• Open Access

Transfer of the Virulence-Associated Protein A-Bearing Plasmid between Field Strains of Virulent and Avirulent Rhodococcus equi

Authors


Abstract

Background

Virulent and avirulent isolates of Rhodococcus equi coexist in equine feces and the environment and are a source of infection for foals. The extent to which plasmid transfer occurs among field strains is ill-defined and this information is important for understanding the epidemiology of R. equi infections of foals.

Objectives

To estimate the frequency of transfer of the virulence plasmid between virulent and avirulent strains of R. equi derived from foals and their environment.

Animals

None.

Methods

In vitro study; 5 rifampin-susceptible, virulent R. equi isolates obtained from clinically affected foals or air samples from a farm with a history of recurrent R. equi foal pneumonia were each mixed with 5 rifampin-resistant, avirulent isolates derived from soil samples, using solid medium, at a ratio of 10 donor cells (virulent) per recipient cell. Presumed transconjugates were detected by plating on media with rifampin and colony immunoblotting to detect the presence of the virulence-associated protein A.

Results

Three presumed transconjugates were detected among 2,037 recipient colonies, indicating an overall estimated transfer frequency of 0.15% (95% CI, 0.03–0.43%). All 3 transconjugates were associated with a single donor and 2 recipient strains. Genotyping and multiplex PCR of presumed transconjugates demonstrated transfer of the virulence-associated protein A-bearing plasmid between virulent and avirulent R. equi.

Conclusions and Clinical Importance

Transfer of the virulence plasmid occurs with relatively high frequency. These findings could impact strategies to control or prevent R. equi through environmental management.

Abbreviations
BHIA

brain heart infusion agar

BHIB

brain heart infusion broth

CFU

colony-forming unit

EIDL

Equine Infectious Disease Laboratory

PBS

phosphate-buffered saline

PCR

polymerase chain reaction

rep-PCR

repetitive sequence-based PCR

TBS

tris-buffered saline

TBST

tris-buffered saline and 0.05% Tween 20

VapA

virulence-associated protein A

Rhodococcus equi, a facultative-intracellular, gram-positive, saprophytic bacterium, is the most frequent bacterial cause of severe pneumonia in foals.[1] The organism has a worldwide distribution and has a strong economic impact on the horse-breeding industry.[2] Exposure occurs early in life and infection is believed to result primarily through inhalation of aerosolized bacteria.[3] Control and prevention of foal pneumonia secondary to R. equi has proven to be challenging and development of an effective vaccine remains elusive to date.[4-8]

Control and prevention of infectious diseases is often considered in terms of the contributions and interactions of host, agent, and environmental factors.[9] Host factors that contribute to the development of disease in individual foals are crucial, but remain ill-defined.[10, 11] From the standpoint of the agent, the pathogenesis of R. equi pneumonia is dependent on the presence an 85- to 95-kDa virulence-associated plasmid and the expression of the virulence-associated protein A (VapA).[2] R. equi isolates from horses or their environments are commonly classified as either virulent or avirulent, based on the presence or absence of VapA. Molecular epidemiological studies demonstrate that there is considerable genetic diversity of isolates obtained from the environment and clinical specimens from foals.[12-16] Thus, it appears that any isolate possessing the virulence-associated plasmid appears capable of causing disease in foals (ie, the critical agent factor appears to be possession of the virulence-associated plasmid). Environmental factors also contribute to the epidemiology of R. equi pneumonia in foals: evidence exists that the airborne concentration of virulent R. equi is associated with the incidence of R. equi pneumonia at the level of both farm and foal.[17, 18] Thus, it is plausible that reducing the environmental burden of virulent R. equi might result in reduced cumulative incidence of R. equi pneumonia at breeding farms.

Both avirulent and virulent isolates are widespread in the environment of foals.[15, 16, 19-23] Because avirulent isolates are more abundant than virulent isolates in equine feces, soil, and air samples collected from horse-breeding farms,[17-23] it is important to understand the extent to which virulent R. equi can transfer their virulence plasmids to avirulent environmental isolates: a relatively high rate of plasmid transfer would confound efforts for environmental control, whereas inefficient or absence of transfer would facilitate efforts to reduce environmental exposure to R. equi. The most common mechanism by which plasmid genes are shared between bacteria is by conjugation.[24] Bacterial conjugation is a complex mechanism by which plasmid DNA can be transferred from donor bacteria to recipient bacteria through cell-to-cell contact.[24] Conjugative transfer genes have been identified on the virulence plasmid,[25] and conjugation of the virulence plasmid has been elegantly demonstrated between isogenic strains of R. equi.[26] To date, however, there are no published reports of the frequency of plasmid transference among virulent field isolates of R. equi to avirulent environmental isolates. Thus, the aim of this study was to estimate the frequency with which the virulence plasmid of R. equi was transferred from virulent isolates of R. equi obtained from clinically affected foals or air samples from their environment to avirulent isolates of R. equi obtained from soil samples at equine farms.

Materials and Methods

Two experiments were conducted. The first entailed estimating the efficiency of transfer between multiple virulent and avirulent field strains of R. equi. The second entailed confirming plasmid transference to transconjugates.

Bacterial Strains

The R. equi strains used in this study were selected based on the presence or absence of the virulence plasmid coding for the gene for VapA, and susceptibility or resistance to rifampin. Five virulent, rifampin-sensitive donors and 5 avirulent, rifampin-resistant recipients obtained from a repository of isolates maintained in the Equine Infectious Disease Laboratory (EIDL) were used in this study. The 5 donor strains (D1, D2, D3, D4, and D5) were selected using the following criteria: (1) being confirmed as virulent R. equi by multiplex PCR for the choE and vapA genes[27]; (2) being confirmed as susceptible to rifampin by antimicrobial susceptibility testing; (3) having growth completely inhibited on brain heart infusion agar (BHIA)1 containing rifampin2 at a concentration of 32 μg/mL; and (4) being obtained from clinically affected foals (D1, D2) or air samples from a farm with a history of recurrent foal pneumonia caused by R. equi (D3, D4, D5). The 5 recipient strains (R1, R2, R3, R4, and R5) were selected using the following criteria: (1) being from environmental samples (ie, recovered from air, soil, or feces); and (2) being confirmed by multiplex PCR testing for the presence of the choE gene and absence of the vapA and vapB genes.[27, 28] Rifampin resistance was generated in the avirulent (recipient) strains by first culturing the individual recipient isolates on BHIA for 48 hours at 37°C. After this incubation period, the isolates were each subcultured onto BHIA containing 2 μg/mL rifampin and allowed to incubate for 48 hours at 37°C. This process was repeated, increasing the concentration of rifampin by a magnitude of 2-fold at each iteration, until each isolate was capable of growth on BHIA containing rifampin at a concentration of 128 μg/mL. The rifampin-susceptible virulent isolates and the newly generated rifampin-resistant isolates were cryopreserved using the CryoCare preservation system.3

Plasmid Transfer

Pure R. equi cultures from bead stocks (5 virulent and 5 avirulent) were thawed and plated onto BHIA and incubated at 37°C for 48 hours to confirm pure growth of R. equi. A single colony from each plate was selected and used to inoculate 25 mL of brain heart infusion broth (BHIB)4 in a 125-mL Erlenmeyer flask; the broth was then incubated at 37°C for 48 hours, with shaking. Ten-fold serial dilutions using phosphate buffered saline (PBS)5 were prepared to determine the concentration of R. equi in each broth culture. The broth cultures were then diluted to 1 × 107 colony-forming units per mL (CFU/mL).

Each donor strain was combined with each recipient strain, creating 25 different individual pairs. Separate 15-mL conical tubes were prepared to contain 4.5 mL BHIB, 500 μL of a donor strain (1 × 107 CFU/mL), and 50 μL of a recipient strain (1 × 107 CFU/mL). The tube contents were then filtered through a commercially available, 13-mm diameter, 0.45-μm pore size nitrocellulose conjugation filter system6 using a sterile 20-mL syringe. The donor (500 μL) and recipient (50 μL) strains also were filtered individually as positive and negative controls, respectively; this allowed us to confirm that the recipient strain consistently remained resistant to rifampin and that none of the donor isolates had spontaneously developed resistance to rifampin. Conjugation filters were then placed (bacterial side up) onto BHIA plates using sterile thumb forceps and incubated at 37°C for 48 hours to allow for mating. After incubation, the filters were removed from the BHIA plates and vortexed for 20 seconds in 1.5-mL tubes containing 1 mL of BHIB. This was performed to suspend the bacteria growing on the filter for quantitative culture and identification of transconjugates using the selective media.

Postplasmid Transfer Quantitative Culture

The suspension from each filter was serially diluted using PBS, and 100 μL of the 10−3, 10−4, and 10−5 dilutions were plated in duplicate on BHIA plates containing 32 μg/mL of rifampin and cultured at 37°C for 48 hours. Donor controls were also plated on BHIA plates without rifampin to provide a positive control for the colony immunoblotting assay. The number of CFUs for each plate was tabulated and recorded, including the positive and negative controls. Plates containing the dilution that yielded between 15 and 150 CFU were tested by a colony immunoblotting assay to identify the formation of presumed transconjugates (ie, R. equi recipients to which the plasmid had been transferred) and to verify donor and recipient controls remained VapA-positive and VapA-negative, respectively.

Colony Immunoblot Assay

The number of presumed transconjugates (VapA-positive, rifampin-resistant) was determined using a previously described modified colony immunoblot assay[15] in which nitrocellulose membranes7 were applied to the postconjugation quantitative culture plates and probed with a primary mouse VapA monoclonal antibody (generously provided by Dr Shinji Takai) and a secondary anti-mouse IgG antibody8 conjugated with horseradish peroxidase. When the substrates 3,3′,5,5′-tetramethylbenzidine9 and hydrogen peroxide were added, the virulent R. equi colonies were identified as those that appeared blue and colonies of avirulent R. equi remained colorless. The frequency of transfer for each pair was calculated using the following equation:

display math

where E is the% frequency of plasmid transfer, V is the number of virulent R. equi colonies identified by the colony immunoblotting, and T is the total number of R. equi colonies identified on the plates before immunoblotting. Descriptive statistics including the median and range were calculated for each of the 25 pairs and for all the pairs combined. The 95% confidence interval for the frequency of plasmid transference was calculated by exact binomial methods.

Virulence Plasmid Transfer Confirmation

Our second experiment was conducted to verify that VapA detected by immunoblotting was the result of plasmid transfer presumably by conjugation rather than spontaneous development of rifampin resistance in the virulent donor. The previously described plasmid transfer procedure was repeated using the mating pair with the highest observed transfer rate. The R. equi isolates identified to be growing on the counterselection media (BHIA plates containing 32 μg/mL of rifampin) after the transfer procedure were then streaked in duplicate using sterile toothpicks onto BHIA plates containing 32 μg/mL of rifampin in a grid pattern to enable location on the duplicated plate of colonies identified as positive on the original plate used for immunoblotting. These duplicate plates where then cultured at 37°C for 48 hours. The plate duplication was performed because colony immunoblotting generally results in colonies being lost; plate duplication ensured availability of counterselected (ie, rifampin-resistant) isolates identified by immunoblot as being VapA-positive for further testing. The VapA-positive, rifampin-resistant isolates identified by immunoblotting were then selected for subculture and streaked onto BHIA containing 32 μg/mL of rifampin and incubated at 37°C for 48 hours.

Repetitive Sequence-Based PCR Genotyping

DNA was extracted from the transconjugate isolates, donor, recipient, and positive (33701+) and negative (33701−) controls using a 10-μL loop of plated culture and a commercial DNA isolation kit10 following the manufacturer's instructions. All DNA was diluted to a concentration of between 25 and 50 ng/μL. Before genotyping, DNA samples from these isolates were tested by multiplex PCR to confirm the presence or absence of the vapA gene.[27] The extracted DNA was amplified using commercial kit11 according to the manufacturer's instructions. The highly discriminating repetitive sequence-based PCR genotyping technique (rep-PCR)[29] was then used to determine the genotypic similarity of the transconjugates, the donor, and the recipient. Briefly, 18 μL rep-PCR master mix 1, 2 μL of primer mix AA, 2.5 μL 10× PCR buffer,12 and 0.5 μL each of dimethyl sulfoxide and DNA polymerase13 were combined for a total of 23.5 μL per reaction for preparation of the master mix. Next, 23 μL of the master mix was added to 2 mL of genomic DNA (either from sample isolates of the donor, recipient or transconjugate, the reference strain 33701, or the manufacturer's proprietary DNA used as a positive control for PCR amplification) for a total of 25 μL per reaction mixture for rep-PCR. A negative control also was included for each rep-PCR assay in which 2 μL of molecular-grade water was added to 23 μL of the master mix. The thermal cycling parameters were as follows: initial denaturation of 94°C for 2 minutes; 35 cycles of denaturation at 94°C for 30 seconds, annealing at 65°C for 30 seconds, and extension at 70°C for 90 seconds; and a final extension at 70°C for 3 minutes. The DNA amplicons were separated using electrophoresis and a microfluidics chip device14 in a bioanalyzer.15 Analysis was performed by the appropriate software (version 3.4).16 Similarity matrices were determined using the Kullback-Leibler correlation coefficient, and the unweighted-pair group method with arithmetic mean was used to create a dendrogram. A computer-generated virtual gel image based on results of the microfluidics was also produced by the software16 to better visualize rep-PCR patterns. Isolates were considered genotypically distinct if there was <90% similarity between isolates from the similarity matrix and if there were at least 2 differences in rep-PCR bands.[29, 30] Isolates were considered closely related (ie, the same strain) if they were ≥90% similar and identical when similarity was >98%.

Results

The formation of transconjugates was identified in 2 of the 25 mating pairs; the other 23 mating pairs had a transfer frequency of 0% at the dilution used for the calculation. The number of recipient colonies on the plates counted for the 25 mating pairs was 2,307 (Table 1). The crude frequency of plasmid transference among recipients was 0.15% (3/2,037 recipient CFU; 95% CI, 0.03–0.43%). A single donor and 2 recipients accounted for the 3 detectable occasions of plasmid transfer (Table 1): mating pairs D3 : R2 and D3 : R4 had a plasmid transfer frequency of 2.2% (2/90 recipient CFU) and 3.2% (1/31 recipient CFU), respectively. The quantitative dilution was 10−4 of the plates on which conjugal plasmid transfer was detected. The median plasmid transference for all pairs was 0% with a range of 0–3.2%. All (100%) donor controls were virulent based on detection of VapA by immunoblotting and remained rifampin-sensitive (ie, there was no growth of donor strains on the rifampin-containing BHIA plates in the quantitative dilutions) throughout the experiments. All (100%) of recipient controls were avirulent based on the absence of VapA detection by colony immunoblotting and remained rifampin-resistant (ie, grew readily on the rifampin-containing BHIA plate in the quantitative dilutions) throughout the experiments.

Table 1. Transfer of the virulence-associated protein A-bearing plasmid between virulent and avirulent Rhodococcus equi isolates.
 R1R2R3R4R5
  1. D1–D5, virulent donor isolates; R1–R5, avirulent recipient isolates.

  2. The frequency of transfer for each pair was calculated using the following equation: E = (V/T) × 100% where E is the % frequency of plasmid transfer, V is the number of presumed transconjugate colonies identified by the colony immunoblotting, and T is the total number of R. equi colonies identified on the plates before immunoblotting. The data are represented above for each mating pair as E (V/T). The total number of recipient colonies on the plates counted (T) for the 25 mating pairs was 2,307. The bolded items are the only non-zero (positive) results.

D10% (0/125)0% (0/82)0% (0/61)0% (143)0% (0/124)
D20% (0/34)0% (0/42)0% (0/187)0% (0/20)0% (0/86)
D30% (0/21) 2.2% (2/90) 0% (0/17) 3.2% (1/31) 0% (0/149)
D40% (0/74)0% (0/101)0% (0/24)0% (0/160)0% (0/94)
D50% (0/50)0% (0/84)0% (0/84)0% (0/39)0% (0/115)

The plasmid transfer experiment was repeated for the mating pair D3 : R4 yielding a transfer frequency of 1.0% (3/313 recipient CFU). All 3 presumed transconjugates and the donor were positive by PCR for the vapA gene and the choE gene, identifying them as R. equi, and the recipient was negative by PCR for vapA (but positive for choE). Rep-PCR of the donor (D3), recipient (R4), and 3 transconjugates (T1, T2, T3) indicated that the presumed transconjugates were not donor isolates that had developed resistance to rifampin (Fig 1). Similarity matrices using the Kullback-Leibler correlation coefficient indicated that the 3 presumed transconjugates were identical (ie, >98% genetic similarity) and that the recipient isolate was closely related (91–95% genetic similarity) to the presumed transconjugates; however, the 3 presumed transconjugates were genotypically distinct (59–74% genetic similarity) from the donor.

Figure 1.

Dendrogram and computer-generated virtual gel image of repetitive sequence-based PCR genotyping of donor, recipient, and presumed transconjugate isolates of Rhodococcus equi. D3, donor isolate; R4, recipient isolate; T1–T3, presumed transconjugate isolates. The% similarity is based on the Kullback-Leibler correlation coefficient, and the unweighted pair group method with arithmetic mean. The computer-generated virtual gel image allows visualization of the rep-PCR patterns and genotypic differentiation of donor, recipient, and presumed transconjugate isolates of R. equi.

Discussion

Results of the study reported here indicate that plasmid transfer can occur between virulent and avirulent field isolates of R. equi at 37°C using a nitrocellulose filter on an agar matrix. Our results are consistent with the recent findings[26] confirming conjugal transfer of the virulence plasmid between isogenic strains of virulent and avirulent R. equi. These findings provide important microbiological evidence for a mechanism of persistence and perpetuation of virulent R. equi in the environment of foals. It is also important to note that the majority of the isolates did not transfer within the detection limits of this study and thus there appears to be variation in the ability of individual isolates to share and receive the plasmid. When comparing these results with the conjugative transfer frequency with which Clostridium perfringens shares its virulence plasmid (viz, 5.7 × 10−2 transconjugates/donor cell), or with which R. erythropolis transfers its megaplasmid to recipient strains (viz, 5 × 10−4 transconjugates/recipient cell), the observed frequency of plasmid transference for R. equi appeared relatively high.[31-33] Our estimated transfer frequencies were also consistent with the recently published report documenting a relatively high plasmid conjugation rate between a single isogenic donor and recipient mating pair of R. equi.[26]

Our results have at least 2 important implications. First, the finding that the virulence plasmid can be shared in vitro with relatively high frequency suggests that eradication of the virulent strains of R. equi in the environment of horse-breeding farms will be very difficult. Efficient mating pairs are likely among the diverse population of R. equi that exists among isolates from foals and their environment.[12-16] To illustrate the impact of plasmid transfer, consider the following hypothetical calculation. A typical defecation by a mature mare weighs ~3 kg, the typical fecal concentration of R. equi is approximately 1 × 103 CFUs/g, and the proportion of virulent R. equi in feces of mature mares is ~16%.[15] Thus, a single defecation would contain an estimated 3 × 106 total R. equi including ~3 × 105 virulent isolates. If the frequency of plasmid transfer was ~0.15% among recipients, then one would expect ~4,000 transfers of the virulence plasmid to occur (0.15% × 2.7 × 106 recipients); moreover, the virulent isolates also will encounter avirulent isolates in the soil when passed in pasture. Thus, perpetuation of the environmental burden of virulent R. equi at equine farms seems likely on the basis of our findings. Consequently, environmental control strategies, like oral administration of gallium nitrate to periparturient mares to reduce fecal shedding of both virulent and avirulent R. equi,17 would need to be implemented each year. Second, the finding that the plasmid can be shared between virulent and avirulent isolates could explain why environmental contamination with virulent R. equi remains so widespread among both farms with and without endemic disease.[2, 13-18]

Variation identified among the individual field strains' ability to share and receive the plasmid was an interesting finding. Of the 5 donor strains examined in the first experiment, a single donor strain (D3) was responsible for all of the observed instances of plasmid transfer. Heterogeneity among donors and recipients is not uncommon in conjugation studies of other bacteria.[34] Thus, it is plausible that strains of virulent R. equi vary in their ability to donate the plasmid. This variability among strains might be related to variations in the putative conjugation genes known to exist on the virulence-associated plasmid itself or on some of the other coinhabiting cryptic plasmids.[25, 35, 36] Alternatively, there might be variation among the recipient strains' ability to receive plasmids. For example, evidence exists that some bacteria exclude or improve transference of plasmids when they bear 1 or more plasmids (a phenomenon known as entry exclusion).[36, 37] Conceivably, some of our avirulent recipient strains could have borne a large plasmid that did not encode the genes for either VapA or VapB and that limited plasmid receipt. Another explanation of variation in frequency of plasmid transfer could be the presence of a pheromone-responsive conjugation system, in which recipient cells produce sex pheromones that attract donor cells and help facilitate conjugation.[38] To date, however, a pheromone-responsive conjugation system has been identified only in enterococci; to the authors' knowledge, pheromone-responsive conjugation has not been evaluated for R. equi. Variation in conjugation frequency among donor and recipient strains of R. equi merits further investigation.

There are several important limitations to this study. First, the efficiency of plasmid transfer was only evaluated using a single physiologic temperature (37°C), a single pH, a single donor : recipient ratio, and a single period of time. The temperature of 37°C was selected in this study because of its clinical relevance being near body temperature and evidence that the vap genes, including vapA (encodes for VapA), are optimally expressed at this temperature. In the environment, however, R. equi would encounter different temperatures. Temperature-sensitive plasmid conjugation has been described for other bacteria,[39] and it was recently reported that the virulence plasmid of R. equi was transferred conjugatively at a higher frequency when cultured at 30°C.[26] These findings indicate that further studies evaluating epidemiologically relevant temperatures are needed. The expression of VapA has also been shown to be greatest in a slightly acidic environment; thus, the influence that variation in pH could have on plasmid transference (and expression of the putative conjugation genes) merits investigation. The donor : recipient ratio of 10 : 1 was selected in this study to help increase the chances of cell-to-cell contact thought to be required for conjugation to occur. Evaluation of different donor : recipient ratios including 1 : 1 and 1 : 10 (the latter being more representative of the environmental ratio) will help characterize the impact that the ratio might have on the frequency of conjugative plasmid transfer. The decision to allow transfer to occur for 48 hours was based primarily on the typical culture period utilized for a slow-growing organism like R. equi. Conjugal transfer rates are significantly increased when isogenic isolates are allowed to mate for 72 hours.[26] Thus, our selected incubation time could have limited our ability to detect less efficient levels of plasmid transfer by conjugation.

Second, our estimated rate of transfer was based on the proportion of presumed transconjugates among avirulent isolates (recipient cells) on an individual dilution plate. We chose to use a dilution method for several reasons. First, we wanted to have accurate colony counts and thus chose to only analyze (by immunoblotting) those plates with an approximate range of 15 and 150 total CFU/plate.[15, 23] Second, we wanted to reduce the total plated CFU load to reduce the odds of identifying donors with spontaneous chromosomal mutations allowing for rifampin resistance. Interestingly, while protecting ourselves from false-positive results by utilizing quantitative dilutions and the specified range of CFU/plate described above, we limited our ability to detect less efficient transfer frequencies. Thus, this could partially explain why we only detected 2 successful transfer pairs out of 25 and 3 transconjugal events among 2,037 colonies. It is noteworthy that despite the differences in experimental design, our estimated frequency of plasmid transfer was similar to that recently reported.[26]

Third, use of rifampin resistance for counterselection had the disadvantage of spontaneous resistance to rifampin developing among donors. Spontaneous genetic mutation for rifampin resistance develops through a single point mutation in the rpoB gene of the bacterial chromosome.[40, 41] Results of multiplex PCR indicated that the transconjugates and donor expressed vapA, but that the recipient did not. Moreover, rep-PCR genotyping indicated that the presumed transconjugates were genotypically quite distinct from the donor and very closely related to the recipient (Fig 1), providing compelling evidence that there was transfer of the VapA-bearing plasmid from donor to recipients. This finding is also consistent with probability-based reasoning. We conducted additional experiments (data not shown) to estimate the rate at which spontaneous mutation of rifampin resistance developed in our donor isolates. We observed that the mutation rate ranged from 1.4 × 10−9 to 1.0 × 10−8, consistent with observations from previous reports.[40, 41] The number of donor bacteria added to each of our plates was approximately 5 × 106 CFU. Based on these findings, we estimated the probability that a single VapA-positive colony identified on the counterselection plate resulted from a spontaneous mutation to rifampin would have been no more than the product of 1.0 × 10−8 (our highest observed rate of spontaneous mutation) and 5 × 106 (the total number of donors mixed with recipients), or 5 × 10−2 (5%). Put another way, the probability that this event was attributable to a mechanism other than spontaneous mutation was 95%. Collectively, the observed frequency of apparent plasmid transference, our use of positive and negative controls, the observed rate of spontaneous rifampin resistance in our donors, and the rep-PCR and PCR results strongly support that our findings are attributable to plasmid transfer (presumably by conjugation).[26]

A fourth limitation of this study is that we did not directly evaluate the presence of the entire virulence plasmid; rather, we used the expression of VapA protein and identification of the vapA gene as an indirect measurement of the presence and functionality of the virulence plasmid. Expression of VapA by R. equi is strongly associated with disease in foals and detection of this protein is commonly used to classify isolates of this bacterium as virulent.[2, 42, 43] Thus, although the plasmid was not identified directly in these presumed transconjugates, substantial clinical, epidemiological, and microbiological evidence support identification of VapA protein and the vapA gene as good measures of virulence. Nevertheless, direct evidence that the whole plasmid was transferred from donors to recipients would have provided more definitive evidence of plasmid transfer. For example, PCR for the traA gene of the virulence plasmid in the presumed transconjugates would have allowed documentation of transfer of the entire plasmid.[28] A fifth limitation of our study is that neither maintenance of the plasmid in the presumed transconjugates nor the ability of the presumed transconjugates to share the plasmid with other avirulent recipients was evaluated. Future studies are needed to evaluate the ability of presumed transconjugates to maintain the plasmid and transfer the plasmid to avirulent isolates of R. equi.

A sixth limitation of our approach was that we did not establish that transference of the plasmid resulted in creating the virulent phenotype using in vitro infections of macrophages or in vivo infections of mice or foals. Cooptive evolution is a mechanism by which virulence plasmids obtained by R. equi through horizontal gene transfer (eg, by conjugation) recruit housekeeping genes from the core rhodococcal genome to help ensure virulence and survival in the host.[44] Confirmation of the acquisition of pathogenicity in the presumed transconjugates identified in this study would likely have supported this cooptive virulence model in R. equi.[44] Evidence exists, however, that plasmid curing eliminates the virulent state and that introduction of the plasmid to isogenic strains establishes virulence.[45] Thus, it is plausible that our presumed transconjugates were indeed virulent. Furthermore, conjugal transfer of the virulence plasmid of R. equi to avirulent, plasmid-cured recipients resulted in re-establishment of the virulent phenotype in both cultured macrophages and infected mice.[26] Seventh, the sample size of donors and recipients in this study was small. Studies evaluating more field-derived strains including fecal and soil isolates should be considered to better characterize the frequency of plasmid transference and variation among donors and recipients in sharing the plasmid. Lastly, we cannot exclude the possibility that plasmid transference occurred by alternative mechanisms including transformation, transduction, or cell-to-cell fusion rather than conjugation. However, we consider these much less likely because conjugation occurs far more commonly in nature and these alternative mechanisms of plasmid transfer were not detected in a recent experiment.[26]

In conclusion, we report that plasmid transfer can occur with relatively high frequency between some virulent and avirulent isolates of R. equi in vitro. These findings suggest a mechanism by which environmental burdens of virulent R. equi are maintained. Future strategies to control or prevent R. equi pneumonia by reducing environmental exposure of foals to virulent R. equi will need to consider the ability of virulent isolates to share their virulence plasmids with avirulent isolates.

Acknowledgments

This work was done in the Equine Infectious Disease Laboratory, College of Veterinary Medicine and Biomedical Sciences, Texas A&M University, College Station, Texas. The authors thank Drs Angela Bordin, Jessica Harrington, and Sara Lawhon, and Ms Courtney Brake, Ms Stephanie Buntain, Ms Kaytee Weaver, and Mr Grant Wicks for technical assistance. Kimberly Bishop and Audrey Morrissey were supported by the NIH and Merck-Merial Foundation Summer Veterinary Medical Scientist Training Program.

Funding: This study was funded by the Link Equine Research Endowment.

Conflict of Interest: Authors disclose no conflict of interest.

Footnotes

  1. 1

    Becton, Dickinson and Company, Sparks, MD

  2. 2

    Sigma Chemical Co, St Louis, MO

  3. 3

    Key Scientific, Stamford, TX

  4. 4

    Becton, Dickinson and Company

  5. 5

    Mediatech Inc, Manassas, VA

  6. 6

    F-Millipore Filters, Millipore Corporation, Bedford, MA

  7. 7

    Bio-Rad Laboratories, Hercules, CA

  8. 8

    Horseradish peroxidase-conjugated goat IgG fraction against mouse IgG, MP Biomedicals Inc, Aurora, OH

  9. 9

    Sigma Chemical Co

  10. 10

    MO BIO Laboratories Inc, Carlsbad, CA

  11. 11

    Bifidobacterium DNA fingerprinting kit, bioMérieux Inc, Durham, NC

  12. 12

    Applied Biosystems Inc, Foster City, CA

  13. 13

    Applied Biosystems Inc

  14. 14

    Caliper Technologies Inc, Hopkinton, MA

  15. 15

    Agilent 2100 Bioanalyzer, Agilent Technologies Inc, Santa Clara, CA

  16. 16

    DiversiLab system, bioMérieux Inc, Durham, NC

  17. 17

    Buchanan BR, Kuskie K, Pusterla N, Cohen ND. Effect of oral gallium nitrate on fecal shedding of Rhodococcus equi in mares. J Vet Internal Med 2011;25:757 (abstract)

Ancillary