Keywords

1 Introduction

The Rhodococcus genus comprises of Gram-positive, nonmotile, nonsporulating, aerobic bacteria, with a high G+C content and a mycolic acid-containing cell wall. Members of Rhodoccocus genus are genetically and physiologically diverse bacteria, widely distributed in soil, water and marine sediments; some Rhodococcus spp. are also pathogens for plants (R. fascians), animals and humans (R. equi). Members of Rhodococcus have also been recovered from extreme environments such as the deep sea, oil-contaminated soils and freeze-thaw tundra on glacial margins (Sheng et al. 2011; Shevtsov et al. 2013; Konishi et al. 2014).

Rhodococcus genus is featured by a broad metabolic versatility and environmental persistence supporting its clinical, industrial and environmental significance. In particular, Rhodococcus strains have peculiar degradative capacities towards a variety of organic compounds, including toxic and recalcitrant molecules like chlorinated hydrocarbons (Grӧning et al. 2014; Cappelletti et al. 2017), herbicides (Fang et al. 2016), 4,4′-dithiodibutyric acid (DTDB) (Khairy et al. 2016) and dibenzothiophene (DBT) (Tao et al. 2011) as well as the ability to resist to various stress conditions (desiccation, radiation, heavy metals) (LeBlanc et al. 2008; Taketani et al. 2013; Cappelletti et al. 2016). They are also able to mediate a broad range of biotransformation, including enantioselective syntheses, and to produce biosurfactants, which facilitate the cell contact with hydrophobic substrates (Martìnkovà et al. 2009). Due to their metabolic flexibility and their tolerance to various stresses, they play an important role in nutrient cycling and have potential applications in bioremediation, biotransformations and biocatalysis.

Because of their biotechnological applications, the massive application of high-throughput genomic technologies has dramatically increased the number of sequenced Rhodococcus genomes, and a great effort has been directed towards the computational analysis of genomic data. Since the first complete genome (of R. jostii RHA1) published in 2006, an exponentially increasing number of genomes of Rhodococcus strains have been sequenced due to the progress of the sequencing technologies and to the reduction of the DNA sequencing costs. In July 2018, 218 complete genomes of Rhodococcus are available in NCBI (www.ncbi.nlm.nih.gov/genomes/GenomesGroup.cgi?taxid=1827), most of them being available as draft genomes. Collectively, the analysis of genome sequences has provided insights into the genetic basis for diversity, plasticity and adaptation characteristic of Rhodococcus spp.

2 The Rhodococcus Genome: General Features and Structure

Rhodococcus genomes show high GC amount (61–71%) and range in size from around 4 Mb to over 10 Mb. In particular, the genome size seems to be at least partially dependent on the Rhodococcus strain lifestyle and niche complexity; indeed, the pathogenic R. equi has a substantially smaller genome than the soil-associated versatile biodegrader R. jostii RHA1 (9.7 Mb) and the other two environmental rhodococci, Rhodococcus erythropolis PR4 (6.9 Mb) and Rhodococcus opacus B4 (8.2 Mb) (Letek et al. 2010). Indeed, obligate pathogens live inside a host, reducing the need to adapt to sudden environmental changes and presence of competitors, which may contribute to genome minimization (Moran 2002).

Among Rhodococcus strains of the same species, the amount of GC in the genome was quite constant, showing the highest values (>70% GC-content) in R. aetherivorans and R. ruber species (Table 1). On the other hand, a remarkable variability was observed in the rhodococcal genome structure and plasmid presence.

Table 1 Genomic features of Rhodococcus strains selected on the basis of the completeness of the genomes sequence and/or on the availability of functional studiesa
Full size table

Chromosomal and plasmid DNA of Rhodococcus strains is either linear or circular. Both R. jostii RHA1 and R. opacus B4 have linear chromosomes, while R. erythropolis PR4 and R. equi chromosomes are circular (Table 1). Interestingly, chromosome topology does not seem to correlate with taxonomy or phylogenetic relationship, as R. equi and R. erythropolis belong to different subclades, and the latter is considered the prototype of the “erythropolis subclade”, which also includes R. opacus (Letek et al. 2010). Further, not all actinomycetales genera present linear chromosomes. Streptomycetes typically have linear chromosomes of very large size (>8.5 Mb), so linearization appears to have occurred independently in different actinobacterial lineages during evolution, apparently in association with increasing genome size (Bentley et al. 2002).

The co-existence of large linear and circular plasmids of different sizes within the same cell is also a characteristic of several actinomycetes genera; however, linear plasmids are mainly described in Streptomyces and Rhodococcus strains. The linearity of DNA molecules in Rhodococcus strains has been mainly associated to the presence telomere-like structures comprising of terminal inverted repeats (TIRs) and proteins bound to their 5′ ends (Kikuchi et al. 1985; Dib et al. 2015). This structure, named as invertron, has been detected in both chromosomes and plasmids of Rhodococcus and has been extensively studied in Streptomyces strains (Ventura et al. 2007). Most commonly, both ends of linear plasmids are the same or featured by only few mismatches, giving recognizable terminal inverted repeats (TIRs). For instance, plasmid pHG207 of R. opacus MR22 carries an imperfect terminal inverted repeat (TIR) of 583/560 bp (Kalkus et al. 1993). The RHA1 replicons have been described as typical actinomycete invertrons, containing two sets of inverted repeats flanking the GCTXCGC central motif with covalently associated proteins (Warren et al. 2004). The chromosomal inverted repeats (of around 10 Kp in RHA1) are much longer than those of the plasmids (e.g. the pRHL1 telomeres are 500 bp). The RHA1 telomeric inverted repeats are similar to those of Streptomycetes, particularly over the first 300 bp. The replication of linear Streptomyces chromosomes and plasmids is initiated from a fairly centrally located replication origin rich in DnaA box sequences and proceeds bidirectionally towards the telomeres. The telomeres themselves are replicated by a mechanism that includes priming from the terminal proteins covalently bound to the 5′ ends. It has been proposed that linear plasmids evolved from bacteriophages and that linear chromosomes originated from single-crossover recombination between an initially circular chromosome and a linear plasmid (McLeod et al. 2006; Ventura et al. 2007).

2.1 Genomic and Metabolic Traits of Reference Rhodococcus Strains

The analysis of single bacterial genomes and its annotation confer a wide range of knowledge in relation to the general genome features and chromosomal structure but also in relation to identification of specific genes in a genome and the detection of regions containing functionally connected genes. Although most of them are still in a draft version, the sequencing of individual Rhodococcus genomes has often provided important inputs for wet-lab experiments aimed at investigating the relevant metabolic capacities shown by the isolates.

Table 1 reports the genome features of representative Rhodococcus strains belonging to different species featured by specific metabolic and physiological traits.

Members of Rhodococcus jostii, R. opacus and R. wratislaviensis species possess genomes of very large size, ranging from 8.5 to over 10 Mb (Table 1). R. jostii RHA1 is considered the model of the genus as its genome was the first completely sequenced from a Rhodococcus strain. R. jostii RHA1 was isolated from lindane-contaminated soil (McLeod et al. 2006) and was widely investigated for the strong ability to degrade a wide variety of aromatic compounds and polychlorinated biphenyls (PCBs), for lipid accumulation and for stress condition resistance (Gonçalves et al. 2006; LeBlanc et al. 2008; Patrauchan et al. 2012; Costa et al. 2015).

R. opacus R7 and R. wratislaviensis NBRC 100605 have the largest Rhodococcus genomes described up to date (10.1 and 10.4 Mbp, respectively) and two of the largest bacterial genome reported in the literature. While NBRC 100605 metabolic features have been scarcely investigated (Guo and Wu 2017), R. opacus R7 has been thoroughly studied considering both the broad degradative abilities towards aromatic and aliphatic hydrocarbons and the associated genome features (Di Gennaro et al. 2001, 2010; Zampolli et al. 2014; Orro et al. 2015; Di Canito et al. 2018). Based on the terminal sequence signatures defining the possible linearity of replicons, R7 possesses one linear chromosome and additional five linear replicons, ranging in size from 20 Kb to 430 Kb (Tables 1 and 2) (Di Gennaro et al. 2014). Other Rhodococcus opacus strains described from a genomic point of view have broad organic compound degradative abilities and capacity to produce and accumulate lipids for possible biodiesel application. In particular, R. opacus PD630 is the most studied bacterium for its ability to produce and accumulate lipids (mostly triacylglycerols, TAGs) using different carbon sources (Alvarez et al. 1996; Castro et al. 2016). Indeed, it was shown to accumulate significant amounts of lipids, namely, 76 and 87%, of its cellular dry weight, when grown on gluconate and olive oil, respectively (Alvarez et al. 1996; Voss and Steinbuchel 2001). R. opacus B4 was the first R. opacus with a completely sequenced genome; it was described to be able to transform and degrade several types of hydrocarbons, to stabilize water-oil phases and to accumulate TAG (Na et al. 2005; Honda et al. 2008; Sameshima et al. 2008). The analysis of R. opacus B4 genome revealed an 8.8-Mb-long genome composed by one linear chromosome, two linear plasmids of pROB series of size range of 2–4 Kb and three circular plasmids of pKNR series of size range of 110–550 Kb (Table 2).

Table 2 Details on the plasmid content of reference Rhodococcus genomes from Table 1 a
Full size table

Members of R. erythropolis are largely distributed in the environment and exhibit a remarkable metabolic versatility related to their capacity to degrade complex compounds, such as quorum-sensing signals N-acylhomoserine lactones, phenols, sterols and fuel derivatives (Table 1). Moreover, many R. erythropolis strains have been isolated from petroleum-contaminated sites and have been described for their ability to degrade various hydrocarbons using multiple catabolic pathways (de Carvalho and da Fonseca 2005). All the R. erythropolis strains whose complete genome has been sequenced showed the presence of plasmids ranging in length from 3 Kb to 261 Kb (Table 2).

The genomes of R. aetherivorans and R. ruber species showed the highest amount of GC, being around 70% in all the strains described up to date. Members of these two species have been described for different biosynthetic and biodegradative capacities. R. aetherivorans BCP1 (formerly Rhodococcus sp. BCP1) is able to degrade a wide range of alkanes, naphthenic acids and chlorinated alkanes (Frascari et al. 2006; Cappelletti et al. 2011, 2015; Presentato et al. 2018a; Ciavarelli et al. 2012) and to produce metal-based nanostructures (Presentato et al. 2016, 2018b, c). BCP1 has a genome of 6.2 Mbp composed by one chromosome and two plasmids (of 100–120 Kb size) (Tables 1 and 2). R. aetherivorans IcdP1 was isolated from an abandoned coking plant; it was proven to be able to degrade numerous high-molecular-weight polycyclic aromatic hydrocarbons and organochlorine pesticides (Qu et al. 2015). Although their genome data are not available, R. aetherivorans IAR1 was described to accumulate polyhydroxybutyrate PHB, while R. aetherivorans IG24 was employed by Merck Corporation for the production of the anti-HIV drug Crixivan™ (Treadway et al. 1999; Hori et al. 2009).

Rhodococcus pathogenic strains belong to R. equi and R. fascians species, which cause disease to animals and plants, respectively. In line with their lifestyle inside a host, their genomes are generally smaller (<6 Mb) compared to the environmental Rhodococcus strains (Table 1). Recently, numerous genomes of both these two species have been sequenced to conduct phylogenomic analyses and to investigate on the evolutionary mechanisms of virulence traits (Letek et al. 2010; Creason et al. 2014a, b). In particular, virulence genes have been associated to plasmids. Several R. equi strains have been described to possess circular plasmids, whereas R. fascians members are featured by linear plasmids. The presence of linear plasmids in R. fascians has been correlated with phytopathogenicity, although it was described not to be strictly necessary (Creason et al. 2014a).

Compared to the other species, only few genomic studies are available for R. qingshengii, R. pyridinivorans and R. rhodochrous. Nevertheless, members of these species have been reported to possess peculiar metabolic activities such as degradative capacity towards rubber, fungicide and other hydrocarbon-related complex molecules. In relation to this, additional genome sequencing efforts promise to expand the knowledge on the genetic and metabolic diversity of Rhodococcus bacterial strains (Table 1) (Xu et al. 2007; Dueholm et al. 2014; Watcharakul et al. 2016).

3 Comparative Genomics of Rhodococcus

Comparative genomics allow providing genomic basis for the differences observed among microorganisms in terms of phenotypes, metabolic capacities and cellular response to specific environmental conditions. Recent studies have described genome-wide comparative analyses of Rhodococcus strains with the aim of defining the phylogeny and evolutionary relationship, determining the Rhodococcus genus core genome, analysing their catabolic potentials and stress response (Orro et al. 2015; Cappelletti et al. 2016). Genome comparative analysis is performed through pairwise or multiple (>3 genomes) whole-genome alignments. Among the applicable programs, Artemis Comparison Tool (ACT), available within the Integrated Microbial Genomes (IMG) system, was used for genome comparison of strain R. opacus M213 with different Rhodococcus strains (Pathak et al. 2016). Mauve (http://darlinglab.org/mauve/mauve.html) performs multiple sequence alignment and has been used for comparative genomic analyses of different Rhodococcus strains, e.g. R. opacus R7 and R. aetherivorans BCP1 (Orro et al. 2015), R. opacus M213 (Pathak et al. 2016) and Rhodococcus fascians strains (Creason et al. 2014b) (Fig. 1).

Fig. 1
figure 1

Whole-genome sequence alignment with program Mauve diagram. The alignment of three pairs of genomes is shown: R. jostii RHA1 and R. opacus R7 (panel A), R. jostii RHA1 and R. aetherivorans BCP1 (panel B), R. opacus R7 and R. aetherivorans BCP1 (panel C). The coloured region represents a block of sequences that is collinear to a corresponding block of sequences in the other genome sequence. [Modified from Orro et al. (2015)]

Full size image

Comparative genome analysis allows to detect genomic regions that are conserved or syntenic (Pathak et al. 2016), representing conserved genome sections that have not undergone internal rearrangements and inversions. This conservation typically reflects phylogenetic correlation and represents the indication of functional evolutionary constraints. In comparative studies concerning Rhodococcus, the conservation of collinear blocks in different R. fascians strains was shown to be in accordance with their phylogenetic relationship (Creason et al. 2014b; Pathak et al. 2016). The 16S rRNA-based affiliation of Rhodococcus sp. M213 to opacus species was challenged by the higher M213 syntenic affiliation with R. jostii RHA1 compared to R. opacus strains B4 and PD630 (Pathak et al. 2016).

Another useful way to compare whole genomes is by computing a value that summarizes their similarity or distance. The average nucleotide identity (ANI) value calculated from whole-genome alignments was described to provide a tool for taxonomy and evolutionary studies and species definition (Creason et al. 2014b; Anastasi et al. 2016). Whole-genome alignment analysis of three Rhodococcus equi strains allowed to detect single nucleotide polymorphisms (SNPs) in specific coding regions that were suggested to possibly reflect differences in the lifestyle of these three isolates (Sangal et al. 2014).

In addition to gene/protein sequence-based alignment, gene content comparison can be done in terms of function using, for instance, the concept of functional categories (available in RAST subsystem) or COG (cluster of orthologous groups of proteins). Typically, statistical tools analyse gene functions (or annotations) as categorized by COGs and determine whether certain functions or functional categories are statistically enriched in certain genomes compared to others. In this respect, the R. opacus M213 genome analysis at functional (COG) level further reinforced the closer relationship of M213 with R. jostii RHA1 and other Rhodococcus strains (e.g. R. wratislaviensis IFP 2016, R. imtechensis RKJ300, Rhodococcus sp. strain DK17) compared to R. opacus strains (Pathak et al. 2016). On the other hand, by comparing the RAST-based functional categories, R. opacus R7 genome resulted to be enriched in genes belonging to all the functional categories, compared to R. aetherivorans BCP1, except for those involved in some central metabolic process (e.g. DNA, iron, potassium and phosphorous metabolism, dormancy and sporulation, motility and chemotaxis) (Orro et al. 2015).

3.1 Phylogenomics

The use of the only 16S rRNA sequence as taxonomic marker was proved to limit the resolution of Rhodococcus phylogenetic studies (Gürtler et al. 2004; Pathak et al. 2016; Creason et al. 2014b) and more in general to lead to artefacts in the bacterial phylogenetic tree construction (Ventura et al. 2007). More recent phylogenetic investigations have been carried out using alternate gene/protein sequences (e.g. functional genes or molecular chaperone-encoding genes) as phylogenetic markers, which often evolve faster than the rRNA operon. This provides a larger phylogenetic resolution that can help to distinguish closely related bacterial species. For this purpose, the alkane monooxygenase (alkB) gene was applicable as phylogenetic marker of Rhodococcus strains (Táncsics et al. 2015).

The availability of genome sequences has provided new opportunities to get information on applicable phylogenetic markers, and new approaches arose to examine rhodococcal phylogeny based upon multiple gene or protein sequences. Among these approaches, the construction of phylogenetic trees based on concatenated sequences for large numbers of proteins (also called multilocus sequence analysis—MLSA) has proven particularly useful for evolutionary relationship studies (Gao and Gupta 2012). Phylogenetic tree based upon multiple conserved (or slow-evolving) genes/proteins was demonstrated to properly assign environmental isolates to existing Rhodococcus species (Orro et al. 2015) and to resolve evolutionary relationships of Rhodococcus strains belonging to the same species (Creason et al. 2014b; Letek et al. 2010). MLSA mainly employed sequence concatenamers of housekeeping genes including 16S rRNA, recA, gyrB, rpoB, rpoC and secY (Orro et al. 2015; Kwasiborski et al. 2015). Orro et al. (2015) assigned the species to two Rhodococcus strains on the basis of MLSA with four of these marker genes (Fig. 2). MLSA provided a framework for defining Rhodococcus genus within the Actinobacteria phylum using seven marker genes (ftsY, infB, rpoB, rsmA, secY, tsaD and ychF) (Creason et al. 2014b). In line with the whole-genome based studies by Letek et al. (2010), this study placed the R. equi species within the Rhodococcus genus, which was in contrast with other inconclusive phylogenetic studies based on 16S rDNA sequences (Goodfellow et al. 1998). More genome sequences and informative marker sequences might help in improving MLSA-mediated taxonomy and discern the finer details within the suborder. In this respect, available genome sequences were used to discover novel molecular characteristics that could be used for evolutionary and systematic studies of Rhodococcus genus. In this regard, novel specific conserved signature indels (CSIs) and conserved signature proteins (CSPs) have been proposed as possible molecular markers on the basis of their conservation among different Rhodococcus species (Gao and Gupta 2012).

Fig. 2
figure 2

Multilocus sequence analysis-based tree of 28 Rhodococcus strains. Phylogenetic tree of 28 Rhodococcus strains based on MLSA using the sequence of the four marker genes, 16S rRNA, secY, rpoC and rpsA. [Modified from Orro et al. (2015)]

Full size image

In addition to phylogenomic analysis based on the concatenation of different taxonomic markers, Rhodococcus genome-based trees have been constructed on the basis of average sequence similarity. The genome similarity is at the basis of the wet-lab DNA-DNA hybridization (DDH) technique, which represents the gold standard for bacterial taxonomy classification (Richter and Rosselló-Móra 2009). In the era of genomics, different attempts have been made to replace the traditional wet-lab DDH with in silico genome-to-genome comparison (Goris et al. 2007; Richter and Rosselló-Móra 2009). Among these, the average nucleotide identity (ANI) estimates the average nucleotide identity between two genomic datasets, and an ANI value in the range of 94–96% has been proposed to be the criterion for bacterial species delineation (corresponding to a wet-lab DDH value of 70%), while ANI values of 70–75% identified members of the same genus (Konstantinidis and Tiedje 2005; Goris et al. 2007; Richter and Rosselló-Móra 2009). Creason et al. (2014b) conducted phylogeny study of Rhodococcus genus by using the average nucleotide identity (ANI) approach and found that the genus Rhodococcus can be represented by as many as 20 distinct species (Fig. 3).

Fig. 3
figure 3

Average nucleotide identity (ANI) dendrogram for 59 isolates of Rhodococcus. Complete genome sequences for 59 isolates of Rhodococcus were used to generate an ANI matrix. The matrix was used to calculate an ANI divergence dendrogram. Branches are coloured based on the ANI threshold values. [Modified from Creason et al. (2014b)]

Full size image

Using 27 de novo sequenced R. equi genomes, the ANI value was 99.13%, well above the consensus 95–96% threshold for prokaryotic species definition (Anastasi et al. 2016). This result highlighted the high genetic homogeneity of R. equi group and was found to correspond to a 16S rRNA sequence identity of 100%.

3.2 Core and Accessory Genome

Whole-genome comparative analysis can be used to obtain information on the pan genome, core genome and accessory genome of a group of bacterial strains belonging to a single species or different species belonging a single genus. The pan genome comprises all the genes present in a given bacterial group (the collection of all genetic material). The core genome is defined as the set of all genes shared by the genomes as orthologs and therefore considered as conserved by all the bacterial strains under analysis (Ventura et al. 2007). The genes included within the core genome are supposed to have strong functional significance for cell physiology and replication. Moreover, as the core genome can be considered to define the members of the bacterial group under analysis, the core genes are suitable as molecular targets to infer phylogeny (Ventura et al. 2007). On the other hand, the accessory or variable genome is represented by the genes, which are confined to a single member of the bacterial group and are therefore specific for a single strain (Ali 2013). The accessory genome therefore represents those gene families that can be associated with phenotypic traits that differentiate each member within a given bacterial population under analysis.

Comparative genomic studies on Rhodococcus strains have assessed the core genome shared by either strains belonging to the same species or strains belonging to different species. The resulting core-genome size and identity reported in each work was dependent on the Rhodococcus strains under analysis. Within the Rhodococcus genome set composed by R. opacus R7, R. aetherivorans BCP1, R. jostii RHA1, R. opacus PD630, R. opacus B4 and R. pyridinivorans SB3094, the core genome corresponded to around 50% of all the identified CDSs in each strain (Orro et al. 2015) (Fig. 4a). This indicated a large genetic variability among these Rhodococcus strains, some of them belonging to different species. The portion of core genes increased by considering Rhodococcus strains belonging to the same species, such as R. opacus R7, B4 and PD630 (Fig. 4a) (Orro et al. 2015), R. erythropolis (Kwasiborski et al. 2015) and R. equi (Anastasi et al. 2016). In particular, the genome of R. erythropolis R138 was compared to the genomes of other ten R. erythropolis strains, either available in the database or newly sequenced. The core-genome genes represented up to 87% of all the CDSs identified in R. erythropolis R138. A large core genome was also identified in a comparative study with 29 R. equi strains, which showed to have ~80% of shared genes in their genome (Anastasi et al. 2016). Some of these comparative studies also found that the core genes were mainly localized on the chromosome, whereas the unique genes (accessory genome) were harboured by the endogenous plasmids in each strain (Orro et al. 2015; Kwasiborski et al. 2015). For instance, considering R. erythropolis strains, up to 97% of the shared sequences were located on the circular chromosome of R138 strain (Kwasiborski et al. 2015). These results are general indications of a core set of functions localized on the Rhodococcus chromosome, whereas plasmids have variable genetic contents. The variable accessory genome underlies the peculiar capacities of each Rhodococcus strain and contributes functions for niche adaptation (Kwasiborski et al. 2015). In line with this, considering R. equi strains, the core genome included traits necessary for bacterial physiology, virulence and niche adaptation, including tolerance to desiccation and oxidative stress (Anastasi et al. 2016). Similarly, R. erythropolis core genome was highly represented by gene categorized within the primary metabolism of amino acids, carbohydrates, cofactors and proteins (Kwasiborski et al. 2015). Interestingly, genes encoding enzymes involved in the degradation of complex molecules (i.e. N-acylhomoserine lactones, phenol, catechol and sterol derivatives) were also shared by the 11 R. erythropolis under analysis by Kwasiborski et al. (2015). On the other hand, functions encoded by unique genes (accessory genome) were associated to degradative enzymes in R. opacus M213 (Pathak et al. 2016), R. aetherivorans BCP1 and R. opacus R7 (Orro et al. 2015). In particular, R. aetherivorans BCP1 possessed unique genes involved in short-chain n-alkanes degradation  which were not present in the other Rhodococcus strains under analysis (Orro et al. 2015) Conversely, R. opacus R7 genome had a high number of unique genes involved in aromatic degradation metabolism. These genetic findings generally correlated with the different types of contaminants persisting in each Rhodococcus strain isolation source. (Di Gennaro et al. 2001; Frascari et al. 2006; Orro et al. 2015).

Fig. 4
figure 4

Venn diagram displaying the core-genome and the unique genes (accessory genome) specific to each strain. The number of genes shared by all strains included within each genome set (i.e. the core genome) is in the centre. Numbers in nonoverlapping portions of each oval show the number of genes unique to each strain. The total number of protein coding genes found in each genome is listed below the strain name. Panel (A), core-genome and unique genes resulting from the comparison of the genomes from R. opacus R7, R. aetherivorans BCP1, R. jostii RHA1, R. opacus PD630, R. opacus B4, R. pyridinivorans SB4094; panel (B), core-genome and unique genes resulting from the comparison of the genomes from R. opacus R7, R. opacus PD630 and R. opacus B4

Full size image

3.3 Functional Genomics

Genome sequence-based analysis relies on homology-based gene annotation and protein domain identification to annotate open reading frames (ORFs) and to perform functional predictions. Bioinformatic tools have been applied for the functional analysis of Rhodococcus genomes to determine the genetic basis possibly involved in relevant metabolic activities and phenotypic traits, e.g. degradation abilities of specific xenobiotics, plant growth-promoting effect and stress tolerance (Francis et al. 2016; Cappelletti et al. 2016; Pathak et al. 2016). For instance, genome-based analyses assessed the occurrence and distribution of key genes and pathways involved in the synthesis and accumulation of triacylglycerols (TAG), wax esters, polyhydroxyalkanoates, glycogen and polyphosphate (Hernández et al. 2008; Villalba et al. 2013). In line with the extensive experimental work demonstrating the ability of Rhodococcus strains to synthesize and accumulate neutral lipids, numerous genes/enzymes predicted to be involved in TAG biosynthesis and degradation and fatty acid β-oxidation were identified in the genomes of R. jostii RHA1, R. opacus PD630, R. opacus B4, R. erythropolis PR4, R. equi 103S and R. fascians F7 (Villalba et al. 2013). The comparative genome analysis of psychrophilic Rhodococcus sp. JG3 and other mesophilic rhodococci allowed to detect the cold adaptive traits which confer JG3 with the ability to survive the extremely arid, cold and oligotrophic conditions of permafrost (Goordial et al. 2016). A bioinformatic analysis of the genomes of 20 Rhodococcus strains allowed identifying numerous biosynthetic gene clusters encoding pathways involved in the possible production of novel secondary metabolites (Ceniceros et al. 2017).

Several works have assessed a large range of metabolic abilities of strains using Phenotype Microarray (PM) technologies. PM results were interpreted using genomic analysis to predict the genes involved in central metabolic pathways, xenobiotic degradation metabolisms and stress response (Letek et al. 2010; Holder et al. 2011; Orro et al. 2015; Cappelletti et al. 2016).

The large-scale functional analysis of Rhodococcus genomes has been further integrated with the results obtained through omic technologies. In particular, high-throughput experiments were conducted to obtain global datasets on the expression of genes (transcriptomic analyses through RNA-seq or microarray) and on the protein profiles induced under specific growth conditions or after specific cell treatments. Transcriptomic studies were performed to define the genes involved in the degradation of aromatic compounds and steroids (chlorate and cholesterol) (Gonçalves et al. 2006; van der Geize et al. 2007; Swain et al. 2012), of PCBs and diesel oil (Puglisi et al. 2010, Laczi et al. 2015) and of synthetic polymers (Gravouil et al. 2017) but also in lipid metabolism and accumulation (Chen et al. 2014; Amara et al. 2016) and isoprene metabolism (Crombie et al. 2015). Proteome studies were combined with genome analysis for the study of the degradation of xenobiotic 4,4-dithiodibutyric acid (DTDB) by R. erythropolis MI2 (Khairy et al. 2016) and the metabolism of short-chain alkanes in R. aetherivorans BCP1 (Cappelletti et al. 2015) and to study the global response to stress conditions such as desiccation and carbon starvation in R. jostii RHA1 (LeBlanc et al. 2008; Patrauchan et al. 2012). In many of these studies, the involvement of multiple homologous genes was identified in specific metabolic pathways, providing clues to the redundancy of the different catabolic pathways in Rhodococcus (Gonçalves et al. 2006; Swain et al. 2012). Further, some omic studies gave indications on the involvement of possible regulators in the specific catabolic pathways under analysis (Crombie et al. 2015). In R. ruber C208, transcriptomic results indicated the involvement of alkane degradation pathway and β-oxidation of fatty acids as the main catabolic route for polyethylene degradation; further, it also indicated metabolic limiting steps which could represent molecular target to optimize the biodegradation process (Gravouil et al. 2017). In addition to catabolic genes/proteins, many genes/proteins involved in stress response (e.g. chaperone-like proteins, superoxide dismutase, catalase/peroxidase) were reported to be induced during the degradation of organic compounds (Tomás-Gallardo et al. 2006; Puglisi et al. 2010; Khairy et al. 2016). The microarray analysis of R. aetherivorans I24 cells exposed to PCBs reported the transcriptional response to be primarily directed towards reducing oxidative stress rather than catabolism (Puglisi et al. 2010).

4 The Genomic Basis of Metabolic Versatility in Rhodococcus

The extraordinary metabolic versatility of Rhodococcus strains is reflected in their genomic attributes, like (1) large genome sizes (up to 10 Mbp) encoding numerous catabolic pathways for a variety of chemical compounds; (2) a significant degree of gene redundancy that ensures functional robustness and free-to-evolve genetic reservoirs; and (3) the occurrence of circular and linear (mega)plasmids that generally evolve more rapidly than the chromosome and consist of an additional pool of DNA that can evolve and can be easily transferred (Redenbach and Altenbuchner 2002; Van der Geize and Dijkhuize 2004). In this context, the broad Rhodococcus adaptability is related to their genome “flexibility” which refers to the genomic rearrangements mainly occurring on the large linear plasmids and due to still mostly unknown molecular mechanisms promoting the frequent non-homologous illegitimate recombination. This last aspect has been discussed in several review articles by Larkin and collaborators (Larkin et al. 1998, 2010; Kulakov and Larkin 2002). Additionally, recent works have underlined how the equipment of genes for the transport of many different substrates (de Carvalho et al. 2014) and the numerous genes encoding oxygenases, typically catalysing the hydroxylation and cleavage of organic compounds, are at the basis of the wide metabolic capacities of Rhodococcus strains. Only a few portion of the oxygenase genes have been acquired through horizontal gene transfer events, while most of them were chromosomally located, suggesting their fundamental role in Rhodococcus physiology (Mcleod et al. 2006).

4.1 Catabolic Gene Redundancy

Since the first complete genome sequence obtained from a Rhodococcus strain, the presence of multiple catabolic genes encoding homologous enzymes, called isozymes, attracted research interest. Recently, the analysis of whole-genome sequences of several rhodococci confirmed the redundancy in catabolic pathways initially only hypothesized for this genus (McLeod et al. 2006). Catabolic gene redundancy is often considered at the basis of the Rhodococcus catabolic versatility, functional robustness, adaptation to polluted and extreme environments and high-performing environmental competition (Mcleod et al. 2006; Pérez-Pérez et al. 2009). Several examples of genetic redundancy originated either from gene duplication or horizontal gene transfer events have been described in Rhodococcus strains, the latter being supported by the proximity of transposase and invertase sequences to duplicated gene clusters (Taguchi et al. 2007). In general, the duplicate genes have been described to evolve and to encode isoenzymes possessing similar sequences but different substrate specificities and induction patterns. On the other hand, some duplicates have been described to be, at some extent, functionally redundant (Zhang 2012), catalysing the same metabolic reaction and presenting overlapping substrate range and inducing profiles. This type of duplication has probably occurred in relative recent period of time or has not been subject to strong selective pressure. For instance, in R. jostii RHA1, it has been hypothesized that the multiple homologous genes involved in the central aromatic pathways have received low selective pressure for functional diversification or gene removal (McLeod et al. 2006).

In this framework, the presence of multiple homologous genes strongly contributes to the wide Rhodococcus versatility in terms of catabolic activities including aliphatic and aromatic hydrocarbon catabolism, chlorinated hydrocarbons transformation and steroid degradation. Further, the presence of multiple functional homologs was found in central metabolic pathways (e.g. tricarboxylic acid cycle) (Van der Geize and Dijkhuize 2004).

While the presence of multiple copies of specific genes was not a universal feature of Rhodococcus, genomic redundancy generally characterizes the members of this genus in relation to metabolic and physiological traits useful for the adaptation to the environmental niche from which each Rhodococcus strain was isolated. For instance, the genome analysis of the psychrophilic Rhodococcus sp. JG3 isolated from a permafrost soil with a possible alkane source showed genomic redundancy in relation to genes involved in the adaptation to cold temperatures such as those associated to osmotic stress and to genes encoding alkane 1-monooxygenases (Goordial et al. 2016).

Functional redundancy has referred to the multiple alkane hydroxylase genes found in Rhodococcus strains featured by versatile alkane degradation capacity and typically isolated from petroleum-contaminated (sediment or marine) sites (van Beilen et al. 2002) (Fig. 5a). In particular, alkB gene encodes the membrane alkane hydroxylase AlkB, which catalyses the initial oxidation of alkanes. Four alkB homologous genes have been detected in R. erythropolis NRRL B-16531 and R. qingshengii Q15 (Whyte et al. 2002); three to five alkB genes were found in different R. erythropolis strains from various contaminated soils and in Rhodococcus sp. strain TMP2 (van Beilen et al. 2002; Takei et al. 2008). The multiple alkB genes found in Rhodococcus strains were mainly localized on the chromosome, and some of them showed high sequence divergence and different flanking regions (Whyte et al. 2002). The different AlkBs often catalysed the oxidation of different ranges or classes of alkanes and were differently expressed depending on the substrate and growth condition (Takei et al. 2008; Laczi et al. 2015). Because of this, the presence of multiple AlkB hydroxylases has been associated to broad metabolic capacities of Rhodococcus strains towards medium- and long-chain alkane and also to branched alkanes. In R. aetherivorans BCP1, two homologous gene clusters encoding alkane monooxygenases of SDIMO (soluble di-iron monooxygenases) family were involved in the oxidation of short-chain alkanes (Cappelletti et al. 2013, 2015). This functional redundancy was correlated with the strong capacity of BCP1 strain to utilize gaseous alkanes and to co-metabolize chlorinated alkanes (Frascari et al. 2006; Cappelletti et al. 2012). In some cases, alkane hydroxylation functional redundancy was also associated to the co-existence in a single Rhodococcus strains of genes encoding different alkane hydroxylases, AlkB, SDIMOs and cytochrome P450 belonging to the CYP153 family. For instance, the versatile and efficient degradation of alkanes by R. erythropolis PR4 was associated to the presence of four alkB genes, two CYP153 genes and other genes coding P450 on its genome, which are differently expressed on alkanes and hydrocarbon mixtures (Laczi et al. 2015).

Fig. 5
figure 5

Examples of genetic redundancy in Rhodococcus. Genetic organization of (a) alkB genes in Rhodococcus strain NRRLB-16531, (b) catabolic island including tpa, pat and pad gene that is duplicated in the two linear plasmids in R. jostii RHA1, (c) the three homologous gene clusters (bphA, etb1A and etb2A genes) encoding dioxygenase systems involved in the initial hydroxylation of substituted benzenes in R. jostii RHA1. The plasmidic or chromosomal localization is indicated in correspondence to each gene cluster. [Modified from Whyte et al. (2002), Hara et al. (2007), Gonçalves et al. (2006), respectively]

Full size image

Catabolic gene redundancy in some Rhodococcus strains has been extensively reported for aromatic compound degradation pathways. In particular, R. jostii RHA1, isolated from γ-hexachlorocyclohexane (lindane)-contaminated site, was described to be featured by high redundancy of catabolic pathways involved in aromatic hydrocarbon catabolism, which are also responsible for the co-metabolic transformation of polychlorinated biphenyls (PCBs). Several homologous genes were predicted to encode enzymes involved in the upper and lower degradation pathways of substituted benzenes, like ethylbenzene (ETB) and biphenyl (BPH). In RHA1, three homologous gene clusters (bphA, etb1A and etb2A genes) were found to encode dioxygenase systems, which are all induced by BPH and ETB by a possible common regulatory system (Fig. 5b). Nevertheless, these dioxygenase systems have shown distinct substrate specificity in terms of both aromatic hydrocarbons and PCB congeners (Iwasaki et al. 2006, 2007; Patrauchan et al. 2008). The gene cluster bphA is localized on the large linear plasmid pRHL1, while etb1A and etb2A are carried on the other linear plasmid pRHL2 and share high sequence similarity. Multiple copies of bph genes (bphC-G) were also predicted to be involved in the benzene ring cleavage downstream of the initial hydroxylation. Among these multiple homologs, only a few numbers of genes were transcriptionally induced during the growth on BPH and ETB, while the most part was expressed at constitutive levels (Gonçalves et al. 2006). In particular, only three out of the eight bphEFG clusters were up-regulated during RHA1 growth on BPH and ETB, independently on their genome localization (Gonçalves et al. 2006; Patrauchan et al. 2008). This indicated that out of the eight bphEFG, five homologous clusters produced paralog enzymes involved in distinct physiological processes (Irvine et al. 2000; Taguchi et al. 2004). Other Rhodococcus strains, isolated from the termite ecosystem typically exposed to plant-derived lignin and aromatics, showed multiple copies of bph gene clusters involved in PCB/biphenyl degradation; many of these were present on linear plasmids and in the proximity of transposase and invertase sequences (Taguchi et al. 2007). In addition to the catabolic enzymes, functional redundancy was found in the regulatory system responsible for the growth of RHA1 on aromatic compounds. In particular, two copies of the gene cluster bphS and bphT encode the two-component systems, BphS1T1 and BphS2T2, which showed high similarity in the amino acid sequence (>92%) and the same substrate spectrum, except for biphenyl (Takeda et al. 2010).

Additional genomic redundancy found in the genome of aromatic Rhodococcus degraders was related to genes involved in the catabolism of phthalate (pad), terephthalate (tpa), catechol (cat), protocatechuate (pca) and benzoate (ben) (Fig. 5c). In particular, two identical copies of a catabolic island including the pad and tpa clusters were found on the linear plasmids of R. jostii RHA1 and Rhodococcus sp. DK17, flanked by transposase-encoding genes (Patrauchan et al. 2005; Choi et al. 2005). The duplicated phthalate-degrading operons resulted to be simultaneously expressed and equally functional during the DK17 growth on phthalate, allowing this strain to achieve the maximal degradation of phthalate (Choi et al. 2007). Genes involved in naphthalene and phthalate were found duplicated on two separate genomic islands in R. opacus M213, and, in Rhodococcus sp. TFB, it was demonstrated that naphthalene degradation probably results from the activities of different isozymes (Tomás-Gallardo et al. 2006). Genetic redundancy in aromatic degradation pathways was observed in R. opacus R7 possibly in relation to both the size of the genome and the degradation abilities towards different aromatic classes (Di Gennaro et al. 2014; Orro et al. 2015) and in R. opacus M213.

The catabolic redundancy was also associated to different substrate specificities and to possible mechanisms of metabolic intermediate detoxification in the steroid metabolism (involving kshAB and kshD genes) by some Rhodococcus strains. Four homologous ksh gene clusters are present in R. jostii RHA1 genome; the cluster 1 was shown to encode enzymes involved in cholesterol catabolism (van der Geize et al. 2007), while the homologous cluster 3 supported the catabolism of cholate (Swain et al. 2012). In R. erythropolis SQ1, ksh cluster 1 was involved in steroid metabolism, while ksh cluster 2 was supposed to have a role in limiting the intracellular accumulation of toxic metabolic intermediates (van der Geize et al. 2008). The genome analysis of R. rhodochrous DSM43269A revealed five kshA homologous genes (kshA1 to kshA5) which were phylogenetically distinct, and each one showed a unique steroid induction pattern and substrate range, ensuring a fine-tuned steroid catabolism (Petrusma et al. 2011).

The redundancy of genes encoding (chloro)phenol hydroxylases supported the ability of R. opacus 1CP to metabolize a large spectrum of phenolic and chloro-phenolic compounds (Grӧning et al. 2014). All the three homologous phenol hydroxylases were able to convert the tested phenolic substrates at significant rates. This was probably due to the broad substrate specificity of these phenol hydroxylases and, therefore, low specialization (Grӧning et al. 2014). Multiple genes (from 3 to 5) encoding phenol hydroxylases (pheA1/pheA2) were identified in other Rhodococcus strains (Grӧning et al. 2014) In particular, in R. jostii RHA1, differences in the transcriptional regulation of two pheA1/pheA2 clusters suggested their activation under different growth conditions (Szőköl et al. 2014).

Interestingly, the functional redundancy found in Rhodococcus was strongly associated to evolutionary mechanisms related to niche adaptation. In this respect, the genome and microarray analysis of R. aetherivorans I24 study of this strain to PCB/biphenyl exposure indicated the involvement of the only core enzymes in the substituted benzene metabolism, while most of the isozymes found in RHA1 were missing (Puglisi et al. 2010). This was associated to the fact that, unlike RHA1, I24 strain was not isolated from a PCB-contaminated site. Therefore, as previously mentioned, despite the gene redundancy is recognized as a trait of Rhodococcus genus, the presence of functional isozymes involved in specific catabolic pathways reflects the selective pressure imposed by the environmental habitat and is also in part associated to the size of the genomes, as in the case of R. jostii and R. opacus strains featured by larger genomes compared to R. aetherivorans and R. ruber (Table 1).

4.2 Plasmids

Plasmids are circular or linear extrachromosomal DNA elements, which are capable of semi-autonomous or self-replication; they do not typically encode essential genes for the host but instead carry genes that may help the organism to adapt to novel environments or nutrient sources (Aminov 2011; Carroll and Wong 2018). In this respect, most of the Rhodococcus genes belonging to the variable accessory genome have plasmidic localization (see Par. 3.2) (Creason et al. 2014a; Orro et al. 2015; Pathak et al. 2016). Plasmids are considered a major driving force in prokaryotic evolution as they can be transferred between cells, as mobile genetic elements, mediating horizontal gene transfer events (Hülter et al. 2017). Accordingly, in bacteria, plasmids are thought to play critical roles in the evolution, propagation and assembly of catabolic pathways, antibiotic and metal resistance mechanisms, antimicrobial biosynthetic pathways and pathogenicity (Meinhardt et al. 1997; Shimizu et al. 2001).

Many Rhodococcus strains harbour plasmids, which can be linear or circular, of small (of few Kbp) or large size (of hundreds of Kbp), with cryptic or catabolic functions. Some Rhodococcus strains have been described to simultaneously possess several plasmids of different types and sizes, e.g. R. erythropolis PR4 contains one linear plasmid, pREL1 (~270 Kb), and two circular plasmids, pREC1 (~100 Kb) and pREC2 (~3.5 Kb) (Sekine et al. 2006); R. opacus B4 possesses two linear plasmids of pROB series (pROB01 of ~560 Kb, pROB01 of ~240 Kb) and three circular plasmids of pKNR series (pKNR of 111 Kb, pKNR01 of 4.4 Kb, pKNR02 of 2.8 Kb) (Na et al. 2005). Many Rhodococcus plasmids, both linear and circular, encode catabolic functions associated to xenobiotic degradation (Shimizu et al. 2001), chemolithoautotrophical growth on gaseous hydrogen and carbon dioxide, hydrogen production (Grzeszik et al. 1997), metal toxicity resistance (Cappelletti et al. 2016) and pathogenicity (Letek et al. 2010). In this respect, deletions in several Rhodococcus plasmids have resulted in the loss of degradative genes and specific growth deficiencies; e.g. deletions on pRHL1 and pRHL2 in R. jostii RHA1 and on pTA422 in R. erythropolis TA421 affected biphenyl degradation (Fukuda et al. 1998; Kosono et al. 1997). In some cases, catabolic plasmids were also found to harbour homologous genes, and their mutation was found to exert moderate or null phenotypic effects (Patrauchan et al. 2005). In this framework, catabolic plasmids are thought to play a key role in Rhodococcus catabolic versatility and efficiency by harbouring unique set of genes involved in specific catabolic pathways but also contributing to Rhodococcus multiple pathways and functional redundancy (Kim et al. 2018). Additionally, Rhodococcus plasmids have been described to have a much higher density of DNA mobilization genes (e.g. insertion sequences, transposase genes), pseudogenes, unique species-specific genes and niche-specific determinants (e.g. genes involved in pathogenesis in R. equi and in peripheral aromatic clusters in R. jostii RHA1 and R. opacus R7) than the corresponding chromosomes. Rhodococcal plasmids are therefore under less stringent selection and are key players in rhodococcal genome plasticity and niche adaptability (Sekine et al. 2006).

Linear plasmids, which are often very large, are widespread and diverse among Rhodococcus strains (Ventura et al. 2007). As they can reach lengths of several hundreds of kilobases, they are frequently referred to as megaplasmids. For instance, pRHL2 from R. jostii RHA1 is 443 kb (Shimizu et al. 2001), pPDG1 from R. opacus R7 is 656 kb (Di Gennaro et al. 2014), and pRHL1 from R. jostii RHA1 is 1.12 Mb (McLeod et al. 2006). The reason of extensive research attention on Rhodococcus large linear plasmids is connected with the fact that they are associated with catabolic genes and that they confer advantageous abilities on their hosts (Meinhardt et al. 1997). For instance, linear plasmids have been described to encode enzymes involved in the catabolism of naphthalene (Kulakov et al. 2005; Orro et al. 2015; Pathak et al. 2016), biphenyl (Taguchi et al. 2004), toluene (Priefert et al. 2004), alkylbenzene (Kim et al. 2002), isopropylbenzene and trichloroethylene (Meinhardt et al. 1997), phthalate (Patrauchan et al. 2005), gaseous n-alkanes (Cappelletti et al. 2015), chloroaromatic compounds (Konig et al. 2004), isoprene (Crombie et al. 2015) and triazine compounds (Dodge et al. 2011) and also in the desulphurization of organosulphur compounds (Denis-Larose et al. 1997), resistance to toxic metals (Meinhardt et al. 1997; Cappelletti et al. 2016), pathogenicity towards bovines (Valero-Rello et al. 2015) and phytopathogenicity (Francis et al. 2012). In many cases, catabolic genes present in linear plasmids contribute to degradation pathways together with genes located on the chromosome (Gonçalves et al. 2006).

As a distinctive feature, linear megaplasmids from Rhodococcus strains are capable of conjugal transfer, being responsible for genetic information sharing through HGT (Meinhardt et al. 1997; Dib et al. 2015). The importance of linear elements is also associated to their higher “flexibility” compared to circular plasmids. In particular, the telomeres are considered to be frequently subject to recombinational events (Volff and Altenbuchner 2000; Chen et al. 2002). In the case of HGT, intermolecular recombination events can take place between plasmids but also between host chromosomes of compatible species. Both the frequent observation of linear plasmids and illegitimate recombination in Rhodococcus have led to the hypothesis of the hyper-recombinational gene storage strategy. This is related to the function of the plasmids as storage of large number of catabolic genes that can represent recombination sources to respond and adapt to novel compounds in the native soil environments.

As for most bacteria, circular plasmids are also very common in Rhodococcus species. Circular plasmids in Rhodococcus have been generally shown to possess smaller size compared to the linear ones (Table 2). However, they often contribute to the catabolic capacities of the Rhodococcus host strain. Several circular plasmids harbour genes encoding part or complete xenobiotic degradation pathways in Rhodococcus strains such as pRTL1 (100 Kb) encoding haloalkane degradation enzymes in R. rhodochrous NCIMB13064 (Kulakova et al. 1995); pREC1 contains a complete set of genes for the β-oxidation of fatty acids (Sekine et al. 2006) and the large circular plasmids pKNR (111 kb) in R. opacus B4 (Honda et al. 2012).

Additionally, many Rhodococcus plasmids are self-transmissible but phenotypically cryptic. The cryptic plasmids described in Rhodococcus strains have principally small size and circular structure, e.g. pREC2 (3.6 Kb) in R. erythropolis PR4, pB264 (4.9 Kb) in Rhodococcus sp. B264-1. In some cases, these plasmids have been isolated to develop vector systems for DNA manipulation and protein expression in Rhodococcus strains. For instance, pKA22 (4.9 Kb) from R. rhodochrous NCIMB13064 (Kulakov et al. 1997), pFAJ2600 (5.9 Kb) from R. erythropolis NI86/21 (De Mot et al. 1997), pAN12 (6.3 Kb) from R. erythropolis AN12 and pKNR01 (4.4 Kb) from R. opacus B4 are some of the small circular cryptic plasmids, described in the literature, that have been used to construct E. coli-Rhodococcus shuttle plasmids (Kostichka et al. 2003). The first two plasmids require the activity of two replication proteins, RepA and RepB, to replicate. As these two replication proteins resemble replication proteins of the theta-type replicating Mycobacterium plasmid pAL5000 (Ventura et al. 2007), pKA22 and pFAJ2600 are categorized as replicons of pAL5000 family. The plasmid pAN12 is of the pIJ101/pJV1 family, which usually replicates thanks to a single replication. Detection of ssDNA intermediates in several of the pIJ101/pJV1 family of plasmids suggested that they replicate by rolling circle mechanism (Kostichka et al. 2003).

5 Towards Genome-Scale Modification of Rhodococcus Through System and Synthetic Biology

Genetic manipulation (or genetic engineering) methods have been applied to Rhodococcusstrains in order to obtain new strains that express additional genetic properties to acquire new catabolic capacities (Xiong et al. 2012, 2016; Venkataraman et al. 2015; Rodrigues et al. 2001; Hirasawa et al. 2001), to perform promoter activity studies (van der Geize et al. 2008; Cappelletti et al. 2011) and to introduce genetic mutations or deletions to determine the phenotypic alterations and the function of a specific gene or gene cluster (van der Geize et al. 2008; Amara et al. 2016).

Genetic manipulation methods are based on the development of efficient Rhodococcus transformation procedure, as biochemical and genetic characteristics of this genus have for long time limited the genetic manipulation of this strain (Sallam et al. 2006; Cappelletti 2010). Although protoplast-mediated transformation methods have also been used for some Rhodococcus strains (Singer and Finnerty 1988; Duran 1998), the use of electroporation procedure showed the highest efficiency, and it is presently the method that is most widely used for Rhodococcus transformation (Shao et al. 1995; Sekizaki et al. 1998; Kalscheuer et al. 1999). Several genetic manipulation strategies during the last three decades led to the development of (1) E. coli-Rhodococcus shuttle vectors from cryptic plasmids of Rhodococcus strains (Kostichka et al. 2003; Na et al. 2005; Matsui et al. 2006), (2) expression vectors for heterologous gene expression and protein production (Kalscheuer et al. 1999; Nakashima and Tamura 2004a, b), (3) random mutagenesis methods using transposons or spontaneous illegitimate recombination (Desomer et al. 1991; Sallam et al. 2006; Crespi et al. 1994) and (4) targeted gene disruption methods based on unmarked mutagenesis deletion systems with SacB as counter-selection (van der Geize et al. 2001).

Recently, advances in genomics and genome editing offered new opportunities to engineer Rhodococcus in a directed and combinatorial manner through a genomic-scale rational design of the cell through system and synthetic biology approaches.

With the aim of developing synthetic biology platforms for Rhodococcus engineering, a first BioBrick™-compatible plasmid system was designed for R. opacus PD630 by Ellinger and Schmidt-Dannert (2017). Generally, the BioBrick™ toolbox allows convenient and reproducible assembly of multigene pathways into series of vectors and the possibility to quickly mobilize any cloned gene into vectors with different features for gene expression and protein purification (Vick et al. 2011). Currently, the majority of the available BioBricks were designed for the Gram-negative model organism Escherichia coli. However, recently, novel BioBrick™ tools have been designed for the engineering of other bacteria, e.g. Bacillus (Radeck et al. 2013). The first BioBrick™-compatible vector designed for a Rhodococcus strain pSRKBB derived from the backbone of the E. coli-Rhodococcus shuttle vector pSRK21, which was modified by removing all BioBrick™-incompatible restriction, while it retained the capacity to replicate in R. opacus PD630 (Ellinger and Schmidt-Dannert 2017) (Fig. 6). This BioBrick-compatible plasmid was demonstrated to enable robust heterologous protein expression in R. opacus PD630 from the lac promoter, giving a first demonstration of the use of Rhodococcus as platform for synthetic biology.

Fig. 6
figure 6

Diagram of BioBrick™-compatible vector pSRKBB for Rhodococcus. The enlarged region includes the lac promoter and the ribosome binding site (RBS) upstream of the multi-cloning site where coding sequences can be cloned for constitutive expression. Further details are provided by Ellinger and Schmidt-Dannert (2017)

Full size image

Moreover, in order to expand the molecular toolkit for Rhodococcus genetic engineering, a recent work analysed the genetic tools to obtain gene expression control in R. opacus PD630 by characterizing a constitutive promoter library, optimizing antibiotic resistance markers and defining the copy numbers of different gene expression plasmids (DeLorenzo et al. 2018). The same work also reported methods for genome editing by using recombinase-based technique for site-specific genomic modifications and by describing the methodology to identify neutral integration sites for stable heterologous expression. Lastly, a tunable and targeted gene repression system was developed by utilizing for the first time in Rhodococcus the CRISPR interference method (DeLorenzo et al. 2018). Taken together, the heterologous gene expression tools along with the CRISPRi method open new prospective for the optimization of the bioconversion abilities of R. opacus and for the possible development of molecular tools for other Rhodococcus species with relevant environmental and/or industrial characteristics.

At the same time, genomic information have been used to develop genome-scale metabolic models, i.e. mathematical representations of the stoichiometry of the biochemical networks occurring in Rhodococcus cell. In this regard, an in silico genome-scale metabolic model (iMT1174) was developed to make metabolic predictions on the behaviour of R. jostii RHA1 in relation to the accumulation of three types of carbon storage (i.e. glycogen, polyhydroxyalkanoate and triacylglycerols), using different carbon sources (glucose or acetate) and under growth conditions typically occurring in activated sludge bioreactor systems for wastewater recovery (Tajparast et al. 2015, 2018). The simulations were compared with the experimentally measured metabolic fluxes through 13C-labelling assays, and the predictive capacity of the model was established. These works represent the first steps towards systems biology approaches enabling to simulate and predict Rhodococcus metabolic fluxes leading to the production of industrially interesting metabolites. More efforts are needed to extend system and synthetic biology tools for genome-scale engineering of Rhodococcus species different from the model ones, i.e. R. opacus PD630 and R. jostii RHA1, in order to get deep into the extraordinary physiological and metabolic diversity featuring different Rhodococcus strains and to provide new opportunities for their utilization in targeted bioconversion and biodegradation processes.

6 Conclusions

Due to the biotechnological significance of this genus, a dramatic increase of sequenced Rhodococcus genomes has been published and become available in the database. The comparative and functional analyses of these genome data provided the opportunity to get insight into the genetic basis of the extraordinary metabolic diversity and versatility of Rhodococcus genus members. The high versatility and adaptability of members of Rhodococcus was found to be the reflection of the size and complexity of their genomes. The co-existence of linear and circular replicons (chromosome and plasmids) and the presence of large catabolic plasmids have strong implications in rhodococcal genome plasticity, fluidity and metabolic diversity. Further, the genetic and functional redundancy is a general characteristic of Rhodococcus genus, which supports metabolic robustness and versatility towards the degradation of different classes of substrates and organic compounds. The specific type of genomic redundancy was found to reflect the peculiar niche adaptation need of each Rhodococcus strain, and, at least in part, it was related to the size of the genome.

Lastly, advances in Rhodococcus genomics and genome editing provided new opportunities for the application of directed and combinatorial approaches to Rhodococcus engineering. Recent breakthroughs in genetic engineering of Rhodococus have included the use of synthetic biology platforms and new approaches for genome editing (CRISPR/Cas9 tool) of R. opacus PD630. Systems biology approaches have also been applied to R. jostii RHA1 to develop genome-scale metabolic models to predict cell metabolic fluxes for the production of relevant metabolites and storage compounds. Despite more efforts are needed to expand system and synthetic biology tools for genome-scale engineering of Rhodococcus species different from the model ones, i.e. R. opacus PD630 and R. jostii RHA1, the novel molecular toolkits along with the extended knowledge on genomic and functional genomics have largely contributed in making Rhodococcus strains relevant for future novel industrial applications.