety by the addition of glycine or diglycine residues (15). A recently proposed enzymatic network for glycine modification of lipid A involves three amino acid lipid modification genes: almE, almF, and almG. In V. cholerae, a two-component regulatory system, CarR-CarS, regulates the transcription of the alm operon (17). CarS (VC1319) is a membrane-bound histidine kinase that phosphorylates the soluble CarR (VC1320) at the D55 amino acid residue and helps to bind to the almEFG promoter. CarR acts as a positive regulator for the operon (17). After the translation of almEFG, glycine moieties are added to AlmF (VC1578) with the help of AlmE (VC1579). Then, AlmG (VC1577), which is a membrane-bound protein, transfers the glycines from AlmF to the negatively charged lipid A molecule of the inner membrane. Subsequently, the glycinylated lipid molecules are transported to the outer surface of the outer membrane and reduce the net negative charge on the outer surface. This protection causes the bacterial surface to reduce its net negative charge and removes the site for CAMP binding (15, 17, 18).
The carRS system also regulates the expression of the vps genes responsible for biofilm formation through an extracellular matrix consisting of Vibrio exopolysaccharide (19). vps genes are clustered in two different loci in V. cholerae, vps-I and vps-II, and are negatively regulated by carRS. In this study, we analyzed the expression of two representative genes of vps clusters. vpsA (VC0917), located in cluster I, encodes a UDP-N-acetylglucosamine 2-epimerase, and vpsL (VC0934), situated in cluster II, encodes a glycosyltransferase. Both are essential for the production of Vibrio polysaccharide and significantly interfere with biofilm formation (20). In this study, we investigated the nucleotide sequences of the genes that are reported to have roles in PB resistance in V. cholerae and the expression of carR, almEFG, vpsA, and vpsL with varying concentrations of extracellular Ca2+ and pH. Also, we analyzed the CarR models and interpreted the possible functional differences between two variants. Further, we hypothesized the possible mode of stabilization of the carRs allele in recent PB-sensitive strains over the carRr variants through in silico analysis.
RESULTS
Nucleotide sequence comparison reveals an SNP in carR.In V. cholerae, the nucleotide sequences of the genes of the two-component system carRS and the almEFG operon, responsible for glycine and diglycine addition to lipid A of the outer membrane, were analyzed mainly for the phenomenon of PB resistance (15, 17, 18). A BLAST search of a total of 12 PB resistance-conferring genes of the 12 representative isolates was performed, along with N16961 as a reference. All the genes of the two phenotypically PB-resistant isolates showed 100% similarity to those of N16961. However, the 10 PB-sensitive strains showed 100% sequence similarity to N16961, except for carR (99.86% identity) (Table 1). These 10 sensitive strains were found to contain a single nucleotide polymorphism (SNP) at nucleotide position 265 that replaced guanine with adenine (G265A) in carR. Based on our results, we hypothesize that the point mutation was responsible for converting a PB-resistant strain into a sensitive one.
Comparison of 12 polymyxin B resistance-conferring genes of all the tested polymyxin B-sensitive strains (n = 10) with those of polymyxin B-resistant N16961
A265G substitution in carR on a PB-sensitive phenotypic background.In order to confirm the role of the A-to-G change in PB susceptibility, we replaced the A residue in nucleotide 265 of carR of the recent PB-sensitive strain IDH07230 with G, constructing the IDH07230MT strain. To strengthen the hypothesis, we made a G265A substitution in carR of the PB-resistant El Tor standard strain N16961 and constructed N16961MT.
Nucleotide 265 of carR regulates PB susceptibility.The newly constructed carR mutants were checked for PB susceptibility and compared with their wild-type counterparts, i.e., IDH07230MT with IDH07230 and N16961MT with N16961. The PB MICs of the wild-type sensitive strain IDH07230 and its mutant, IDH07230MT, were determined and found to give MIC values of 1.0 μg ml−1 and 32.0 μg ml−1, respectively (Fig. 1). When the cells were plated on PB-supplemented medium (50 U ml−1), IDH07230MT was found to grow well, indicating its resistance to PB (50 U ml−1).
Polymyxin B MIC assays of wild-type (IDH07230 and N16961) and mutant (IDH07230MT and N16961MT) strains. The arrows indicate the MICs (in micrograms per milliliter) on Etest gradient polymyxin B strips; MIC values are shown below the images. Assays were carried out with at least three biological repetitions.
N16961 and its isogenic mutant, N16961MT, were found to give MIC values of 96.0 μg ml−1 and 4.0 μg ml−1, respectively, which confirmed that the mutant N16961MT is sensitive to PB (Fig. 1). This result was also verified with a PB (50 U ml−1) growth assay, in which the N16961MT strain showed no growth on PB-containing medium. Thus, we confirmed that nucleotide 265 in carR has a significant role in PB susceptibility in recently emerged V. cholerae O1 El Tor isolates.
Molecular modeling reveals differences between CarRr and CarRs structures.The CPHmodels server for protein homology modeling identified the response regulator Regx3 of Mycobacterium tuberculosis (Protein Data Bank [PDB] ID, 2OQR) as the template structure for homology-based three-dimensional (3D) model building. Though the template protein is a dimer, we used the monomeric form of the protein for homology modeling (chain A of 2OQR). The dimerization probability for CarR is much lower, in consideration of the fact that the probability of homodimer assembly for proteins with similar sequences is very low if the sequence identity falls below 40% (21). Moreover, to check the conservation of residues that contribute to H bonding between the monomers, we analyzed the biological assembly of 2OQR and identified 16 potential residues (D186, R113, K101, V99, D97, D96, G94, E84, A80, T79, V78, I76, R68, D52, E24, and E8) that form H bond interactions (see Fig. S1a in the supplemental material). Of these 16 residues, 11 are conserved in carR (marked in red in Fig. S1b). In this H bond interaction model, some critical negatively charged residues (D186, E84, and E24) that are important to form a carR homodimer are absent. Our PSI-BLAST-based search of PDB data also retrieved the same template structure, which had 37% sequence identity with our query sequence. The generated model consists of one N-terminal domain and one C-terminal domain connected by a central helix (Fig. 2A). Ramachandran plot analysis also showed that 99.5% of the residues are in the allowed region.
(A) Homology model of CarR protein. (B) Negatively charged residues positioned at the central helix and adjacent loop region of the CarRr protein. (C) H bond interaction of D89N residue with the adjacent D100 and F101 residues from the loop region of CarRs protein.
Analysis of the CarRr protein has exposed the fact that residue D89 is situated in the central helix region, surrounded by three negatively charged residues (D85, D86, and D87), as well as a nearby loop region that also contains two aspartate residues (D99 and D100) (Fig. 2B). This shows the presence of a repulsive force between the central helix and the adjacent loop residues. For CarRs protein, the N89 residue is able to form an H bond interaction with the D100 and F101 residues (Fig. 2C). This H bond interaction may be able to hold the loop near the central helix. To further study the effects of these interactions on the overall structure of the CarRr and CarRs proteins, we performed molecular dynamics simulation (MDS) analysis. The comparative root mean square deviation (RMSD) profile analysis of 10-ns MDS of CarRr and CarRs protein structures showed that the wild-type (CarRr) protein structure was quite flexible, while the mutant (CarRs) structure became rigid after 2 ns (Fig. 3A). This may be the effect of the specific D89N mutation, which is able to hold the adjacent loop region near the central helix, resulting in a more stable folding pattern. Due to the repulsive forces acting between the central helix and the adjacent loop region, the CarRr protein showed more flexibility. To understand the molecular level changes behind these structural alterations, we analyzed the minimum energy conformations of both structures extracted from the 10-ns MDS trajectory. For the CarRr protein, it showed the same extended form of folding seen in the modeled structure (Fig. 3B), but for the CarRs protein, it proved to be a compact subunit (Fig. 3C). This revealed that the H bond interactions of the N89 residue of CarRs with the adjacent loop residues D100 and F101 restricts the flexibility of the C-terminal domain, which in turn comes closer to the N-terminal domain and forms a compact subunit. In the case of the CarRr protein, the repulsive forces between the central helix and the adjacent loop are unable to stabilize the C-terminal domain, resulting in an extended format.
(A) Comparative RMSD plot of the CarRr and CarRs proteins. It is evident from the plot that the CarRs protein stabilized after 2,000 ps, whereas the CarRr protein retained its flexibility. (B and C) Minimum energy conformations of CarRr (B) and CarRs (C) proteins extracted from the 10-ns molecular dynamics simulation trajectory showing their most stable conformations.
An A265G SNP has emerged in V. cholerae carR and possesses selective advantage in nature over other carR alleles.Fixed-effects likelihood (FEL) analysis suggested that among the 12 different carR alleles, including our carRs target sequence (CARR_IDH07230) (marked “Test” in Fig. S2 in the supplemental material), from different Vibrio strains, the particular G265A mutation that occurs at the site (codon position 89) has pervasive positive/diversifying selection with significant omega (nonsynonymous/synonymous rate ratio) and P values of “infinity” and 0.001, respectively (see the supplemental material) (22). This prediction and the corresponding statistical values suggest that the G265A mutation, which is also reflected in a D89N amino acid change, has positive selection in nature. Therefore, this particular change should be beneficial to the mutant strain for its long-term survival in nature over the other Vibrio strains.
D89N mutation’s effect on CarRs stability.CarRs stability was measured using the DynaMut and SDM2 servers. The predicted change in folding, the free energy (ΔΔG), was 0.474 kcal/mol, which demonstrated the higher stability of the CarRs protein over the CarRr protein (23). Moreover, DynaMut also provided interatomic interactions of residues at the specific site of mutation for the two proteins, which also showed increased electrostatic interaction of ASN89 with the neighboring ASP100 and ASP86. These additional interactions further stabilize the central helix, as shown in our previous analysis. The SDM2 server also predicted the same effect. For our D89N (CarRs) protein, we received a pseudo-ΔΔG score of 0.13, which also supports the increased stability of the CarRs protein due to the D89N mutation (see Fig. S3 in the supplemental material).
A single point mutation in carR downregulates the expression of almE, almF, and almG.The carR-carS two-component system has been reported to regulate the expression of the almEFG operon mRNA (18). Hence, we checked whether the point mutation at nucleotide position 265 in carR was able to change its regulatory function. To verify this, we quantified almE, almF, and almG at the mRNA level by quantitative reverse transcription real-time (qRT)-PCR. Expression of the genes in the almEFG operon was found to be significantly lower (P < 0.05) in both our sensitive strains, IDH07230 and N16961MT. Wild-type N16961 and the IDH07230MT mutant showed high levels of mRNA expression. However, the level of carR expression was found to be the same in all the strains (Fig. 4).
Relative expression levels of carR, almE, almF, and almG mRNA levels measured via qRT-PCR in clinically isolated wild-type and corresponding mutant strains to compare the effects of the G265A mutation in the carR gene on transcription of almE, almF, and almG. The data were normalized to recA expression using the Pfaffl method, with expression of the wild-type N16961 set to 1.0. The graph represents the mean expression levels from three independent experiments performed in triplicate. *, P < 0.05. One-way analysis of variance (ANOVA) and Dunnett’s multiple-comparison tests were used to determine the statistical significance. The error bars represent standard deviations from the mean.
Expression of carR, almE, almG, vpsA, and vpsL was decreased with increased Ca2+ concentration and increased at high pH.Since almE and almG are parts of the almEFG operon with a common promoter, we chose the two genes as representatives in the expression analysis. The mRNA expression of carR was reduced significantly (P < 0.05) when the extracellular concentration of Ca2+ was increased from 1 mM to 10 mM and 100 mM. Similarly, almE and almG expression levels for CarRs strains were significantly lower (P < 0.05) than for CarRr variants and followed a declining trend with increased Ca2+ concentrations for all the strains. vpsA and vpsL were also shown to have decreased transcription levels at a high Ca2+ concentration (100 mM). The isogenic V. cholerae strains IDH07230 and IDH07230MT were found to have significantly lower (P < 0.05) vpsA and vpsL transcription than isogenic N16961 and N16961MT strains at all the tested concentration of Ca2+. Thus, the SNP in carR is responsible for the variation in the transcription of almEG and is not dependent on Ca2+ levels. The variation observed in almEG transcription is due to the low expression of carR, which itself is downregulated with increased Ca2+ concentration (see Fig. S4 in the supplemental material).
pH also affects the transcription of carR, almEG, vpsA, and vpsL. carR expression was found to increase with an increase in pH from 5.0 to 9.0 (the difference was not significant). As the carR levels are high at higher pH, the expression levels of the carR-regulated genes almEG in resistant and sensitive strains were also found to be higher. There was no significant difference between resistant and sensitive strains in vpsA and vpsL transcription (P > 0.05) with the increase in pH (5.0 to 9.0) (see Fig. S5 in the supplemental material).
DISCUSSION
The two epidemic biotypes of V. cholerae, classical and El Tor, differ considerably in their molecular and epidemiological attributes. Genetically, they have been found to carry different alleles of the same set of virulence-related genes, like ctxB and tcpA, or to harbor additional genetic elements, like Vibrio seventh pandemic islands (VSPs) (24). It has been observed that the classical biotype causes a more severe form of cholera than El Tor, whereas the El Tor biotype is more fit to survive under varying environmental conditions (1, 25). The recently emerged El Tor biotype was found to have not only greater fitness in the environment, but also ctxB of the classical biotype, leading to its increased pathogenicity (26). The occasional appearance of other variant strains with changes in the leader peptide region of the classical ctxB (ctxB7) was also seen in El Tor vibrios from a cholera outbreak in Haiti during 2010 (2). Weill et al. (9) suggested that the ctxB7 allele emerged in South Asia, an observation that is consistent with the first appearance of ctxB7 in Kolkata, India, during 2006; it then migrated to West Africa, Haiti, and East Africa in the form of large outbreaks (27–30). Similarly, another important phenotypic change observed was PB susceptibility in the El Tor biotype. It was first seen in Kolkata during 2012 and subsequently spread throughout India and Africa as stable clones (7–10). Moreover, it was found that, similar to 2012 El Tor isolates from Kolkata, the PB-sensitive Yemen isolates also carried the D89N substitution in CarR (9). These observations suggested that the PB-sensitive El Tor strains probably emerged first in Kolkata and then dispersed in different places.
Conventionally, PB resistance has been used as a useful marker to differentiate between the classical and El Tor biotypes. Recent reports suggest that El Tor vibrios modify lipid A with glycine or diglycine residues to increase the net positive charge to inhibit the binding of the CAMP PB to cause resistance to the drug (15). The CarR-CarS two-component system has proven to contribute significantly to this modification process. CarR is a transcriptional regulator that acts on the promoter region of almEFG (18) and is involved in the final steps of glycine modification. Our analysis of sequences of the gene has found a single mutation only on carR of an El Tor PB-sensitive strain. We also confirmed that the mutation is responsible for the PB sensitivity of the recent Indian El Tor strains. Sequence analysis of the carR gene found in classical biotype strain O395 revealed that the new variant El Tor sensitive strains differ from O395 and N16961. Hence, this mutation is considered to be significant and is capable of altering PB susceptibility in endemic El Tor vibrios.
Herrera et al. (17) characterized the CarR (CarRr) protein found in a PB-resistant strain and showed that the CarRr protein is phosphorylated at the D55 residue and becomes activated by CarS, i.e., the D55 residue in CarRr is an important site of phosphorylation for the activation of the protein function. We hypothesize that another D residue at amino acid position 89 (D89) may also be phosphorylated and that phosphorylation strongly activates CarRr to perform its downstream functions in PB-resistant strains. Consequently, the D89N substitution of CarR (CarRs) in PB-sensitive strains may eliminate a phosphorylation site (D89), making it functionally weak.
Our structural domain models of CarR present in El Tor strains have predicted a unique scenario. The model shows that the D89N substitution brings two domains of CarR closer, constricting the interacting cleft, which results in the reduction of DNA binding specificity.
The mRNA expression studies fully support our hypotheses. The genes almE, almF, and almG were found to be downregulated in CarRs protein-carrying strains, and because of this, the CarRs strains are unable to modify lipid A, making them sensitive to PB. It has been reported that the expression of the carRS two-component system is decreased when subjected to high Ca2+ concentrations (19). Our results corroborate this finding. Further, we observed that almEFG expression differed in CarRr and CarRs strains when the extracellular Ca2+ concentrations and pH were changed. This change is supposed to be due to the variation in carR mRNA levels. The downregulation of biofilm formation genes, vpsA and vpsL, at high Ca2+ concentrations and with pH variations suggested that there is no role for the carR SNP and that the differential expression is due to the difference in CarR mRNA levels only (see Fig. S4 and S5).
The D89N substitution has been found to be more stable in nature than other mutations in the carR allele because this particular mutation has pervasive positive/diversifying selection, which possibly has a selective advantage and environmental stability in subsequently emerged V. cholerae strains.
Bilecen et al. (18) reported that CarRr significantly contributes to intestinal colonization in mouse models with a strain-specific approach. In this paper, we have shown the biological significance of the CarR SNP found in PB-sensitive strains by altering their environmental parameters, like concentrations of divalent cation and pH, but in vivo analysis, such as the expression of genes after colonizing the gut in mouse and rabbit models, remains to be done.
MATERIALS AND METHODS
Bacterial strains, plasmids, and culture conditions.The clinical V. cholerae O1 strains used in this study are streptomycin resistant and ampicillin sensitive (Table 2). Cells were grown at 37°C in Luria Bertani (LB) broth and on LB agar for nonselective passages. LB agar containing ampicillin (100 μg ml−1) and/or streptomycin (100 μg ml−1) was also used to culture cells according to the requirement. The suicide vector pCVD442 (Ampr) (31) maintained in Escherichia coli SM10λpir was used in this study to construct vectors and mutant strains. This E. coli strain with the vector constructs was used in the conjugal transfer of genes grown at 37°C on LB agar plates containing ampicillin (100 μg ml−1). Additionally, a minimal medium containing 5% sucrose was used to screen sucrose-resistant mutants. To study PB susceptibility, culture medium supplemented with PB (50 U ml−1) was used. Mueller-Hinton agar (MHA) was used to determine the antimicrobial MICs for V. cholerae strains.
Bacterial strains and plasmids used in this study
Analysis of WGS.Twelve V. cholerae O1 strains isolated from cholera patients in Kolkata were phenotypically confirmed for PB sensitivity (Table 3). Whole-genome sequences (WGS) of these isolates were used for comparison with standard PB-resistant El Tor strain N16961. WGS of these 12 representative strains from the DNA Data Bank of Japan (DDBJ) server (32) and that of N16961 (33) from NCBI (GenBank accession number AE003852.1) were used. A set of 12 putative genes that may be associated with PB susceptibility were compared using Basic Local Alignment Search Tool (BLAST) of the National Center for Biotechnology Information (NCBI) (Table 1). A PB-sensitive V. cholerae strain, IDH07230, was randomly selected for the subsequent experiments.
Characteristics of the strains used in this study
Construction of recombinant plasmids.pCVD442 was used to construct pVA1 and pVA2 (Table 2). Briefly, the pCVD442 vector was digested at its multiple-cloning site with SacI (New England Biolabs [NEB]) according to the manufacturer’s instructions and was treated with calf intestinal alkaline phosphatase (CIP) (NEB) to remove the 5′ phosphate end to prevent self-ligation. The insert VC1320a was made by PCR amplification of the VC1320 (carR) gene from the PB-resistant strain N16961 with primers vprA_SacI_F/vprA_SacI_R (Table 4), covering the entire region of the gene flanked by SacI restriction sites. VC1320b was prepared with the same primers using the PB-sensitive strain IDH07230. These PCR-amplified inserts contain the carR gene (a different allele) flanked by SacI restriction sites. The two inserts, VC1320a and VC1320b, were digested separately with SacI and ligated with the SacI-digested and CIP-treated vector pCVD442 to obtain pVA1 and pVA2, respectively. Insertion of the precise sequences was confirmed by sequencing with primer set Pcv_Ins_F2/Pcv_Ins_R (Table 4). The constructs were then transformed into conjugative E. coli SM10λpir (31).
Primers used in this study
Incorporation of chromosomal point mutation.The donor SM10λpir strains containing the constructs were selected by growing the cells on Luria agar (LA) with ampicillin (100 μg ml−1), as the vector contains an Ampr marker. The recipient V. cholerae strains were grown on LA with streptomycin (100 μg ml−1), as they are innately resistant to the drug. The donor and recipient strains were mated and transferred as described previously (34). The exconjugants (V. cholerae containing the vector construct) were selected on LA plates containing both ampicillin (100 μg ml−1) and streptomycin (100 μg ml−1). A further passage on LA containing 5% sucrose was made to cure the suicide vector (31, 35). Several colonies from the sucrose plate were picked and tested for insertion of the referred point mutation into the genomic DNA by nucleotide sequencing of the specific gene.
Nucleotide sequencing.Primers vprA_F(S) and vprA_R(S) were designed to cover the entire coding region of the target gene, VC_1320 (carR) (Table 4). The desired region was PCR amplified and purified using a PCR purification kit (Fermentas). The purified products were directly sequenced (Applied Biosystems) and analyzed using DNASTAR software. Searches for identical sequences were also made using the BLAST tool available at NCBI.
Polymyxin B MIC assay.Strains were grown in tryptic soy broth (TSB) with an optical density of 0.5 McFarland standard and spread on MHA plates. Etest gradient PB strips (bioMérieux, Durham, NC) were placed on the plates and incubated at 37°C for 20 to 24 h. The breakpoints were interpreted according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guideline for Enterobacteriaceae, as no interpretive standard was available for V. cholerae (36).
Polymyxin B 50-U ml−1 plate assay.The MIC results were validated using LB agar plates with 50 U ml−1 PB. TSB cultures were spread over the plate and observed for growth after incubation. PB-resistant strains showed confluent growth, whereas sensitive strains did not grow. Wild-type N16961 and O395 were selected as positive- and negative-control strains, respectively (37).
Analysis of translated amino acid sequences of carR.The carR nucleotide sequences of strains N16961 and IDH07230 were translated using the ExPASy translation tool (Swiss Institute of Bioinformatics). ExPASy-translated sequence of N16961 was aligned with coding DNA sequence (CDS) found in GenBank (NCBI) to match the consistency of ExPASy translation.
Molecular modeling.The crystal structure of CarR is not available, but to study the effect of amino acid substitution on its structure, a 3D model was essential. To construct a homology model of the CarR protein, the CPHmodels-3.2 server was used; it determines the template structure based on profile-profile alignment guided by secondary-structure and exposure predictions. The predicted template was cross-checked, using PSI-BLAST, against the PDB. The stereochemical parameters of the generated 3D model of the CarR protein were evaluated through Ramachandran plot analysis using the RAMPAGE server. The model of the CarRs protein was built using the mutagenesis tool in PyMOL software.
Molecular dynamics simulation.MDS was performed using GROMACS 4.5 software (38) for the proteins present in both the PB-resistant (CarRr) and sensitive (CarRs) variants. The purpose of doing this exercise was to refine the modeled structures and to check the structural alteration upon D89N mutation. The lowest potential energy (PE) confirmations were extracted from the 10-ns MDS trajectory and further refined by energy minimization. The topology for each protein was generated using a GROMOS96 43A1 force field with simple point charge (SPC) water model. The simulation was performed within a cubic water box; periodic boundary conditions were applied, with a distance of 0.9 nm between the macromolecule and the walls of the periodic box. Water molecules were added to fill the water box, and Na+ ions were added to the system to neutralize the overall charge. To eliminate any residual strain between the protein-ligand complex and the water molecules, steepest descent followed by conjugate gradient energy minimization was performed without any restraints for the whole system; the minimization is converged when the maximum force is smaller than 10 kJ mol−1. A separate 500-ps volume (constant number; volume and temperature) equilibration followed by 1,000-ps pressure (constant number; pressure and temperature) equilibration of the system was performed through position-restrained simulation by restraining bond parameters only to relax the system. During this process, a LINCS algorithm was used to constrain the bond length and bond angles (39), while 3-site SPC water molecules were restrained through the SETTLE algorithm (40). The V-rescale temperature-coupling and the Parrinello-Rahman pressure-coupling constants were set to 0.1 and 2, respectively, to keep the system at a constant temperature of 300 K and a pressure of 1 bar. For long-range electrostatics, the particle mesh Ewald (PME) method was used, while a 10-Å cutoff was applied to truncate the short-range components. During this step, a 14-Å cutoff was applied to Van der Waals interactions. Finally, a 10-ns production MD was performed by releasing the position restraints to analyze the behavior of the system. The RMSD plot and the H bond interaction plot were generated from the trajectory output file using Microsoft Excel. Molecular interactions at different time points were visualized through PyMol software.
Further, to determine the carRs stability in V. cholerae in nature, we performed FEL analysis on Datamonkey (http://www.datamonkey.org) (22) with 12 different carR alleles of V. cholerae, including carRr and carRs alleles retrieved from NCBI GenBank. The FEL method directly estimates nonsynonymous and synonymous substitution rates at each site in a sequence alignment in order to identify sites under positive or negative selection. The selective pressure at the protein level is usually measured by the nonsynonymous/synonymous rate ratio (omega, or dN/dS), with omega values of <1, 1, and >1 indicating purifying (or negative) selection, neutral evolution, and diversifying (or positive) selection, respectively. The omega ratio is commonly calculated as an average over sites (41). The multiple-sequence alignment required to run FEL was performed through the muscle server (https://www.ebi.ac.uk/Tools/msa/muscle/).
To check D89N protein stability, the DynaMut and SDM2 servers were used. DynaMut integrates graph-based signatures, along with normal mode dynamics to generate a consensus prediction of the impact of a mutation on protein stability. Here, we used DynaMut to predict the effect of D89N mutation of the V. cholerae CarR protein. Another server, SDM2, was used to reconfirm our prediction of the mutational effect on protein stability. SDM2 calculates the stability difference score (pseudo-ΔΔG), and the negative and positive values correspond to mutations predicted to be destabilizing and stabilizing, respectively (42).
RNA isolation.LB cultures were grown overnight at 37°C and diluted 1:100 in fresh LB medium to an optical density at 600 nm (OD600) of 0.6. One milliliter of the culture was taken, and cells were harvested. Total RNA was isolated using TRIzol reagent (Life Technologies) (43). The integrity of RNA isolation was checked by electrophoresis using 1% agarose gels. The concentration of total RNA was determined in a NanoDrop spectrophotometer and normalized to a concentration of 400 ng μl−1. DNase I treatment was made to remove the traces of DNA with Ambion DNase I (RNase free; Life Technologies) according to the manufacturer’s instructions. After DNA digestion, total RNA was cleaned using an RNeasy minikit (Qiagen).
qRT-PCR.One microgram of total RNA was used to synthesize total cDNA using a RevertAid first-strand cDNA synthesis kit (Thermo Fisher Scientific). The resulting cDNA was quantified and normalized to a concentration of 1 μg μl−1. Real-time PCR was performed in a Light Cycler 480 instrument II (Roche Diagnostics) with SYBR green master mix (Thermo Scientific), using gene-specific primer sets (Table 4). The results were deduced from experiments performed in triplicate. All the samples were finally normalized to the expression of their housekeeping gene, recA, and relative quantification was carried out by the Pfaffl et al. method (45).
ACKNOWLEDGMENTS
This research was supported in part by the Indian Council of Medical Research (ICMR); the Japan Initiative for Global Research Network on Infectious Diseases (J-GRID) from the Ministry of Education, Culture, Sport, Science and Technology in Japan; the Japan Agency for Medical Research and Development (AMED) (grant number JP18fm0108002); and the National Institute of Infectious Diseases (NIID), Japan. P.S., S.S., and R.N.S. acknowledge Council of Scientific and Industrial Research (CSIR) fellowships [no. 09/482(0060)/2014-EMR-I and 09/482(0071)/2019-EMR-I] and Indian Council of Medical Research (ICMR) fellowship no. 3/1/3/JRF-2015/HRD-LS/88/40189/82, respectively. A.G. was J. C. Bose Chair Professor of the National Academy of Sciences, India.
FOOTNOTES
- Received 11 February 2020.
- Accepted 18 February 2020.
- Accepted manuscript posted online 24 February 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
