Antimicrobial Resistance in Rhodococcus equi

ABSTRACT

Pneumonia caused by Rhodococcus equi remains an important cause of disease and death in foals. The combination of a macrolide (erythromycin, azithromycin, or clarithromycin) with rifampin has been the recommended treatment for foals with clinical signs of infection caused by R. equi since the early 1980s with, until recently, only rare reports of resistance. Resistance to macrolides and rifampin in isolates of R. equi cultured from horses is increasing, with isolates resistant to all macrolides and rifampin now being cultured from up to 40% of infected foals at some farms. This text reviews the available data regarding antimicrobial resistance in R. equi, with emphasis on the molecular mechanisms of the recent emergence of resistance to macrolides and rifampin in equine isolates of R. equi.

INTRODUCTION

Rhodococcus equi, a Gram-positive facultative intracellular pathogen, is one of the most important causes of disease in foals between 3 weeks and 5 months of age. R. equi has also emerged as a common opportunistic pathogen in immunocompromised people, especially those infected with the human immunodeficiency virus (1–3). Infection in both foals and people is most commonly characterized by life-threatening pyogranulomatous pneumonia, but extrapulmonary infections are also common (3, 4). In foals, extrapulmonary disorders might occur concurrent with or independent of pneumonia, and some foals have multiple extrapulmonary disorders concurrently (5, 6). Because ultrasonographic screening for early detection has become routine practice at many farms endemic for pneumonia caused by R. equi, the most frequently recognized form of R. equi infection on those farms is a subclinical form in which foals develop sonographic evidence of peripheral pulmonary consolidation or abscessation without manifesting clinical signs (7, 8).

R. equi is commonly cultured from the submaxillary lymph nodes of pigs with granulomatous lymphadenitis. However, the causative role of R. equi in these lesions is unclear, because it can be isolated from 3 to 5% of apparently healthy pigs, and experimental infection studies have failed to reproduce granulomatous lymphadenitis (9–11). R. equi is also occasionally isolated from abscesses or granulomas in the lymph nodes of cattle (12). It has been cultured from rare cases of pulmonary or extrapulmonary infections in cattle, goats, camelids, dogs, and cats (13–17).

Since the 1980s the standard treatment recommendation for foals infected with R. equi has been the combination of a macrolide (erythromycin initially and, more recently, clarithromycin or azithromycin) and rifampin with, until recently, very few documented instances of resistance to these drugs. This text reviews the available data regarding antimicrobial resistance in R. equi, with emphasis on the molecular mechanisms of the recent emergence of resistance to macrolides and rifampin in equine isolates of R. equi.

IN VITRO ACTIVITY

In vitro Susceptibility Testing of R. equi

Antimicrobial susceptibility testing can be done using a variety of methods, with broth dilution, disk diffusion, and concentration gradient test (Etest) being commonly used by veterinary diagnostic laboratories. The decision regarding which method to use is based on cost, ease of use, and flexibility to meet the needs of the laboratory. The methods all assess inhibition of growth rather than killing of the bacterium as the endpoint. In vitro susceptibility tests must be performed using standardized procedures to provide valid and reproducible results. Although numerous studies have investigated the in vitro activity of a variety of antimicrobial drugs against R. equi over the years, different methods, media, and incubation conditions have been used. A protocol for antimicrobial susceptibility testing of R. equi by broth microdilution was recently adopted by the Veterinary Antimicrobial Susceptibility Testing subcommittee of the Clinical and Laboratory Standards Institute (CLSI) and will be included in the upcoming CLSI document (18). The approved protocol recommends a standard inoculum of 1 × 10⁵ CFU/ml, an incubation temperature of 35 ± 2°C, the use of cation-adjusted Mueller-Hinton broth supplemented with 2% (vol/vol) lysed horse blood, and reading of the results after 24 h (19). This protocol was recently evaluated in 18 laboratories. All 18 participating laboratories were able to perform broth microdilution according to guidelines established by the CLSI, with over 98% of all MIC determinations for the Escherichia coli quality control strain being within the acceptable range (18). There was more interlaboratory variability for R. equi, with 72 to 100% of all MICs being within the acceptable range (modal MIC ± 1 dilution), with only one laboratory being below 80% (18).

The results generated by in vitro susceptibility tests, whether they are determined by disk diffusion, concentration gradient, or dilution methodologies, are usually presented to the clinician by designating the pathogen as susceptible, intermediate, or resistant. Despite adopting a standardized protocol for antimicrobial susceptibility testing, the CLSI has not established interpretive criteria for the classification of isolates of R. equi as susceptible, intermediate, or resistant. Therefore, interpretive criteria reported by diagnostic laboratories are extrapolated from criteria established for other bacterial agents and, for most drugs, based on therapeutic concentrations achievable in people or other animal species. The results of broth macrodilution, disk diffusion, and Etest for in vitro susceptibility testing of macrolide-susceptible and macrolide-resistant isolates of R. equi were compared recently. Categorical agreement between methods ranged between 85.1 and 100% depending on the drug tested (20). Overall, the agreement between Etest and disk diffusion was better than the agreement between broth macrodilution and the agar-based methods (20). Etest tended to overestimate MICs relative to broth macrodilution for clarithromycin and gentamicin and underestimate MICs for erythromycin and doxycycline (20).

MIC and MBC of Drugs Used Against R. equi

Drugs with the highest in vitro activity against R. equi based on MIC are clarithromycin, rifampin, imipenem, telithromycin, erythromycin, gentamicin, apramycin, vancomycin, azithromycin, gamithromycin, doxycycline, enrofloxacin, and linezolid (Table 1). Trimethoprim-sulfonamide combinations are also active in vitro against the majority of isolates. Not all macrolides are equally active, with tildipirosin, tilmicosin, tulathromycin, spiramycin, and tylosin being poorly active against R. equi in vitro. Most drugs including macrolides and rifampin exert only bacteriostatic activity against R. equi, with only amikacin, gentamicin, enrofloxacin, and vancomycin being bactericidal (21). Combinations including a macrolide (erythromycin, clarithromycin, or azithromycin) and either rifampin or doxycycline, and the combination doxycycline-rifampin are highly synergistic against R. equi in vitro (21–23). In contrast, combinations containing amikacin and erythromycin, clarithromycin, azithromycin, or rifampin and the combination gentamicin-rifampin are antagonistic in vitro (21–23). However, the clinical significance of these in vitro findings has not been established.

TABLE 1.

MIC data for R. equi isolates susceptible to macrolides and rifampin^a

Antimicrobial classes/agents	n	MIC (µg/ml)
		MIC90b	MIC50c	Range
Macrolides/azalides/ketolides
Azithromycin	264	1	1	0.06–4
Clarithromycin	264	0.06	0.06	0.008–0.5
Erythromycin	263	0.5	0.5	0.06–1
Gamithromycin	30	1	1	0.5–1
Spiramycin	200	32	32	2–64
Telithromycin	25	0.25	0.25	0.25–0.5
Tildipirosin	40	32	16	8–32
Tilmicosin	278	64	32	0.5–>64
Tulathromycin	278	>64	64	8–>64
Tylosin	200	32	32	1–64
Rifamycins
Rifampin	194	0.12	0.06	0.008–0.25
β-lactams
Ampicillin	264	8	4	0.06–8
Amoxicillin-clavulanic acidd	264	8	4	0.06–16
Penicillin G	200	4	4	0.03–8
Cefazolin	64	16	≤2	≤2–>16
Cefoperazone	200	32	16	2–64
Cefotaxime	200	8	8	0.12–16
Ceftazidime	64	> 32	> 32	≤0.25–>32
Ceftiofur	278	16	8	0.25–16
Cefquinome	200	4	2	0.12–8
Imipenem	200	0.12	0.12	0.015–0.25
Aminoglycosides
Amikacin	64	4	≤2	≤2–8
Apramycin	200	0.5	0.25	0.12–8
Gentamicin	264	0.5	0.5	0.12–1
Quinolones
Enrofloxacin	264	1	1	0.25–4
Nalidixic acid	200	128	128	16–256
Tetracycline
Tetracycline	264	8	4	0.5–16
Doxycycline	278	1	1	0.25–2
Other classes
Chloramphenicol	342	16	8	≤4–32
Clindamycin	264	8	2	0.8–8
Florfenicol	200	16	16	4–32
Linezolid	78	1	1	0.5–2
Quinupristin/dalfopristin	200	18	8	0.5–16
Trimethoprim/sulfamethoxazolee	342	1	0.5	0.06–>4
Vancomycin	278	0.5	0.5	0.12–1

^a Adapted from references 27, 43, 49, 57.

^b MIC that inhibits at least 90% of the isolates tested.

^c MIC that inhibits at least 50% of the isolates tested.

^d Expressed as MIC of amoxicillin.

^e Expressed as MIC of trimethoprim.

Mutant Prevention Concentration (MPC)

Given the recent emergence of resistance to macrolides and rifampin among isolates of R. equi (see below), the relative propensity of currently available antimicrobial agents to selectively enrich for resistant mutant subpopulations among R. equi isolates has been studied. A common way to compare drugs for selective enrichment of resistant mutants is based on measurement of the MPC, which is defined as the drug concentration that prevents selective enrichment of first-step resistant mutants within a large susceptible bacterial population. The range of concentrations between the MIC and the MPC is known as the mutant selection window, which represents the danger zone for emergence of resistant mutants (24). Minimizing the length of time that the drug concentrations remain in the mutant selection window may reduce the likelihood of development of resistance during therapy. Of 10 antimicrobial agents studied (erythromycin, clarithromycin, azithromycin, rifampin, amikacin, gentamicin, enrofloxacin, vancomycin, imipenem, and doxycycline), rifampin had the highest MPC, indicating that rifampin monotherapy is likely to select for resistance (25). However, combining rifampin with erythromycin, clarithromycin, or azithromycin resulted in a profound and significant decrease in MPC (25). The MPC was well above clinically achievable plasma concentrations for most antimicrobial drugs studied (25).

Activity of Drugs Against Intracellular R. equi

The ability of R. equi to survive and replicate in macrophages is the basis of its pathogenicity, and strains unable to replicate intracellularly are avirulent for foals (26). Many drugs active against R. equi in vitro have been hypothesized to be ineffective in vivo because of poor cellular uptake and resulting low intracellular concentrations. Studies with facultative intracellular bacterial pathogens have shown that evaluation of the bactericidal activity of antimicrobial agents against intracellular bacteria is more closely associated with in vivo efficacy than with traditional in vitro susceptibility testing. In one study, clarithromycin was more active than azithromycin, erythromycin, and gamithromycin against intracellular R. equi (27). More recently, equine monocyte-derived macrophages were infected with virulent R. equi and exposed to erythromycin, clarithromycin, azithromycin, rifampin, ceftiofur, gentamicin, enrofloxacin, vancomycin, imipenem, or doxycycline at concentrations achievable in plasma at clinically recommended dosages in foals. Enrofloxacin, gentamicin, and vancomycin were significantly more active than other drugs against intracellular R. equi, whereas doxycycline was the least active drug (21).

EMERGENCE OF RESISTANCE TO MACROLIDES AND RIFAMPIN IN ISOLATES FROM HORSES

The combination of a macrolide such as erythromycin, clarithromycin, or azithromycin with rifampicin has been the mainstay of therapy in foals infected with R. equi since the early 1980s (28–30), with only one report of macrolide resistance in foals before 1999 (31). During the same period, reports of rifampin resistance were rare and typically associated with the use of the drug in monotherapy (31–33). Over the past 15 years, however, the incidence of macrolide-resistant R. equi from foals has increased considerably, at least in the United States. The overall prevalence of macrolide and rifampin-resistant isolates in Texas and Florida between 1997 and 2008 was 4%, with most resistant isolates being identified after 2001 (34). In the same study, the odds of death were approximately 7 times higher in foals infected with resistant isolates (34). More recently, it has been documented that mass antimicrobial treatment of subclinically affected foals has selected for antimicrobial resistance over time, with isolates of R. equi resistant to all macrolides and rifampin now being cultured from the environment and from up to 40% of pneumonic foals at one farm (35). While there was considerable chromosomal heterogeneity among susceptible R. equi isolates, macrolide- and rifampin-resistant isolates from the farm were all closely related and formed two distinct genotypic clusters (35). To date, isolates of R. equi resistant to macrolides and rifampin have been identified from at least five U.S. states. However, the true prevalence of macrolide- and/or rifampin-resistant R. equi in the United States and elsewhere is unknown. Reports of macrolide resistance in veterinary isolates of R. equi outside the United States have been extremely rare. There is one report of isolation of a macrolide- and rifampin-resistant strain from a foal in China (36). Resistance to macrolides or rifampin has been reported in 3 to 4% of isolates cultured from people (37).

MOLECULAR BASIS OF DRUG RESISTANCE

Macrolides, Lincosamides, and Streptogramin B

Three main mechanisms account for acquired macrolide resistance in bacteria: rRNA methylation, active efflux, and enzymatic inactivation (38). rRNA methylation and active efflux are the mechanisms responsible for the majority of resistant isolates. Most macrolide-resistance genes are associated with mobile elements and thus have the capacity to spread among strains, species, and bacterial ecosystems. rRNA methylation, encoded by erythromycin-resistant methylase (erm) genes, results in cross-resistance to the macrolides, lincosamides, and streptogramin B. Efflux of macrolide antimicrobial agents is mediated by members of the ATP binding cassette family of proteins or by major facilitator superfamily transporters. These proteins pump antimicrobial agents out of the cell or cellular membrane, thereby allowing the bacterial ribosomes to function again. The third and less common mechanism of resistance is due to enzymatic inactivation of macrolides by bacteria (39). A very small proportion of macrolide-resistant bacteria do not carry any of the known acquired macrolide-resistance genes described above. These isolates typically have mutations in the V domain of the 23S rRNA genes and/or the genes coding the ribosomal proteins L4 and L22 (40).

Macrolide resistance in macrolide-resistant isolates of R. equi cultured in the United States is caused by erm(46), an erythromycin-resistant methylase gene that has been identified only in R. equi to date (41). The erm(46) gene encodes a predicted methyltransferase that targets the 50S subunit of the bacterial ribosome and is most similar (68 to 69% nucleotide sequence identity) to the mycobacterial rRNA methyltransferases encoded by erm(38), erm(39), and erm(40). The erm(46) gene is flanked upstream by an open reading frame encoding a putative AAA-family P-loop ATPase/nucleotide kinase domain, and downstream by an open reading frame encoding a putative integrase/Tra5-like transposase with closely related homologs in other Actinobacteria. Erm(46) and its mycobacterial counterparts form a distinct branch within an Erm subclade populated by enzymes from various Actinobacteria genera including Streptomyces, Corynebacterium, and Micrococcus (41), suggesting that the R. equi erm(46) gene has an actinobacterial origin and was likely horizontally acquired via a transposable mobile element.

There was complete agreement between the macrolide resistance phenotype and detection of the erm(46) gene, with 100% of resistant (n = 62) and susceptible (n = 62) isolates testing positive and negative, respectively (41). Expression of erm(46) in the macrolide-susceptible strain 103⁺ conferred high-level (>256 µg/ml) resistance to all macrolides tested (azithromycin, clarithromycin, erythromycin, gamithromycin, tildipirosin), lincosamides, and streptogramin B (41). Expression of the gene conferred a lower level of resistance (8 to 16 µg/ml) to ketolides and to the combination of quinupristin and dalfopristin. Expression of the gene did not confer resistance to aminoglycosides, tetracyclines, glycopeptides, β-lactams, fluoroquinolones, and rifampin. Mating experiments confirmed horizontal transfer of erm(46) from resistant to susceptible strains of R. equi, with transfer frequencies ranging from 3 × 10^–3 to 1 × 10^–2 (41). To the authors’ knowledge, the presence of erm(46) in R. equi isolates resistant to macrolides cultured outside the United States has not been examined. Macrolide resistance in an isolate of R. equi cultured from a foal in China was associated with an A to G mutation at position 2063 in domain V of the 23S rRNA gene (36). It is unknown if the isolate was also carrying erm(46), given that the report was published before the discovery of the gene.

The vast majority of macrolide-resistant R. equi isolates identified so far are also resistant to rifampin. There are no known mechanisms of cross-resistance between macrolides and rifampin. Rifampin resistance in R. equi has been shown to be the result of mutations in the rpoB gene (see below). Selection pressure caused by the combined use of rifampin with a macrolide for the treatment and prevention of R. equi infection at endemic horse breeding farms might have coselected for the acquisition of erm(46) alongside an rpoB mutation in specific strains.

Rifampin

Rifampin resistance in several bacterial genera typically results from the substitution of a limited number of highly conserved amino acids in the RNA polymerase β subunit encoded by the rpoB gene. This gene is divided into three regions: clusters I, II, and III. In Mycobacterium spp., the vast majority of substitutions conferring rifampin resistance are found in cluster I within an 81-bp resistance-determining region corresponding to codons 507 to 533 (E. coli numbering) of the rpoB gene. Most substitutions associated with rifampin resistance detected to date in R. equi are single substitutions within the same resistance-determining region, with mutations at codons 526 or 531 being the most commonly reported (Table 2) (32, 42, 43). The level of antimicrobial resistance has been reported to be dependent on both the location and the nature of the base substitution. However, different isolates with the same substitution have been found occasionally to have considerably different MICs (43). Rare isolates of R. equi resistant to rifampin do not have mutations in the 81-bp resistance-determining region (43). It is possible that such isolates have mutations in other regions of the rpoB gene, as reported in a small percentage of rifampin-resistant strains of Mycobacterium tuberculosis. Alternatively, resistance in these isolates of R. equi might be caused by one or more mechanism(s). Several rifampin-inactivating enzymes have been identified in environmental and pathogenic Gram-positive bacteria such as Bacillus spp., Nocardia spp., Mycobacterium smegmatis, and Listeria monocytogenes (44–46). A gene encoding a monooxigenase-like protein was identified in one strain of R. equi and shown to confer low-level resistance to rifampin by inactivating the drug (47). The role played by rifampin-inactivating enzymes in rifampin resistance in foals in a clinical setting is unknown.

TABLE 2.

Mutations in the rpoB gene associated with rifampin resistance in R. equi^a

Originb	Country	Rifampin MIC (µg/ml)	Substitution
			Codonc	Amino acid exchange
Human	Thailand	8	509	Ser → Pro
Horse	United States	64	513	Gln → Leu
Horse	Germany, France	2–4	516	Asp → Val
Horse	France, United States	128–256	526	His → Asp
Horse	France	8	526	His → Asn
Human, in vitro	Thailand, NAd	≥256	526	His → Tyr
In vitro	NA	≥256	526	His → Arg
Horse	France, United States	8–128	531	Ser → Leu
Horse	United States	≥256	531	Ser → Phe
Human	Thailand	64	531	Ser → Trp

^a Adapted from references 32, 42, 43.

^b Species of origin or in vitro mutations generated in the laboratory.

^c E. coli numbering.

^d NA: not applicable.

Fluoroquinolones

Fluoroquinolones are rarely used to treat infections caused by R. equi in foals because of the risk of arthropathy (48). R. equi is highly resistant to first-generation quinolones such as nalidixic acid. However, resistance to higher-generation fluoroquinolones in clinical use is rare in equine isolates of R. equi, with less than 5% of isolates being classified as resistant to enrofloxacin or ciprofloxacin using CLSI breakpoints for other bacterial agents (MIC ≥ 4 µg/ml as resistant), and with most isolates classified as being of intermediate susceptibility (MIC 1 to 2 µg/ml) to enrofloxacin and susceptible to ciprofloxacin (43, 49, 50). Resistance to ciprofloxacin appears to be more common in isolates cultured from people, with approximately 85% of isolates classified as susceptible (37, 51). Bacterial resistance to fluoroquinolones can be the result of chromosomal mutations coding for modifications in target subunits of bacterial topoisomerases II and IV, by active efflux, or by alterations in the expression of outer membrane proteins.

The mechanisms of fluoroquinolone resistance have not been studied in clinical isolates of R. equi. However, in vitro selection for resistance to ciprofloxacin in R. equi is typically associated with one of seven identified single amino acid substitutions in the quinolone resistance-determining region of DNA gyrase subunit A (52, 53). Mutants with amino acid substitutions at Ser-83 of GyrA were particularly resistant (MIC > 64 µg/ml) (53). In the same study, a single amino acid in the quinolone resistance-determining region of gyrB was identified and associated with a lower level of resistance (MIC=4µg/ml) (53). Some ciprofloxacin-resistant mutants did not have substitutions in the quinolone resistance-determining region of gyrA or gyrB, indicating that other fluoroquinolone resistance mechanisms such as efflux pumps or alteration in membrane permeability might also occur. The MPC for enrofloxacin and ciprofloxacin ranged between 32 and 64 µg/ml, which is well above concentrations achievable in vivo, indicating that monotherapy with fluoroquinolones might result in emergence of resistant mutants (25, 52).

Other Antimicrobial Agents

Resistance to aminoglycosides in clinical use such as gentamicin and amikacin and to glycopeptides such as vancomycin is extremely rare (20, 37, 43, 49). In one study, an inducible glycopeptide-resistance operon (vanO) was described in a single isolate of R. equi from soil (54). The vanO operon had unique gene organization compared to the vanA operon in enterococci and displayed structural similarities to putative operon-like clusters detected in actinomycetes (54). The vanO operon is located on the chromosome, and attempts at transferring vancomycin resistance by conjugation or transformation were not successful (54). The molecular mechanisms of antimicrobial resistance to other antimicrobial agents in R. equi have not been studied. The genome of the R. equi reference strain 103S contains an array of putative antimicrobial-resistance determinants, including a variety of antibiotic-inactivation enzymes, β-lactamases, and multidrug efflux systems (55). However, the functionality of these putative determinants has not been confirmed. Isolates of R. equi have been reported to be resistant to a variety of antimicrobial agents and drug classes. These findings must be interpreted in the context of the lack of CLSI-approved breakpoints for R. equi. Therefore, breakpoints for the classification of isolates of R. equi as susceptible, intermediate, or resistant are CLSI breakpoints for other species, and bacterial agents used have not been consistent between studies.

R. equi isolates are typically resistant or of intermediate susceptibility to most β-lactam antimicrobial agents, with the exception of the carbapenems such as imipenem or meropenem (Table 1). Resistance to carbapenems is rare, and low-level imipenem resistance has been associated with an altered penicillin-binding protein pattern (56). Most isolates of R. equi are of intermediate susceptibility to tetracyclines and chloramphenicol, and resistance is not uncommon. However, resistance to doxycycline is extremely rare (<1% of isolates) (20, 43, 49). Most isolates of R. equi are susceptible to combinations of trimethoprim-sulfonamide, but resistance is not uncommon.

CONCLUSIONS

Our understanding of the molecular mechanisms of macrolide and rifampin resistance in R. equi has improved considerably in the past few years. However, further work is needed to assess the overall prevalence of macrolide and rifampin resistance in isolates of R. equi. Resistance to other antimicrobial agents such as tetracyclines, chloramphenicol, trimethoprim-sulfa, and fluoroquinolones is not uncommon, and more work is needed to characterize the molecular mechanisms of resistance to these drugs.