ABSTRACT
Rhodococcus equi is a facultative intracellular, Gram-positive, soilborne actinomycete which can cause severe pyogranulomatous pneumonia with abscessation in young horses (foals) and in immunocompromised people, such as persons with AIDS. All strains of R. equi isolated from foals and approximately a third isolated from humans contain a large, ∼81-kb plasmid which is essential for the intramacrophage growth of the organism and for virulence in foals and murine in vivo model systems. We found that the entire virulence plasmid could be transferred from plasmid-containing strains of R. equi (donor) to plasmid-free R. equi strains (recipient) at a high frequency and that plasmid transmission reestablished the capacity for intracellular growth in macrophages. Plasmid transfer required living cells and cell-to-cell contact and was unaffected by the presence of DNase, factors pointing to conjugation as the major means of genetic transfer. Deletion of a putative relaxase-encoding gene, traA, located in the proposed conjugative region of the plasmid, abolished plasmid transfer. Reversion of the traA mutation restored plasmid transmissibility. Finally, plasmid transmission to other Rhodococcus species and some additional related organisms was demonstrated. This is the first study showing a virulence plasmid transfer in R. equi, and it establishes a mechanism by which the virulence plasmid can move among bacteria in the soil.
INTRODUCTION
Rhodococcus equi is a Gram-positive, saprophytic actinomycete that if inhaled or ingested by susceptible individuals may become an intracellular pathogen capable of survival and replication inside host macrophages (23, 36). Typically, the bacterium causes severe pyogranulomatous pneumonia in foals (young horses) aged 2 to 4 months, frequently accompanied by the development of extensive pulmonary and extrapulmonary abscesses. The organism also causes disease in pigs, presenting most commonly as chronic submaxillary lymphadenitis with tuberculosis-like lesions (43). Lately, reports of R. equi pneumonia in HIV-infected or otherwise immunocompromised people have become common (57).
All strains of R. equi isolated from foals and many recovered from humans carry a large ∼81-kb plasmid (52, 53) encoding 73 open reading frames (ORFs), termed pVAPA1037 (31). This plasmid is genetically distinct from that carried by pig R. equi isolates (pVAPB1593) and some others derived from R. equi-infected people. The main region of difference among these plasmids is in the pathogenicity island (PAI) region. Outside of the PAI, the plasmid backbones are virtually identical (31). The plasmid of foal isolates is essential for disease development in the equine host and in murine model systems of R. equi infection (18, 26, 56). Furthermore, plasmid curing results in loss of the bacterium's capacity to replicate in macrophages (18, 24). On the basis of sequence homology searches and location, the genes on the virulence plasmid are divided into four groups, namely, the PAI region genes, the plasmid replication and partitioning-related genes, the genes of unknown function, and the conjugation-related genes. To date, only genes located within the PAI region have been evaluated to any extent, with three reported to play key roles in virulence (10, 26). Specifically, PAI-encoded and surface-localized, virulence-associated protein A (vapA) and its positive regulators (virR and ORF8) are crucial for resistance to innate macrophage defenses and for bacterial multiplication in vivo (26, 44, 47). An R. equi vapA deletion mutant is incapable of intracellular replication and unable to establish a persistent infection in severe combined immunodeficient (SCID) mice (26). Given that virulence genes are often located on mobile genetic elements and that bioinformatic analysis identifies orthologs of conjugation genes on the R. equi virulence plasmid (31, 51), we reasoned that this plasmid was likely conjugative.
Conjugation is one of the three primary methods of DNA transfer and is a means by which bacteria spread antibiotic resistance genes, virulence factors, as well as other genes promoting survival advantages in specific environments. Bacterial cell-to-cell contact is required for conjugation, the molecular details of which are very well studied in Gram-negative bacteria (30). For these, relaxase enzyme-mediated plasmid transfer has been described, wherein a relaxase-bound single-stranded DNA intermediate is transferred unidirectionally from donor to recipient bacteria (37). Relaxase-mediated plasmid transfer appears to be conserved in some Gram-positive bacteria as well (28, 29). In the present work, we examined the self-transmissibility of the R. equi virulence plasmid (pVAPA1037) and its potential for transfer to other genetically related bacteria. We also assessed the requirement for virulence plasmid-encoded TraA, a putative relaxase, in plasmid transfer.
MATERIALS AND METHODS
Bacterial strains.The strains and plasmids used in this study are listed in Table 1. R. equi 103+ was originally isolated from a foal with R. equi pneumonia (18). R. equi strains ATCC 33701 and ATCC 33705 were obtained from the American Type Culture Collection (ATCC; Manassas, VA) and were originally isolated from a foal and a pig with R. equi disease, respectively. Plasmid-cured derivatives of R. equi strains 103+, ATCC 33701, and ATCC 33705 were created by serial subculture at 37°C until the virulence plasmid was lost to yield strains 103− (7), ATCC 33701−, and ATCC 33705−, respectively. Loss of the virulence plasmid was confirmed by PCR analysis demonstrating an absence of amplification of several different regions of the virulence plasmid (data not shown). Subsequently, these virulence plasmid-free derivatives were chromosomally marked with a dihydrofolate reductase (dhfr) gene conferring trimethoprim resistance (designated Trimr or T). The vapA deletion mutant (ΔvapA) on the 103+ strain background was derived as described earlier (26), wherein the vapA gene was replaced by a gene [aac(3)-IV] bestowing apramycin resistance (designated Aprr or A). The complemented vapA deletion mutant strain (ΔvapA/vapA) in which a wild-type copy of the vapA gene was integrated on the chromosome of the ΔvapA mutant strain was constructed as previously reported (25) and is both Aprr and hygromycin resistant (designated Hygr or H). The hygromycin-marked deletion mutant of vapG (ΔvapG) on the 103+ background was created as described earlier (10). Rhodococcus sp. strains ATCC 4277, ATCC 15610, and ATCC 21505 were a kind gift from Ellen Neidle, University of Georgia. These organisms are listed in the ATCC catalogue as Rhodococcus erythropolis, Rhodococcus rhodochrous, and Nocardia globerula, respectively, but subsequent 16S RNA sequence analysis of an ∼500-bp amplicon of the 16S RNA gene of each (obtained with primers 16S-F and 16S-R; Table 2) by Accugenix Inc. (Newark, DE) was able to confirm their identities to the Rhodococcus genus level only. A phylogenetic tree constructed using the Accugenix Inc. sequencing data shows the relatedness of these organisms and demonstrates that they are of a species distinct from R. equi (see Fig. S1 in the supplemental material). Thus, in this paper, we identify these organisms as Rhodococcus sp. followed by their ATCC number. For the studies described herein, the chromosomes of these organisms were marked with the integrating plasmid (pSET152), conferring Aprr. Mycobacterium smegmatis strains MKD8 and Jucho J4 were generously provided by Keith Derbyshire, Wadsworth Center (New York State Department of Health). Dietzia sp. strain ATCC 184 was purchased from ATCC, and though catalogued as Rhodococcus rhodochrous, it was later identified as a Dietzia sp. following 16S RNA sequence analysis by Accugenix Inc.
Bacterial strains and plasmids used in this study
Primers used in this study
Media and growth conditions.Rhodococcus spp. were grown in brain heart infusion (BHI) broth or agar (Difco Laboratories, Sparks, MD) or Luria-Bertani (LB) broth or agar (Difco Laboratories) at 30°C or 37°C and supplemented with appropriate antibiotics as necessary. In certain experiments, bacteria were grown on minimal M9 medium (Difco Laboratories) supplemented with thiamine (0.1 mM) and with sodium acetate (25 mM) as the carbon source. This medium is referred to as minimal medium (MM). Apramycin was used at a final concentration of 80 μg/ml, hygromycin was used at a final concentration of 180 μg/ml, streptomycin was used at a final concentration of 50 μg/ml, and trimethoprim was used at a final concentration of 50 μg/ml. M. smegmatis cells were grown in LB medium (broth or agar).
Sequence analysis.An NCBI alignment tool, COBALT (40), was employed for multiple-sequence alignment of various TraA sequences. To identify similar sequences allowing subsequent construction of a phylogenetic tree, the predicted protein sequence of R. equi TraA was used to query the NCBI database with BLASTp (4), and from the results, several sequences of a variety of bacteria were selected for further comparison. Multiple-protein-sequence alignments were generated using the ClustalW program (8), a component of Geneious Pro software (version 5.6; Biomatters Ltd., New Zealand). A nonrooted neighbor-joining tree was created using the Jukes-Cantor model with resampling via the bootstrap method (10,000 replicates). Escherichia coli TraI served as the outgroup. The GenBank accession numbers for the proteins used are as indicated for the following species: Arthrobacter aurescens, AAS20144; Amycolicicoccus subflavus, YP_004495502; Corynebacterium aurimucosum, ZP_06042366; Corynebacterium pseudogenitalium, ZP_07715455; Dietzia cinnamea, ZP_08025317; E. coli F_TraI, AAC44186; E. coli R388_TrwC, CAA44853; Gordonia westfalica, NP_954808; Rhodococcus equi TraA, NP_066783; Rhodococcus erythropolis pREA400_TraA, ABF48485; Rhodococcus erythropolis pREC_orf16, YP_345077; Rhodococcus pyridinivorans, ZP_09310308; and Streptococcus agalactiae, CAA33713.
Plasmid construction.To mark the chromosome of recipient cells, a new version (pVM6) of the previously described integrating vector pSET152 (25) was used, wherein the apramycin resistance gene was replaced by the dhfr gene of mouse origin (GenBank accession number AA607882). To do this, mouse dhfr gene fragment 3 (F3) was PCR amplified with primers DHFR3-F and DHFR3-R (Table 2) using pUAB400 (49) as the template, and the product was digested with BamHI and PstI and ligated to similarly digested pUAB300 (49). pUAB300 already contains dhfr fragments 1 and 2, and ligation of F3 produced the full-length dhfr gene and yielded plasmid pCVM3. pCVM3 was then used as the template to PCR amplify the full-length dhfr gene, with primers DHFR(Full)-F and DHFR(Full)-R (Table 2). The amplicon was subsequently digested with XbaI and BglII and cloned in pSET152, which had previously been digested with same restriction enzymes, to create pVM3. pVM3 was digested with PshA1 to remove the apramycin resistance gene cassette, and the resultant large fragment was purified and self-ligated to produce pVM6.
Construction of the suicide plasmid to make the traA (bp 15092 to 19360) deletion mutant was begun by amplifying a 700-bp DNA fragment upstream of the traA gene of R. equi strain 103+ (bp 14374 to 15292), using the primer pair traA L-F and traA L-R (Table 2). The amplicon was digested with EcoRI and XbaI and ligated to similarly digested pUC19 to generate pUC-L. Next, an 800-bp amplicon corresponding to the virulence plasmid sequence downstream of the traA gene (bp 19338 to 20049) was generated using R. equi 103+ total DNA as the template and primers traA R-F and traA R-R (Table 2). The fragment was digested with XbaI and PstI and ligated to similarly digested pUC-L to create pUC-LR. Subsequently, a DNA fragment encoding the apramycin resistance gene was amplified from pSET152 using primer pair Apr F Xba and Apr R Xba, digested with XbaI, and then ligated between the cloned flanks at the XbaI site of pUC-LR, yielding pVM16. Finally, a fragment containing the apramycin resistance gene cassette flanked by upstream and downstream traA sequences was generated from pVM16 through EcoRI and PstI digestion, made blunt ended by treatment with the Klenow fragment, and then ligated to SmaI-digested pSelAct-Hyg to generate the traA deletion suicide vector pVM17. pSelAct-Hyg was prepared from parental plasmid pSelAct (54), which was first digested with BlpI and BsiWI to remove the fragment (941 bp) containing the apramycin resistance gene cassette, and then the larger fragment (∼4.8 kb) was purified and treated with the Klenow fragment to make it blunt ended. Next, the hygromycin resistance gene was removed from pMV261 with CviAII and made blunt ended by treatment with the Klenow fragment and ligated to the 4.8-kb fragment of pSelAct to generate pSelAct-Hyg. pSelAct-Hyg has a hygromycin resistance gene and carries the codA-upp selectable markers; the latter allow counterselection against single-crossover intermediates when plating on medium supplemented with 5-fluorocytosine (5FC) (54).
To begin construction of the plasmid for complementation of the traA mutation, the traA gene was PCR amplified with primers traA-F and traA-R, digested with EcoRV and PciI, and ligated to similarly digested pSET152-HMH downstream of a constitutive Mycobacteria sp. hsp60 promoter to make pVM12, an integrating plasmid with traA. The Hsp60 promoter was then replaced by a 500-bp fragment upstream of the traA gene presumed to contain the native traA promoter. For this, the 500-bp fragment was PCR amplified with primer pair traAnative-F and traAnative-R from the R. equi wild-type plasmid template and digested with XbaI and KpnI. The digested fragment was then ligated to similarly digested pVM12 to give pVM26. Finally, pVM26 was digested with MscI to remove the ϕC31 integrase gene and was self-ligated to produce the suicide plasmid pVM38 containing the full-length traA gene with the native upstream sequence.
Construction of the traA deletion mutant and complementing strain.The suicide vector pVM17, containing flanking regions of the traA ORF cloned on either side of an apramycin resistance gene cassette and carrying the codA-upp counterselectable marker genes on its backbone, was used to create the traA deletion mutant of R. equi lacking pVAPA_080 (15092 to 19360 bp; previously identified as ORF30) (51). For this, the method described by van der Geize et al. (54) employing codA-upp as a counterselection was used with little modification. In brief, pVM17 was electroporated into R. equi 103+ and transformants were selected on BHI plates supplemented with apramycin and hygromycin. Transformants were screened by PCR for integration of the full pVM17 plasmid into the virulence plasmid at either of the flanks of the traA ORF (data not shown). These transformants were confirmed to be sensitive to 5FC due to the presence of the codA-upp genes and were treated as single-crossover intermediates. Two such colonies were serially subcultured in BHI containing apramycin at 30°C to facilitate a second recombination event between the wild-type traA and its mutated version. Recombination resulted in looping out of the intervening vector sequence, including the codA-upp genes, thus removing the susceptibility of the cells to 5FC. To recover the mutants so generated, dilutions of the subculture were plated on MM agar supplemented with apramycin and 5FC (100 μg/ml). Subsequently, the loss of hygromycin resistance in the Aprr 5FCr colonies was confirmed by streaking the colonies on BHI agar containing hygromycin. Deletion of the traA ORF was confirmed by PCR analysis using various primer pairs. To generate the traA complementing strain, the traA mutant was transformed with pVM38 and single crossovers were selected on BHI agar with Apr and Hyg and confirmed by PCR analysis.
Electroporation of R. equi.For preparing electrocompetent cells, R. equi strains were grown in 200 ml BHI broth to an A600 of ∼0.8. Cells were collected by centrifugation and washed twice with an equal volume of cold distilled water. Cells were then resuspended in a cold 10% glycerol in distilled H2O at a 1:20 dilution of the initial culture volume. Aliquots (400 μl) of the cells were made and stored at −80°C for further use. For electroporation, cells were mixed with ∼0.5 μg of plasmid DNA and placed in a prechilled 0.2-cm electroporation cuvette (Bio-Rad). Electroporation was performed using a Gene Pulser apparatus (Bio-Rad) set at 2.5 kV, 25 μF, and 1,000 Ω. One milliliter of BHI broth supplemented with 0.5 M sucrose was added to the cuvette immediately after electroporation for better recovery of transformants. Bacteria were then incubated for 1 h at 30°C and subsequently plated on BHI agar supplemented with appropriate antibiotics.
Bacterial mating procedures.In each mating experiment, equal numbers of differentially marked donor and recipient strains were mixed. Depending on the experiment, the donor was either a virulence plasmid-containing vapA deletion strain (ΔvapA/A), a vapA-complemented version of the latter (ΔvapA/vapA/AH), or a hygromycin-marked vapG deletion mutant (ΔvapG/H); each one was derived from wild-type strain 103+. The predominant recipient was a dhfr-marked, virulence plasmid-cured derivative of wild-type strain 103+. Other recipients used included dhfr-marked, virulence plasmid-containing wild-type strain 103+; dhfr-marked, virulence plasmid-free versions of wild-type R. equi strains ATCC 33701 and ATCC 33705; acc(3)IV-marked Dietzia sp. ATCC 184; acc(3)IV-marked Rhodococcus sp. strains ATCC 21505, ATCC 15610, and ATCC 4277; and streptomycin (Sm)-marked M. smegmatis strains MKD8 and Jucho J4. Broth cultures of the strains to be mated were grown overnight at 37°C with appropriate antibiotics and were adjusted to an optical density at 600 nm (OD600) of 1.0 on the following day. Approximately 107 CFU of both donor and recipient cells was mixed in a small volume (∼5 μl), and the mixture was spotted on BHI agar for 72 h at 30°C. After such time, the cell mixtures were scraped and resuspended in 2 ml of phosphate-buffered saline (PBS), and serial dilutions (up to 10−7) were plated on selective agar medium containing specific antibiotics to select for the presence of transconjugants, evidenced by dual antibiotic resistance (see Fig. S2 in the supplemental material). Dilutions were similarly plated on single-antibiotic-containing plates to determine the available numbers of donor or recipient cells. Conjugation frequencies were calculated by dividing the number of transconjugants by the total number of recipient cells. PCR analysis was used to confirm the presence of both antibiotic resistance gene cassettes in the transconjugants. Transfer of the entire plasmid was verified by PCR amplification of several regions of the virulence plasmid. In parallel with the solid medium mating experiments, equal numbers of donor and recipient cells were inoculated in 2 ml BHI broth and kept for 72 h at 30°C. Subsequently, the cells were pelleted and resuspended in 2 ml PBS, dilutions were made, and plating was done as described above to calculate the frequency of plasmid transfer in liquid. In addition, in some experiments, bacterial matings were performed in the presence or absence of 10 μg/ml of DNase (Qiagen, MD) on solid medium. Conjugation frequencies were calculated and compared with those obtained under DNase-free conditions. To evaluate cell viability as a condition of plasmid transfer, donor and recipient strains were heat killed by exposure to 100°C (on a heating block) for 45 min, with vortexing every 5 min after adjusting the OD600 to 1.0. Killing was confirmed by plating for viability. Dead donor and dead recipient cells were mated with live recipient and live donor cells, respectively. As a control, live donor and live recipient cells were mated in parallel, as per usual. Matings were kept at 30°C for 72 h and resuspended in PBS, and dilutions were made to calculate the conjugation frequency as described above.
Intracellular growth of R. equi in bone marrow-derived macrophages.Bone marrow-derived macrophages were isolated from the femurs and tibias of adult female BALB/c mice as described by Coulson et al. (10). These macrophages were seeded in 24-well tissue culture-treated plates at 2 × 105 cells/well and kept overnight at 37°C with 5% CO2. On the next day, monolayers were washed once with warm Dulbecco modified Eagle medium (DMEM), and fresh complete medium (DMEM with 10% fetal calf serum, 10% colony-stimulating factor 1, and 2 mM glutamine) was added. R. equi strains grown overnight were washed with PBS and diluted in the same buffer to an A600 of 1.0 (∼2 × 108 cells/ml). Bacteria were added to the wells at a multiplicity of infection (MOI) of 10:1 (bacteria/macrophage) and incubated for 1 h at 37°C. Monolayers were then washed four times with warm DMEM to remove the unbound bacteria. After washing, DMEM was replaced with complete DMEM containing 20 mg/ml of amikacin sulfate (to prevent extracellular bacterial growth). At 1, 24, and 48 h postinfection, macrophage monolayers were repeatedly washed with warm DMEM and then lysed by adding 500 μl of sterile distilled water and incubating at 37°C for 20 min with subsequent vigorous pipetting. Serial dilutions of the lysate were plated on BHI agar and incubated at 37°C for 48 h, after which colonies were counted. In these experiments, each infection was carried out in triplicate wells.
Statistical analysis.Statistical analyses were performed using the SigmaPlot program (version 11.0; Systat Software Inc., San Jose, CA). For comparing the mean transfer frequency of two matings, a t test was performed. When more than two samples were analyzed, a one-way analysis of variance (ANOVA) was used. Significance was set at a P value of <0.05.
RESULTS
The R. equi virulence plasmid carries genes with putative conjugative function.Bioinformatic analyses of the virulence plasmid of R. equi strain 103+ identified a region extending from ORFs pVAPA_0010 to pVAPA_0120 (previously referred to as ORFs 25 to 34), (51) that appeared to contain genes (putative helicase [Orf pVAPA_0030], traA, topA, traG) (Fig. 1A) related to those involved with conjugative functions in other organisms (31). For example, DNA helicases are required for unwinding of double-stranded DNA and are known to be encoded on conjugative plasmids (20, 30). In addition, relaxase enzymes with strong homology to the predicted amino acid sequence of R. equi TraA are known to bind to the oriT region of conjugative plasmids and initiate the nicking and unwinding of the plasmid to be transferred (21). Further analysis of the R. equi TraA revealed features typical of the MOBF family of relaxases (16), specifically, the presence of both an N-terminal relaxase domain with the conserved aspartate and histidine residues (D131, H199, H210, and H212 of R. equi TraA) and a second C-terminal helicase domain (Fig. 1B). Another identifying characteristic of this MOBF class of relaxases is the existence of two catalytically active tyrosines in the catalytic center (16). When the R. equi TraA N-terminal sequence (1 to 100 amino acids) was searched, we found four tyrosine residues (Y54, Y71, Y75, and Y78) (not shown), yet it remains to be established whether any of these tyrosines are needed for the catalytic function of the R. equi TraA.
(A) Schematic representation of the conjugation region of the R. equi virulence plasmid (pVAPA1037). ORFs pVAPA_0010 to pVAPA_0120 are shown. The names of the ORFs with homology to known genes involved in conjugation are displayed above the respective ORFs. Other ORFs in this region with no known homology are shaded in black. (B) Multiple-protein-sequence alignment (COBALT, NCBI) between the N-terminal amino acids (92 to 313) of R. equi TraA and other closely related and well-studied TraA proteins of the MOBF family. The active-site residues of the E. coli F plasmid relaxase domain (D81, H146, H157, H159) and the aligning residue in other relaxases are indicated by the use of bold font. All aligned residues are shaded gray. Single amino acid gaps in alignment are designated with dashes. The numbers to the left and right of each line correspond to the residue numbers of the first and last amino acids of the particular sequence in that line, respectively. R. eryth., R. erythropolis. (C) A neighbor-joining nonrooted phylogenetic tree of relaxase proteins of various bacterial species was constructed using a ClustalW alignment. Bootstrap values are indicated at the nodes (10,000 replicates). E. coli F plasmid TraI relaxase serves as the outgroup.
To investigate the phylogenetic relationship of R. equi TraA to other closely related relaxases, a nonrooted neighbor-joining tree was created (Fig. 1C) using the Jukes-Cantor model. Results showed that R. equi TraA was more closely related to R. erythropolis pREC16 TraA than R. erythropolis pREA400 TraA, the latter being the only other Rhodococcus sp. relaxase studied in detail thus far. R. equi TraA forms a separate cluster with Arthrobacter aurescens TraA, Gordonia westfalica TraA, and pREC16 TraA. In contrast, pREA400 TraA of R. erythropolis is grouped separately with Dietzia cinnamea and Corynebacterium sp. TraA. Both of these clusters are distantly related to the E. coli F plasmid-encoded relaxase TraI (Fig. 1C).
ORF pVAPA_0100 displayed homology with a DNA topoisomerase-like protein (topA). This finding is significant because the conjugative plasmid RP4 similarly contains a gene (traE) encoding topoisomerase activity (32), and thus, ORF pVAPA_0100 may encode a functional homolog of traE. Homologs of the ORF pVAPA_0120 (traG of IncP plasmids or traD of IncF plasmids) gene encode coupling proteins, essential constituents of all conjugative systems. These proteins are required to connect two indispensable conjugation components (the relaxosome complex and the secretion machinery) (34).
Genetic exchange occurs between strains of R. equi.The aforementioned genetic sequence similarities led to the hypothesis that the R. equi virulence plasmid was conjugative. In order to establish whether virulence plasmid transfer was possible, it was necessary to make use of existing strains or construct bacterial strains which could serve as a donor or a recipient to facilitate tracking of the virulence plasmid from one to the other. One of the donors that we used was the apramycin-resistant R. equi vapA deletion mutant (ΔvapA/A) (26) in which the virulence plasmid PAI-encoded vapA gene was replaced by a gene [aac (3)IV] conferring resistance to apramycin. In most experiments, we used the virulence plasmid-cured derivative of R. equi strain 103+ (7) as the recipient, which was subsequently marked by insertion on the chromosome of the dhfr gene, imparting resistance to trimethoprim (103−/T). To determine if the virulence plasmid of R. equi was transferrable, the virulence plasmid-carrying donor strain (ΔvapA/A) and plasmid-cured derivative recipient (103−/T), each marked with a unique antibiotic resistance gene cassette, were mixed together on antibiotic-free solid medium for 72 h and then selection for resistance to both apramycin and trimethoprim was performed (see Fig. S2 in the supplemental material). Numerous bacterial colonies resistant to both antibiotics arose, a finding consistent with virulence plasmid transfer from the donor to the recipient strain. It is important to note that in control experiments using only the donor or the recipient strain, no colonies appeared on solid medium containing both antibiotics, excluding the possibility of dual antibiotic resistance due to spontaneous mutation. PCR analysis confirmed the presence of both the apramycin and trimethoprim resistance genes in these dually antibiotic-resistant colonies, whereas the donor and recipient strains possessed one or the other (see Fig. S3 in the supplemental material). Calculations of transfer frequency demonstrated that the antibiotic markers were transferred at a high frequency, ranging from 2.80 × 10−2 to 3.98 × 10−1 (mean ± standard deviation [SD], 2.68 × 10−1 ± 2.08 × 10−1) (Table 3, experiment A).
Conjugation frequencies of various mating experiments
Since we initially used the vapA deletion strain ΔvapA/A as the donor, we thought it important to determine whether the absence of vapA in this strain affected plasmid transfer. We thus compared the transfer frequencies obtained using the ΔvapA/A strain or the same strain carrying an integrated wild-type copy of vapA (pSET152-vapA) on the chromosome (ΔvapA/vapA/AH) as donors in parallel matings using 103−/T as the recipient. The transfer frequencies were calculated and found to be statistically equivalent (P = 0.912; Table 3, experiment B). Similar findings were obtained when the ΔvapG/H strain was used as the donor (data not shown). Therefore, it was determined that the use of any of these strains as donors should effectively represent the transfer capacity of a wild-type R. equi strain. In subsequent experiments, unless otherwise noted, the ΔvapA/A strain was used as the donor.
To confirm that transfer was not limited merely to the region of the virulence plasmid where the apramycin resistance gene cassette [aac(3)-IV] resided and that the entire virulence plasmid was transferred, additional PCR analyses were performed using 5 different sets of primers (namely, REVP1 and REVP1c, REVP2 and REVP2c, REVP6 and REVP6c, RETrb1 and REVP1c, and VapH F and VapH R; Table 2) that annealed to the virulence plasmid at various locations. In each case, an amplicon of the expected size was obtained from the donor and putative transconjugant strains (see Fig. S3 in the supplemental material), findings indicative of complete plasmid transfer.
Optimal conditions of plasmid transfer.Having established that complete virulence plasmid transfer occurs in R. equi, experiments were done to optimize the environmental parameters (medium, temperature, and time) favoring plasmid transfer. Thus, the donor and recipient cells were mixed together on both complete medium (BHI) and minimal medium (MM) and the plates were incubated at 37°C or 30°C for 72 h, after which the end transfer frequencies were calculated. Results presented in Fig. 2A show that transfer was approximately 100-fold more efficient at the lower temperature (30°C versus 37°C). Medium composition had little effect on transfer frequency (Fig. 2A).
Effects of environmental parameters on plasmid transfer frequency. (A) Effects of temperature (30°C and 37°C) and medium composition (BHI medium versus MM). R. equi strains ΔvapA/A (donor) and 103−/T (recipient) were mixed on BHI medium or MM in duplicate and kept at 30°C or 37°C for 72 h. Subsequently, serial dilutions were plated on Apr-Trim agar selection medium to calculate the transfer frequency. (B) Plasmid transfer frequency increases with time of donor and recipient contact. R. equi strains (donor and recipient) were mixed for various times (6, 24, 48, and 72 h) before plating on selective medium. Virulence plasmid transfer frequencies are indicated for various times of cell-to-cell contact. Error bars indicate standard deviations from the means of three independent experiments.
The effect of time on plasmid transfer frequency was evaluated by varying the amount of time (6, 24, 48, and 72 h) that the donor and recipient strains were incubated together. At the end of each time period, plasmid transfer frequencies were calculated as described in Materials and Methods. As shown in Fig. 2B, dually antibiotic-resistant colonies could be detected in as little as 6 h, with the number increasing in a time-dependent manner; i.e., the transfer frequency increased with an increase in the incubation time. PCR analysis was done on dually antibiotic-resistant clones that arose at each time point postincubation, and results confirmed transfer of the full plasmid (data not shown).
R. equi virulence plasmid transfer occurs via conjugation.As genetic material transfer may occur by various mechanisms (e.g., conjugation, transformation, transduction, cell-to-cell fusion), a series of experiments was performed to differentiate between transmission of the virulence plasmid via conjugation and transmission by the other possible means. In transformation, naked DNA transfer occurs without the need for living donor cells. We postulated that if plasmid transfer was mediated via transformation, it would be unaffected by the death of the donor cells. Thus, the ΔvapA/A donor strain was heat killed prior to incubation with the recipient strain. We found that killing the donor cells abolished plasmid transfer altogether (Table 3, experiment C), a finding consistent with conjugation as a means of transmission. To further exclude the likelihood of transformation as the mode of plasmid transfer, DNase (10 μg/ml) was added to a mixture of live donor and recipient cells in order to degrade any naked DNA that might be present. The addition of DNase did not affect the plasmid transfer frequency (P = 0.538; Table 3, experiment D), a result suggesting the absence of exposed DNA and supportive of conjugal plasmid transfer. In other experiments, the incubation of the donor and recipient cells was done in liquid medium, and the plasmid transfer frequency dropped dramatically by 100- to 1,000-fold in comparison to that which occurred during incubation on solid medium (P = 0.04; Table 3, experiment E). This result showed that extended cell-to-cell contact aided by cultivation on solid medium promoted more efficient plasmid transfer, data consistent with conjugative transmission.
To demonstrate the unidirectionality of plasmid transfer (i.e., from donor to recipient) and thereby exclude the possibility of cell-cell fusion as an event leading to the presence of both selection markers in a single cell, a nonconjugative, compatible (26), episomal plasmid, pMV261, carrying a hygromycin resistance gene cassette was introduced into the donor and recipient strains. Each of these Hygr versions of the donor (ΔvapA/A/H) and recipient (103−/T/H) strains were then mixed and incubated with strains 103−/T and ΔvapA/A, respectively, and recombinants were selected as per usual (as shown schematically in Fig. S2 in the supplemental material). The resultant dually antibiotic-resistant clones (Aprr Trimr) were then screened for hygromycin resistance by patching onto hygromycin-containing BHI medium. None of the clones resulting from the mating involving a Hygr donor was found to be hygromycin resistant, while all the colonies arising from matings involving an Hygr recipient were found to be resistant to hygromycin (data not shown), findings which exclude cell-cell fusion as the mode of virulence plasmid transfer. Cumulatively, the results of these experiments indicated that virulence plasmid transmission required living cells, was optimal on solid medium, was not affected by DNase, and was not the result of donor and recipient cell-cell fusion, and thus, the data support conjugation as the means of virulence plasmid transmission.
Conjugal transfer restores the virulence-associated phenotype.The capability for intramacrophage growth is a parameter of R. equi virulence. It has been demonstrated that strains carrying the pVAPA1037 virulence plasmid are able to replicate in murine or equine macrophages and are also able to grow in mice and foals (10, 18, 56). In contrast, the respective plasmid-free derivatives cannot replicate in macrophages and are efficiently cleared in vivo (10, 18, 56). Once conjugative transfer of the R. equi virulence plasmid was confirmed, we reasoned that such should restore the virulence phenotype to the avirulent, plasmid-cured R. equi recipient strain 103−/A. To verify this, a virulent donor (strain ΔvapG/H) was mated with the avirulent strain 103−/A and transconjugants were selected as described above. One transconjugant clone (strain 103−/ΔvapG/AH) was chosen for subsequent analysis. Murine bone marrow-derived macrophages were infected with the donor, recipient, and transconjugant strains. As is characteristic, the donor strain multiplied in macrophages, increasing in number approximately 18-fold over a 48-h period. The plasmid-free recipient strain (103−/A), however, did not replicate. Transfer of the virulence plasmid restored the capacity for intramacrophage growth, as the transconjugant strain (103−/ΔvapG/AH) displayed growth kinetics identical to that of the donor (Fig. 3). This finding confirmed that transmission of the virulence plasmid was associated with acquisition of the virulence phenotype.
Transfer of the virulence plasmid restores the capacity for intramacrophage growth. Monolayers of murine bone marrow-derived macrophages were infected with R. equi ΔvapG/H (donor), 103−/A (recipient), and 103−/ΔvapG/AH (transconjugant) at an MOI of 10:1. Following uptake and washing to remove unbound bacteria, amikacin was added to kill any remaining extracellular bacteria. Triplicate monolayers were lysed at 1 h, 24 h, and 48 h postinfection. Lysates were serially diluted and plated to determine the associated numbers of CFU. The data shown here are representative of three independent experiments. Error bars indicate standard deviations from the mean.
R. equi virulence plasmid transfer is dependent on traA.traA is predicted to encode a relaxase enzyme that would be responsible for cleavage of the plasmid and leading the single-strand intermediate into the recipient cell (21). It was therefore predicted that traA would be essential for conjugal transfer of the virulence plasmid. To verify a role for the traA homolog in conjugative transmission of the R. equi virulence plasmid, a marked deletion mutant of traA on the 103+ strain background was created using a two-step allelic replacement strategy (as diagramed in Fig. S4 in the supplemental material) and described in Materials and Methods. PCR analysis confirmed that the traA gene had been replaced by a 937-bp DNA fragment encoding the apramycin gene. For example, primer pair REVP6 and REVP6c, which anneals to sites internal to the deleted region, produced the expected 436-bp amplicon in the parent (103+; Fig. 4B, lane WT [wild type]) and no product in the mutant (ΔtraA) or plasmid-cured negative-control strain 103− (Fig. 4B). Additionally, a primer pair specific for the apramycin resistance gene cassette (Apr F Xba and Apr R Xba) was also used and produced the expected amplicon in the mutant and not in the parent strain (Fig. 4B). The results of the mating experiments presented in Fig. 4C showed that when the traA mutant was used as the donor, no transconjugants were obtained, indicating that the conjugation machinery was disrupted and supporting a role for traA during conjugation in R. equi. To confirm the latter, a complementation analysis was done. For this, a plasmid (pVM38) containing the full-length traA gene and its native promoter was inserted at the mutant locus to regenerate a functional traA gene by a single-crossover event, as diagramed in Fig. 4A. PCR analysis with primer pair REVP6 and REVP6c was done to confirm the reconstruction (Fig. 4B, lane C). Additional PCR amplifications showed the presence of both the apramycin resistance gene cassette and the hygromycin resistance gene cassette in the complemented strain (Fig. 4B). When this traA-reconstructed strain was used as the donor, conjugative capacity was restored, thereby confirming a role for traA in virulence plasmid transfer (Fig. 4C).
Complementation of the traA gene deletion mutant. (A) Schematic diagram depicting the single-crossover event which regenerated the full-length traA gene. Recombination between homologous plasmid and chromosomal sequences is depicted by the crossed lines (X) between the suicide plasmid (pVM38) and the traA deletion mutant. (B) PCR analysis confirmed restoration of the traA gene in the complemented strain which carries both the apr and hyg cassettes. Total DNA from R. equi strains 103+ (wild-type control strain; lane WT) and 103− (plasmid-cured strain; lane −), the ΔtraA/A mutant (lane Δ), and complemented strain ΔtraA/traA/AH (lane C) served as the templates. Lane M, molecular size marker. (C) Phenotypic complementation was confirmed by mating experiments using the complemented strain as the donor.
The R. equi virulence plasmid shows entry exclusion.Conjugative plasmids frequently display entry exclusion. Entry exclusion is defined as the property of the plasmid which restricts the cells that contain the plasmid to serve as recipients in subsequent rounds of conjugation (15). To determine if the R. equi virulence plasmid demonstrates this characteristic, we marked the chromosome of the virulence plasmid-containing wild-type R. equi strain 103+ with the dhfr gene (as described in Materials and Methods) and used this Trim-resistant strain as the recipient in mating experiments with the apramycin-resistant strain ΔvapA/A serving as the donor. A dramatic decrease in conjugation frequency (∼500-fold) resulted when the virulence plasmid-containing strain (103+/T) was the recipient in comparison to that observed when the virulence plasmid-free strain 103−/T served as the recipient (Table 3, experiment F), suggesting that the prior presence of the virulence plasmid rendered the strain less amenable to serving as a recipient. Traditionally, the extent of entry exclusion is expressed as an exclusion index, which is the ratio of the transfer frequency observed with a plasmid-free recipient to that observed with a plasmid-containing recipient (15). For the R. equi virulence plasmid, the calculated exclusion index was 663 ± 460, a value indicating that entry exclusion is strongly displayed by this plasmid.
The R. equi virulence plasmid has a limited host range.The host range for conjugative plasmids is determined by the array of organisms that can serve as recipients or the variety of hosts within which the plasmid can replicate. In order to begin to determine the host range of the R. equi virulence plasmid, matings were set up to evaluate plasmid transfer to bacteria of varied relatedness to R. equi strain 103+. We first assessed the ability of the plasmid to transfer and replicate in plasmid-cured derivatives of two closely related R. equi strains, ATCC 33701 and ATCC 33705, originally of equine and swine origin, respectively. As earlier, the chromosomes of these strains were marked by insertion of a dhfr gene to confer trimethoprim resistance, allowing them to serve as recipients in matings with strain ΔvapA/A as the donor. The results showed that the virulence plasmid could be transferred among these R. equi strains with frequencies not statistically different from the frequency for strain 103+ (P = 0.075 and P = 0.123, respectively) (Table 4). Next, recipient strains of Rhodococcus spp. other than R. equi were constructed by inserting pSET152, an integrating plasmid carrying an apramycin resistance gene, on their chromosomes. These strains were then used in mating experiments with R. equi ΔvapG/H acting as the donor, and transconjugants were selected on Apr-Hyg agar. Results showed that the virulence plasmid was transmitted to Rhodococcus sp. ATCC 15610 and Rhodococcus sp. ATCC 21505 at frequencies comparable to the frequency for R. equi strain 103− (P = 0.223 and P = 0.823, respectively) (Table 4). A lower frequency of plasmid transfer resulted when using Rhodococcus sp. ATCC 4277 as the recipient (Table 4). The reduced transfer may reflect instability of the virulence plasmid in that background or might be due to entry exclusion if Rhodococcus sp. ATCC 4277 carries an endogenous plasmid, a circumstance unknown at present. Finally, virulence plasmid transfer to related but non-Rhodococcus sp. bacteria was assessed. Subsequent to construction of an apramycin-resistant recipient strain, efficient plasmid transfer from R. equi ΔvapG/H to Dietzia sp. ATCC 184 was demonstrated (Table 4). Matings of R. equi to the actinomycete Mycobacterium smegmatis, wherein strains MKD8 and Jucho J4 were used as recipients, failed to yield transconjugants (Table 4), despite the fact that these strains have been shown to be capable of serving as recipients in conjugation experiments with M. smegmatis donor strains (41). Collectively, these results show that the R. equi virulence plasmid is a limited-host-range plasmid and is transmissible to closely related bacteria, with some exceptions.
R. equi virulence plasmid host range determination
DISCUSSION
Conjugation is one of the methods by which genetic material is transferred between bacteria. Molecular analyses of conjugal plasmids in Gram-negative organisms are well documented, but conjugation in Gram-positive bacteria is much less understood (21, 30). In general, the process of conjugation is initiated by relaxase-mediated nicking of DNA at the plasmid oriT (origin of transfer), followed by unidirectional transfer of single-stranded DNA from the donor to a recipient cell subsequent to establishment of cell-to-cell contact. Available information suggests that the initial nicking of DNA and its unidirectional transfer to the recipient are similar in both Gram-negative and Gram-positive bacteria. The major difference between the two is believed to be in the way in which cell-to-cell contact is initiated (21). In Gram-negative bacteria, a pilus-like structure is formed in donor bacteria, and it interacts with recipients and pulls them closer (21). In Gram-positive bacteria, to our knowledge, there is no report of involvement of a pilus-like structure; rather, other mechanisms facilitate the physical association of donors and recipients. For example, pheromone-mediated conjugative plasmid transfer occurs in enterococci (9, 12, 21), wherein recipient cells excrete a pheromone that, when bound by donor cells, initiates a sequence of events resulting in aggregation of donor and recipient cells and the formation of a mating channel through which plasmid DNA is transferred to recipient cells (12). Aggregation-mediated transfer is documented in Bacillus thuringiensis and lactic acid bacteria (LAB) (5, 6, 45, 55). In B. thuringiensis, two aggregation phenotypes (Agr+ and Agr−) have been identified in donor and recipient cells, respectively, but their molecular basis is not understood (21). In LAB, aggregation is reported to be caused by the interaction of two cell surface components, one encoded by a plasmid gene (clu) found in the donor and the other encoded by a chromosomal gene (agg) present in both donor and recipient (5, 21, 55). Further investigation is needed to discern the specific components involved in virulence plasmid transfer in R. equi.
Conjugation systems are divided into six different families on the basis of their associated relaxase type. These relaxase-dictated families are MOBF, MOBH, MOBQ, MOBC, MOBP, and MOBV (16). The predicted relaxase protein encoded by the R. equi virulence plasmid traA gene displayed the characteristics of the MOBF family of relaxases, determined by the presence of two distinct domains (relaxase and helicase) as well as the existence of a 3H-motif signature [H(×3)R(×3)PXXHXH(×4)N] (14) present in motif III of the relaxase domain. Two tyrosine residues in the catalytic core, an aspartate residue in motif II, and a histidine triad in motif III of the relaxase domain are also reported to be features of the MOBF family of relaxases. As documented in studies of the TrwC relaxase of the R388 plasmid from E. coli, D85 abstracts a proton from Y18 (and presumably also from Y26) and helps these tyrosine residues to act as nucleophiles, which then attack scissile DNA phosphodiester bonds. The histidine triad (H150, H161, and H163) is reported to be involved in Mg2+ ion coordination, which is essential for the DNA cleavage reaction (16). In the predicted R. equi TraA protein, we found the aspartate residue (D131) and histidine triad (H199, H210, H212) to be conserved. While there are 4 tyrosine residues within the first 100 amino acids of the N terminus of R. equi TraA which might be a part of the enzyme catalytic core, this needs to be experimentally confirmed.
Although conjugation has been documented in Rhodococcus erythropolis AN12, wherein transfer of the megaplasmid pREA400 was shown to be traA (relaxase) dependent (11, 58, 59), this work is the first to demonstrate conjugative transfer in R. equi, in which traA is similarly essential to the process. Conjugation has the propensity to spread antibiotic resistance as well as other genes of importance to bacterial survival among various species. Significantly, there are several reports of virulence genes residing on conjugative plasmids (1, 13, 27, 39, 48), and thus, conjugation provides a means to spread these virulence traits between organisms, including even the possibility of plasmid transfer between bacteria residing inside a eukaryotic host. For example, it has been demonstrated that Salmonella spp. can conjugate inside epithelial cells cultured in vitro (9). Similarly, Lim and colleagues (33) showed that specially constructed E. coli strains can conjugate within the cytoplasm of in vitro-cultured mouse cells. High-frequency conjugative transfer of an R plasmid from Escherichia coli donors to Yersinia pestis was also shown in the digestive tract of rat fleas during the normal course of infection (22). Importantly, we documented that conjugative transfer of pVAPA1037 does indeed bestow virulence properties to an avirulent plasmid-free R. equi recipient strain (Fig. 3).
In our studies, transfer of the R. equi virulence plasmid was most efficient on solid medium and at 30°C versus 37°C. The effect of incubation temperature on transfer frequency was also reported in Mycobacterium smegmatis and Salmonella enterica serovar Typhi (19, 41). In studies of conjugation in M. smegmatis, temperatures ranging from 25°C to 40°C were evaluated, and 30°C was found to be optimal for transfer (41). Similarly in S. enterica serovar Typhi, 27°C was more conducive for plasmid transfer than 37°C (19). The lower temperatures perhaps better mimic the natural environmental temperatures (e.g., soil and water) at which these organisms generally exist and where conjugation most often occurs. We also found that extending the mating time resulted in increased plasmid transfer frequency, as has been demonstrated in other conjugative systems (3, 46). Some of the increase that we observed may have been the result of outgrowth of the transconjugants, but it largely reflects greater plasmid transfer, because the matings were done on complete medium without selection, wherein no strain (donor, recipient, or transconjugant) had a growth advantage. Although transfer frequency was significantly less efficient in liquid medium than on solid agar, it was still high (∼10−4 transconjugants/recipient cell), likely indicating a stable mechanism of attachment of donors and recipients.
Entry exclusion is a feature common to conjugative plasmids and is the phenomenon which renders the cells containing conjugative plasmids (donors) poor recipients to further rounds of conjugation. It has been shown that entry exclusion plays an important role in plasmid survival and stability and it circumvents competition in a host among plasmids of identical backbones (15). Entry exclusion involves inhibition of the physical entry of incoming DNA. In E. coli, this is achieved through the activity of an inner membrane protein (like TraS encoded in the F plasmid), which blocks DNA transfer when present in recipient cells after interaction with some donor protein (such as TraG encoded in the F plasmid) (15). Even though the R. equi virulence plasmid does not have a homolog for these genes, it nevertheless displayed the property of entry exclusion. The calculated entry exclusion index for the R. equi virulence plasmid is slightly greater than the range (100 to 300) reported for the F plasmid (15).
On the basis of replicative capacity in different organisms, conjugative plasmids are divided into broad-host-range and narrow-host-range groups. We found that the R. equi virulence plasmid is efficiently transferred among different R. equi isolates and to some non-R. equi Rhodococcus spp., such as strains ATCC 21505 and ATCC 15610, as well as to related actinomycetes, such as Dietzia sp. ATCC 184. However, it appears not to be readily transmissible to or maintainable in all Rhodococcus spp. (i.e., ATCC 4277) or the genetically similar M. smegmatis. These results suggest a restricted host range for the virulence plasmid.
To conclude, our results demonstrate that the virulence plasmid of R. equi is transferable via conjugation. This finding is important in light of the fact that R. equi is primarily found in the soil, where it has ample access to other organisms to which it may pass virulence genes through conjugation. Perhaps such transfers could explain the occasional presence of typically nonpathogenic bacteria in human infections (35, 42). This finding also warrants caution in reporting an infection to be caused by R. equi only on the basis of the presence of plasmid markers, as other related organisms might acquire this plasmid via conjugation. Our data are also important in light of the fact that conjugation has been shown to facilitate the formation of biofilms in some organisms (17). Biofilms are multicellular structures which are relevant clinically because of their extreme resistance to antibiotic therapy and are thus difficult to treat, even if the identical sessile cells are antibiotic susceptible (38). Significantly, biofilm formation by R. equi was recently reported, where biofilms formed on catheters were found to be the source of Rhodococcus bacteremia in cancer patients (2). Studies are under way in our laboratory to investigate the possible relationship between conjugation and biofilm formation in R. equi.
ACKNOWLEDGMENTS
We thank Meredith Tatum and Kimberly Goldbach for their technical assistance. We also thank Miriam Braunstein, Garry Coulson, and Vincent Starai for critical reading of the manuscript and for their valuable comments.
This work was supported by funds from the National Institutes of Health (grant R01 AI060469).
FOOTNOTES
- Received 7 July 2012.
- Accepted 2 October 2012.
- Accepted manuscript posted online 5 October 2012.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01210-12.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
