The complete genome sequence and annotation of M. haemocanis extends our understanding of the biology of hemoplasmas and provides clues about the growth requirements for in vitro cultivation of these bacteria. Based on the metabolic pathway predictions and specific metabolic deficiencies, a more comprehensive medium can be designed. To date, only three other species of hemoplasmas have been entirely sequenced[23–27]. The genome features of M. haemocanis, including its small size, low G + C content and use of UGA codon to encode tryptophan, are similar to those of other hemoplasmas and are typical of members of the genus Mycoplasma. It is believed that the reduced metabolic pathways of hemoplasmas are probably a consequence of the adaptation to the nutrient-rich blood environment[23, 24]. The predicted metabolic pathways of M. haemocanis are very similar to those of M. haemofelis having orthologs for all the CDSs identified in the genome of this feline hemoplasma; this is not surprising since both species are obligate red cell pathogens that reside in the blood of their hosts. As suggested for other hemoplasmas, it is likely that M. haemocanis takes advantage of the erythrocyte’s metabolism, scavenging nutrients, which leads to diminished erythrocyte life-span and exacerbation of anemia during acute disease.
Additional primary virulence factors were not identified in the genome of M. haemocanis. The o-sialoglycoprotein endopeptidase, related to the cleavage of glycophorin A, is conserved among hemoplasmas; the superoxide dismutase (SOD), identified in M. haemofelis[24, 26] is also present in M. haemocanis, but not found in any other sequenced mycoplasma. Although SOD may protect these bacteria from superoxide anion toxicity faced in the blood environment, it is unlikely that this enzyme plays a determinant role in the primary pathogenicity associated with M. haemofelis infection or in the opportunistic infection caused by M. haemocanis.
As with other hemoplasmas, M. haemocanis contains an abundance of paralogous gene families (63.7% of all its CDSs) and the presence of strategically located tandem repeats. Although there is evidence supporting the role of paralog genes and the presence of tandem repeats in the development of antigenic diversity in Mycoplasma species[38, 39], additional studies are needed to verify the ability of hemoplasmas to undergo antigenic variation. The presence of irregular cyclic episodes of bacteremia in splenectomized dogs reported following experimental infection with M. haemocanis, and the possibility that such cycles are due to phase variation is also an area of active investigation in our laboratory.
Comparison of the genomes of M. haemocanis and M. haemofelis revealed remarkable genetic similarities. Most of the coding and non-coding sequences were conserved and topography of genes within their chromosomes was similar. Even the paralogous gene families were conserved between the two species; the only exceptions were one family with 8 members in M. haemocanis, and four small families of M. haemofelis with 8, 5, 4 and 3 members, and two with 2 members. The major difference in the paralogous families is the number of duplicate genes inside each of the common families. Thus, as with other bacteria that cannot survive without their host, it appears that maintaining paralogous gene families to generate antigenic variants is a high priority for the hemoplasmas too. On the other hand, CDSs that are unique to M. haemocanis or M. haemofelis might represent a set of proteins related to differences in virulence and/or related to host specificity. Most of these unique proteins are hypothetical. Although we attempted to improve the function prediction accuracy using the ESG software, most of the probabilities assigned were less than 50% and results remained inconclusive (Additional file2: Table S2). Regarding the subcellular localization of the unique CDSs, it is important to mention that the PSORTb software only predicts cytoplasmic membrane localizations when 3 or more transmembrane helices are present within the sequence, otherwise unknown localization is returned. Therefore, these predictions based on strict criteria might have underestimated the potential for membrane localization of these CDSs.
CDSs with known function that are unique to M. haemocanis do not appear to have a significant impact on its pathogenicity since they code for an enzyme involved in sugar transport and for ribosomal proteins. On the other hand, M. haemofelis possesses a type II restriction enzyme and two C-5 cytosine-specific DNA methylases (C5 Mtase); the restriction endonuclease is located in the same operon as one of the C5 Mtase, indicating that this operon is functional. Moreover, this endonuclease/methyltranferase pair is not present in any of the other hemoplasmas and the restriction enzyme is absent in the strain Langford 1 of M. haemofelis. DNA methylation has been associated with virulence in other bacteria; however, the function of these pair in M. haemofelis Ohio2 is unknown.
As mentioned previously, the hemoplasmas cannot be cultivated in vitro. This has resulted in a lack of detailed phenotypic and genotypic characterization, which has hampered our ability to correctly classify these organisms within the Mycoplasmataceae family. In addition, the 16S rRNA gene failed to provide sufficient resolution to separate M. haemocanis and M. haemofelis as different species of Mycoplasma[5, 17]. To date, the genotypic evidence for species differentiation of these two hemoplasmas is solely based on phylogenetic studies using a 177 bp fragment of their RNase P genes[18, 44]. Herein, we performed a phylogenomic comparison between M. haemocanis and strains of M. haemofelis to resolve this long lasting controversy. In recent years, the sequencing of entire genomes has allowed the in silico evaluation of genomic similarities between different organisms. ANI and tetranucleotide signatures have been used as surrogates to previous methods of species circumscription, such as 16S rRNA gene phylogeny and DNA-DNA hybridization. With both ANI and tetranucleotide indexes below the proposed thresholds for species definition, our results show that the M. haemocanis strain Illinois and M. haemofelis (strains Langford and Ohio2) are different species of mycoplasmas infecting two distinct animal species. This conclusion is also supported by the transmission studies done more than 50 years ago.
Taken together our results suggest that, although sharing very similar genomes, M. haemocanis and M. haemofelis are different mycoplasmal species infecting dogs and cats, respectively. The set of unique proteins may be a target for vaccine development against these hemoplasmas, especially for the feline hemoplasmosis that can cause acute disease in immunocompetent hosts.
Nucleotide sequence accession number
The genome of M. haemocanis strain Illinois was deposited in GenBank under the accession number CP003199.1.