All the plasmids and strains used in this study are listed in Table 1. R. anatipestifer CH-1 (GenBank accession number: NC_018609) was isolated from the brains of diseased ducks from Chengdu, Sichuan, China, identified by our laboratory (Wang et al., 2014) and cultured in tryptic soybean broth (TSB, Oxoid) or tryptic soy agar (TSA, Oxoid) medium at 37°C in 5% CO₂. Escherichia coli strains were grown in Luria-Bertani (LB, Oxoid) broth or LB agar (1.5% agar powder, Solarbio) at 37°C. The antimicrobial agents that were added to the media used for strain construction and selection were purchased from Sigma, and they were added to reach the following final concentrations: ampicillin (Amp), 100 μg/ml; cefoxitin (Cfx), 1 μg/ml; chloramphenicol (Cm), 25 μg/ml; kanamycin (Kan), 50 μg/ml; and spectinomycin (Spc), 80 μg/ml.

The DNA fragments were obtained by polymerase chain reaction (PCR). PCR was carried out in a Hybaid PCR thermocycler (Bio-Rad) using the DNA Polymerase (Takara) according to manufacturer’s instructions. All primers used in this study were synthesized by Invitrogen.

To monitor the mRNA expression levels, quantitative real-time (qRT)-PCR was used. Total RNA was extracted from R. anatipestifer CH-1 strains in the logarithmic phase using RNAiso Plus (Takara) according to the manufacturer’s instructions. The RNA was then reverse transcribed into cDNA using the PrimeScript^TM RT Reagent Kit (Takara) with gDNA Eraser (Takara). PCR with SYBR^® Premix Ex Taq^TM II (Takara) was performed using the gene-specific primers P1-16S rRNA-F/P2-16S rRNA-R P3-raeBRT-F/P4-raeBRT-R and P5-B739_0874-F/P6-B739_0874-R. In the qRT-PCR analysis, 16S rRNA gene was used as an internal control. The relative gene expression was calculated using the 2^{-Δ ΔCt} method with Bio-Rad CFX Manager software (Schmittgen and Livak, 2008). Experiments were performed in triplicate.

Recombinant suicide plasmids pRE112::B739_0873USD and pRE112::B739_0872-B739_0873USD were used to delete the B739_0873 and both the B739_0872 and B739_0873 genes, respectively, by allelic exchange, according to previously described methods (Luo et al., 2015). Briefly, the primers P7-raeBup-F/P8-raeBup-R, P13-raeA-raeBup-F/P14-raeA-raeB up-R, and P11-raeBdown-F/P12-raeBdown-R were used to amplify the upstream and downstream homologous arm regions of the B739_0873 and B739_0872-B739_0873 genes in the R. anatipestifer CH-1 genome, respectively. The primers P9-raeBspc-F/P10-raeBspc-R and P15-raeA-raeBspc-F/P10-raeBspc-R were used to amplify Spc resistant (SpcR) cassettes from the plasmid pYES1. Each set of three PCR fragments was integrated using overlap PCR (Xiong et al., 2006) with primer pairs P7-raeBup-F/P12-raeBdown-R and P13-raeA-raeBup-F/P12-raeBdown-R. The fusion segments were then cloned into pRE112 to construct pRE112::B739_0873USD and pRE112::B739_0872-B739_0873USD. Subsequently, the recombinant plasmids were introduced into R. anatipestifer CH-1 by conjugation as described previously (Liao et al., 2015). The transconjugants were selected on TSA plates containing Spc (80 μg/ml) and Cfx (1 μg/ml), and they were then confirmed using PCR with the conserved 16S rRNA gene primers P1-16S rRNA-F/P2-16S rRNA-R and the corresponding identifying primers P16-raeBIdent-F/P17-raeBIdent-R and P18-raeA-raeBIdent-F/P19-raeA-raeBIdent-R. The resultant gene-deletion mutant strains were designated RA-CH-1 ΔB739_0873 and RA-CH-1 ΔB739_0872 ΔB739_0873.

To confirm that the RA-CH-1 ΔB739_0873 and RA-CH-1 ΔB739_0872 ΔB739_0873 phenotypes were due to the B739_0873 gene deletion, the E. coli–R. anatipestifer shuttle plasmid pLMF03 was used to construct the recombinant plasmid pLMF03::B739_0873 that contained an intact B739_0873 gene. Briefly, the primers P20-raeB-F/P21-raeB-R were used to amplify the B739_0873 open reading frame (ORF), which was ligated into pLMF03 at NcoI and XhoI sites to generate pLMF03::B739_0873. For the complementation analysis, the pLMF03::B739_0873 plasmid was introduced into the two R. anatipestifer CH-1 mutant strains by conjugation, as described previously (Wang et al., 2017). The transconjugants were selected on TSA plates containing Kan (50 μg/ml) and Cfx (1 μg/ml), and they were then confirmed using PCR with the conserved 16S rRNA gene primers P1-16S rRNA-F/P2-16S rRNA-R and the Cfx-identifying primers P24-cfx-F/P25-cfx-R. The resultant complemented mutant strains were designated RA-CH-1 ΔB739_0873 pLMF03::B739_0873 and RA-CH-1 ΔB739_0872 ΔB739_0873 pLMF03::B739_0873.

To evaluate the growth rates under non-competitive conditions, we monitored the growth curves for the wild-type strain, the RA-CH-1 ΔB739_0873 mutant, and the RA-CH-1 ΔB739_0873 complemented strain, according to the previously described method (Luo et al., 2015). In vitro competition experiments were performed with the wild-type strain and the RA-CH-1 ΔB739_0873 mutant (Perez et al., 2012). Briefly, both strains were mixed in a 1:1 ratio when they were in the exponential phase. Subsequently, approximately 10^-5 cells from the mixtures were added to 10 ml TSB and grown at 37°C. After 16 h, 10-fold serial dilutions of the cells were spread onto both TSA and TSA containing 80 μg/ml Spc in duplicate and incubated overnight at 37°C. The competition index (CI) was defined as the ratio between the number of mutant and wild-type CFUs. All experiments were performed in triplicate.

A twofold serial dilution assay was used to measure the minimal inhibitory concentration (MIC) of antimicrobial agents for R. anatipestifer CH-1 strains according to the Clinical and Laboratory Standards Institute guidelines (CLSI, 2015). The general method from the CLSI document was used and there was no specific guidance document for R. anatipestifer. The E. coli American Type Culture Collection (ATCC) 25922 strain was used for quality control. Antimicrobial agents were serially diluted twofold in TSB broth with concentrations ranging between 1 and 512 μg/ml. The turbidity of the inoculum was adjusted to 10⁷ CFU/ml and 100 μl was added into every well. The 96-well microtiter plates were incubated at 37°C for 24 h. The lowest concentration that inhibited bacterial growth was considered to be the MIC. All tests were performed in triplicate.

To determine whether the B739_0873 protein was driven by PMF, the PMF inhibitor carbonyl cyanide m-chlorophenylhydrazone (CCCP, Sigma) was used to attempt to inhibit B739_0873 activity. First, a series of CCCP concentrations (1.25, 2.5, 5, and 10 μM) were prepared to determine the optimal concentration for the full survival of the wild-type strain, the RA-CH-1 ΔB739_0873 mutant and the complemented strain RA-CH-1 ΔB739_0873 pLMF03::B739_0873. Subsequently, the MICs of aminoglycosides and detergents for these R. anatipestifer CH-1 strains were determined with or without CCCP.

To further confirm that the B739_0873 protein was the inhibited efflux pump, a gentamicin accumulation assay was performed. Cells of R. anatipestifer CH-1, RA-CH-1 ΔB739_0873 and RA-CH-1 ΔB739_0873 pLMF03::B739_0873 were grown in TSB broth and harvested at the exponential phase of growth, and then washed twice at room temperature with 50 mM phosphate buffer (pH 7.0), containing 1 mM MgSO₄ and 0.4% (wt/vol) glucose. The washed cells were resuspended in the same buffer at a density of 1 mg (dry weight) per ml, including 32 μg/ml gentamicin. To determine the dry weight, cells harvested at the exponential phase of growth by centrifugation for 10 min at 8,000 rpm, removed the supernatant and dried to a constant weight at 60°C. At the 24-min time point, 5 μM CCCP was added to the cells suspensions to assess energy-dependent efflux. The samples were collected every 8 min, centrifuging at 12,000 g and 4°C for 1 min and the samples were immediately diluted into ice-cold phosphate buffer, followed by two washes and sonication on ice. Gentamicin uptake was measured using a Gentamicin ELISA Test Kit following the manufacturer’s instructions (Reagen).

Subsequently, we investigated whether five conserved amino acid residues (Asp 400, Asp 401, Lys 929, Arg 959, and Thr 966) (Guan and Nakae, 2001; Su et al., 2006) in the R. anatipestifer CH-1 B739_0873 protein are vital for it to function. The point mutations were introduced into plasmid pLMF03::B739_0873 using the overlap PCR method (Xiong et al., 2006). The mutant sites were designed in primers. As illustrated in Table 2, primer pairs P26-raeB_D400A-R/P27-raeB_D400A-F, P28-raeB_D401A-R/P29- raeB_D401A-F, P30-raeB_K929E-R/P31-raeB_K929E-F, P32-raeB_R959A-R/P33-raeB_R959A-F, and P34-raeB_T966E-R/P35-raeB_T966E-F, were reverse complementary. The upstream fragments were amplified from the R. anatipestifer CH-1 genome using primer pairs P20-raeB-F/P26-raeB_D400A-R, P20-raeB-F/P28-raeB_D401A, P20-raeB-F/P30-raeB_K929E-R, P20-raeB-F/P32-raeB_R959A-R, and P20-raeB-F/P34-raeB_T966E-R, and the downstream fragments were amplified using primer pairs P27-raeB_D400A-F/P21-raeB-R, P29-raeB_D401A-F/P21-raeB-R, P31-raeB_K929E-F/P21-raeB-R, P33-raeB_R959A-F/P21-raeB-R, and P35-raeB_T966E-F/P21-raeB-R. The two PCR fragments were integrated using overlap PCR with primer pair P20/P21 to generate versions of the B739_0873 gene, each with a single mutation site. The B739_0873 mutant fragments were purified by MiniBEST DNA fragment Purification Kit (Takara) following the manufacturer’s instructions and then cloned into the pLMF03 plasmid to generate mutant recombinant plasmids, pLMF03::B739_0873_D400A, pLMF03::B739_0873_D401A, pLMF03::B739_0873_K929E, pLMF03::B739_0873_R959A, and pLMF03::B739_0873_T966E. They were introduced into RA-CH-1 ΔB739_0873 by conjugation, as described previously (Wang et al., 2017). The transconjugants were selected on TSA plates containing Kan (50 μg/ml) and Cfx (1 μg/ml), and they were then confirmed using PCR with the conserved 16S rRNA gene primers P1-16S rRNA-F/P2-16S rRNA-R and the Cfx-identifying primers P24-cfx-F/P25-cfx-R. The resultant complemented mutant strains were designated RA-CH-1 ΔB739_0873 pD400A, RA-CH-1 ΔB739_0873 pD401A, RA-CH-1 ΔB739_0873 pK929E, RA-CH-1 ΔB739_0873 pR959A, and RA-CH-1 ΔB739_0873 pT966A.

The recombinant plasmid pET32a (+)::B739_0873 was used to express B739_0873 protein. Briefly, the primers P22-Truncated raeB-F/P23-Truncated raeB-R were used to amplify the truncated B739_0873 ORF, which was ligated into pET32a (+) at BamHI and XhoI sites to generate pET32a (+)::B739_0873.

Strain E. coli BL21 (DE3) pET32a (+)::B739_0873 was grown overnight at 37°C in LB broth containing 100 μg/ml Amp. Subsequently, the LB broth containing 100 μg/ml Amp was inoculated with the overnight culture to an optical density at 600 nm (OD₆₀₀) of 0.05 and grown at 37°C. Expression was induced by adding 0.6 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at an OD₆₀₀ of 0.6 for 6 h at 37°C. Cells harvested by centrifugation for 10 min at 8,000 rpm and 4°C were suspended in lysis buffer (20 mM Tris–HCl, pH 8.0) and then sonicated on ice. After centrifugation at 10,000 rpm and 4°C for 10 min, the supernatant of the lysate was loaded onto an equilibrated (with 20 mM Tris–HCl, pH 8.0) nickel-nitrilotriacetic acid (NTA) column (7sea Biotech, China). By washing with lysis buffer (20 mM Tris–HCl, pH 8.0) containing 20 and 50 mM imidazole, non-specific contaminants were removed. Subsequently, the B739_0873 protein was eluted with lysis buffer (20 mM Tris–HCl, pH 8.0) containing 200 mM imidazole. Then the purified protein was placed in the dialysis membrane (Solarbio), soaked in phosphate-buffered saline (PBS) buffer (pH 7.4) and changed the buffer every other 1, 2, 4 h, the final dialyzed overnight, to eliminate any residual imidazole.

The purified B739_0873 His-tagged protein was used to generate a polyclonal antibody. Approximately 2 mg of B739_0873 protein emulsified in complete Freund’s adjuvant (Sigma) was used to immunize four rabbits (the local rabbit industry in Ya’an, Sichuan) via intradermal injections. Subsequently, booster doses of 3 and 4 mg of B739_0873 protein were prepared in incomplete Freund’s adjuvant (Sigma), and the immunization was given after 2 and 3 weeks, respectively, using subcutaneous injections. A week after the last immunization, the antibody was collected from the ear vein of the rabbits and frozen at -80°C.

Strains RA-CH-1 ΔB739_0873 pLMF03, RA-CH-1 ΔB739_0873 pLMF03::B739_0873, RA-CH-1 ΔB739_0873 pD400A, RA-CH-1 ΔB739_0873 pD401A, RA-CH-1 ΔB739_0873 pK929E, RA-CH-1 ΔB739_0873 pR959A, and RA-CH-1 ΔB739_0873 pT966A were grown at 37°C in TSB containing 1 μg/ml Cfx. Bacteria were harvested by centrifugation until they grew to exponential phase, suspended in the PBS buffer and then sonicated on ice. The sonicated cells were suspended in loading buffer and heated for 5 min at 100°C. Proteins were separated by 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) using TGX Stain-Free^TM FastCast^TM Acrylamide Kit (Bio-Rad) and subsequently transferred to a polyvinylidene difluoride (PVDF) membrane (Bio-Rad). Non-specific binding sites were blocked with 5% skim milk (Solarbio) in TBS-Tween 20 (0.05%). The PVDF membranes were probed with B739_0873 rabbit polyclonal antibody (1:200), followed by a 1:3,000 dilution of a goat anti-rabbit IgG alkaline phosphatase-conjugated secondary antibody (Bio-Rad). The binding of antibodies to protein was revealed using a substrate for horseradish peroxidase (HRP)-based chemiluminescence Western blot detection following the manufacturer’s instructions (Takara).

All animals were handled in strict accordance with good animal practices, as defined by the local animal welfare bodies. The protocol for the animal work to be performed at Sichuan Agriculture University was reviewed and approved by the Sichuan Agriculture University ethics committee in September 2014.

Statistical analysis was performed using GraphPad Prism version 7 software for Windows. The significance of the between-group differences was ascertained using Student’s t-test. A value of P < 0.05 was considered significant.

The B739_0873 gene of R. anatipestifer CH-1 consists of an ORF with 3,156 bp, which encodes a 1,051-amino acid protein with a putative molecular mass of 115 kDa. The B739_0873 protein is annotated as a cation/multidrug efflux pump in the National Center for Biotechnology Information (NCBI) database. A protein–protein Basic Local Alignment Search Tool (BLASTP) analysis of the B739_0873 amino acid sequence of different R. anatipestifer strains indicated 99–100% identity. Homologs of B739_0873 are also present in Riemerella columbina, Riemerella columbipharyngis, Flavobacteriaceae bacterium 3519-10, Salinimicrobium terrae, Chryseobacterium treverense, Chryseobacterium solincola, Salegentibacter salegens, and Salegentibacter agarivorans, and identities range from 77 to 83%. However, the function of all these homologs has not previously been characterized.

Further analysis of the B739_0873 protein revealed that it was located in the cytoplasmic membrane and there were 12 trans-membrane domains (TMD) and two extremely large periplasmic loops between helix 1 and 2 and between helix 7 and 8. These results are consistent with the general features of RND efflux pumps (Tseng et al., 1999). The Clustal comparison showed that B739_0873 shares 28% identity with E. coli AcrB and 27% identity with Pseudomonas aeruginosa MexB. Furthermore, the essential residues in E. coli AcrB (Asp 407, Asp 408, Lys 940, Arg 971, and Thr 978) (Su et al., 2006; Takatsuka and Nikaido, 2006) and P. aeruginosa MexB (Asp 407, Asp 408, Lys 939, Arg 969, and Thr 976) (Guan and Nakae, 2001) that are responsible for the proton relay pathway are well conserved in B739_0873 (corresponding to Asp 400, Asp 401, Lys 929, Arg 959, and Thr 966).

Analysis of the upstream regions of the B739_0873 gene showed that the B739_0872 gene encodes a protein belonging to the MFP family (Dinh et al., 1994), the B739_0871 gene encodes a protein belonging to the outer membrane factor (OMF) family (Paulsen et al., 1997) and the B739_0870 gene encodes a tetracycline resistance repressor protein (TetR) family transcriptional regulator (Ramos et al., 2005). These three genes occur together with the B739_0873 gene on the R. anatipestifer CH-1genome (Figure 1). Sequence alignments showed that B739_0870 shares 31% identity with E. coli AcrR, B739_0872 shares 24% identity with E. coli AcrA and 25% identity with P. aeruginosa MexA, and B739_0871 shares 20% identity with E. coli TolC and 23% identity with P. aeruginosa OprM.

Overall, the bioinformatic analysis suggests that B739_0873 is a putative RND efflux pump. Thus, we designated the B739_0870-0871-0872-0873 genes as raeR-raeC-raeA-raeB and the B739_0870-0871-0872-0873 proteins as RaeR-RaeC-RaeA-RaeB described below.

To investigate the role of raeB, a mutated raeB was constructed in R. anatipestifer CH-1. To determine whether the inactivation of raeB affected the growth state of R. anatipestifer CH-1, both the growth curves and the CI values were evaluated. The results showed that RA-CH-1 ΔraeB had the same growth rate and CI value as the wild-type strain (Supplementary Figures S1, S2). To determine whether the inactivation of raeB affected the virulence of R. anatipestifer CH-1, the median lethal dose (LD₅₀) was measured as described previously by Wang et al. (2017). The results showed that there were no significant LD₅₀ changes between the wild-type strain and RA-CH-1 ΔraeB (data not shown).

In a second set of experiments, the susceptibility of the RA-CH-1 ΔraeB mutant and the wild-type strain to a variety of antimicrobial agents with dissimilar structures (including Amp, acriflavine, azithromycin, aztreonam, carbenicillin, cefradine, ceftiofur, cefuroxime, cephalothin, chloromycetin, ciprofloxacin, enrofloxacin, erythromycin, ethidium bromide, florfenicol, gentamicin, Kan, lincomycin, nalidixic acid, rifampicin, sodium dodecyl sulfate (SDS), streptomycin, sulfamethoxazole, tetracycline, trimethoprim, Triton X-100, and vancomycin) was compared. It was shown that deletion of raeB increased susceptibility to all the aminoglycosides tested [streptomycin (eightfold), Kan (eightfold), and gentamicin (16-fold)] and all the detergents tested [Triton X-100 (fourfold) and SDS (16-fold)] (Table 3). There were no MIC changes between the wild-type strain and the mutant strain for any of the tested cephalosporins, chloramphenicols, cationic dyes, quinolones, glycopeptides, lincosamides, macrolides, penicillins, rifampicin, sulfonamides, and tetracyclines (Supplementary Table S1). To exclude the possibility that gene inactivation had a polar effect on the transcription of adjacent genes, RT-PCR was performed to measure the mRNA levels of the downstream gene, B739_0874, and it was shown that there was no significant difference in B739_0874 gene transcription between the wild-type strain and the RA-CH-1 ΔraeB mutant (data not shown). This result revealed that the changes were caused solely by raeB. Complementation of the RA-CH-1 ΔraeB mutant with plasmid pLMF03::raeB restored resistance to aminoglycosides and detergents (Table 3). This result indicated that raeB is involved in aminoglycoside and detergent resistance.

To address whether raeB is regulated by aminoglycosides and detergents, sub-inhibitory concentrations of aminoglycosides and detergents were added to an R. anatipestifer CH-1 growth culture and the transcription of raeB was measured using qRT-PCR. As shown in Figure 2, the level of raeB expression was up-regulated by two- to sevenfold after treatment with Triton X-100, SDS, streptomycin, Kan, or gentamicin.

The final concentration of CCCP in TSB broth was 5 μM, which did not affect the growth of the wild-type strain, the RA-CH-1 ΔB739_0873 mutant and the complemented strain RA-CH-1 ΔB739_0873 pLMF03::B739_0873. The results showed that the addition of CCCP (5 μM) decreased the MICs of streptomycin (eightfold), gentamicin (16-fold), and Kan (eightfold) for the wild-type strain and RA-CH-1 ΔraeB pLMF03::raeB. Similarly, the MICs of SDS and Triton X-100 for the wild-type strain and RA-CH-1 ΔraeB pLMF03::raeB decreased by 4- and 16-fold, respectively, in the presence of CCCP. In contrast, CCCP addition did not modify the MICs of streptomycin, gentamicin, Kan, SDS, and Triton X-100 for RA-CH-1 ΔraeB. The MICs of streptomycin, gentamicin, Kan, SDS, and Triton X-100 for the wild-type strain and RA-CH-1 ΔraeB pLMF03::raeB in the presence of CCCP were consistent with those evidenced for RA-CH-1 ΔraeB in the absence or presence of CCCP. These data indicated that an aminoglycoside and detergent efflux pump exists in R. anatipestifer CH-1.

As shown in Figure 3, at the 24-min incubation time point, the accumulation level of gentamicin achieved a steady-state for all strains. The gentamicin accumulation level in the mutant strain RA-CH-1 ΔraeB was about six times higher than that in the wild-type strain and RA-CH-1 ΔraeB pLMF03::raeB. After CCCP was added to the cells containing gentamicin at 24-min, the accumulation of gentamicin increased in the wild-type strain and RA-CH-1 ΔraeB pLMF03::raeB; these accumulation levels were lower than that in RA-CH-1 raeB. In contrast, under our conditions, CCCP had no significant effect on the level of gentamicin accumulation in the mutant strain RA-CH-1 ΔraeB. In addition, the wild-type strain, RA-CH-1 ΔraeB, and RA-CH-1 ΔraeB pLMF03::raeB with no CCCP added at 24-min served as controls. These data indicated that RaeB pumped out gentamicin in an energy-dependent process, presumably coupled to the PMF.

We investigated whether amino acid residues Asp 400, Asp 401, Lys 929, Arg 959, and Thr 966 are vital for the function of the RaeB protein of R. anatipestifer CH-1, five mutant recombinant plasmids, pLMF03::raeB_D400A, pLMF03::raeB_D401A, pLMF03::raeB_K929E, pLMF03::raeB_R959A, and pLMF03::raeB_T966E, were constructed. These plasmids were introduced into RA-CH-1 ΔB739_0873, and the MICs for RA-CH-1 ΔraeB pD400A, RA-CH-1 ΔraeB pD401A, RA-CH-1 ΔraeB pK929E, RA-CH-1 ΔraeB pR959A, and RA-CH-1 ΔraeB pT966A were examined. It was shown that none of these mutant strains exhibited restored resistance to aminoglycosides and detergents (Table 3). To exclude the possibility that these results occurred because these mutant raeB genes were not expressed, we detected the expression of these mutant proteins by immunoblotting with antibody against the RaeB protein. The immunoblot analysis showed that all these RaeB mutant proteins expressed in RA-CH-1 ΔraeB (Figure 4A) were detectable and indistinguishable from the RaeB protein expressed in RA-CH-1 ΔraeB (Figure 4B).

We confirmed that raeB is co-transcribed with raeA (Supplementary Figure S3) as previously reported by Liu et al. (2017). To understand more about raeB, the RA-CH-1 ΔraeA ΔraeB mutant was constructed. Antibiotic susceptibility testing showed that this mutant displayed MICs that were decreased by eightfold for streptomycin, 16-fold for gentamicin, eightfold for Kan, 16-fold for SDS, and fourfold for Triton X-100 (Table 3). However, complementation of the RA-CH-1 ΔraeA ΔraeB mutant with plasmid pLMF03::raeB did not restore resistance to any of these compounds (Table 3). These results indicate that both raeA and raeB are required for aminoglycoside and detergent resistance.