Mycoplasmas are responsible for a wide range of diseases in humans and animals, impacting health and economic activity – especially in the agricultural sector. They are cell wall-less bacteria of the class Mollicutes that are among the smallest known organisms capable of self-replication (Razin et al., 1998), with reduced genome sizes that have resulted in severely curtailed biosynthetic capabilities (Rivera and Cedillo, 2015). Consequently, many mycoplasmas lead a parasitic lifestyle, in which they are dependent on their host for essential nutrients. This symbiotic relationship presents both a challenge and an opportunity for the development of new treatments for Mycoplasma infections. The challenge arises from the close relationship between pathogen and host, which makes it difficult to specifically target the Mycoplasma. In addition to this, antimicrobial resistance against commonly used cell wall synthesis inhibitors is a serious concern (Meyer and Van Rossum, 2014; Chernova et al., 2016) However, the mycoplasma’s dependence on certain host-derived growth factors offers a unique opportunity for the discovery of inhibitors of a process or pathway that is central to this interaction, and that would be detrimental to the survival of the organism.

In this context, the mechanism by which mycoplasmas obtain the central metabolic cofactor coenzyme A (CoA) presents itself as an especially relevant case study. First, the enzymes involved in CoA biosynthesis have long been regarded as possible targets for antimicrobial development given their essential role in central energy and fatty acid metabolism, and the prediction that the bacterial enzymes could selectively be inhibited based on their divergence from those of their eukaryotic hosts (Spry et al., 2008; Moolman et al., 2014; Balish and Distelhorst, 2016). Second, CoA is also required for membrane biosynthesis and therefore growth of Mycoplasma (Pollack et al., 1997). Third, in vitro studies have shown that some of the genes encoding the enzymes responsible for CoA biosynthesis in mycoplasmas are dispensable. The extreme example is the organism with the “minimal bacterial genome” discovered following the genome reduction of Mycoplasma mycoides, which did not have any of the enzymes required for producing its own CoA (Yus et al., 2009; Hutchison et al., 2016). While de novo CoA biosynthesis may therefore also be a worthwhile target for the development of new anti-mycoplasma agents, our current lack of knowledge of the relevance and importance of this process in mycoplasmas, and whether it could be bypassed within their pathogenic context, prevents further exploration of the potential of this pathway.

In the universal, canonical CoA biosynthesis pathway five enzymes are responsible for the conversion of the precursor substrate pantothenic acid (Pan, or vitamin B₅) into CoA (Figure 1A). The first enzyme is pantothenate kinase (PanK), followed by CoaBC – a bifunctional protein in bacteria that has both phosphopantothenoylcysteine synthetase (PPCS) and phosphopantothenoylcysteine decarboxylase (PPCDC) activities. This is followed by phosphopantetheine adenylyltransferase (PPAT), with the last enzyme in the pathway being dephospho-coenzyme A kinase (DPCK; Strauss, 2010). An important possible deviation from the canonical pathway entails the use of the CoA degradation product pantetheine (PantSH) as an alternative substrate by PanK in a CoA salvage mechanism that allows for the bypassing of the PPCS and PPCDC activities of the CoaBC protein (Figure 1A). However, this is only possible for a certain subset of PanK enzymes, which occur as three distinct types: type I (PanK_I) and type III PanK (PanK_III) enzymes (encoded for by the coaA and coaX genes, respectively) that are found in prokaryotes, and type II PanKs (PanK_II) found in eukaryotes and selected Staphylococcus and Bacillus species. In particular, PanK_III enzymes are not able to use PantSH as substrate and organisms with this type of PanK can therefore not make use of the CoA salvage pathway and bypass CoaBC. In addition, there has also been some evidence that certain cells are able to take up more advanced CoA biosynthetic intermediates, suggesting the possible existence of mechanisms whereby CoA could be obtained through the action of only PPAT and DPCK, or even DPCK alone (Sibon and Strauss, 2016). A schematic summary of all these mechanisms is presented in Figure 1B.

With this as background, we set out to establish the genetic capacity of Mycoplasma to synthesize their own CoA de novo using Pan as substrate, or through some truncated variations of the canonical pathway. Such an analysis is particularly warranted due to the inconsistent presence of the genes that code for the CoA biosynthesis enzymes across Mycoplasma genomes (Daugherty et al., 2002; Spry et al., 2008; Yus et al., 2009; De Jonge et al., 2013). While this could be a reflection of the parasitic lifestyle of these organisms, it could also be a result of insufficient annotation, since Mycoplasma genomes typically contain a large number of hypothetical protein-coding genes. Distinguishing between these possibilities is crucial if CoA biosynthesis is to be considered to hold any realistic potential for anti-mycoplasma drug development.

In this study, we determine the existence of genes encoding the CoA biosynthesis pathway enzymes among Mycoplasma genomes using existing functional annotations as well as sequence, family, motif, and domain analysis of protein products. We discovered an unexpectedly differentiated prevalence of these genes across Mycoplasma species, with a few specific combinations emerging. We therefore used phylogenetic analysis to determine the evolutionary relationship of the Mycoplasma species and the functional relationship of their CoA biosynthetic enzymes using 16S rRNA and amino acid data, respectively. This allowed us to uncover the enzyme targets that would hold the most potential for the development of new broad-spectrum or species-specific anti-mycoplasma agents.

Our study of CoA biosynthesis in Mycoplasma found that their genomes contain genes encoding the CoA pathway enzymes (Figure 1A), but all four genes were not present in every species and different combinations were found across genomes. Among the evaluated genomes available on NCBI, the typical gene annotations of coaA/coaX, coaBC (or coaB/coaC), coaD, and coaE were seldom found and therefore were of no use in confirming the presence of these genes within Mycoplasma genomes. Information from the KEGG and SEED databases was also limited since not all species – or strains of a species – are represented within these databases.

A description of protein products was, however, more consistent and this allowed the identification of associated genes which were then mostly annotated with a locus tag number. Where no protein products were indicated, enzyme homologs could be identified within several genomes using a combination of NCBI blastp or psi-blast searches using known protein sequences from a genetically close species as query. To further confirm, the identity of the gene its protein product was evaluated in terms of protein family, domain, and motif information using the web-based CDD, InterPro, and MEME Suite recourses.

Of the 62 Mycoplasma species evaluated in this study, 10 contained all four enzyme-encoding genes of the CoA synthesis pathway in their genomes. These 10 species were only found in the Hominis and Pneumoniae groups in the 16S phylogeny. Except for M. iowae, which contained separate PPCS- and PPCDC-encoding genes, the rest all had a bi-functional CoaBC-encoding gene which is more typical of prokaryotes. Although all strains of M. iowae contain separate genes, differences among strains were found in M. fermentans. The open reading frame of all CoaBC-encoding genes always overlapped with that of PanK, which could imply that the genes are transcribed as a single operon and are therefore under the same selective pressure. However, where the PPCS- and PPCDC encoding genes are separated, the PanK-encoding gene is found thousands of base pairs upstream, but at no fixed distance and transcription as a single operon can therefore be excluded.

Except for M. sturni, the cellular localization of the PanK enzymes could not be predicted using InterPro, but prokaryotic PanKs are typically intracellular proteins. However, the PanK of M. sturni was predicted to be a transmembrane protein (Supplementary Table 5) with the extracellular domain containing the PAN- and INTERFACE-motifs that are specifically involved in interaction with the PanK_III substrate, Pan (Yang et al., 2006). It could therefore bind extracellular Pan and convert it to 4′-phosphopantothenic acid (P-Pan) for further processing by CoaBC, which was also predicted to be an extracellular, membrane-bound protein. However, if the PanK enzymes in other Mycoplasma species are in fact situated in the cytoplasm, then the product of PanK, P-Pan, will have to be exported before it can be used as a substrate by CoaBC. Unassisted export of this substrate seems highly unlikely since it is very polar (consisting of a phosphate group and a carboxylic acid moiety) and would thus require a dedicated transporter to cross the membrane. Alternatively, P-Pan might be obtained from the host if readily available in the environment. If this is possible, then the PanK enzyme might not even be required for CoA synthesis in these organisms, but have an entirely different role. How and where the subsequent processing of the CoaBC product, 4′-phosphopantetheine (PPantSH), will take place is not clear since the cellular localization of the Mycoplasma PPAT could not be predicted. Prokaryotic PPAT enzymes are typically located in the cytoplasm and if this is also the case for Mycoplasma then PPantSH would have to be imported after being released from the CoaBC enzyme.

In eight of the evaluated species, a CoaBC-encoding gene was absent with only PanK-, PPAT-, and DPCK-encoding genes present. In the 16S phylogeny these species were only found in the Pneumoniae and Hominis groups, but none in the Spiroplasma group. This combination of enzymes would seem to indicate that CoA biosynthesis occurs via the salvage pathway (Figure 1B). However, Mycoplasma PanKs are all predicted to be PanK_III type enzymes, none of which are known to accept PantSH – the substrate of the salvage pathway – as a substrate (Strauss, 2010). When comparing the PAN-motif (hGhDR) in Mycoplasma PanK_III enzymes to that described for PanK_IIIs from other bacteria, the arginine residue is replaced by leucine or isoleucine in Mycoplasma (Figure 8). This arginine residue plays a prominent role in the active site of the PanK enzyme, since it is directly involved in an interaction with the substrate, Pan. Therefore, the substitution of this charged residue with an uncharged hydrophobic residue will most likely have major consequences with regards to the binding of Pan. The INTERFACE-motif in Mycoplasma similarly deviates from the general residue sequence. Since these two motifs together form a substrate-binding site in PanK_IIIs, the substitution of residues in both motifs may allow the enzymes to use PantSH as a substrate so that CoA biosynthesis could take place via the salvage pathway. It is also possible that the substitutions in these motifs have no impact on PanK substrate specificity, and that like other PanK_IIIs these enzymes are also not able to phosphorylate PantSH. They might thus have a role unrelated to CoA synthesis as suggested previously, with these species only relying on the PPAT and DPCK proteins to obtain CoA. Unfortunately, our attempts to date to experimentally evaluate the substrate specificity of the Mycoplasma PanKs were unsuccessful due to difficulties in obtaining soluble protein for use in activity assays.

The majority of Mycoplasma species evaluated in this study only contained genes encoding for PPAT and DPCK. These species were all situated in either the Hominis or Spiroplasma groups, but none in the Pneumoniae group. If these genes do produce functional proteins, PPAT would have to obtain its PPantSH substrate in the absence of PanK. Recently, it has become evident that certain cells, including some micro-organisms, might be able to obtain PPantSH from their surroundings to fulfil their CoA needs (Sibon and Strauss, 2016). Also, in Drosophila melanogaster Schneider 2 cells, PPantSH is able to cross the membrane without the help of a transporter (Srinivasan et al., 2015). How Mycoplasma would obtain PPantSH remains an open question. However, all indications are that PPantSH is readily available in the environment. In addition to being an intermediate product of the de novo CoA biosynthetic pathway (required by the host), PPantSH can also be produced via the degradation of CoA and the acyl carrier proteins involved in lipid biosynthesis (Jackowski and Rock, 1984b). Furthermore, E. coli was found to excrete PPantSH, but also never re-imported it (Jackowski and Rock, 1984a). Thus, PPantSH supplied by the host or surrounding microbial communities would allow Mycoplasma to indeed salvage PPantSH from the environment for conversion to CoA by the action of PPAT and DPCK (Figure 1B).

The most common enzyme-encoding gene that could be identified among the investigated mycoplasmas was that for DPCK. With some exceptions, this gene was present in all of the mycoplasmas. Eleven of the species contained only a DPCK-enzyme encoding gene and these species were distributed over the Hominis and Pneumoniae groups. Five of these were HAD-DPCK-encoding genes and these were only found among the Hominis group members. Mycoplasma conjunctivae was the only species to contain a HAD-DPCK-encoding gene in combination with another CoA-synthesis pathway gene (PPAT).

The fact that only the final enzyme in the CoA synthesis pathway could be identified for some species suggests that the DPCK enzymes scavenge their substrate – dephospho-coenzyme A (dPCoA) – either from the host or the surrounding environment (Figure 1B). The DPCK enzymes of Mycoplasma gallisepticum and Mycoplasma imitans were both predicted to contain signal peptides and be located extracellularly, which would allow easy access to dPCoA from the host or environment. The cellular localization of the rest could not be predicted, but given that DPCK enzymes are typically found intracellularly, this would require the Mycoplasma to have a (as of yet unidentified) dPCoA transport system. The presence of only DPCK and their concomitant requirement for the uptake of dPCoA has also been predicted for various Rickettsia and Chlamydia spp. (Strauss, 2010; Driscoll et al., 2017). Furthermore, the characterization of a mitochondrial transporter of dPCoA in D. melanogaster suggests that a prokaryotic version of such a transporter might exist (Vozza et al., 2017). The obvious dependency of mycoplasmas on their host for many metabolites and intermediates – due to their restricted metabolic biosynthetic potential – highlights the necessity for a vast set of transporters. However, while many Mycoplasma transporters are vaguely annotated based on similarities to established transporter classes or membrane segments, little is known regarding their actual substrate preferences or specificities (Großhennig et al., 2013).

The role of the HAD domain in the bifunctional HAD-DPCK enzyme is not clear. As indicated before, the HAD protein is predicted to belong to the Cof subfamily of proteins which consists of phosphatases. These types of phosphatases are promiscuous and seem to be associated with numerous vitamin and cofactor metabolic processes (Lawhorn et al., 2004; Kuznetsova et al., 2006). While the phosphatase activity might suggest a role in regulation (by removing phosphates required for biosynthesis) this has not yet been experimentally tested.

Only eight Mycoplasma species with no CoA biosynthesis enzyme-encoding genes could be identified. Five of these form part of the hemotropic Mycoplasma cluster (Table 1; Brown, 2010), which is a division within the Pneumoniae group while the rest were from the Hominis group. The mycoplasmas in the hemotropic cluster are also referred to as hemoplasmas and are blood-borne pathogens that infect the erythrocyte (Messick, 2004). The absence of genes in the eight species was confirmed by a comparative genomics study that showed them to not contain any enzymes involved in CoA metabolism, the pentose-phosphate pathway or pyruvate dehydrogenase complex (Guimaraes et al., 2014). In the case of the hemoplasma, the nutrient-rich environment of the blood provides the ideal conditions for additional degenerative evolution, and thereby metabolic reduction. However, it is interesting to note that the blood-borne Plasmodium parasite that causes malaria also infects erythrocytes, and yet CoA biosynthesis is the only vitamin-requiring pathway that is essential in this organism (Spry et al., 2008). Clearly, the need for CoA depends on the organism and its metabolic requirements, and for many pathogens, this remains largely unexplored.

The phylogeny produced using 16S rRNA gene sequences reflected the evolutionary relationships of the investigated mycoplasmas as reported in the literature (Weisburg et al., 1989; Johansson et al., 1998; Maniloff, 2002; Brown, 2010; Wium et al., 2015). Except for the Spiroplasma group, where all species only contained genes encoding for PPAT and DPCK, the evolutionary relationships of species were not found to correlate with the combination of CoA synthesis genes contained within their genomes.

Apart from the DPCK phylogeny, the distribution of species in the phylogeny of individual enzymes did not resemble the evolutionary relationships found in the 16S rRNA phylogeny. Moreover, the functional relationship of one enzyme in the CoA synthesis pathway also did not determine which of the other enzymes would be present in a species. However, members of the Spiroplasma group did produce distinctive clades in both the PPAT and DPCK phylogenies. With a few exceptions, this was also the case for members of the Hominis group with only PPAT and DPCK proteins.

From our results, it can be concluded that most Mycoplasma species do have the genetic capacity to synthesize CoA, but there is an unpredictability according to the universally accepted 16S rRNA phylogeny among species in terms of which enzyme-encoding gene combinations can be expected, and therefore what the required substrate for CoA synthesis would be in each case. This variation may be entirely dependent on the nutritional conditions at the site of infection. Their parasitic lifestyle on the one hand, and their nutritionally exacting nature on the other, typically causes Mycoplasma to exhibit a rather strict host, organ, and tissue specificity (Razin, 2006). They are therefore reliant on the nutritional milieu provided by the host or the microbial community at the site of infection, which would in turn influence their biosynthetic capacity or requirements. We note, for example, that those species with the most complete pathways mainly have non-mammalian hosts. It is possible that in these hosts the ability to obtain CoA via truncated pathways is limited due to low levels of circulating CoA biosynthesis intermediates. This observation suggests that instead of correlating with evolutionary relationships, the combination of CoA synthesis genes is more likely to correlate with the mycoplasmas’ immediate nutritional environment. However, testing such a proposal will only be possible once more information becomes available on the levels of circulating metabolites in the various hosts.

Despite the variation in the CoA synthesis capacity of Mycoplasma, there is the potential for the CoA biosynthetic pathway as a drug/vaccine target. A comprehensive investigation into the CoA synthesis pathway of the targeted Mycoplasma will, however, have to be performed in order to identify the genetic variation present within the genome of the specific species (or within strains of the same species, as these might also show differences). The most consistently present enzyme-encoding gene was that of DPCK, suggesting that the development of a broad-spectrum anti-mycoplasmal agent acting on CoA biosynthesis should target this enzyme. The potential of DPCK as a selective target is further underscored by the fact that the main DPCK activity of the human host is found on a bifunctional PPAT-DPCK protein called CoA synthase (COASY), indicating that selective targeting may be possible. On the other hand, for those species in which no CoA biosynthesis enzyme-encoding genes could be identified, the likely requirement for a specific CoA transporter also offers an opportunity for the development of new therapies that would target this protein and/or its function. Overall, our findings suggest that CoA biosynthesis in Mycoplasma holds many significant opportunities for further study and investigation, specifically in the search for new treatments of Mycoplasma infections.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AB and ES contributed to conception and design of the study. TR performed all experimental work and analysis and wrote first draft of the manuscript. AB supervised the project with the assistance of ES and revised the manuscript for submission with support from ES. All authors contributed to the article and approved the submitted version.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Postgraduate student funding was received from the National Research Foundation and the Faculty of Science, Stellenbosch University.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.