ABSTRACT
Rhodococcus equi is a leading cause of severe pneumonia in foals. Standard treatment is dual antimicrobial therapy with a macrolide and rifampin, but the emergence of macrolide- and rifampin-resistant R. equi isolates is an increasing problem. The objective of this study was to determine the effect of macrolide and/or rifampin resistance on fitness of R. equi. Three unique isogenic sets were created, each consisting of four R. equi strains, as follows: a susceptible parent isolate, strains resistant to macrolides or rifampin, and a dual macrolide- and rifampin-resistant strain. Each isogenic set’s bacterial growth curve was generated in enriched medium, minimal medium (MM), and minimal medium without iron (MM-I). Bacterial survival in soil was analyzed over 12 months at −20°C, 4°C, 25°C, and 37°C, and the ability of these strains to retain antimicrobial resistance during sequential subculturing was determined. Insertion of the mobile element conferring macrolide resistance had minimal effect on in vitro growth. However, two of three rpoB mutations conferring rifampin resistance resulted in a decreased growth rate in MM. In soil, macrolide- or rifampin-resistant R. equi strains exhibited limited growth compared to that of the susceptible R. equi isolate at all temperatures except −20°C. During subculturing, macrolide resistance was lost over time, and two of three rpoB mutations reverted to the wild-type form. The growth of rifampin-resistant R. equi colonies is delayed under nutrient restriction. In soil, possession of rifampin or macrolide resistance results in decreased fitness. Both macrolide and rifampin resistance can be lost after repeated subculturing.
IMPORTANCE This work advances our understanding of the opportunistic environmental pathogen Rhodococcus equi, a disease agent affecting horses and immunocompromised people. R. equi is one of the most common causes of severe pneumonia in young horses. For decades, the standard treatment for R. equi pneumonia in horses has been dual antimicrobial therapy with a macrolide and rifampin; effective alternatives to this combination are lacking. The World Health Organization classifies these antimicrobial agents as critically important for human medicine. Widespread macrolide and rifampin resistance in R. equi isolates is a major emerging problem for the horse-breeding industry and might also adversely impact human health if resistant strains infect people or transfer resistance mechanisms to other pathogens. This study details the impact of antimicrobial resistance on R. equi fitness, a vital step for understanding the ecology and epidemiology of resistant R. equi isolates, and will support development of novel strategies to combat antimicrobial resistance.
INTRODUCTION
The Gram-positive bacterium Rhodococcus equi is a common and important cause of foal pneumonia (1, 2). The disease is endemic on many horse-breeding farms, and the resulting costs from veterinary care and mortality of some foals are high (3). R. equi pneumonia also has a long-term effect on the equine industry because foals that recover from the disease are less likely to race as adults (4).
The standard treatment for R. equi foal pneumonia is dual antimicrobial therapy with a macrolide and rifampin; effective alternatives to this combination are lacking. The World Health Organization classifies these agents as critically important for human medicine (5); therefore, these drugs should be used judiciously. Control of R. equi on farms where it is endemic currently relies upon early detection of disease using thoracic ultrasonography and early antimicrobial treatment. Because the disease is insidious, such that clinical signs generally are not apparent until disease is advanced, this approach permits earlier disease detection, when treatment might be more effective. Thoracic ultrasonographic screening appears to have decreased R. equi-associated mortality on farms where it is endemic compared to that in historical controls; however, antimicrobial treatment of foals presenting with ultrasonographic abnormalities but lacking clinical signs of R. equi pneumonia (i.e., fever, anorexia, lethargy, tachypnea, cough) did not significantly reduce recovery time, and many foals with ultrasonographic lesions recover without treatment (6–8).
With increased use of macrolides and rifampin for chemoprophylaxis or treatment of subclinically affected foals, a significant increase in macrolide- and rifampin-resistant R. equi isolates has been documented (9). Isolates of R. equi resistant to all macrolides and rifampin have been cultured from up to 40% of the foals at a farm in Kentucky (10). These findings are particularly concerning, as the odds of death are approximately 7-fold higher for foals infected with resistant strains than for those infected with susceptible isolates (9).
Rifampin resistance in R. equi results from mutations in the beta subunit of the RNA polymerase (rpoB) gene (11). To date, four single base pair mutations in the rpoB gene have been described (11). Macrolide resistance in R. equi, in contrast, is conferred by acquisition of an integrative conjugative element (ICE) carrying a new rRNA methylation gene, designated erm(46) (12).
Antibiotics target important functions, such as cell wall synthesis, regulation of chromosome supercoiling, RNA transcription, and protein synthesis. Thus, it is not surprising that the development of antimicrobial resistance often results in a decrease in biological fitness (13). The effects of macrolide and rifampin resistance on fitness and virulence have been studied in other bacterial species, but not in R. equi.
rpoB is highly conserved among bacterial species and is involved in transcription and elongation; therefore, mutations in the rpoB gene could alter these essential functions (14, 15). Mutations of the rpoB gene conferring rifampin resistance in Mycobacterium tuberculosis decrease fitness and survival in macrophages (16), but the effect of rifampin resistance on the fitness of Staphylococcus aureus varies depending on the specific mutation (17). Macrolide-resistant Campylobacter jejuni isolates have impaired fitness in vitro and decreased infectivity in chickens (18), while macrolide resistance in Enterococcus faecalis is associated with increased virulence in swine (19).
Because the effects of resistance to a given antimicrobial agent on fitness are complex and variable, the impacts of macrolide or rifampin resistance on fitness of other bacterial species cannot be extrapolated to R. equi. The objective of this study was to determine the effects of macrolide- or rifampin-resistance traits on overall fitness of R. equi in different media and in soil over time. We hypothesized that macrolide- and rifampin-resistant R. equi isolates will exhibit reduced bacterial fitness compared to that of susceptible wild-type R. equi isolates.
RESULTS
Resistance to macrolides or rifampin impacts in vitro growth of Rhodococcus equi.Figure 1 illustrates the growth in varied media of susceptible (n = 30) and macrolide- and rifampin-resistant (n = 30) R. equi strains obtained from foals with pneumonia. In brain heart infusion (BHI) broth, the mean ± standard error (SE) of the growth rate of macrolide- and rifampin-resistant isolates (0.167 ± 0.008 h−1) was similar to that of susceptible isolates (0.165 ± 0.008 h−1; P = 0.863). In contrast, the mean growth rate of the resistant isolates was significantly reduced in minimal medium (MM) (0.339 ± 0.009 h−1 versus 0.362 ± 0.010 h−1, respectively; P = 0.030) and in minimal medium without iron (MM-I) (0.048 ± 0.008 h−1 versus 0.250 ± 0.012 h−1, respectively; P < 0.001).
Effect of macrolide and rifampin resistance on in vitro growth of clinical R. equi isolates. The in vitro growth of 30 susceptible (S; diamonds) and 30 dual macrolide- and rifampin-resistant (R; triangles) clinical R. equi isolates was followed over 30 h at 37°C in (a) brain heart infusion broth, (b) minimal medium broth, and (c) minimal medium without iron.
Next, we wanted to determine if the impaired bacterial growth shown in Fig. 1 was a result of macrolide, rifampin, or a combination of macrolide and rifampin resistance by creating and assessing the in vitro growth of three unique isogenic sets. In BHI, two out of three macrolide-resistant R. equi strains (2M and 3M) with insertion of the ICE erm(46) displayed a significant reduction in maximum in vitro growth compared to that of the susceptible parent R. equi isolate (P = 0.013 and P = 0.023, respectively; Fig. 2). Compared to that of the susceptible R. equi isolate, strains 1R and 2R displayed a reduction in maximum growth that was mimicked by the dual-resistant R. equi strains in those isogenic sets (P < 0.01; Fig. 2).
In vitro growth curves for three R. equi isogenic sets grown overnight in brain heart infusion broth. (a) Isogenic set 1, (b) isogenic set 2, and (c) isogenic set 3. Each isogenic set consisted of a susceptible wild-type R. equi isolate (WT, diamonds), an R. equi strain resistant to macrolides (M; circles) or rifampin (R; squares), and an R. equi strain resistant to both macrolides and rifampin (RM; triangles). In vitro growth was recorded, in triplicate, every 30 min over the course of 30 h at 37°C, but for clarity, only hourly data are shown. Each data point shown is the mean (± standard deviation [SD]) of three independent experiments.
When bacterial growth was analyzed in MM (Fig. 3) and MM-I (Fig. 4), resistance to macrolides had no impact on in vitro growth of R. equi. In contrast, rifampin-resistant R. equi strains 2R and 2RM demonstrated impaired growth in MM (P = 0.011 and P = 0.005, respectively). In MM-I, R. equi strain 3R had significantly reduced maximal growth compared to that of the susceptible parent R. equi isolate (P = 0.033).
In vitro growth curves for three R. equi isogenic sets grown in minimal medium broth. (a) Isogenic set 1, (b) isogenic set 2, and (c) isogenic set 3. Each isogenic set consisted of a susceptible wild-type R. equi isolate (WT; diamonds), an R. equi strain resistant to macrolides (M; circles) or rifampin (R; squares), and an R. equi strain resistant to both macrolides and rifampin (RM; triangles). In vitro growth was recorded, in triplicate, every 30 min over the course of 30 h at 37°C, but for clarity, only hourly data are shown. Each data point shown is the mean (±SD) for three independent experiments.
In vitro growth curves for three R. equi isogenic sets grown overnight in minimal medium broth without iron. (a) Isogenic set 1, (b) isogenic set 2, and (c) isogenic set 3. Each isogenic set consisted of a susceptible wild-type R. equi isolate (WT; diamonds), an R. equi strain resistant to macrolides (M; circles) or rifampin (R; squares), and an R. equi strain resistant to both macrolides and rifampin (RM; triangles). In vitro growth was recorded, in triplicate, every 30 min over the course of 30 h at 37°C, but for clarity, only hourly data are shown. Each data point shown is the mean (±SD) of three independent experiments.
Macrolide and rifampin resistance is lost after repeated subculturing.Table 1 demonstrates that resistance to macrolides and rifampin is lost with repeated in vitro passaging. Only R. equi strain 1R retained resistance to rifampin, whereas both 2R and 3R became susceptible after passage 10. rpoB sequencing and the determination of MICs demonstrated that this phenotypic change in rifampin resistance resulted from the reversion of the rpoB mutation to the wild-type sequence (Fig. S1).
Number of R. equi colonies resistant to rifampin or macrolides during subculturing of R. equi isogenic sets
The percentage of resistant R. equi colonies declined over time among all three macrolide-resistant R. equi strains (Table 1). By passage 90, only 1%, 3%, and 12% of colonies remained macrolide-resistant for R. equi strains 1M, 2M, and 3M, respectively, compared to the 100% resistance observed at passage 10 with these strains. PCR analysis and determination of the MICs confirmed the loss of resistance to macrolides and established that macrolide-sensitive R. equi colonies at passages 10 and 90 lacked erm(46), whereas the resistant R. equi colonies retained the gene.
Resistance to macrolides, rifampin, or both impacts survival of R. equi in the soil.Figure 5 illustrates the impact of macrolide and rifampin resistance on R. equi in the soil by determining the CFU/ml of each strain at −20°C, 4°C, 25°C, and 37°C over the course of 12 months. At −20°C, there was no significant effect of resistance on R. equi CFU/ml (P = 0.057), no significant effect of time (P = 0.368), and no significant interaction between the effects of resistance and time (P = 0.655). At 4°C, there was a significant main effect of resistance (P < 0.001), with the wild-type R. equi culture having a higher marginal mean CFU/ml than those of either the macrolide- or rifampin-resistant strains. There was also a significant effect of time (P < 0.001), with the marginal means of all time points being different from one another. There was no significant interaction between the effects of resistance and time at 4°C (P = 0.481). At 25°C, there was a significant effect of resistance (P = 0.002), with the wild-type R. equi culture having a higher marginal mean CFU/ml than those of all of the resistant isolates. There was also a significant effect of time (P < 0.001), with the marginal means at 14 days and at 2 months being significantly higher than those at 6 and 12 months. There was no significant interaction between the effects of resistance and time at 25°C (P = 0.458). At 37°C, there was a significant interaction between the effects of resistance and time (P < 0.001); the wild-type R. equi culture had a higher mean CFU/ml than those of all of the resistant strains at 14 days, but there were no significant differences between strains at any of the remaining time points.
Impact of macrolide- and rifampin-resistance on R. equi in the soil. Normalized log of the number of CFU of wild-type (WT; diamonds), macrolide-resistant (M; circles), rifampin-resistant (R; squares), and dual macrolide- and rifampin-resistant (RM; triangles) R. equi strains grown in soil. CFU of each strain at (a) −20°C, (b) 4°C, (c) 25°C, and (d) 37°C at multiple time points over the course of 12 months are shown. Each data point shown is the mean (±SD) of three independent experiments. Results of the factorial analysis of resistance and time are summarized in the text.
DISCUSSION
These data support our hypothesis that macrolide- and rifampin-resistant R. equi isolates would exhibit reduced bacterial fitness compared to that of susceptible wild-type R. equi by demonstrating substantially decreased fitness of resistant R. equi strains under a number of conditions. Our data illustrate that resistance to macrolides and rifampin reduced the fitness of clinical R. equi isolates under nutrient constraints. The defect in in vitro growth of dual-resistant clinical R. equi isolates was more severe in the absence of iron. R. equi is a facultative intracellular pathogen of host macrophages (20), an environment deprived of iron; to compensate, the bacterium utilizes several mechanisms for iron acquisition (21, 22). Due to the low-iron environment in macrophages, it is possible that dual macrolide and rifampin resistance diminishes the ability of R. equi to survive in macrophages. Thus, while not addressed in this study, in the absence of antimicrobial pressure, it is reasonable to suspect that these resistant bacterial isolates might be outcompeted in the host by susceptible R. equi strains.
The R. equi isogenic sets developed for this study also demonstrate effects of resistance to macrolides or rifampin on in vitro growth of R. equi that differ from those observed in the dual-resistant clinical isolates. In BHI, two of three R. equi isogenic isolates carrying the erm(46) gene conferring macrolide resistance showed a slight reduction in maximum in vitro growth, in contrast with the clinical dual-resistant isolates, which did not demonstrate a reduction in maximum growth compared to that of susceptible R. equi strains. However, in MM and MM-I, the isogenic macrolide-resistant strains did not demonstrate a reduction in maximum growth, suggesting that, when nutrients are readily available, the presence of erm(46) alone may have a negative effect on bacterial fitness that is absent under nutrient constraints (Fig. 3 and 4). In addition, bacterial fitness could be impacted by differences in chromosomal backgrounds or additional acquired plasmids in different strains. R. equi carries a virulence plasmid that is required for replication in macrophages and disease presentation in foals (20, 23–25). On this plasmid, the virulence-associated protein A (vapA) gene may interfere with normal phagosomal acidification, creating an environment conducive to intracellular survival and replication (26–29). Two additional genes housed on the plasmid, virR and virS, regulate vapA expression under particular environmental cues, such as changes in temperature and pH (30, 31). VirR and VirS proteins also influence the expression of ∼18% of chromosomal genes, impacting the cell’s regulatory network to permit adaption to the macrophage environment (32). It is of relevance to note that all strains of each isogenic set carry the virulence plasmid, and the interactions between the virulence plasmid and chromosomal regulation might contribute to differences observed in in vitro growth of macrolide-resistant R. equi isolates in BHI compared to MM. Additionally, while the specific location of the ICE erm(46) in R. equi is not described, it could be housed within a transposon that also possesses unknown genes influencing bacterial fitness under different environmental conditions. Our laboratory previously demonstrated that the erm(46) gene is transferrable to susceptible R. equi isolates, providing macrolide resistance (12). If the ICE erm(46) gene is transferrable to other bacterial species, R. equi could serve as an environmental reservoir for macrolide resistance.
When analyzing the effect of the rpoB mutations, conferring rifampin resistance, on in vitro growth of R. equi, there was a difference in growth. Since rifampin targets bacterial transcription, an essential function, it is not surprising that mutations in the evolutionarily conserved rpoB gene would alter transcription efficiency, elongation, and/or termination, and therefore impact the fitness of the organism. As the rpoB mutations varied among the three isogenic sets (Fig. 6), it was of interest to note that not all rpoB mutations affected R. equi fitness to the same extent. This phenomenon has been previously observed in Escherichia coli and Mycobacterium tuberculosis (14–16, 33, 34). This finding implies that specific rifampin-resistance mutations might differentially affect intracellular growth of the organism in the low-iron environment of macrophages. It is important to note, though, that while all the rpoB mutations in the isogenic sets in this study were different, they were still all located in the same codon. Effects on in vitro growth might vary if a mutation occurs at a different amino acid position. Interestingly, the results observed with the rifampin-resistant R. equi strains were mimicked by the dual-resistant R. equi strains, demonstrating that the observed phenotype was a result of the rpoB mutations and not of the presence of the ICE element carrying erm(46).
Characterization of rpoB mutations conferring rifampin resistance in each isogenic set of R. equi strains (1, 2, and 3). (a) Base pair sequencing at amino acid position 526, shaded in gray, for the wild-type susceptible R. equi sequence compared to the point mutations that confer rifampin resistance in the 3 strains (1R, 2R, and 3R). (b) Amino acid change resulting in the rpoB mutation for each isogenic set. Each rifampin-resistant strain in each isogenic set contains a different rpoB mutation.
Upon subculturing in the absence of antimicrobials, R. equi resistance to macrolides or rifampin could be lost. All of the macrolide-resistant R. equi strains became susceptible at a similar frequency over time, and only one of three rpoB mutations was retained in rifampin-resistant strains. These data suggest that resistant R. equi strains might revert back to a susceptible variant in the absence of selective pressure, but also that certain rpoB mutations may be more stable than others. It would be of interest to determine if macrolide or rifampin resistance was similarly lost in MM and MM-I, and, if resistance were lost under nutrient restriction, whether it would occur at the same rate as that observed in BHI. Additionally, future studies are needed to determine if clinical dual-resistant R. equi strains behave similarly in subculture.
As R. equi is an environmental bacterium, determination of whether resistance to macrolides, rifampin, or both affects bacterial survival in the soil is important. Our data demonstrate that macrolide- and/or rifampin-resistant R. equi strains exhibit similar fitness in the soil at cooler temperatures (−20°C and 4°C). However, when there is an increase in soil temperature, the susceptible R. equi isolate is more fit than the macrolide- or rifampin-resistant or dual-resistant R. equi strains. This finding suggests that in the absence of antibiotic pressure in the host, resistant R. equi isolates would likely be outcompeted by the susceptible R. equi strains. It would be intriguing to repeat this experiment in nonsterile soil to elucidate interactions with other soil components that also impact the growth of susceptible and resistant bacteria.
Overall, these results demonstrated that resistance to macrolides, rifampin, or both reduced the fitness of R. equi isolates under the conditions examined. Thus, reducing antimicrobial treatment of subclinically affected foals might permit susceptible R. equi strains to outcompete resistant strains within the soil. This decrease in environmental antimicrobial resistance would likely improve outcomes for R. equi-infected foals and benefit human health.
MATERIALS AND METHODS
Bacterial strains, culture media, and growth conditions.Bacterial strains used in this work are listed in Table 2. R. equi strains were grown at 37°C in brain heart infusion (BHI) broth, Mueller-Hinton (MH) broth (Becton, Dickinson, Sparks, MD), minimal medium (MM) broth, or minimal medium without iron (MM-I) broth. The MM contained K2HPO4 (5 g), NaH2PO4·H2O (1.5 g), MgSO4·7H2O (0.2 g), (NH4)2SO4 (1.0 g), thiamine (0.01 mM), lactic acid (1.5 g), and trace element solution (200 μl), brought to a final volume of 1 liter with double-distilled water (ddH2O) and filter sterilized with a 0.22-μg filter. Trace elements included EDTA (5 g), ZnSO4·7H2O (2.2 g), CaCl2·2H2O (0.643 g), MnCl2·4H2O (0.506 g), FeSO4·7H2O (0.499 g), (NH4)6Mo7O24·4H2O (0.110 g), CuSO4·5H2O (0.157 g), and CoCl2·6H2O (0.116 g). The pH was adjusted to 6.0 and the final volume was adjusted to 100 ml with ddH2O. Preparation of MM-I was identical, except the trace elements lacked FeSO4·7H2O. For agar plating, technical agar (15 g/liter) was added.
R. equi bacterial strains used in this studya
Nalidixic acid-novobiocin-actidione-tellurite (NANAT) agar was used for the selective isolation of R. equi modified from the procedure described by Woolcock and colleagues (35). Specifically, NANAT medium consisted of 800 ml of ddH2O and 100 ml of store-bought beef broth, peptone from Glycine max (soybean) type IV powder (20 g), NaCl (5 g), yeast extract (5 g), d-(+)-glucose anhydrous (5 g), Na2O4S2 (0.2 g), Na2S2O3·5H2O (1.2 g), K2HPO4 (2 g), and NaHCO3 (2 g). After autoclaving and allowing the medium to cool to 55°C, novobiocin sodium salt (≥90% high-performance liquid chromatography [HPLC], 25 mg/ml), 3.5% potassium tellurite solution, nalidixic acid sodium salt (20 mg/ml), and amphotericin B solution (10 mg/ml) were added. When necessary for study conditions, antibiotics were added at the following concentrations: erythromycin (ERY), 8 μg/ml; rifampin, 25 μg/ml; and zeocin, 25 μg/ml.
Antimicrobial susceptibility testing.Inocula were prepared from overnight cultures in MH broth by the direct colony suspension method according to the guidelines established by the Clinical and Laboratory Standards Institute (CLSI) (36). The MIC of erythromycin was determined using Etest strips (bioMérieux, Durham, NC) according to the manufacturer's recommendations at concentrations between 0.016 and 256 mg/liter. The MIC of rifampin was determined by the broth macrodilution method in glass tubes in accordance with the guidelines established by the CLSI (36) at 2-fold dilutions between 0.03 and 256 mg/liter. Control strains tested in parallel and on each test occasion for both methods were Staphylococcus aureus ATCC 29213 and Enterococcus faecalis ATCC 29212. MICs for the isogenic sets can be found in Table S1.
Creation of rifampin-resistant R. equi strains.Rifampin resistance, conferred by a rpoB mutation, was induced by growing wild-type R. equi on a BHI plate at 37°C for 24 to 48 h, followed by resuspension of the culture into 5 ml phosphate-buffered saline (PBS). Approximately 1 × 109 R. equi CFU/ml were plated onto BHI agar plates containing 25 μg/ml of rifampin, followed by incubation at 30°C for 72 h. After incubation, the resulting R. equi colonies were single-colony purified on BHI plus 25μg/ml rifampin. Identification of the rpoB gene was achieved by sending the PCR product from primer pair rpoB-F and rpoB-R (Table 2) for Sanger sequencing (Eurofin Genomics, Louisville, KY). Each resultant mutant underwent antimicrobial susceptibility testing to confirm rifampin resistance. Three rifampin-resistant R. equi strains were created; the rpoB mutation position and resultant amino acid change for each strain are illustrated in Fig. 6.
Creation of transconjugant isolates.Mating was performed as described by Tripathi et al. (37). R. equi strain 103S erm(46) Zeor, possessing the mobile element erm(46) and chromosomally marked with the Streptoalloteichus hindustanus ble gene (conferring resistance to zeocin), was used as the donor (38). For transfer of erm(46), equal numbers of six recipient R. equi isolates (isolates 1, 1R, 2, 2R, 3, or 3R) and the zeocin-marked donor strain were used (Table 2). The donor and recipient strains were grown overnight at 37˚C in BHI broth supplemented with the appropriate antibiotic. The next day, the optical density at 600nm (OD600) was adjusted to 1.0 (∼2 × 108 CFU/ml). Approximately 107 CFU/ml of both donor and recipient bacteria were mixed in a small volume (5 to 10 μl) of BHI broth, spotted on BHI agar, and incubated for 72 h at 30°C. Afterwards, the cell mixture was scraped from the plates and resuspended in 1 ml PBS. Serial dilutions (up to 10−6) of the resuspended cells were plated on BHI agar containing erythromycin. Putative transconjugant colonies were screened for the presence of the transferred mobile element by picking and patching in the presence and absence of zeocin, followed by PCR analysis to confirm the absence of zeocin resistance (primers Zeo-F/Zeo-R, Table 3) and presence of erm(46) (Erm46-F/Erm46-R, Table 3) and vapA (VapA-F/VapA-R, Table 3).
Primers utilized for R. equi strain confirmation and sequencing
In vitro growth analysis.From a collection of frozen R. equi isolates obtained from transtracheal aspirates from foals diagnosed with R. equi pneumonia, 60 unique clinical isolates were selected. A total of 30 isolates were classified as susceptible to both macrolides and rifampin (MICs, ≤2 mg/liter and ≤1 mg/liter, respectively) and 30 were classified as resistant to both macrolides and rifampin (MICs, ≥8 mg/liter and ≥4 mg/liter, respectively) based on MIC as determined by CLSI guidelines. In addition, three of the susceptible isolates were chosen as the parent wild-type strains for the development of the isogenic sets described above.
Overnight cultures of these clinical isolates and the three isogenic sets were grown in BHI, MM, or MM-I and then diluted in fresh medium to an OD600 of approximately 0.05. Triplicate 400-μl aliquots of the bacterial suspensions were transferred to a flat-bottomed 48-well plate (Corning; Corning NY) and incubated at 37°C with shaking. Bacterial growth was monitored by measuring the OD600 every 30 min in an automated plate reader. Three independent experiments were performed in each medium type.
Soil plot.Soil samples were obtained from a horse farm in Athens, GA, and sterilized by autoclaving for 30 min at 121°C. One gram of sterile soil was aliquoted into 5-ml Eppendorf tubes, contaminated with 1 × 108 organisms from susceptible, macrolide-resistant, rifampin-resistant, or dual-resistant R. equi strains from each isogenic set in triplicate, and mixed for 10 min by shaking at four different temperatures (−20°C, 4°C, 25°C, and 37°C). Aliquots from each strain/soil sample were then collected at time 0, day 14, and at 2, 6 and 12 months, suspended with 1.5 ml PBS, serially diluted, and plated onto NANAT agar containing appropriate antibiotics to determine the CFU/ml. At time 0, 6, and 12 months, two R. equi colonies per triplicate from each condition underwent PCR analysis using primer pairs for vapA and erm(46) (Table 3), as well as Sanger sequencing of the rpoB gene.
Bacterial subculture analysis.To examine the stability of in vitro antimicrobial resistance, macrolide- or rifampin-resistant R. equi strains from the isogenic sets were inoculated into antibiotic-free MH broth and grown at 37°C. The broth was subcultured every 2 days in fresh MH broth at a 1:400 dilution for a total of 90 passages. At passages 10, 20, 30, 50, 70, and 90, the cultures were serially diluted and plated on MH and MH-plus-ERY (for macrolide-resistant strains) or MH-plus-RIF (for rifampin-resistant strains) media. The total numbers of CFU/ml on each type of plate were determined. Sequentially, 100 colonies from the antibiotic-free MH plates underwent picking and patching onto MH in the presence and absence of the respective antibiotic to determine the percentage of antimicrobial-resistant colonies at each passage. The presence of erm(46) and the rpoB mutation was confirmed by PCR analysis and Sanger sequencing of five randomly selected R. equi colonies from passages 10 and 90; additionally, the MIC of 10 randomly selected R. equi colonies for each strain was determined.
Statistical analysis.The maximum growth rate during exponential growth (µ) and maximum bacterial cell density reached during the growth curve (A) were estimated from fits of the OD600 values using the R package growthrates. In vitro growth rates or maximal cell densities were compared by one-way analysis of variance (ANOVA) from three independent experiments. R. equi soil data were analyzed using a linear mixed-effects model with resistance status and time modeled as fixed effects and bacterial isolate modeled as a random effect to account for repeated measurements. When appropriate, pairwise comparisons were performed using the Šidák procedure. All tests assumed a two-sided alternative hypothesis, and P values of <0.05 were considered significant. Statistical analyses were performed using the commercially available statistical software Stata/SE 14 (StataCorp LLC, College Station, TX).
ACKNOWLEDGMENT
We thank Bradley Voss for his assistance with data analysis.
FOOTNOTES
- Received 3 November 2018.
- Accepted 16 January 2019.
- Accepted manuscript posted online 25 January 2019.
Supplemental material for this article may be found at https://doi.org/10.1128/AEM.02665-18.
- Copyright © 2019 American Society for Microbiology.
