The pathogenic actinobacterium Rhodococcus equi: what's in a name?

Rhodococcus equi is the only recognized animal pathogenic species within an extended genus of metabolically versatile Actinobacteria of considerable biotechnological interest. Best known as a horse pathogen, R. equi is commonly isolated from other animal species, particularly pigs and ruminants, and causes severe opportunistic infections in people. As typical in the rhodococci, R. equi niche specialization is extrachromosomally determined, via a conjugative virulence plasmid that promotes intramacrophage survival. Progress in the molecular understanding of R. equi and its recent rise as a novel paradigm of multihost adaptation has been accompanied by an unusual nomenclatural instability, with a confusing succession of names: "Prescottia equi", "Prescotella equi", Corynebacterium hoagii and Rhodococcus hoagii. This article reviews current advances in the genomics, biology and virulence of this pathogenic actinobacterium with a unique mechanism of plasmid-transferable animal host tropism. It also discusses the taxonomic and nomenclatural issues around R. equi in the light of recent phylogenomic evidence that confirms its membership as a bona fide Rhodococcus.

Rhodococcus equi

Rhodococcus equi is a high‐G+C Gram‐positive, facultative intracellular coccobacillus that parasitizes macrophages, causing pulmonary and extrapulmonary pyogranulomatous infections in different animal species and people (Prescott, 1991; von Bargen and Haas, 2009; Vazquez‐Boland et al., 2013). Since its discovery in 1923 by H. Magnusson in Sweden as the causative agent of purulent bronchopneumonic disease in foals (Magnusson, 1923) (Fig. 1A), R. equi is well known in veterinary medicine as a major horse pathogen (Muscatello et al., 2007; Giguere et al., 2011). In humans, it mostly affects immunocompromised individuals, notably HIV‐infected patients, where the infection resembles pulmonary tuberculosis (Yamshchikov et al., 2010). R. equi is ubiquitous in soil, multiplies in herbivore manure and the large intestine, and spreads in the farm habitat presumably via faecal‐oral cycling (Muscatello et al., 2007; Vazquez‐Boland et al., 2013). Lung infections are likely contracted through inhalation of airborne dust particles carrying R. equi (Muscatello et al., 2006; Cohen et al., 2008; Petry et al., 2017).

Figure 1

R. equi lung infection and colony morphology. A. Purulent bronchopneumonia in foal with multifocal abscesses. Courtesy of Dr. U. Fogarty, Irish Equine Centre. B. Typical mucous appearance of R. equi colonies (LB agar incubated at 30°C for 48 h).

Initially named Corynebacterium equi by H. Magnusson himself, R. equi was transferred in 1977 to the genus Rhodococcus (Goodfellow and Alderson, 1977), currently within the Nocardiaceae in the order Corynebacteriales. R. equi shares a protective mycolic acid‐containing cell envelope with other members of this group of Actinobacteria. Like other rhodococci, it is strictly aerobic and non‐motile, forms orange‐salmon pigmented colonies (Fig. 1B) and shows coccus‐to‐rod or (occasionally) branched cell shape transition (Jones and Goodfellow, 2012). The genus Rhodococcus comprises at least 57 species and an ever‐growing number of unclassified isolates. Many of these are of considerable significance for the environmental, pharmaceutical and energy sectors owing to their versatile catabolic and biocatalytic properties (van der Geize and Dijkhuizen, 2004). Two rhodococcal species are recognized as pathogenic, Rhodococcus fascians, which causes leafy gall in plants (Stes et al., 2013), and R. equi.

R. equi genome

The only complete and manually curated genome sequence available for R. equi is from strain 103S (= NCTC 13926 = DSM 104936), a prototypic equine clinical isolate (Letek et al., 2010). The reference 103S genome (NCBI RefSeq NC_014659.1, accession FN563149.1) consists of a circular chromosome of 5.04 Mbp with 4,598 predicted genes and a G+C content of 68.8% (Fig. 2A). A second key genome component is the virulence plasmid, which carries the vap pathogenicity island (PAI) (Takai et al., 2000). In 103S, it is a circular plasmid of 80.6 kb designated pVAPA1037 (reference sequence accession AM947677) (Letek et al., 2008). The R. equi chromosome appears to be genetically stable, as indicated by the rarity of DNA mobility genes or insertion sequences (Letek et al., 2010) and absence of significant recombination (Anastasi et al., 2016). A small number of pseudogenes (14 in 103S, most in horizontally acquired regions) suggests that it is under strong selection.

Figure 2

Genomic relatedness of R. equi isolates. Modified from Anastasi et al. (2016). A. Circular diagram of R. equi 103S (= NCTC 13926, DSM 104936) chromosome (5.02‐Mpb, outer ring with forward and reverse strands) compared to draft genomes of representative isolates from different sources and genetic lineages (inner rings). BLASTn alignments, red colour indicates > 98% sequence identity. HGT regions in 103S (arrow heads) coincide with gaps in the DNA alignments, indicating they are strain‐specific or less conserved. B. Core‐genome maximum‐likelihood phylogeny of R. equi isolates in A. Top, unrooted tree; reference genome isolate 103S and type strain of the species are indicated. Bottom, tree rooted with the closely related species, R defluvii Ca11T. The star‐like topology in the early branchings of the R. equi lineage suggests that the species’ diversification occurred through rapid clonal radiation from the common progenitor. See also Fig. 5.

Figure 5

Whole‐genome Corynebacteriales ML tree. Nodes indicate bootstrap values. Tree constructed with five R.equi genomes, 47 non‐equi Rhodococus genomes including representatives from the major 16S rRNA gene clades (Goodfellow et al., 1998; McMinn et al., 2000; Jones and Goodfellow, 2012; Ludwig et al., 2012), and 57 genomes from 11 Corynebacteriales genera. Rooted with Streptomyces albus NBRC 1304T (outgroup). (T) indicates type strain. Genome used for R. equi type strain DSM 20307T = ATCC 6939T is assembly acc. no. LWTX00000000 (Anastasi et al., 2016). Major genera are highlighted in different colour. Black arrowheads indicate misclassifications revealed by the phylogenomic analysis. One of them is R. rhodnii NRRL B‐16535T (GenBank assembly acc. no. GCA_000720375.1); this probably represents a sequence mislabelling or strain mixup. Modified from Anastasi et al. (2016).

Comparative genomic analyses show that R. equi is genetically homogeneous and clonal, with a large core genome equivalent to ≈80% of an isolate's gene content. It is a well‐defined taxon with an Average Nucleotide Identity (ANI) of 99.13% and 100% 16S rDNA sequence identity. In a core‐genome phylogenomic tree, R. equi isolates radiate at a short genetic distance from each other (0.001–0.002 substitutions per site) (Fig. 2B), consistent with a relatively recent evolutionary origin and a rapid clonal diversification from the common progenitor.

Like many other bacteria, R. equi has an open pangenome. Although non‐core genes only represent ≈20% of each strain's gene content, a significant proportion of the accessory genome (60%) is only present in one or two isolates, accounting for the intraspecific variability. R. equi genome evolution is primarily driven by gene gain/loss processes, with a significant contribution of horizontal gene transfer (HGT) events (Letek et al., 2010; Anastasi et al., 2016) (Fig. 2A). Phages are abundant in R. equi (Summer et al., 2011; Petrovski et al., 2013; Salifu et al., 2013) and probably play an important role in HGT‐driven genome plasticity.

Core R. equi traits

Comparative genomic studies confirmed that most traits predicted to be important for R. equi biology and niche adaptation belong to the core genome (Anastasi et al., 2016). This includes all putative pathogenicity determinants identified in the 103S chromosome, notably a number of mycobacterial virulence gene homologs (Letek et al., 2010). Genes involved in tolerance to desiccation and oxidative stress, presumably important for survival in dry soil and transmission via aerosolized dust, also belong to the core genome. There is also a conserved intrinsic resistome with several putative β‐lactamases, aminoglycoside phosphotransferases and multidrug efflux pumps. These probably contribute to the variable susceptibility reported for R. equi to diverse antimicrobials, as observed, for example, with β‐lactams and quinolones (Nordmann and Ronco, 1992; Mascellino et al., 1994; Soriano et al., 1998; Makrai et al., 2000; Jacks et al., 2003; Letek et al., 2010; Yamshchikov et al., 2010).

A distinctive characteristic of R. equi is a complete absence of phosphoenolpyruvate:carbohydrate transport system (PTS) components, consistent with an eminently asaccharolytic metabolism. Among the rhodococci, only its close relative Rhodococcus defluvii (Kämpfer et al., 2014) also lacks a PTS sugar transport system (Anastasi et al., 2016), indicative of specific gene loss in the common ancestor of the R. equi‐R. defluvii sublineage (see below Fig. 5). The absence of PTS homologues is rather unique within the Actinobacteria; among the few examples are Mycobacterium tuberculosis, also a parasite of macrophages, and the obligate intracellular pathogen Tropheryma whipplei (Barabote and Saier, 2005; Letek et al., 2010). This suggests that loss of PTS sugar transport might be associated with adaptation to intracellular parasitism in this bacterial group.

Recently, genes encoding putative non‐PTS transporters for glucose (GlcP) and ribose (RbsCB) have been identified in the R. equi core genome. Both permeases seem to be functional, although utilization by R. equi 103S of ribose and, particularly, glucose is inefficient (and variable for the latter) compared to preferred carbon sources such as lactate or acetate (Letek et al., 2010; Anastasi et al., 2016). Since R. equi assimilates carbon principally via short‐chain organic acids and lipid catabolism, these two sugar transporters might act as occasional ‘nutritional fitness’ enhancers in specific habitats.

In addition to monocarboxylate and dicarboxylate transporters, the R. equi core genome encodes an extensive array of lipases (both secreted and intracellular) and β‐oxidation enzymes. There are also three complete mce (‘mammalian cell entry’) systems, which form channel mechanisms specialized in lipid transport (Ekiert et al., 2017), for example, cholesterol (Mohn et al., 2008; Pandey and Sassetti, 2008). Similar to M. tuberculosis (Muñoz‐Elias and McKinney, 2005), R. equi virulence requires the glyoxylate shunt enzyme isocitrate lyase (ICL) (Wall et al., 2005). ICL mediates the diversion of TCA cycle intermediates for gluconeogenesis and carbohydrate biosynthesis from acetylCoA generated through fatty acid β‐oxidation or acetate oxidation. This indicates that, like the tubercle bacillus, R. equi utilizes lipids as in vivo growth substrate. Interestingly, although not a fermentative organism, R. equi possesses a putative bifunctional D‐xylulose 5‐phosphate (X5P)/fructose 6‐phosphate (F6P) phosphoketolase (Xfp) (Meile et al., 2001). This enzyme may provide flexibility in carbon and energy metabolism by converting pentose phosphate pathway (PPP) and glycolytic intermediates into acetyl phosphate (and acetate/acetyl‐CoA) (Ingram‐Smith et al., 2006).

R. equi appears to be particularly well adapted for growth on exogenous L‐lactate, with a dedicated transporter (LldP) and determinants for its conversion into acetate, either directly (L‐lactate monooxygenase) or via pyruvate (lutABC operon [Chai et al., 2009]) combined with pyruvate decarboxylation via pyruvate dehydrogenase [cytochrome]) (Letek et al., 2010). Moreover, R. equi has denitrification capacity, with a NarK nitrate/nitrite transporter, NarGHIJ nitrate reductase and a NirBD nitrite reductase. It also has the ability to grow on urea as the sole nitrogen source through the action of a urease and an ATP‐dependent urea carboxylase (Letek et al., 2010; Anastasi et al., 2016).

Another core feature is the disruption of the thiCD locus by an HGT island, rendering R. equi auxotrophic to thiamin (Letek et al., 2010; Anastasi et al., 2016). Apart from this, R. equi is otherwise not nutritionally demanding and can grow vigorously in the presence of just inorganic N (e.g. in the form of ammonium chloride) and an organic acid as a carbon source. Together with its alkalophily (optimal growth between pH 8.5 and 10) (Letek et al., 2010), the nutritional and metabolic profile of R. equi may confer a competitive advantage in manure and the large intestine, its natural reservoirs, where there is an easy access to microbiota‐derived thiamine and lactate/short‐chain fatty acid fermentation products (Letek et al., 2010; Anastasi et al., 2016). Via a NiFe‐type hydrogenase, R. equi has the potential to utilize H2, released through microbial metabolic activity, potentially contributing to survival in the intestinal habitat.

R. equi illuminates rhodococcal genome evolution

R. equi possesses a more compact genome compared to environmental rhodococci, exemplified by Rhodococcus erythropolis PR4 (6.52 Mb) and, particularly, Rhodococcus jostii RHA1 (7.80 Mb) or Rhodococcus opacus B4 (7.25 Mb), for which complete genomes are also available (McLeod et al., 2006; http://www.nite.go.jp/index-e.html). Analysis of gene duplication and HGT events, together with the slow rate of gene decay in the 103S chromosome, indicated that the genome size differences are due to genome expansion in the environmental species rather than genome contraction in R. equi (Letek et al., 2010). Genome expansion is due to amplification of paralogous families and acquisition of HGT DNA and extrachromosomal genes, often as part of plasmids as large as 1 Mb in size. These plasmids are particularly rich in HGT DNA (up to 50%), contain a much higher density of mobility genes and pseudogenes, unique species‐specific genes and niche‐adaptive determinants, specifically catabolic (McLeod et al., 2006; Letek et al., 2010). The metabolic complexity of the environmental Rhodococcus spp. is a likely reflection of the isolation criteria, seeking for specific abilities such as degradation of multiple aromatic pollutants, biotransformation or production of secondary metabolites (van der Geize and Dijkhuizen, 2004; Larkin et al., 2005; Yamashita et al., 2007; Kitagawa and Tamura, 2008; Holder et al., 2011; Foster et al., 2014). For example, compared to R. equi, the polychlorinated biphenyl degrader R. jostii RHA1 contains a much larger complement of unique metabolic genes, aromatic gene clusters (29 vs only three), non‐ribosomal peptide synthases (24 vs 11) and polyketide synthases (7 vs 1 in R. equi) (McLeod et al., 2006; Letek et al., 2010).

Circular and linear genomes: a matter of size

The determination of the complete 103S genome sequence made it apparent that Rhodococcus spp. differ in chromosome topology despite being monophyletic. While R. equi 103S and R. erythropolis PR4 both possess covalently closed circular chromosomes, R. jostii RHA1 and R. opacus B4 have linear ones. Remarkably, not only the four species belong to a same subdivision of the genus Rhodococcus, but R. erythropolis and R. jostii/R. opacus even belong to sister sublineages within the same terminal clade (no. 2, see below Fig. 5) (Anastasi et al., 2016). Since the four chromosomes share the same overall structure and synteny (Letek et al., 2010; Anastasi et al., 2016), the only obvious difference is a comparatively larger size for R. jostii and R. opacus (≥7.25 Mb), similar to Streptomyces spp (≥8 Mb), which also possess linear genomes. Thus, actinobacterial chromosome linearization appears to occur as a function of increasing size rather than phylogenetic background. This mirrors the situation with the rhodococcal plasmids, which independently of the host species tend to be linear above 100 kb (Larkin et al., 2010; Valero‐Rello et al., 2015).

Plasmid‐determined virulence

A distinguishing feature of the genus Rhodococcus is the characteristic presence of large circular or linear conjugative plasmids carrying niche‐adaptive DNA (Larkin et al., 2010). While these regions encode catabolic and detoxification pathways in rhodococcal species isolated from xenobiotic‐contaminated ecosystems (McLeod et al., 2006; Sekine et al., 2006), in the pathogenic species R. equi and R. fascians they encode virulence (i.e. host‐adaptive) determinants (Letek et al., 2008; Francis et al., 2012; Valero‐Rello et al., 2015). In the case of R. equi, the plasmid's HGT‐acquired vap PAI (Fig. 3) supports intramacrophage survival and is essential for animal host colonization (Coulson et al., 2010).

Figure 3

The three host‐specific R. equi virulence plasmids. Comparison of pVAPA (equine type) and pVAPB (porcine type) circular virulence plasmids and the recently characterized linear pVAPN plasmid (ruminant type) with closest homologs from environmental biodegrader Rhodococcus spp. Regions of significant similarity are connected with grey stripes. The vap PAIs are shaded in light blue. Gene colour code: Hypothetical proteins (gray), conjugation or DNA replication/recombination/metabolism (red), DNA mobility genes (magenta), transcriptional regulators (blue), secreted proteins (dark green), membrane proteins (pale green), metabolic functions (yellow), vap family genes (black) and pseudogenes (brown).Green and pale red bars below the genes indicate conjugation and replication/partitioning functional modules respectively; dashed underline indicates HGT region. Modified from Valero‐Rello et al. (2015)

The vap PAI encodes a set of homologous secreted virulence‐associated proteins (Vap) (Letek et al., 2008; Valero‐Rello et al., 2015) (Fig. 4A) that fold in a cork‐shaped eight‐stranded antiparallel β‐barrel structure (Geerds et al., 2014; Whittingham et al., 2014). One of them, designated VapA in the equine‐type plasmid pVAPA, is essential for pathogenesis (Jain et al., 2003; Gonzalez‐Iglesias et al., 2014; Valero‐Rello et al., 2015). The exact mechanism of action of VapA and homologous proteins remains unknown but is thought to be related to the biogenesis of the modified, Rab7‐positive endosome where the bacterium replicates within macrophages (aka the “R. equi‐containing vacuole’’, RCV) (Fernandez‐Mora et al., 2005; Sydor et al., 2013; Rofe et al., 2017). Consistent with this, VapA has recently been found to localize to the membrane of the RCV (Wright et al., 2018). Non‐vap genes are also present in the PAI (Valero‐Rello et al., 2015; MacArthur et al., 2017), notably the vir operon, which encodes two key regulators (VirR and VirS) required for vap PAI gene activation and virulence (Byrne et al., 2007) (Fig. 4A). Another vir operon product, IcgA, has been shown to modulate intracellular growth of R. equi (Wang et al., 2014).

Figure 4

Structure and evolution of the host‐specific vap PAIs. Modified from Valero‐Rello et al. (2015). A. Genetic structure of the vap PAIs from pVAPA (15.1 kb), pVAPB (21.5 kb) and pVAPN (15.9 kb). PAI genes in grey (non‐vap genes, in darker shade the vir operon) or black (vap genes). Genes outside the PAIs in white. PAI boundaries indicated by yellow arrowheads. The figure schematizes the evolutionary relationships of the vap genes as inferred from phylogenetic analysis, gene duplication/loss analysis (panel B) and genetic structure comparison. Straight lines connect allelic variants of same vap gene ancestor; those of vapA have red surround, curved lines indicate vap gene duplications. Crosses denote vap genes that were lost. Asterisks indicate pseudogenes. B. Gene duplication and loss in R. equi vap multigene family. Constructed with notung v2.6 from a vap gene ML tree. The analysis indicates that the common ancestor of the three host‐specific PAIs contained seven vap genes which evolved by gene duplication from a single ancestor vap gene. C. Fate of the vap PAI during host‐driven R. equi virulence plasmid evolution.

A number of core chromosomal metabolic genes appear to have been co‐opted within the regulatory network of the vap PAI and exhibit expression patterns similar to those of the plasmid virulence genes. Two of these genes, encoding chorismate mutase and anthranilate synthase enzymes involved in aromatic amino acid biosynthesis, appear to facilitate intracellular survival in macrophages (Letek et al., 2010).

Plasmid‐mediated host tropism: a novel paradigm

Three R. equi virulence plasmid types have been identified to date, pVAPA and pVAPB associated with equine and porcine isolates, respectively, and pVAPN (‘N’ for no‐A/B) associated with ruminants (bovines, ovines and caprines) (Ocampo‐Sosa et al., 2007). pVAPA and pVAPB are variants of a same circular replicon which differ in vap PAI structure (Letek et al., 2008), whereas pVAPN is an unrelated linear plasmid, with again a specific vap PAI (Valero‐Rello et al., 2015; MacArthur et al., 2017) (Fig. 3). pVAPA/B/N type‐host mismatch virtually never occurs among equine, porcine and ruminant isolates, suggesting stringent host‐driven exclusion of non‐adapted plasmids (Ocampo‐Sosa et al., 2007; unpublished data from JV‐B laboratory). Phylogenomic analyses did not find any association between host and chromosomal genotype but, instead, clear evidence of active exchange of the pVAPA/B/N plasmids across the R. equi population, with corresponding host jumps (Anastasi et al., 2016; MacArthur et al., 2017). The R. equi virulence plasmid appears to be easily lost in the absence of host selection (Takai et al., 1994; Ocampo‐Sosa et al., 2007) but can be readily reacquired via conjugation (Tripathi et al., 2012; Valero‐Rello et al., 2015). The available evidence supports a model whereby R. equi species‐specific infectivity is mediated by the virulence plasmids, with dynamic plasmid loss‐regain allowing flexible adaptation to saprotrophic life in the environment and parasitization of different animal hosts.

Contrasting with their selectivity for certain animal species, the three host‐adapted plasmids are commonly found in human isolates (Ocampo‐Sosa et al., 2007; Anastasi et al., 2016). This suggests that animals are the source of infection for people, establishing R. equi as a novel zoonotic pathogen. It also implies that humans are essentially opportunistic hosts for R. equi (Ocampo‐Sosa et al., 2007; Vazquez‐Boland et al., 2013, 2010). The situation appears to be analogous for other animal species which seem to be also accidental hosts for R. equi, as for example suggested by recent virulence plasmid characterization studies from dog isolates (Bryan et al., 2017).

Evolution of R. equi virulence

As mentioned above, the pVAPA/B/N plasmids each carry a type‐specific vap PAI. The major differences lie in the vap multigene family (Fig. 4A). Phylogenetic reconstruction of vap multigene family evolution indicates that the nearest common ancestor of the vap PAI contained seven vap genes (Valero‐Rello et al., 2015). These progenitor vap alleles originated via gene duplication from an ancestor vap determinant (Fig. 4B). This proto‐vap gene was probably horizontally acquired because obvious homologs are absent from other Actinobacteria while they are found in bacteria from different phyla and even fungi, yet remaining relatively uncommon (Whittingham et al., 2014; Valero‐Rello et al., 2015).

A likely hypothetical scenario is that the proto‐vap, in combination with some non‐vap determinant present in the common PAI ancestor, acquired at some stage the ability to promote intracellular survival. Perhaps initially a defence mechanism against predation by bacterivore environmental protozoa, eventually this also allowed the host bacterium to escape phagocytic killing by macrophages, paving the way to becoming an animal pathogen. Indeed, a critical intracellular survival determinant is obviously present in the extant vap PAIs, because the three host‐adapted plasmids promote intracellular survival in cultured macrophages (Giguere et al., 1999; Coulson et al., 2010; Gonzalez‐Iglesias et al., 2014; Valero‐Rello et al., 2015; Willingham‐Lane et al., 2016).

Cumulative epidemiological and experimental evidence indicates that the intracellular survival‐promoting function is primordial and dissociable from host tropism, because the three plasmid types promote virulence in accidental (non‐adapted) animal hosts (e.g. mice, apparently also humans). This critical vap determinant is probably the common ancestor of vapA of the equine pVAPA type and its allelic variants vapN of the ruminant pVAPN type (Valero‐Rello et al., 2015) and vapK1/2 (and putative duplicate thereof, vapB) of the porcine pVAPB type (Valero‐Rello et al., 2015; Willingham‐Lane et al., 2018) (Fig. 4A and B).

Subsequently, host‐tropic properties evolved in the common ancestor of the vap PAI, presumably through adaptive evolution of the vap multigene family in equines, swines and ruminants. The process appears to have started in the pre‐pVAPA/B plasmid, followed by horizontal transfer of the PAI from the pVAPA lineage to the pVAPN replicon (Valero‐Rello et al., 2015) (Fig. 4C). The perfect conservation of the vap PAIs within each host‐adapted virulence plasmid type indicates they are under strong selection, likely driven by species‐specific host factors yet to be identified. The conservation of the DNA mobility gene remnants flanking the vap PAIs and the pseudogenes in each PAI type suggests that the PAI diversification process is relatively recent (MacArthur et al., 2017).

Common rhodococcal strategy for rapid niche adaptation

The R. equi virulence plasmids share similar backbones with other plasmids found in environmental rhodococci. Thus, the pVAPA/B replicon is homologous to that of pREC1 from the alkane degrader R. erythropolis PR4 (Sekine et al., 2006) or pKNR from the organic solvent‐tolerant R. opacus B4 (Honda et al., 2012) (Fig. 3). All these circular plasmids possess a conjugation machinery based on a MOBf (TrwC)‐type relaxase (Garcillan‐Barcia et al., 2009), designated TraA, together with a type IV secretion system (T4SS) which forms the transport channel. pVAPN, on the contrary, is closely related to the linear plasmid pNSL1 from the environmental Rhodococcus sp. NS1 (Zhu et al., 2010) (Fig. 3). Like the circular plasmids, pVAPN and pNSL1 share a conserved backbone but differ in a unique variable region (VR) adjacent to the replication/partitioning region (Fig. 3). Self‐transmissibility relies in this case on a relaxase/T4SS‐independent mechanism mediated by a TraB translocase, a novel conjugation system first characterized in the Streptomyces linear plasmids. The TraB protein is evolutionarily related to FtsK/SpoIIIE involved in chromosome segregation (Guglielmini et al., 2013) and forms a hexameric channel through which dsDNA is conducted in an ATP‐dependent manner (Vogelmann et al., 2011). While not obviously similar, the pVAPN and pNSL1 replicons are phylogenetically related to other rhodococcal linear plasmids (Valero‐Rello et al., 2015).

The VRs of all the rhodococcal plasmids, whether circular or linear, are typically flanked or contain DNA mobility gene remnants including a variety of recombinases and transposases (Letek et al., 2008; Valero‐Rello et al., 2015) (Fig. 4A). Integrative elements thus appear to play a key role in the formation and plasticity of the VRs. An intriguing feature is the conservation of the rep‐parA module across the pVAPA/B and pREC1 circular plasmids and the pVAPN and pNSL1 linear plasmids (Fig. 3). The rep‐parA module is detected as HGT DNA in pREC1 and pNSL1 and adjacent to it there is a putative phage excisionase gene that is also conserved in the circular and linear plasmids. This suggests that the rep‐parA module adjoining the VRs itself forms part of a gene cassette that is horizontally exchangeable between different rhodococcal replicons despite their different ancestry (Letek et al., 2008; Valero‐Rello et al., 2015).

The R. equi virulence plasmids consolidate the notion that rapid niche adaptation through shared sets of self‐transferable extrachromosomal replicons is a key common attribute of the actinobacterial genus Rhodococcus.

The lingering problem of R. equi taxonomy and nomenclature

Although sharing obvious physiological, compositional and genetic features with the other rhodococci, specifically the common plasmid‐driven niche specialization strategy, the taxonomic status of R. equi within the genus Rhodococcus has been repeatedly questioned (McMinn et al., 2000; Jones and Goodfellow, 2012). These taxonomic difficulties are mirrored in the nomenclature of the species. To several previous validly published names, i.e. Corynebacterium equi Magnusson, 1923, Nocardia restricta (Turfitt 1944) McClung 1974 and R. equi (Magnusson, 1923) Goodfellow and Alderson 1977, additional names have been recently added in rapid succession, as discussed below.

At the root of the nomenclatural instability are a number of 16S rRNA phylogenetic studies, which placed R. equi at the periphery of the genus Rhodococcus (McMinn et al., 2000; Gurtler et al., 2004) or among the Nocardia (Rainey et al., 1995; Goodfellow et al., 1998). Their significance was, however, unclear because of the low bootstrap values (Rainey et al., 1995), or because Nocardia branched off from within the Rhodococcus radiation (Rainey et al., 1995; Goodfellow et al., 1998; McMinn et al., 2000) instead of forming a deep monophyletic lineage near the base of the Corynebacteriales (Ludwig et al., 2012).

A recent publication by Jones et al. (2013a) based on numerical phenetic and genotypic (PCR fingerprinting, 16S rDNA sequence) clustering further complicated the situation. While, as expected from their taxonomic status as a separate species, R. equi strains grouped in a distinct cluster, perfectly equivalent to that formed by other rhodococcal clades, this fact was used to justify the reclassification of R. equi into a new genus, as ‘Prescottia equi’ gen. nov., comb. nov. (Jones et al., 2013a). The name ‘Prescottia’ was proposed in honour of the R. equi research pioneer John F. Prescott from University of Guelph in Canada. However, the new genus name was found to be illegitimate according to the International Code of Nomenclature of Prokaryotes (aka the Bacteriological Code) (Lapage et al., 1992) because it already designated the Orchidaceae plant genus Prescottia Lindley 1824. The same authors corrected the mistake by proposing ‘Prescotella’ gen. nov., and ‘Prescotella equi’ comb. nov. as its sole species (Jones et al., 2013b).

The proposal to transfer R. equi to a new genus was presented to the international R. equi community at the 5th Havemeyer Workshop on R. equi held in Deauville (France) on 9–12 July 2012 and was largely met with disapproval for several reasons (Cauchard et al., 2013). First, it clashed with the conclusions of the recently published 103S genome study, which found that R. equi was genomically well classified as a Rhodococcus sp. (Letek et al., 2010). Second, strong concerns were voiced that changing such a well‐consolidated name in veterinary and medical science would cause considerable confusion. Another important caveat was whether the methodology used in the Jones et al. (2013a) study had adequate resolution to define taxa above the species level according to modern phylotaxonomic criteria. The opinion was expressed that reassigning R. equi to a novel genus would only be justified upon systematic re‐examination of the taxonomy of the entire genus Rhodococcus using phylogenomic approaches (Cauchard et al., 2013). We transmitted these concerns on behalf of the international R. equi community to leading members of the International Committee for Systematics of Prokaryotes (ICSP) in correspondence exchanges in March 2013.

Probably due to the stir caused by the proposed name changes, the nomenclature of the species was re‐examined and it became apparent that R. equi had an earlier heterotypic synonym in Corynebacterium hoagii (Morse 1912) Eberson 1918 (Tindall, 2014a). Evidence for the identity of R. equi and the nomenclatural type C. hoagii as the same species was provided by 16S rRNA gene phylogenies used in the description of the closely related species R. defluvii (Kämpfer et al., 2014). This meant that not only ‘Prescotella equi’ was an illegitimate name but that the name R. equi also contravened itself the principle of priority of the Bacteriological Code (Tindall, 2014a). Since it was evident that R. equi is not a Corynebacterium, and Kämpfer et al. (2014) confirmed that the differentiation of R. equi from other members of the genus Rhodococcus was unsupported by chemotaxonomic and phylogenetic evidence, these authors proposed to retain R. equi within the genus Rhodococcus but with the priority epithet hoagii, as Rhodococcus hoagii comb. nov.

Truly, a ‘fine mess’, to quote the authors who prompted this ‘nomenclatural storm’ by proposing the new genus names ‘Prescottia/Prescottella’ (Goodfellow et al., 2015).

R. equi is genomically a bona fide Rhodococcus

Anastasi et al. (2016) performed a detailed whole genome sequencing (WGS) phylogenetic analysis of a collection of representative R. equi isolates, including the type strain DSM 20307T (= ATCC 6939T), and available sequences from other rhodococcal and Corynebacteriales species (Fig. 5). An important first conclusion is that R. equi is a strictly monomorphic taxon (Fig. 2), thus settling the question of whether R. equi is phylogenetically heterogeneous. This had been often raised in the literature (McMinn et al., 2000; Gurtler et al., 2004; Jones and Goodfellow, 2012), curiously based on just a single 16S rDNA study that reported similarity values of 97.9–98.3% between the R. equi‐type strain ATCC 6939T/ DSM 20307T and some of the (only 10) isolates analysed (McMinn et al., 2000). The reason for such rRNA sequence variability is unclear, but might be due to possible strain misidentifications (Vazquez‐Boland et al., 2010) and/or 16S rDNA sequencing errors. In the genomes analysed by Anastasi et al. (2016), representing a diversity of genetic backgrounds and isolation sources, all 16S rDNA sequences were 100% identical.

Secondly, the phylogenomic studies disambiguated the taxonomic relationship of R. equi with other Rhodococcus spp. All R. equi isolates group in a well‐supported monophyletic cluster (no. 3 or ‘equi’ subclade) deeply rooted in the rhodococcal phylogeny which also contains R. defluvii Ca11Ts as the closest relative, Rhodococcus triatomae BKS15‐14 (a species not previously resolved into any specific 16S rRNA gene clade) (Ludwig et al., 2012), and an unclassified isolate (Anastasi et al., 2016) (Fig. 5). Two other subclades are generally congruous with the 16S rDNA groupings ‘rhodochrous’ (subclade 1) and ‘erythropolis’ (subclade 2) (McMinn et al., 2000; Jones and Goodfellow, 2012; Ludwig et al., 2012). Subclade 1 splits into two distinct sublineages, one encompassing R. ruber, and the other, the type species of the genus, R. rhodochrous and Rhodococcus pyridinivorans. Subclade 2 also splits into two sublineages, one containing R. opacus, R. jostii, Rhodococcus imtechensis and Rhodococcus wratislaviensis; the other, R. erythropolis and Rhodococcus qingshengii (Fig. 5). The remaining major phyletic lines correspond to R. rhodnii LMG 5362 and R. fascians isolates, defining two novel subclades (nos. 4 and 5 respectively). Subclade 5 ‘fascians’ branches off at an early bifurcation in the Rhodococcus phylogeny (Anastasi et al., 2016) (Fig. 5).

The general conclusion of these WGS studies is that R. equi is a prototypic Rhodococcus, and that if a new genus were to be created to accommodate the species, the same treatment would need to be applied for each of the major rhodococcal lineages (Anastasi et al., 2016). Such an atomization of the genus seems unjustified because Rhodococcus forms a distinct and biologically coherent and uniform monophyletic grouping comparable in rank and diversity to other well‐established Corynebacteriales genera, such Corynebacterium, Gordonia or Mycobacterium (Anastasi et al., 2016) (Fig. 5). It would also defeat the very purpose of bacterial nomenclature in facilitating the coherent study of evolutionarily and biologically related organisms assembled under a common taxon name.

Corynebacteriales phylogenomics

The phylogenomic analyses by Anastasi et al. (2016) also clarified the evolutionary relationships of the rhodococci with other Corynebacteriales, in particular Nocardia, inconsistently resolved by previous 16S rDNA phylogenies (Rainey et al., 1995; Goodfellow et al., 1998; McMinn et al., 2000; Jones and Goodfellow, 2012; Ludwig et al., 2012; Kämpfer et al., 2014). Rhodococcus and Nocardia form two clearly distinct clades within a well‐supported phyletic line that also comprises Smaragdicoccus niigatensis DSM44881T, classified in the Nocardiaceae (as is Rhodococcus), as well as Mycobacterium spp. and Amycolicicoccus subflavus DQS3‐9A1T, a single‐species genus of the Mycobacteriaceae (Fig. 5). Another major Corynebacteriales line of descent is formed by members of the genera Gordonia and Williamsia, classified in the Nocardiaceae, and Tsukamurella of the monogeneric Tsukamurellaceae (Fig. 5). Two additional major lines of descent are clearly defined, one encompassing Corynebacterium, Turicella otitidis ATCC15513T (Corynebacteriaceae) and Dietzia (family Dietziaceae), the other corresponding to Segniliparus spp. (family Segniliparaceae) (Anastasi et al., 2016) (Fig. 5).

The phylogenomic data, therefore, indicate that the current Nocardiaceae taxon is polyphyletic and call for a possible reclassification of the Corynebacteriales into four families: a) Mycobacteriaceae, including the genera Mycobacterium, Amycolicicoccus, Smaragdicoccus, Rhodococcus and Nocardia; b) Gordoniaceae, with the genera Gordonia, Williamsia and Tsukamurella; c) Corynebacteriaceae, with the genera Corynebacterium, Turicella and Dietzia; and d) Segniliparaceae.

Conservation of Kämpfer et al. 2014

While, as discussed above, there appears to be no reasonable grounds for transferring R. equi to a new genus, the problem remains with the epithet hoagii. Though valid and legitimate in strict nomenclatural terms, the name R. hoagii is met with rejection in the context where the organism is relevant. The name R. equi is very well established, widely accepted and in widespread use, not only among the veterinary community and equine industry but also the medical profession where the bacterium is recognised as a human opportunistic pathogen. The epithet equi suitably encapsulates the very essence of R. equi and its significance for the communities concerned and the public. In contrast, the hoagii epithet has remained largely in disuse, restricted to an obscure type strain characterized by features such as production of oxoalkylxanthines and pregnadienes, with no obvious connection with the identity of R. equi as a well‐known pathogen.

Apart from the confusion already generated due to the use of the new epithet hoagii in gene repositories and genomic databases, the application of R. hoagii is likely to cause significant problems of traceability and interpretation of the literature. We believe that this falls under the concept of nomen perplexum, one of the exceptions allowing rejection of a bacterial name (Rule 56a.4 of the Bacterial Code, perplexing name: ‘a name whose application is known but which causes uncertainty in bacteriology’) (Lapage et al., 1992). This situation would be definitely compounded if in addition the genus name were to be changed (see below).

Where there is no doubt that the epithet hoagii meets the provisions for rejection is under Rule 56a.5 nomen periculosum, ‘a name whose application is likely to lead to accidents endangering health or life or both or of serious economic consequences’ (Lapage et al., 1992). We cannot think of a more accurate adherence to the notion of a perilous name. There are real chances of potential misdiagnoses or inaccurate risk appraisals due to confusion generated as a result of the introduction of the name R. hoagii to designate a pathogenic microbe with a previously well consolidated and recognized name.

The above points have been already evoked by Garrity (2014) in his Request for an Opinion (RfO) to the Judicial Commission of the ICSP for conservation of R. equi as the valid and legitimate name for the taxon. We entirely align ourselves with the views expressed by our colleague. However, Garrity's RfO was formulated to reject the name Corynebacterium hoagii and since then the name Rhodococcus hoagii has been validated by inclusion in a Notification List of approved names (Oren and Garrity, 2014). A formal request – which we hereby are formulating – is therefore needed for the conservation of Rhodococcus equi (Magnusson, 1923) Goodfellow and Alderson 1977 and rejection of Rhodococcus hoagii (Morse 1912) Kämpfer et al. 2014.

Concluding remarks

The same arguments as above apply for the reassignment of the species to a new genus ‘Prescotella’, still advocated by its proposers (Goodfellow et al., 2015) despite its evident biological inadequacy, the confusion that it will create, the rejection by the R. equi community (Cauchard et al., 2013), and the implications for the nomenclature of the whole genus Rhodococcus (Letek et al., 2010; Anastasi et al., 2016).

Probably as a side effect of the nomenclatural debate stemming from the proposed reclassification of R. equi into a new genus ‘Prescottia/Prescotella’ which turned out to be illegitimate, a significant potential new difficulty arose with the realization that Rhodococcus Zopf 1891 may be illegitimate itself, because a later homonym of the algal genus Rhodococcus Hansgirg 1884 (Tindall, 2014b). Can one reasonably conceive changing well‐established bacterial names such as Staphylococcus, Escherichia or Salmonella? The situation with R. equi is essentially analogous. The bacterial genus Rhodococcus Zopf 1891 has a comparable standing in the scientific literature and a potential change may have disastrous consequences. The same applies to the name R. equi in veterinary and medical microbiology. This extends to the terminology of the infectious disease itself: ‘rhodococcal pneumonia’, ‘rhodococcal infection’ and ‘foal rhodococcosis’ are all well established and commonly used terms in the veterinary, medical and professional literature.

The phylogenomic confirmation of R. equi as a bona fide Rhodococus contributes to the picture of a highly adaptable genus of Actinobacteria with a diversity of lifestyles, from saprotrophic biodegraders to plant pathogens and animal intracellular parasites. Further illustrating the unique versatility of this group of bacteria, the major Rhodococcus phyletic line within which R. equi evolved contains species with both circular and linear chromosomes. The genus Rhodococcus as it currently stands, including R. equi, thus serves to consolidate the important notion that genome topology is primarily a consequence of genome size and has no intrinsic taxonomic value. Together with the shared plasmid‐driven niche‐adaptive strategy, it showcases the extraordinary flexibility of the bacterial genome to ensure rapid accommodation to different ecological scenarios.