Molecular and infection biology of the horse pathogen Rhodococcus equi

Authors


  • Editor: Neil Fairweather

Correspondence: Kristine von Bargen, Cell Biology Institute, University of Bonn, Ulrich-Haberland Street 61a, 53121 Bonn, Germany. Tel./fax: +0049 228 73 6340; e-mail: bargen@uni-bonn.de

Abstract

The soil actinomycete Rhodococcus equi is a pulmonary pathogen of young horses and AIDS patients. As a facultative intracellular bacterium, R. equi survives and multiplies in macrophages and establishes its specific niche inside the host cell. Recent research into chromosomal virulence factors and into the role of virulence plasmids in infection and host tropism has presented novel aspects of R. equi infection biology and pathogenicity. This review will focus on new findings in R. equi biology, the trafficking of R. equi-containing vacuoles inside host cells, factors involved in virulence and host resistance and on host–pathogen interaction on organismal and cellular levels.

Introduction

The soil bacterium Rhodococcus equi has long been known as a pulmonary pathogen of foals and, more recently, of AIDS patients. Similar to other pathogenic members of the actinomycetales such as Mycobacterium tuberculosis or Nocardia asteroides, R. equi is an intracellular pathogen that establishes a specific niche inside the host cell by arresting the normal pathway of phagosome maturation.

This review compiles recent findings on R. equi with a focus on infection biology, pathogenicity and host–pathogen interaction. Factors known to be involved in virulence (pathogen) or resistance to R. equi infection (host), respectively, will be discussed and molecular tools to study their functions will be presented. Because research on immunity to R. equi infections is so extensive to require coverage in a review of its own, it will not be discussed here.

Taxonomy

Rhodococcus– a genus with many talents

The genus Rhodococcus was introduced by Zopf in 1891 and redefined in 1977 by Goodfellow and Alderson based on computer-aided taxonomic analysis (Goodfellow & Alderson, 1977). Rhodococci are Gram-positive, aerobic and nonmotile soil organisms that can be found in various environments from soil and ground water to insect guts and plant surfaces (Bell et al., 1998). Rhodococcus species are metabolically diverse (Larkin et al., 2005) and some are able to metabolize aromatic hydrocarbons or normally toxic chemicals such as polychlorophenols or nitroaromatic compounds, and therefore have a potential in bioremediation. Other strains synthesize commercially valuable products such as acrylamide produced by Rhodococcus rhodochrous isolates or are capable of biotechnologically relevant biotransformations (Bell et al., 1998).

The genus Rhodococcus belongs to the nocardioform actinomycetes (Goodfellow, 1989) and comprises >20 species, including several species that have only recently been reclassified as Rhodococcus and some isolates that probably ought to have an own species status (Gurtler et al., 2004; Rehfuss & Urban, 2005; Yassin, 2005; Ghosh et al., 2006; Li et al., 2007; Xu et al., 2007). The complex mycolic acid-containing cell wall of rhodococci marks them as member of the group of mycolata, which also incorporates the genera Mycobacterium, Nocardia, Corynebacterium, Tsukamurella, Dietzia, Williamsia, Turicella, Gordonia, and Skermania (Gurtler et al., 2004). Rhodococcal mycolic acid acyl chains contain 30–54 carbon units, a size range that is in between that of mycolic acids from corynebacteria (22–38 carbon units) and mycobacteria (60–90 carbon units) (Sutcliffe, 1998). A current model of the Rhodococcus cell envelope (Sutcliffe, 1998) proposes a complex cell wall of peptidoglycan that is, toward the extracellular space, attached to arabinogalactan polysaccharides. Mycolic acids are thought to be ester-linked to the arabinan branches perpendicular to the plane of the plasma membrane, forming a single-layered permeability barrier whose gaps are possibly filled by unbound lipid compounds containing mycolic acids. Structural characteristics of the rhodococcal arabinogalactan suggest a lower density of bound mycolic acids in rhodococci than in mycobacteria (Sutcliffe, 1998). The mycolic acid layer might interact with surface amphiphiles to form an impermeable and asymmetric bilayer, which is often surrounded by capsular material (Fig. 1). Water-filled channels provide a pathway through the permeability barrier for hydrophilic solutes (Riess et al., 2003).

Figure 1.

 Ultrastructural morphology of Rhodococcus equi bacteria in transmission electron microscopy. Virulent, plasmid-positive R. equi 103+ of a late logarithmic culture were fixed, stained with ruthenium red and prepared for transmission electron microscopy as in Sydor et al. (2007). Note the thick layer of polysaccharide capsule material that is surrounding the complex cell wall and stained by the cationic dye. Scale bar=500 nm.

Rhodococcus species include symbionts (Rhodococcus rhodnii), pathogens of animals (e.g. R. equi), plants (Rhodococcus fascians) and humans (e.g. R. equi, R. rhodochrous, Rhodococcus erythropolis and several unidentified Rhodococcus species) (Bell et al., 1998).

Rhodococcus equi

Rhodococcus equi is the Rhodococcus species with the highest pathogenic potential for animals, including humans (Prescott, 1991). It was first described in 1923 by Magnusson as Corynebacterium equi, based on an isolate from the lung of a foal with pyogranulomatous pneumonia (Magnusson, 1923). In 1977, it was reclassified as R. equi (Goodfellow & Alderson, 1977).

Rhodococcus equi strains have a circular chromosome of c. 5 Mb (Sanger Institute, 2008) and can possess different plasmids. Some of these are involved in virulence and will be discussed below. The classification of R. equi as a member of the genus of Rhodococcus has been a matter of discussion (Gurtler et al., 2004); yet analysis of the recently sequenced genome (Sanger Institute, 2008) confirms this taxonomic affiliation in finding the highest similarities to the genome of Rhodococcus sp. RHA1 (Vazquez Boland et al., 2009).

Morphological and biochemical aspects

Rhodococcus equi morphology varies from bacillary to coccoid, depending on the growth conditions: while bacteria are rod shaped after 4 h of growth in culture broth, they become coccoid after a day of growth in liquid media or on blood agar (Fuhrmann et al., 1997). Rhodococcus equi have no flagellae (Prescott, 1991), but some strains have pili or appendages (Yanagawa & Honda, 1976; Nordmann et al., 1994). Rhodococcus equi grow in irregular, smooth and mucoid colonies that develop a shade of salmon pink to yellow after a week of growth (Prescott, 1991). A positive result for acid fastness found in some studies likely depends on the growth conditions and the technique used (Prescott, 1991). Rhodococcus equi are Gram-positive, obligate aerobic bacteria. They are catalase-positive, mostly urease-positive and oxidase-negative and their optimal growth temperature is between 30 and 37 °C (Prescott, 1991). Rhodococcus equi produce soluble ‘equi factor(s)’ associated with a phospholipase and a cholesterol oxidase activity (Linder & Bernheimer, 1982). This factor cooperates with phospholipase D of Corynebacterium pseudotuberculosis, β-toxin of Staphylococcus aureus, or a hemolysin of Listeria monocytogenes to produce complete hemolysis of sheep and cattle erythrocytes (Prescott, 1991). Nutrient requirements are simple, and carbon can be used from many different sources (Prescott, 1991) including simple organic acids such as proprionate or acetate (Hughes & Sulaiman, 1987), which are abundant in herbivore manure (Barton & Hughes, 1984; Prescott, 1987). A predilection of R. equi for lipids as carbon source is also suggested by analysis of the R. equi chromosome, which encodes – just as in mycobacteria – many genes involved in lipid metabolism and no genes for sugar transport (Vazquez Boland et al., 2009).

Rhodococcus equi can cope with some extreme environmental conditions such as low pH and oxidative stress (Benoit et al., 2000, 2002). Their complex hydrophobic cell wall is characterized by mycolic acid-containing lipids and lipoglycans such as trehalose dimycolate and lipoarabinomannan (R. equi LAM or ReqLAM) (Barton et al., 1989; Sutcliffe, 1998; Garton et al., 2002). Rhodococcus equi is surrounded by an antigenically variable, thick and lamellar polysaccharide capsule (Prescott, 1991) (Fig. 1). Twenty-seven different capsule serotypes have been described (Prescott, 1991).

Current genetic tools for R. equi analysis

Rhodococcus equi are routinely transformed by electroporation (Sekizaki et al., 1998). Conjugation with and gene transfer from Escherichia coli S17-1 are possible (Meijer & Prescott, 2004). Chromosome and plasmid DNA of R. equi can be isolated with conventional protocols or a modified alkaline lysis technique (Takai et al., 1991c).

Several E. coliR. equi shuttle vectors have been developed. Most of them are based on origins of replication of mycobacterial plasmids such as pAL5000 or pMF1 and some, for example pREV2 and pREV5, are mutually compatible in R. equi (Giguere et al., 1999; Navas et al., 2001; Jain et al., 2003; Mangan et al., 2005; Sydor et al., 2007). The pRE-7 plasmid (Zheng et al., 1997) is based on the pOTS origin of replication of the R. equi virulence-associated protein A (VapA) encoding virulence-associated plasmid (VAP), and is, therefore, incompatible with this virulence plasmid. pRF30 originated from a plasmid of R. fascians (Sekizaki et al., 1998). An alternative to R. equi mutant complementation using plasmids is the single-site integration of integrase-based vectors into the bacterial chromosome, allowing for a more stable expression of the complementing gene even in the absence of selective pressure (Hong & Hondalus, 2008).

Rhodococcus equi libraries of random mutants for phenotype screening have been created by transforming a Himar1 minitransposon (Ashour & Hondalus, 2003) or complexes of transposon and transposase (transposomes) (Mangan & Meijer, 2001; Sydor et al., 2007). Targeted gene mutation in R. equi has been performed using homologous recombination with suicide plasmids (Navas et al., 2001; Jain et al., 2003; Ren & Prescott, 2004; Wall et al., 2005), some of which allow for selection of single vs. double-crossover events. A recently developed strategy of unmarked in-frame gene deletion is based on counterselection by conversion of 5-fluorocytosine into a cytotoxic compound (van der Geize et al., 2008). Most mutational analyses are carried out with the R. equi strain 103+, because it has a low number of illegitimate recombination events compared with other strains (Ren & Prescott, 2004).

In summary, all tools required for genetic manipulation of R. equi are available, although it should be mentioned that kanamycin resistance – a major marker for actinomycete plasmids – does not work well in R. equi due to a relatively high endogenous resistance level (Tsukamura, 1988).

Infections by R. equi

Rhodococcus equi infection in horses

Foals between 1 and 4 months of age are frequently infected by R. equi (Zink et al., 1986; Yager, 1987) and have been in the center of host–pathogen interaction studies for years. In this animal host, R. equi are the most common cause of lung abscesses (Lavoie et al., 1994). Before rifampin/erythromycin therapy was introduced, mortality of foals with R. equi pneumonia could reach 80%, but was reduced to as little as 12% by the use of these antibiotics in treatment (Hillidge, 1987). Infection with R. equi is believed to occur by inhalation of aerosolized dust contaminated with these bacteria (Muscatello et al., 2007). Transmission of R. equi pneumonia from foal to foal has not been documented; yet, the bacteria can be found in the breath of sick foals (Muscatello et al., 2009), which leaves a possibility of a contagious epidemiology. Virtually all isolates from foals possess a plasmid that contains the gene for the VapA, whereas virulence plasmids are rarely found in isolates from locations that are not associated with horses (Takai et al., 1996c).

Nowadays, the standard case of R. equi infection is a subclinical infection with mild neutrophilic leucocytosis and hyperfibrinogenaemia, which can be associated with abscessation or pulmonary changes (Muscatello et al., 2007; Vazquez Boland et al., 2009), possibly due to a better clinical monitoring of foals and more effective antibiotic treatment regimens. The severe variant of R. equi pneumonia characterized by massive abscessation of the lung with fever, neutrophilia, mucopurulent respiratory discharge and cough (Yager, 1987), described in earlier studies, is rarely observed anymore (Muscatello et al., 2007). Inhaled R. equi are phagocytosed by alveolar macrophages but may not be killed (Hietala & Ardans, 1987; Zink et al., 1987). In the early stage of the disease, pulmonary lesions develop and alveoli fill with neutrophils, macrophages and giant cells. Many of these cells contain intracellular R. equi (Johnson et al., 1983a). As disease progresses, the lung parenchyma becomes necrotic (Johnson et al., 1983a; Zink et al., 1986), and bronchial and mesenteric lymph nodes are affected (Zink et al., 1986; Yager, 1987). In many cases, granulomatous foci in the lung open up, and R. equi spreads through the body and affects other organs (Prescott, 1991). Severe diarrhea with an ulcerative enteritis and mucosal invasion of R. equi is observed frequently, particularly in chronic disease (Cimprich & Rooney, 1977; Zink et al., 1986). This might be due to the ingestion of sputum containing large numbers of bacteria. Accordingly, intestinal lesions can be experimentally induced by oral infection of foals over a prolonged period of time (Johnson et al., 1983b). Osteomyelitis, arthritis and lesions of the liver and the kidneys are observed occasionally (Giguere, 2000). Immune complex deposition in the synovial membranes causes polysynovitis in some foals and might also contribute to the development of uveitis, anemia or thrombocytopenia in infected animals (Giguere, 2000).

Adult horses are rarely infected (Hondalus, 1997). They – as well as the majority of foals – develop a protective immunity and quickly clear infection when re-exposed to virulent R. equi (Lopez et al., 2002). However, although not causing severe disease, R. equi could be detected in a number of adult racehorses with respiratory problems and constraints in their general fitness (Fogarty, 2008). In a small percentage of these horses, abscess formation was detected, which was associated with pyrexia and a degree of inappetence. Infection appeared to coincide with increased environmental contamination/exposure. Reduction of the environmental challenge was often sufficient to resolve the problem whereas horses with abscessation needed and responded to antimicrobial therapy (Fogarty, 2008). Rhodococcus equi has been occasionally associated with equine abortion (Patterson-Kane et al., 2002; Szeredi et al., 2006; Nakamura et al., 2007) and one case of fetal pneumonia in an aborted fetus has been described (Fitzgerald & Yamini, 1995).

Rhodococcus equi infection in other species

Humans: Immunocompetent humans are rarely affected by R. equi (Bell et al., 1998), while a compromised cell-mediated immunity predisposes one to R. equi infection (Prescott, 1991), which is reflected in >200 case reports (Kamboj et al., 2005). The first human case was described in 1967 in a man with plasma cell hepatitis receiving an immunosuppressant therapy (Golub et al., 1967). The increasing frequency of reports on human infection with R. equi since then might reflect an actual increase in cases or an increase in awareness of this opportunistic pathogen, which might have often been dismissed as a contaminant in bacteriological examinations, or the syndromes caused may have led to its misdiagnosis as M. tuberculosis (Doig et al., 1991; Nzerue, 1993). The development of R. equi as an opportunistic pathogen was accelerated with the spread of the AIDS pandemic, with the first case described in 1986 (Samies et al., 1986; McNeil & Brown, 1992; Linder, 1997; Torres-Tortosa et al., 2003). Rhodococcus equi infect AIDS patients with a mortality rate of up to 55% (Bell et al., 1998). As with other immunocompromised individuals, infection mostly results in pneumonia with fever, cough and chest pain, but can also spread to other organs and cause bacteremia (Bell et al., 1998). The lung is the organ affected in most cases (Bell et al., 1998); yet, sometimes, infections occur elsewhere without pneumonia (Fierer et al., 1987; Obana et al., 1991). The predominance of lung infections suggests air-borne transmission. However, as the same plasmid type was found in isolates from pigs, wild boars and humans in a Hungarian study, a food-borne transmission might be possible (Makrai et al., 2002, 2005, 2008).

Various other immunosuppressive conditions such as therapies in connection with malignancies, transplantations, therapies based on steroids or diabetes mellitus have been reported in association with R. equi pneumonia (Bell et al., 1998; Hsueh et al., 1998).

Only rare cases of infections with Rhodococcus species other than R. equi have been reported in humans with intact or compromised immune system (Bell et al., 1998). Although this rare occurrence might be due to difficulties in the identification of these unusual opportunistic pathogens, it still highlights the particular pathogenic potential of R. equi.

Other mammals: R. equi are often isolated from pigs, where they can cause a chronic cervical lymphadenitis with tuberculosis-like lesions, but can also be found in submaxillary lymph nodes and tonsils of healthy animals (Prescott, 1991). Isolates from pigs are intermediately virulent for mice and foals and mostly possess a virulence plasmid encoding a 20-kDa VapB (Takai et al., 1996b, 2000a). Rhodococcus equi also cause tuberculosis-like lesions in lymph nodes of cattle and in the livers of young goats (Prescott, 1991). Other animals occasionally infected by R. equi are sheep, llama, cats and dogs (Muscatello et al., 2007).

Epidemology of R. equi in horse infections

Rhodococcus equi infections of foals occur worldwide. Because there is a lot of variation in the number of cases, depending on the particular farm, the region or the year examined, only very few estimates that cover a large geographic area exist. For example, in Australia, 1–10% of foals are affected with a mortality rate usually <1% due to antibiotic treatment; yet, in some farms, it may be as high as 20% or more (Muscatello et al., 2006b). In a study of Cohen et al. (2005), on the basis of non-randomized samples, 47% of 138 American horse-breeding farms were affected by R. equi pneumonia, with 13.3% of all foals on affected farms being infected and a mortality rate of 8%.

Factors that seem to be associated with increased incidences of R. equi pneumonia on horse-breeding farms include a high density and population size of foals, a large farm size and high numbers of airborne virulent R. equi, which correlates with low soil moisture, high temperatures and a poor pasture grass cover (Muscatello et al., 2007). Farms with endemic R. equi pneumonia are heavily contaminated with virulent R. equi, while the others are not, although avirulent R. equi are frequently found in environment and feces on every farm (Takai, 1997). This implies that foals on farms with endemic R. equi pneumonia are frequently exposed to high numbers of virulent bacteria (Takai et al., 1991b; Takai, 1997). Yet, the actual proportion of virulent strains in the environment is no indication for the prevalence of R. equi pneumonia (Martens et al., 2000; Takai et al., 2001a; Muscatello et al., 2006a; Cohen et al., 2008). Similarly, the relative proportion of virulent R. equi in dams' feces is not indicative of the development of R. equi pneumonia in their foals (Grimm et al., 2007).

In the first weeks of a foal's life, ingestion of R. equi often leads to colonization of the intestines. Foals shed large quantities of R. equi as compared with adults, but the number of bacteria in feces declines after 7 weeks of age (Takai et al., 1986b). This might be due to the development of gut immunity or the change from an aerobic to a rather anaerobic gut interior (Takai et al., 1986b), although a higher isolation rate from foal feces compared with those of older foals or adult horses has also been described for the anaerobe Clostridium difficile (Baverud et al., 2003). Ingestion and inhalation of R. equi do not usually result in disease, but in immunization, unless the foal is exposed multiple times to high numbers of bacteria (Johnson et al., 1983b; Takai et al., 1986a). As a result, antibodies against R. equi are widespread in horse populations (Hietala et al., 1985; Prescott, 1991).

Experimental models for R. equi infection

Experimental infection of animals other than foals fails to produce typical pulmonary lesions with abscessation: piglets develop pulmonary lesions but without any macrophage-rich abscesses, and their infection is finally cleared (Zink & Yager, 1987). Guinea pigs show a suppurative pulmonary response, but no abscessation (Ishino et al., 1987). Normal, healthy mice clear intravenous and intranasal infections with R. equi at bacterial cell numbers high enough to induce pneumonia in foals, demonstrating the relative insensitivity of mice to R. equi (Mutimer & Woolcock, 1982). However, bacterial clearance vs. development of pneumonia is dose dependent. Mortality, bacterial burden and development of lesions similar to those seen in foals can be dramatically increased when mice are treated with the immunosuppressive drugs cyclophosphamide or cortisone acetate (Mutimer & Woolcock, 1982; Bowles et al., 1989a, b).

Although some studies on R. equi virulence are based on the lethality of different strains for severe combined immunodeficiency mice lacking functional B and T cells (Lopez et al., 2008), the animal model most frequently used is based on the persistence (not the multiplication) of R. equi in organs of the immunocompetent mouse (Prescott et al., 1997b; Giguere et al., 1999; Jain et al., 2003; Ren & Prescott, 2004; Pei et al., 2007b): R. equi are injected intravenously and mice are killed after several days. Organs of mice are dissected, and persistence of R. equi strains is assessed by bacterial cell number determination. Clinical isolates of R. equi harboring a VAP persist for several days, whereas avirulent strains without plasmid are rapidly cleared. Plasmid-negative strains will, therefore, be addressed as ‘avirulent R. equi’ in the following, inspite of a certain pathogenic potential.

Although the mouse model does not recapitulate all features seen in infected foals, it has nevertheless served well in the identification of many R. equi factors involved in (Table 1) or dispensable for (Table 2) virulence.

Table 1.   Genes of Rhodococcus equi involved in virulence
Knockout
effect
Gene/operon knock
out
Gene locationExperimental
model
Description/putative functionReferences
  • *

    Slight attenuation.

  • Hypervirulent knock out.

  • Hypervirulent upon complementation of knock out.

AttenuationaceAChromosomeMouse, foalIsocitrate lyase, involved in lipid metabolism and carbon assimilation via the glyoxylate shuntWall et al. (2005)
htrAChromosomeMouseSerine proteasePei et al. (2007b)
virS (orf8)Virulence plasmidMouseTwo-component system response regulatorRen & Prescott (2004)
narGChromosomeMouseNitrate reductase, respiration in the absence of oxygenPei et al. (2007b)
pepDChromosomeMouse*Peptidase DPei et al. (2007b)
vapAVirulence plasmidMouse, foalLipid-modified surface proteinJain et al. (2003)
virR (orf4)Virulence plasmidMouseLysR-type transcription regulatorRen & Prescott (2004)
HypervirulencephoP/R operonChromosomeMouseMagnesium-dependent two-component regulatory system PhoP/PhoQRen & Prescott (2004)
sodCChromosomeMouseSuperoxide dismutase, oxidative stress detoxificationPei et al. (2007b)
Table 2.   Genes of Rhodococcus equi that likely are not involved in virulence
Gene/operon knock outGene locationExperimental
model
Description/putative functionReferences
choEChromosomeMouse, foalCholesterol oxidasePei et al. (2006)
clpBChromosomeMousePutative heat-shock proteinPei et al. (2007b)
fadD8ChromosomeMouseAcyl-CoA synthase, lipid metabolismPei et al. (2007b)
fbpAChromosomeMouseMycobacterial antigen85 homologue, mycolyl transferase, presumedly involved in capsule synthesisSydor et al. (2007)
fbpBChromosomeMouseMycobacterial antigen85 homologue, mycolyl transferasePei et al. (2007b)
furAChromosomeMouseIron-dependent transcription regulatorPei et al. (2007b)
galEChromosomeMouseGalactose epimerasePei et al. (2007b)
glnA1ChromosomeMouseGlutamine synthase, nitrogen metabolismPei et al. (2007b)
iupABCChromosomeMouseOperon involved in iron uptakeMiranda-Casoluengo et al. (2005)
orf10Virulence plasmidMouseUnknown function, strongly upregulated transcription in macrophagesRen & Prescott (2004)
purCChromosomeMouse1-phosphoribosylaminoimidazole-succinocarboxamide synthase, purine biosynthesisPei et al. (2007b)
regX3ChromosomeMouseResponse regulatorPei et al. (2007b)
sigFChromosomeMouseAlternative σ factorPei et al. (2007b)

Diagnosis, treatment and immunization

Diagnosis of R. equi infection

Diagnosis of R. equi infection on the basis of cultures from transbronchial aspirates is not fully reliable and should only be used in combination with other diagnostic tools such as microscopic detection of pleomorphic bacteria in cells from horse respiratory tract secretions (Giguere & Prescott, 1997; Muscatello et al., 2007). Serologic analysis, for example by agar gel immunodiffusion, is ineffective in diagnosing R. equi infection, particularly because antibodies against R. equi are generally present in the horse population (Giguere, 2000). A good monitoring tool on farms with endemic R. equi pneumonia is the analysis of white blood cell concentrations (Giguere et al., 2003) and the cultivation of bacteria from transtracheal washes and identification of R. equi via PCR or cytological examinations from such material (Giguere, 2000; Heidmann et al., 2006).

Treatment of R. equi infection and vaccination trials

Because of the intracellular nature of the pathogen, R. equi pneumonia is treated with lipophilic drugs over a prolonged period of time (Prescott, 1991). Treatment often consists of combined application of macrolides such as erythromycin and rifampicin (Giguere, 2000), which are concentrated in granulocytes and macrophages (Giguere & Prescott, 1997; Heidmann et al., 2006). However, the minimal inhibitory concentrations of rifampin and erythromycin for R. equi strains isolated within the last 10 years have been rising (Buckley et al., 2007), and the emergence of some strains resistant to different antibiotics have been reported (Nordmann et al., 1992; Takai et al., 1997; Asoh et al., 2003; Niwa et al., 2006). These findings highlight the need for strategies alternative to antibiotic treatment to prevent or cope with R. equi infection, such as application of gallium nitrate (Harrington et al., 2006).

Many attempts to develop immunization agents have been made. Immunization with orally applied live or inactivated wild-type strains of R. equi can protect foals against R. equi pneumonia (Chirino-Trejo et al., 1987; Varga et al., 1997; Hooper-McGrevy et al., 2005). Efforts to develop safe attenuated R. equi vaccination strains were unsuccessful (Pei et al., 2007a; Lopez et al., 2008), although some promising results were obtained in mice, with an attenuated Salmonella strain expressing the R. equi VapA (Oliveira et al., 2007).

The immunization with partially purified R. equi exoenzymes, cholesterol oxidase and phospholipase C, did not confer protection to foals (Machang'u & Prescott, 1991), and neither did a Triton X-114 extract of R. equi antigens including VapA (Prescott et al., 1997a). Passive immunization with hyperimmune plasma yielded variable results: in some studies, it protected foals from R. equi infection (Martens et al., 1989; Madigan et al., 1991; Prescott et al., 1997a), in others it merely reduced the severity of disease without decreasing foal mortality (Caston et al., 2006) or it was completely ineffective (Hurley & Begg, 1995). In spite of these limitations, hyperimmune plasma is commercially available and being used on many farms (Giguere & Prescott, 1997).

The compartmentation of R. equi in host macrophages

The R. equi-containing vacuole (RCV)

As a facultative intracellular bacterium, R. equi survives and multiplies inside its host cells, which are mostly mononuclear phagocytes (Hondalus et al., 1993). Normally, particles are wrapped into the phagocyte membrane to form a new organelle, the phagosome. This phagosome matures into a phagolysosome via an early phagosome and a late phagosome stage, each of them being characterized by their own marker molecules, just like locations on a map (Haas, 2007). Maturation is achieved by consecutive interaction, first with early, and then with late, compartments of the endocytic pathway. Killing occurs in this phagosome by the combined action of a low pH (4.0–5.0), many hydrolytic enzymes (proteases, lipases, DNAses, RNAses and more), and the production of reactive oxygen metabolites (such as superoxide radicals, hydrogen peroxide, peroxynitrite or nitric oxide). Intracellular pathogens have developed various strategies to escape phagolysosome formation and killing (Haas, 2007), and so the question arises as to which strategy is used by R. equi.

Once R. equi have been internalized, maturation of their phagosomes initially proceeds as for other phagosomes containing harmless microorganisms: the early endocytic markers, early endosome antigen 1, transferrin receptor (TfR), ras-like protein from rat brain (Rab)5, coronin and phosphatidylinositol-3-phosphate, are acquired and lost with normal kinetics, followed by the appearance of some late marker molecules, such as lysosome-associated membrane protein (LAMP)1, LAMP2, Rab7 and, slightly delayed as compared with control phagosomes, the lipid lysobisphosphatidic acid (LBPA)/bis monoacylglycerol phosphate (BMP) (Fernandez-Mora et al., 2005) (Fig. 2). However, maturation of most phagosomes containing virulent R. equi does not progress into a late endocytic organelle; these phagosomes do not acidify and do not acquire the proton-pumping vacuolar ATPase (v-ATPase), at least during the first 24 h postinfection (Fernandez-Mora et al., 2005; Toyooka et al., 2005). RCVs do not fuse with lysosomes, because they remain negative for ferritin or other fluid-phase markers preloaded into lysosomes, lysosomal acid β-galactosidase and cathepsin D (CathD) (Hietala & Ardans, 1987; Zink et al., 1987; Fernandez-Mora et al., 2005). Acquisition of LAMP proteins and Rab7 by most RCVs might, therefore, occur by delivery of these proteins from the Golgi apparatus and cytosol, respectively (Fernandez-Mora et al., 2005).

Figure 2.

 Trafficking of vacuoles containing heat-killed (R.e.dT), avirulent plasmid-negative (R.e.−) or virulent plasmid-positive (R.e.+) Rhodococcus equi in macrophages. Left section: maturation of fluid-phase endosomes and the interactions of the endocytic pathway with normally maturing phagosomes containing heat-killed R. equi. Right section: the vacuoles of avirulent plasmid-negative and virulent plasmid-positive R. equi and their positioning in the endocytic system. Maturation of phagosomes containing avirulent R. equi follows the same normal pathway as that of those containing heat-killed bacteria, although with delay. Phagosomes containing virulent R. equi initially mature normally and acquire and loose early markers during their interaction with early endosomes. However, no fusion with lysosomes occurs and v-ATPase or lysosomal degrading enzymes such as CathD are not acquired. Rhodococcus equi multiply in a multilobed, nonacidified vacuole with numerous internal vesicles and the host cell is finally destroyed in a necrotic process.

Successful establishment of the RCV depends on the VAP and viability of the bacteria (Fig. 2). Early after infection, RCVs of avirulent, plasmid-negative strains fuse only slightly more frequently with lysosomes than those of virulent, plasmid-positive bacteria. At this time, presence or absence of v-ATPase on the RCV is the only distinctive feature (Fernandez-Mora et al., 2005). At prolonged incubation times, however, differences become more pronounced: RCVs containing avirulent plasmid-cured strains fuse with lysosomes and acidify, suggesting that phagosome maturation is not inhibited but delayed (Fernandez-Mora et al., 2005). In equine alveolar macrophages, fusion of RCVs with lysosomes is equally inhibited with formalin-killed or viable R. equi, i.e. these dead bacteria behave like live ones (Zink et al., 1987). In murine macrophages, however, frequency of phagolysosome formation of RCVs containing heat- or formalin-killed R. equi resembles that of those containing plasmid-negative R. equi at 2 h postinfection (Fernandez-Mora et al., 2005); yet, they contain more acid β-galactosidase and CathD (Fernandez-Mora et al., 2005).

Early after infection, virulent R. equi reside in phagosomes with loose membranes (equine macrophages) (Zink et al., 1987) or equally often in compartments with loose or tightly apposed membranes (murine macrophages) (Fernandez-Mora et al., 2005). After an initial lag-phase, virulent bacteria multiply intracellularly in macrophages from mouse peritoneum, mouse bone marrow, horse lung or in murine J774E macrophages (Hietala & Ardans, 1987; Zink et al., 1987; Hondalus & Mosser, 1994; Fernandez-Mora et al., 2005), whereas plasmid-negative strains do not multiply but survive intracellularly for about 24 h (Hondalus & Mosser, 1994; Giguere et al., 1999; Fernandez-Mora et al., 2005). As infection progresses, RCVs of virulent R. equi become loose, multilobed and fill with vesicles and internal membrane material (Fig. 3a) (Fernandez-Mora et al., 2005). The phagosome membrane remains morphologically intact and no escape into the cytoplasm of the host cell has been observed (Hietala & Ardans, 1987; Zink et al., 1987; Fernandez-Mora et al., 2005), at least not for the time period in which the host cell is still intact.

Figure 3.

 Morphological appearance of the RCV in murine macrophages. J774E mouse macrophages were infected with virulent Rhodococcus equi 103+ for (a) 7 h or (b) 25 h and prepared for transmission electron microscopy as in Lührmann et al. (2004). Film negatives were scanned and digitally optimized in light and contrast. (a) At 7 h postinfection, the RCV is filled with numerous vesicles and membranous material. Scale bar=1 μm. (b) Although this macrophage is irreversibly damaged and almost completely lacks cytosolic material and organelles, the membrane around the RCV can be clearly seen. Scale bar=2 μm. Membranes of the RCVs are indicated by black arrowheads.

How the RCV is established and maintained is unknown. No obvious candidate for trafficking diversion, such as a phospholipase or a protein kinase, is encoded in the VAP. It is likely that the exclusion of the proton-pumping v-ATPase plays an important role in RCV establishment and that it is not just an epiphenomenon. This claim is substantiated by previous studies on endosome maturation, in which acidification plays an important role (Clague et al., 1994; Di et al., 2006), and by studies that show a central role of v-ATPase in the fusion of endocytic organelles, even beyond its role in acidification (Peters et al., 2001; Peri & Nusslein-Volhard, 2008).

Infection of macrophages with R. equi is cytotoxic, particularly at high bacterial loads (Hietala & Ardans, 1987; Zink et al., 1987; Lührmann et al., 2004). Degeneration of host cells becomes detectable at about 8 h postinfection (Fig. 3b) (Zink et al., 1987; Lührmann et al., 2004), and is due to necrosis rather than apoptosis (Hondalus & Mosser, 1994; Lührmann et al., 2004). Rhodococcus equi cytotoxicity strictly depends on bacterial viability and is increased by the possession of a VAP (Lührmann et al., 2004), but, so far, no R. equi gene has been identified to be involved in cytotoxicity. As macrophages infected with R. equi have fewer lysosomes than noninfected cells (Hietala & Ardans, 1987; Zink et al., 1987), enhanced secretion of lysosomes or degranulation of lysosome material into the macrophage cytoplasm might also contribute to macrophage necrosis in response to R. equi infection (Zink et al., 1987).

Port of entry of R. equi and consequences for trafficking

Uptake of R. equi is enhanced when the bacteria are complement-opsonized and internalized by the complement receptor CR3 (Mac-1) (Hondalus et al., 1993). Recognition of ReqLAM by mannose-binding protein may also lead to opsonization of R. equi by C3b and internalization via Mac-1 (Garton et al., 2002). However, binding and uptake of R. equi occur readily in the absence of active complement (Fernandez-Mora et al., 2005), possibly mediated by a direct binding of the mannose receptor to R. equi mannose surface compounds (Meijer & Prescott, 2004). Another adherence mechanism, independent of complement, depends on an R. equi hemagglutinating factor (Fuhrmann & Laemmler, 1996). Adherence to HeLa cells correlates with hemagglutination and with a dry colony morphology of R. equi, but not with virulence (de la Pena-Moctezuma & Prescott, 1995; Bern & Lammler, 1996).

Internalization of antibody-opsonized R. equi via the Fcγ-receptor results in increased phagolysosome formation and killing of the bacteria by macrophages, suggesting that the mode of entry of R. equi into the host cell is important for infection success (Hietala & Ardans, 1987), and, possibly, also contributes to the occasional therapeutic success of hyperimmune serum applications.

Virulence determinants

VAPs

As has been shown in the early 1990s, virulence of R. equi parallels the possession of extrachromosomal VAPs (Takai et al., 1991a, c, 1992, 1993; Tkachuk-Saad & Prescott, 1991). More recent work suggests that the type of plasmid possessed by a certain R. equi strain determines its host specificity, as described by the plasmid-typing scheme TRAVAP (Ocampo-Sosa et al., 2007). This scheme is based on the presence or absence of the plasmid genes traA [a conserved gene for plasmid transfer shown to be functional in other Rhodococcus species (Yang et al., 2007)], and the genes vapA or vapB: whereas strains with the characteristic traA+vapA+vapB predominate in isolates from sick foals, those with traA+vapAvapB+ prevail in those from predominately asymptomatic pigs (Ocampo-Sosa et al., 2007). The traA+vapAB characteristic might define a third type of virulence plasmid and is frequently found in diseased cattle; traAvapAB (no plasmid) is typical for nonpathological, environmental isolates.

VapA-encoding plasmids

Of the various R. equi plasmids, those harboring vapA have been analyzed most intensely due to their central role in virulence in foals. In 1991, Takai et al. (1991a, c) identified a highly immunogenic 15–17-kDa protein associated with virulence in foals, which was, therefore, named VapA (Tan et al., 1995) and, which was encoded on 80–90-kb virulence plasmids (Takai et al., 1991c; Tkachuk-Saad & Prescott, 1991; Sekizaki et al., 1995). These plasmids may exist in multiple copies (Rodriguez-Lazaro et al., 2006) and are lost after repeated passages of the bacteria in broth culture at 38 °C (Chirino-Trejo & Prescott, 1987; Takai et al., 1991c). The plasmid is essential for multiplication in macrophages (Hondalus & Mosser, 1994; Giguere et al., 1999) and prolonged inhibition of phagosome maturation (described above), and it enhances cytotoxicity (Lührmann et al., 2004). Strains isogenic to virulent ones but cured of the plasmid are avirulent in foals and mice (Table 3) and do not multiply in macrophages (Takai et al., 1995b; Giguere et al., 1999).

Table 3.   Median lethal doses (LD50) and minimal infection doses (MID) of different Rhodococcus equi strains with VapA-encoding (traA+vapA+vapB), VapB-encoding (traA+vapAvapB+) or no plasmid (traAvapAB) for mice and foals (Wada et al., 1997;Takai et al., 2000a; Makrai et al., 2005)
R. equi plasmid typetraAvapABtraA+vapA+vapBtraA+vapAvapB+
Mice LD50>1 × 1081 × 1061 × 107
Foal MID>1 × 109≤1 × 104>1 × 106

Twelve different, although remarkably conserved, types of VapA-encoding plasmids have been described on account of their endonuclease restriction pattern, and some seem to be accumulated geographically (Takai et al., 2001b, 2003; Ribeiro et al., 2005; Venner et al., 2007). Two of them have been sequenced (Takai et al., 2000b) and were almost identical. They contain 73 predicted ORFs, including eight pseudogenes (Letek et al., 2008). The plasmids can be divided into four major portions: one is involved in replication, one in conjugation, a third section serves unknown functions and the fourth, with a low-G+C content, is likely a pathogenicity island and was probably acquired from an unknown source (Takai et al., 2000b; Letek et al., 2008). The island comprises 26 putative ORFs, many of which are of unknown function (Takai et al., 2000b; Letek et al., 2008). Nine ORFs encode members of the family of Vaps (Takai et al., 2000b; Byrne et al., 2001; Russell et al., 2004; Polidori & Haas, 2006; Letek et al., 2008). Three Vaps, i.e. VapF, pseudo-VapE [also designated as VapX (Letek et al., 2008)] and VapI, are frame-shift mutants and thought to be nonfunctional (Takai et al., 2000b; Russell et al., 2004; Polidori & Haas, 2006). The Vaps presumed to be functional (A, C, D, E, G and H) have export signal sequences of 26–37 amino acids, suggesting that they are secreted (Takai et al., 2000b; Letek et al., 2008). In fact, VapC, D and E are released into the culture broth by virulent R. equi as three of about 10–12 strongly immunogenic proteins (Byrne et al., 2001; Kohler et al., 2003).

The function of VapA is not known and neither is that of the other Vap proteins. The VapA protein is firmly attached to the surface of the bacteria (Takai et al., 1992), probably by means of one or several covalent lipid modification(s) (Tan et al., 1995). However, VapA does not have a cysteine-containing ‘lipobox’, the typical acylation and cleavage site of lipoproteins of Gram-positive bacteria (Sutcliffe & Russell, 1995). Acylation or differential cleavage of VapA might be responsible for the multiple VapA forms from 15 to 22 kDa typically seen on SDS-gel electrophoresis (Sekizaki et al., 1995; Tan et al., 1995).

VapA is essential for R. equi virulence (Jain et al., 2003) and likely involved in the establishment of the RCV (M. Polidori & A. Haas, unpublished data). Yet, while the VapA gene is indispensable, it cannot replace the whole plasmid with respect to virulence in mice or foals (Giguere et al., 1999).

Regulation of VapA expression

Expression of VapA is induced by temperatures >32 °C, low pH, oxidative stress and low levels of divalent cations such as Fe2+, Ca2+ and Mg2+ (Takai et al., 1992, 1996a; Benoit et al., 2001, 2002; Jordan et al., 2003; Ren & Prescott, 2003), conditions that might be encountered within the phagosome (Ren & Prescott, 2003), similar to expression of other genes of the pathogenicity island, particularly other vap genes. Acidification should not be a long-term signal, as R. equi localizes to a pH-neutral phagosome (Sydor et al., 2007). Increased temperatures have the strongest impact on the induction of VapA expression (Ren & Prescott, 2003). Broth composition might also be involved in regulating the expression of VapA: when grown in brain heart infusion broth additionally containing yeast extract, VapA is synthesized even at 30 °C (Byrne & Hines, 2002).

Orf 4 and 8, genes that regulate VapA, are found in the virR operon (Russell et al., 2004). Orf 4 codes for a LysR-type transcription regulator (VirR), and ORF 8 is homologous to response regulators of two-component regulatory systems (VirS) (Takai et al., 2000b; Russell et al., 2004). However, the second system component, a membrane-associated kinase for sensing and transmission of environmental signals, is not encoded in the VAP and might be localized to the R. equi chromosome (Takai et al., 2000b). Whereas the vapA gene is not expressed from its own promoter when transformed into plasmid-cured strains (Russell et al., 2004), protein expression can be partially restored by additionally transforming virR, and expression is close to wild-type levels when the whole virR operon is transformed (Russell et al., 2004). Rhodococcus equi mutants lacking either VirR or VirS are fully attenuated in mice (Ren & Prescott, 2004). Interestingly, however, the knock out of virR or virS results in the upregulation of mRNA levels of all vap genes, including vapA (Ren & Prescott, 2004). This finding led to the assumption that regulation of the genes encoded in the pathogenicity island is mainly based on a negative transcription regulatory network (Ren & Prescott, 2004). Why the VapA protein is not expressed when virS is deleted (Russell et al., 2004), while vapA mRNA increases (Ren & Prescott, 2004), remains to be investigated and may be explained by a complex regulatory mechanism involving transcriptional and translational expression control.

Another factor that might be involved in VapA expression is IdeR, a chromosomally encoded protein that might regulate gene expression depending on the availability of iron (Boland & Meijer, 2000). The promoter region of vapA contains a putative IdeR consensus sequence, suggesting its regulation by this type of transcription repressor (Ren & Prescott, 2003). Finally, VapA expression is regulated by transcript stability (Byrne et al., 2008). A monocystronic vapA transcript likely arises from the processing of a polycistronic mRNA that comprises the ORFs vapA, vapI, vapC and vapD (Byrne et al., 2008). While the monocystronic vapA transcript has an average half-life of 7.5 min, the remaining vapICD transcript has a half-life of only 1.8 min, which might be explained by a need for different expression levels, resulting from differences in protein localization either on the bacterial surface (VapA) or in the culture supernatant (VapC and VapD) (Byrne et al., 2008) or by different protein functions.

VapB-encoding plasmids and the evolution of the Vap plasmids

Rhodococcus equi strains isolated from humans and healthy or tuberculous pigs often possess a VAP coding for a surface-localized 20-kDa protein, which has been referred to as VapB on account of its high homology to the VapA protein (Byrne et al., 2001). VapA and vapB sequences are strongly related to each other (83.6% identity) (Byrne et al., 2001) and so are the plasmids encoding them (Takai et al., 1993; Letek et al., 2008). They belong to the CURV family of plasmids, which share a similar modular composition (conjugation–unknown function–replication/partition–variable region) (Letek et al., 2008). Whereas the plasmid backbone (comprising the C, U and R regions) is highly conserved, not only between the VapA- and VapB-encoding plasmids but also between them and the pREC1 plasmid of nonpathogenic R. erythropolis, the highly dissimilar variable regions likely determine niche specificity. In the case of Vap plasmids, this variable region contains the ‘pathogenicity islands.’ The variable regions of pREC1 as well as those of VapA- and VapB-encoding plasmids were probably acquired by lateral gene transfer and inserted into what might have been an integration ‘hot spot’ offered by the plasmid backbone (Letek et al., 2008).

Although quite different, some elements of the pathogenicity islands of the VapA- and VapB-encoding plasmids are conserved: the virR operon (also harboring VapH or J, respectively), an Lsr2 homologue, a putative SAM-dependent methyltransferase, a putative major facilitator superfamily transporter and a likely secreted AroQ-type chorismate mutase (Letek et al., 2008). In addition to VapB, the VapB-encoding plasmid codes for four Vap proteins (VapJ, VapL, VapM and two copies of VapK), suggesting a central role for the Vap protein family in host tropism (Letek et al., 2008), possibly by fulfilling analogous but yet unknown functions in different hosts (Byrne et al., 2001). The members of the vap gene family possibly emerged from four ancestral vap genes of the pathogenicity island from the last common ancestor of the VapA- and VapB-encoding plasmid by gene duplications and subsequent mutations under a niche-specific evolutionary pressure (Letek et al., 2008).

VapB-expressing isolates are intermediately virulent in mice and foals, and the number of bacteria needed for lethality in mice is about 10 times higher than that of VapA-expressing strains (Takai et al., 1996b, 2000a) (Table 3).

Vaps and VAPs – putative functions, possible experimental approaches

So far, three types of VAPs have been identified, two of which have been sequenced and further investigated (Ocampo-Sosa et al., 2007; Letek et al., 2008). Whereas possession of certain VAPs seems to be specific for strains infecting foals, pigs or cattle (Ocampo-Sosa et al., 2007), it is likely that R. equi infections of humans are not determined by particular plasmids but by the basal and chromosomally determined pathogenic potential of R. equi (see Chromosomal virulence factors and Some speculations on the evolution of R. equi virulence sections). Accordingly, isolates with or without different VAPs have been found in this host (Takai et al., 1995a).

The function of the Vap protein family members is unknown and no significant structural homologies to known proteins are found. Although the VapA-encoding VAP harbors many different Vaps, deletion of VapA, C, D, E and F can be functionally complemented by VapA alone (Jain et al., 2003). Because all putative functional Vaps are secreted or surface-associated, a direct interaction with host target molecules seems likely. Vap proteins might be the key for the uptake into or maintenance of an unusual macrophage compartment that does not fuse with lysosomes and does not acidify. They might interact with the phagosomal membrane to cause phagosome maturation arrest or may be indispensable for nutrient acquisition. Whatever its function may be, the VapA protein seems to be important both early and late in the infection of macrophages (our own observations, unpublished data). Sequence analysis has shown that VapA and VapB-encoding plasmids evolved from a common ancestor plasmid (Letek et al., 2008). The predominance of certain plasmid types in strains infecting particular species suggests that VapA and VapB are factors that determine host tropism. However, the failure of VapA deletion strains to multiply in macrophages from mice suggests a role of the Vap proteins beyond determination of the host species. Moreover, it is not known whether host specificity is determined by VapA/VapB or whether it is rather determined by the whole pathogenicity island, the entire plasmid or even the chromosomal background of the strain(s), which evolved from ancestors carrying one or the other plasmid type and adapted to specific hosts after plasmid acquisition. This might be investigated by curing an isolate harboring a horse-associated VapA-expressing plasmid, to conjugate this strain with a strain possessing a VapB-encoding plasmid and to investigate the virulence of these hybrid rhodococci for foals and pigs. A second approach may be to analyze whether host specificity of VapB-deletion strains can be changed from pigs to foals by complementing with VapA or whether their virulence in mice can be altered from intermediate to high by doing so. The final unraveling of Vap protein function will certainly offer highly interesting insights into R. equi virulence mechanisms.

Chromosomal virulence factors

Chromosomally encoded factors are also involved in R. equi virulence. A partial (Rahman et al., 2003) and the complete sequencing of the R. equi genome sequence [unpublished data (Sanger Institute, 2008)] identified many genes homologous to known or suspected virulence factors of other pathogens, especially of M. tuberculosis. Sick foals sometimes yield plasmid-negative strains, stressing the role of the chromosome, although this might be due to a subsequent loss of the plasmid in culture (Morton et al., 2001). Different isolates of R. equi show enormous differences in pathogenicity traits such as cytotoxicity (Lührmann et al., 2004) or intracellular multiplication (Fernandez-Mora et al., 2005). To be specific, several chromosomally encoded factors that might be involved in R. equi pathogenicity have been investigated:

  • Structural components: Similar to M. tuberculosis, constituents of the R. equi cell wall, such as mycolic acid-containing glycolipids, might contribute their virulence. A correlation between virulence of different R. equi isolates and mycolic acid carbon chain length has been observed (Gotoh et al., 1991; Ueda et al., 2001). However, these strains came from completely different sources, i.e. genetic backgrounds. We have recently characterized R. equi mutants with altered biogenesis of mycolic acid compounds, which are more frequently delivered to lysosomes than the wild type, but are not affected in their general fitness (T. Sydor & A. Haas, unpublished data). ReqLAM, another component of the cell wall, induces a similar cytokine production profile in equine macrophages as do complete viable bacteria and might, therefore, be involved in pathogenesis (Garton et al., 2002). The polysaccharide capsule of R. equi has long been regarded as a potential virulence factor due to its important role in the virulence of other bacterial species (Hondalus, 1997). Given the immunostimulatory activity of capsular polysaccharides, a role was also suggested by the predominance of two serotypes (1 and 2) between clinical isolates of R. equi worldwide (Prescott, 1991). However, there is no connection between serotype and virulence, as bacteria of the same serotype may be virulent or avirulent, or between serotype and infected species (Hondalus, 1997). A mutant defective in capsule integrity is as virulent as wild-type R. equi (Sydor et al., 2007). Therefore, the capsule is apparently of little relevance to virulence and might instead be crucial for survival in dry soil.
  • Factors involved in nutrient acquisition: R. equi isocitrate lyase is essential for intracellular multiplication (Wall et al., 2005). This finding and that R. equi grows well on lipid compounds in its natural habitat suggests that R. equi relies on host cell lipids in the phagosome lumen as sources of carbon and other elements as well. Isocitrate lyase (gene: aceA) is the key enzyme of carbon assimilation from lipids via the glyoxylate shunt. An aceA deletion mutant of R. equi multiplies in mouse macrophages early after infection but is killed by 48 h. The mutant is partially attenuated in mice and even more pronouncedly in foals. After experimental intrabronchial infection, the mutant can be found in the foal lung but no fever or lesions develop (Wall et al., 2005). A successful pathogen has to ensure the supply of iron as an enzyme and electron chain cofactor. However, iron in the mammalian body is strictly limited and associated with iron-binding host proteins. Rhodococcus equi can mobilize iron from different sources, including ferrated deferoxamine, bovine transferrin and bovine lactoferrin (Jordan et al., 2003). Rhodococcus equi are also able to use heme and hemoglobin as iron sources, and this ability depends on the iup (iron uptake) ABC operon (Miranda-Casoluengo et al., 2005). However, an iupABC mutant is fully virulent in mice (Miranda-Casoluengo et al., 2005).
  • Nitrate reductase (NarG) is a chromosomally encoded virulence factor in the mouse model (Pei et al., 2007b). NarG is involved in respiration in the absence of oxygen and could be important for coping with microaerophilic conditions in the course of infection (Pei et al., 2007b). An R. equi deletion mutant of the high-temperature requirement A protein (htrA) is fully attenuated in mice (Pei et al., 2007b). This serine protease might be involved in resistance to oxidative stress, and homologous genes are virulence genes of many Gram-negative bacteria (Pei et al., 2007b). A peptidase D (PepD) deletion mutant was only slightly attenuated. In M. tuberculosis, expression of the pepD homologue is controlled by the mprA-mprB regulatory system, which is important for persistent infection, and by an alternative σ factor, which is involved in intracellular bacterial growth (Manganelli et al., 2001; He & Zahrt, 2005), indicating a possible role of PepD in virulence.Two factors have been identified that cause rather hypervirulence than attenuation in the knockout mutant (the PhoP/R operon) or the complemented strain of a deletion mutant [superoxide dismutase C (SodC)]:
  • The R. equi phoP gene is homologous to that of M. tuberculosis (Ren & Prescott, 2004), which is part of the magnesium-dependent two-component regulatory system PhoP/PhoQ and is involved in the virulence of M. tuberculosis (Perez et al., 2001). Transcription of some genes of R. equi VapA-encoding plasmids is regulated by magnesium (Ren & Prescott, 2003) and the knock out of the phoP/R operon in R. equi leads to upregulation of various genes of the pathogenicity island (Ren & Prescott, 2004). Why the phoP/R knockout mutant is hypervirulent remains to be investigated.
  • The second hypervirulent strain is a complemented SodC mutant of R. equi (Pei et al., 2007b). SodC is involved in the bacterial oxidative stress response and increases resistance of M. tuberculosis to killing by activated macrophages (Piddington et al., 2001).

Secreted cholesterol oxidase (ChoE), one enzyme of the so-called ‘equi factors,’ has long been suspected to be involved in virulence (Hondalus, 1997). It was presumed that it could damage host membranes in combination with bacterial phospholipases and thus contribute to R. equi virulence (Hondalus, 1997; Navas et al., 2001). However, virulent as well as avirulent strains secrete this enzyme and actually no environmental ChoE-negative R. equi strain has ever been found (Prescott et al., 1982). A definitive ‘no’ for ChoE as virulence factor came from knock out analysis (Pei et al., 2006).

In summary, chromosome and plasmid genes cooperate in pathogenicity and virulence. The fact that, under certain circumstances (e.g. immune suppression) and in certain hosts (e.g. foals and AIDS patients), plasmidless strains can cause life-threatening disease suggests that there is a basic virulence potential in many strains of R. equi. The possession of the VAP likely exacerbates this potential, possibly by (transcriptional or structural) cross-talk of plasmid-encoded factors with chromosomally encoded factors, and it also contributes to host tropism. The exciting possibility that VapA and VapB themselves may be host tropism factors requires further study.

Some speculations on the evolution of R. equi virulence

Rhodococcus species have long been renowned for their unusual metabolic capabilities toward many different substrates found in their natural habitats. These capabilities are reflected in R. equi's relatively large genome of c. 5 Mb (Sanger Institute, 2008). As the bacteria are predominately acquired from the environment and not directly from another host, they face a different evolutionary pressure than host-transmitted pathogens do. The different environmental requirements for both multiplication in the host and survival under extreme environmental conditions (e.g. high temperatures, low humidity and competition for nutrients) require a corresponding versatile genetic equipment. Reduction of genome size and plasticity as it is seen with pathogens transmitted from host to host would, therefore, be nonproductive (Casadevall, 2008). The wax-like outer surface layer likely evolved to partition and funnel compounds such as hydrophobic hydrocarbons toward the bacterial membrane for metabolism (Bell et al., 1998) and to protect R. equi from dehydration in their natural soil environment (Harland et al., 2008). The surface compounds could have secondarily developed into virulence factors. It can be further speculated that the acquisition of a VAP precursor from an unknown source built on some pre-existing potential for pathogenicity. This potential becomes manifested in the fact that in some cases plasmidless strains also cause disease, and that even plasmid-negative R. equi slow down phagosome maturation and persist within macrophages and, possibly, soil amoeba for a certain time (Hondalus & Mosser, 1994; Fernandez-Mora et al., 2005; Thomas et al., 2008). While R. equi is perfectly capable of living detached from its host, acquisition of the plasmid might have enabled the bacteria to persist inside the animal when inhaled or ingested. Spontaneous modifications of the original VAP might have led to specification for different host organisms (Letek et al., 2008). Specific adaptation of R. equi to grazing animals is likely due to the predilection of these bacteria for volatile fatty acids, which are abundant in herbivore manure. Whereas the bacteria are frequently found in excrements of herbivores and omnivores, they are absent in those of carnivores or humans (Prescott, 1991). This predilection for certain nutrients results in a permanent exposition of grazing animals to R. equi and might have been the prerequisite for the development of the former ‘soil-only’ bacterium into a pathogen for these particular hosts. Although infected foals may present the most striking clinical manifestation and have therefore been studied most intensely, this acute-type illness might be more exception than rule in R. equi infections. The usual clinical picture may be a persisting but rather inconspicuous infection such as in pigs, cattle (Madarame et al., 1998; Flynn et al., 2001) or adult horses (Fogarty, 2008), which are often not recognized until postmortem inspection and therefore tend to escape attention. In this context, it should be noted that the disease caused by plasmid-bearing R. equi in swine and cattle is in many aspects reminiscent of tuberculosis in man, while there are neither virulence plasmids nor a VapA homologue in mycobacteria (Table 4). In both cases, infection results in a long-time colonization, often within granulomatous structures, which does not necessarily kill the host. Rhodococcus equi infections in foals, however, cause a short and fulminant disease, which leads to massive multiplication of the bacteria but likely not to persistence. Just as in human tuberculosis, heavy coughing caused by the pathogens serves distribution – tubercle bacilli are spread predominately through the air, while R. equi is spread by coughing up and swallowing of large bacteria masses in sputum, followed by the multiplication in and release from the foal intestines (Takai et al., 1986a, b) before these animals finally either succumb to the infection or become immune for the rest of their lives. It should have been this multiplication step that made maintenance of the VapA-expressing VAP an evolutionarily attractive feature for these bacteria.

Table 4. Rhodococcus equi vs. Mycobacterium tuberculosis biology
 Rhodococcus equiMycobacterium tuberculosis
Systematic positionGram-positive–ActinomycetalesMycobacteriaceaeRhodococcus, 5-Mbp DNA with high G+T content (68.8%)Gram-positive–ActinomycetalesMycobacteriaceaeMycobacterium, 4.4-Mbp DNA with high G+T content (65.6%)
Cell wall mycolic acidsMycolic acids of carbon length 22–34 with up to four double bonds plus an α-branch length of 12–18 C-atomsMycolic acids of carbon length 54–64 with two double bonds plus an α-branch length of 20–26 C atoms. Oxidative modifications that are absent from R. equi lipids, such as cyclopropane, keto and epoxy modifications
Virulence-associated plasmidsYes, virulence for foals, pigs and possibly cattle strictly plasmid-associated, for humans probably independent of plasmid, for other host species not sufficiently definedNone in M. tuberculosis. A virulence-plasmid has been found in M. ulcerans, however
Illness caused, symptomsGranulomatous lesions of lymph nodes or silent infections in pigs and cattle. Suppurative bronchopneumonia of the foal, large abscesses in the lung and bronchial lymph nodes. Diseased foals often also develop ulcerative colitis. Symptoms include high fever, increased respiratory rate with bronchovesicular sounds over large and wheezing over small airways, cough. Death by asphyxiation. Mortality in foals can be as high as 80%. Chronic infections in the foal have not been described so far, but may occurChronic cough with blood-tinged sputum, fever, pronounced weight loss (wasting disease).
Primary tuberculosis: Usually pulmonary disease, can be disseminated (militiary tuberculosis) and include lymphatic and gastrointestinal systems and meninges.
Secondary tuberculosis: reactivation of ‘dormant’ mycobacterial foci by immunosuppression (malnutrition, HIV infection, etc.) or immunopathology causing bacterial multiplication and active tuberculosis. Mortality depends on geographical localization, social and medical care situation, nutrition, as well as immune status (HIV) and can be >50% in untreated cases
Host rangeSpecific to foals, pigs and cattle. Might also infect other animals such as goats, cats or dogs. Infection of the adult horse is rare. Infection of humans almost exclusively in immunosuppressed individuals, often resulting in pneumoniaSpecific to humans. Some mycobacteria, such as M. bovis or M. avium ssp. paratuberculosis, can also infect ruminants and occasionally cause human infections. Mycobacterial illness of the horse is very rare
TransmissionInhalation of contaminated soilPerson-to-person, inhalation of bacteria-containing droplets
Host cellsAlveolar macrophages, possibly alveolar epithelial cellsAlveolar macrophages, dendritic cells, possibly alveolar epithelial cells
Host cell deathNecrosis resulting in granulomatous inflammationApoptosis or necrosis
ReservoirSoil, manure, widespread in grazing animals and their environmentHumans only, no environmental isolate known
Experimental model organismsMice. They are normally resistant, but differently virulent bacteria are eradicated at different kinetics. Nude mice (T-cell-deficient) and severe combined immune-deficient (SCID)/beige mice are very susceptible to progressive R. equi disease.
Also foals
Guinea pigs are innately susceptible (model for acute infection). Mice and rabbits (model for chronic infection) are differently susceptible, partly depending on their particular genotypes. Five intranasal virulent bacteria can cause lung lesions. However, none of the models reproduces authentic illness genuinely. Latent infection (90% of infected humans) is not reproduced in any of the experimental models
Important factors contributing to virulence (but not necessarily exclusive virulence factors)VapA, mycolic acid-containing compounds, isocitrate lyase (aceA); for more (see Table 1)Some examples: complex glycolipids such as lipoarabinomannan and trehalose dimycolate (cord factor); mycolyltransferase fbpA; PcaA, a methyltransferase that forms cyclopropane residues in mycolic acid chains; protein kinase G (PknG), ESAT-6/CF-10 protein secretion system, ESTA-6/CFP-10 TLR counter-regulation, ESX secretion system, 19-kDa lipoprotein; outer membrane protein A (OmpA)
Intracellular compartmentation and doubling timePhagosome is arrested in its maturation between an early and a late endocytic step; neutral in pH, no transferrin receptor, no lysosomal contents; phagosomes containing plasmidless bacteria slowly develop into phagolysosomes. Doubling time in macrophages after an initial lag period of a few hours is 6–8 h, in rich broth media 2 hPhagosome that is similar in composition to an early endosome or recycling endosome; near-neutral in pH, transferrin receptor enriched, no lysosomal contents. Doubling time in macrophages is 25–35 h, on nutrient agar 12–18 h

Host genes relevant for the outcome of R. equi infection

Rhodococcus equi successfully infects certain host animals and cell types with high specificity, strongly suggesting that not only bacterial virulence factors, but also particular host (cell) factors determine success or failure of R. equi invasion. Most of the foals on endemic farms do not develop disease, suggesting a predisposition of only a certain subpopulation of foals to R. equi infection. On the suborganismic level, R. equi largely successfully infect monocyte and macrophage cells (Meijer & Prescott, 2004), whereas bacteria are quickly killed by neutrophils from adult horses and foals (Hondalus, 1997). Neutrophils might be involved in infection clearance in mice because experimental elimination of neutrophils results in more severe disease (Martens et al., 2005). Rhodococcus equi are killed more efficiently by macrophages of foals exposed to the bacteria than by those of nonexposed foals, and opsonization of R. equi increases phagosome–lysosome fusion (Hietala & Ardans, 1987).

The search for host factors involved in clearing R. equi infection is hampered by the lack of an equine genome sequence and tools for genetic manipulation. Therefore, analysis of genetic dispositions from the natural host is restricted to case observations, whereas targeted gene manipulations are carried out in the mouse model. Yet, the mouse does not naturally develop R. equi pneumonia, which will impede the identification of the factors that account for the particular susceptibility of foals to R. equi infection. These factors might be:

  • The inducible nitric oxide synthase (i-Nos/NOS2) and NADPH oxidase. Multiplication of R. equi in murine macrophages can be completely prevented by macrophage activation using interferon-γ and lipopolysaccharide (Darrah et al., 2000). This process involves NOS2 and the NADPH oxidase as shown using knockout mice (Darrah et al., 2000). Infection control by activated macrophages has been ascribed to the direct bactericidal effect of peroxynitrite, resulting from the reaction of NO radicals (produced by NOS2) with superoxide anions (generated by NADPH oxidase) (Darrah et al., 2000). In vitro, R. equi are resistant to NO or superoxide, but are sensitive to treatment with peroxynitrite (Darrah et al., 2000). NADPH-oxidase knockout mice die nearly as quickly as those with NOS2 knock out, although NADPH-oxidase-deficient mouse macrophages are still capable of controlling R. equi multiplication if they are activated (Darrah et al., 2000). However, NOS2 activity is not required for the intracellular antirhodococcal mechanisms of human monocytes or macrophages (Vullo et al., 1998). Although isolated macrophages from AIDS patients have a decreased ability to kill R. equi and produce no NO upon experimental infection, inhibition of NO pathways of monocytes/macrophages from healthy humans does not alter their efficient killing activity against R. equi (Vullo et al., 1998).
  • The divalent cation transporter natural resistance-associated macrophage protein 1 (NRAMP1). This protein is located on the phagosome membrane in macrophages, and certain human nramp1 alleles are associated with an increased susceptibility to different intracellular pathogens including M. tuberculosis (Nevo & Nelson, 2006). Some variants of the gene make horses more susceptible to R. equi infection (Halbert et al., 2006), and transcription of NRAMP1 is downregulated in equine macrophages infected with R. equi (Meijer & Prescott, 2004). However, NRAMP1 knockout mice are not more susceptible to R. equi infection (Cohen et al., 2004). It may still be an important factor in foals, because a gene that is not important for coping with infection in an animal naturally resistant to R. equi infection could still be decisive for infection outcome in the natural host.
  • The iron-binding transferrin protein. This protein can be found in biochemical polymorphisms in many species, including the horse (Mousel et al., 2003). Certain transferrin alleles might be responsible for increased susceptibility of some foals to R. equi infection (Mousel et al., 2003).
  • Galectin-3 (Ferraz et al., 2008). Surprisingly, knock out of this β-galactoside-binding lectin increases resistance of mice to R. equi infection. This might be explained by downregulatory functions of galectin-3 on immune responses, which is suggested by an increased, although delayed, inflammatory response of galectin-3 knockout mice as compared with wild-type mice (Ferraz et al., 2008).

Conclusions

The last two decades have seen considerable progress in R. equi research. Plasmids involved in virulence and in host tropism have been identified and sequenced (Sanger Institute, 2008), and investigation into the Vap family of proteins has shed light on the evolution and virulence mechanisms of R. equi. The R. equi chromosome has been sequenced and annotated and its analysis will soon be published, accelerating further understanding and providing a basis for a more targeted analysis of virulence and metabolism-involved genes.

Yet, although some players thought to be involved in R. equi pathogenicity have been confirmed (Table 1) or dismissed (Table 2) as virulence factors, virulence mechanisms are still mostly obscure (Box 1). This might be illustrated best by what we know about VapA: it is an essential virulence factor but does not confer virulence when expressed in a plasmid-negative background. It is surface-localized and probably lipid-modified. Yet, its actual function is still enigmatic, which is true for many genes encoded by the VAP.

Table Box 1.. Critical questions of Rhodococcus equi infection biology
• How is phagosome maturation diverted by R. equi?
• What is the molecular role of the VapA protein in host specificity and virulence?
• Which factors, in addition to VapA, are required for virulence phenotypes and how do they functionally cooperate?
• What are the sources of nutrients for the pathogen in the host?
• What is the structure of the rhodococcal cell wall and how does it contribute to ecology and virulence of the bacteria?
• What can we learn from R. equi about pathogenicity of the close relative Mycobacterium tuberculosis and vice versa?
• Which immunological and environmental factors contribute to the particular susceptibility of foals and pigs to R. equi?
• Are there any more suitable experimental models for the foal infection than mice or foals?
• How can a reliable vaccine for foals younger than 3 months be designed?
• Is direct transmission between animals possible, and if so, is it epidemiologically relevant?

The establishment of the unusual RCV inside its host cells is likely a key event in successful infection. Similar to the vacuole of the intensely studied M. tuberculosis, it does not fuse with lysosomes and lacks the proton pump. However, the RCV is characterized by several rather late endocytic markers (Rab7, LBPA/BMP and LAMP1) and a neutral pH, while the M. tuberculosis-containing vacuole resembles more an early endocytic compartment (Rab5, TfR, LAMP1, pH 6.2), suggesting different mechanisms, although there likely are some common principles of compartment establishment.

Another important research area for the future will be to analyze chromosomally encoded proteins and their interplay with R. equi virulence plasmid factors. The R. equi genome reveals an enormous amount of surface-associated proteins (Vazquez Boland et al., 2009) that possibly interact with host cells in the course of infection. Factors involved in phagosome establishment and host cell cytotoxicity will have to be identified. Techniques of random or targeted gene manipulation of R. equi that have been established in the last years (Ashour & Hondalus, 2003; Jain et al., 2003; Sydor et al., 2007; van der Geize et al., 2008) will help to reach this goal.

Acknowledgements

We thank the Deutsche Forschungsgemeinschaft (SFB670) and the German National Academic Foundation for supporting our studies on R. equi phagosome biology and immune activation. We acknowledge our colleagues who generously permitted references to their unpublished work and the expert technical support of Sabine Spürck and Elisabeth Krämer. We thank Dr Ulrich Schaible for his help with Table 4.

Ancillary