Comparative genomic analyses of Mycoplasma hyopneumoniae pathogenic 168 strain and its high-passaged attenuated strain

Genomic features of M. hyopneumoniae 168-L and its global comparison with pathogenic strain 168

The complete genome of M. hyopneumoniae 168-L consists of a 921,093 bp (GC content 28.46%) single circular chromosome (GenBank accession number CP003131). A total of 689 protein-encoding genes were predicted. The average protein size is 378 amino acids and the mean coding percentage is 84.8%. Approximately 51% of genes were assigned to specific functional clusters of orthologous groups (COGs), and 28% were assigned an enzyme classification (EC) number (Figure 1). Comparison with the M. hyopneumoniae 168 genome (GenBank accession CP002274) revealed a highly conserved gene content and order between the two strains. The 168-L genome is 4,483 bp smaller than that of 168 (925,576 bp), because there are 60 insertions and 43 deletions (indels; insertions and deletions of any size) in 168-L relative to 168 (see Additional file 1: Table S1; Additional file 2: Table S2; Additional file 3: Table S3). Among these, 33 indels are located in predicted CDSs, and 70 are in noncoding regions. Besides these indels, 227 single nucleotide variations (SNVs) were identified between 168 and 168-L (Additional file 4: Table S4). While 31 SNVs were mapped to intergenic regions, 196 were in coding regions, inducing amino acid substitutions, frame shifts, and translational stops.

ISMHp1-Related genetic variations between 168 and 168-L

The difference between the genome sizes of strains 168 and 168-L is mainly due to differences in the duplication of Insertion Sequence (IS) elements. IS elements are distributed stochastically across the entire genome of both strains. The 168-L genome contains nine complete and one disrupted IS elements, which is almost identical to that of 168 except for slight differences in ISMHp1. There are nine complete copies of ISMHp1 in 168-L, but 12 copies in 168. The difference in ISMHp1 copy number between 168 and 168-L is due to three complete ISMHp1 deletions (located 690 kb, 870 kb, and 900 kb from oriC) and one complete ISMHp1 inversion, which was originally located at 378 kb from oriC, but was inverted in 168-L (1656 bp, located 372 kb from oriC) (Figure 2a).

Other than the IS elements, notable large-scale genomic differences were also indicated. Compared to strain 168, a genomic deletion of approximately 1.36 kb (locus 1: between MHP168L_311 and MHP168L_729) was identified, which had been substituted with an approximately 2.32 kb novel insertion sequence (locus 2) that was joined to a complete ISMHp1 element in 168-L (Figure 2b). This 168-L-related insertion fragment was also observed in strains 7448 and 232.

Molecular analysis of integrative conjugative element (ICE)

The integrative conjugative element (ICE) is a mobile DNA that is probably involved in genomic recombination events and in pathogenicity. The ICEH elements are more divergent than the typical similarity of other chromosomal locus in M. hyopneunomiae[25], suggesting an accelerated evolution of these constins [26]. During a survey of specific sequences, a specific 26.9-kb region with similarity to the integrative conjugal element of M. fermentans (ICEF) [27] was found in strain 168, which was designated ICEH (for integrative conjugal element of M. hyopneumoniae). Unlike ICEH in strains 7448 and 232, which consist of nineteen and twenty two CDSs, respectively, the ICEH168 consist of 20 CDSs (Additional file 5: Table S5). The organization of these elements is very similar. Some CDSs present similarity to tra genes, which are usually associated with the bacteria conjugative plasmids such as traK, traI, traE[26]. The ICEH168 has three tra genes, with one traG and two copies of the traE gene. Besides, a CDS encoding for a single strand binding protein (SSB) that is essential for the transfer process is also observed.

The ICE analysis of three M. hyopneumoniae genomes (7448, J and 232) carried out previously, revealed that the ICEH is present in the two pathogenic strains (7448 and 232) but is absent from the non-pathogenic one (J strain) [26]. Interest has therefore shifted to questions of whether the ICEH is present in the attenuated vaccine strain 168-L. Interestingly, the ICEH was also observed in strain 168-L. Moreover, the ICEH168 and ICEH168-L are almost the same, except for a missense mutation (G192E) identified in ICEH-ORF3 (MHP168_235). Our analyses indicate that the ICEH may not only present in pathogenic strains of M. hyopneumoniae.

Mutations affecting epithelium adhesion

In 168-L, three transversions were identified in the R1 region, near the C-terminus, of P97 (MHP168_110/MHP168L_110), which encodes cilium adhesin. In M. hyopneumoniae, attachment to the respiratory epithelium is mainly mediated by the membrane protein P97 [4]. This protein is located on the outer membrane surface, and its role in adherence has been firmly established. To bind cilia, a minimum of eight tandem copies of the pentapeptide sequence (AAKPV/E) in R1 are required [5]. Notably, all three transversion mutations were located in the tandem repeat unit (AAKPV/E), causing an E863V substitution. Significant alteration in this critical repeat unit might partly affect the adhesion reaction in 168-L.

Previous studies have demonstrated that P102 binds fibronectin and contributes to the recruitment of plasmin(ogen) to the M. hyopneumoniae cell surface [15]. P102 is commonly linked to P97 cilium adhesin, forming a two-gene operon [32]. Both P97 and P102 have several paralogs within the M. hyopneumoniae genome. However, the paralogs have only part of the complete sequence. Interestingly, P102, the companion gene in this operon, was truncated at 564 bp by a single base insertion in strain 168. Another intact copy of P102 (99% identity) was found 85 kb from this operon. Conversely, in strain 168-L, the original truncated P102 was reverted to the intact one, while another intact copy of P102, 85 kb away, was truncated.

The P146 adhesin-like protein of M. hyopneumoniae shows strong similarity to the LppS lipoprotein of Mycoplasma conjunctivae, which is involved in in vitro adhesion [16]. In addition, the N-terminus region of P146 also shows strong similarity to the P97 adhesin, and has a strongly hydrophobic region (amino acids 7–29), indicating a transmembrane region, and suggesting that the protein is expressed on the surface of M. hyopneumoniae cells [33]. Compared with its counterpart MHP168_676 in 168, MHP168L_676 from 168-L has an in-frame insertion of one amino acid (Q) at the N-terminus of P146. The enormous intra-specific diversity shown for the P146 encoding gene is at least partly because of differences between several repeat regions present in the gene, most notably a polyserine chain of variable length, and a [Q]_n[(P/S)Q]_m repeat region [34]. Interestingly, this one in-frame insertion (Q) was located in the [Q]_n[(P/S)Q]_m repeat region. Polyserine chains often function as a spacer region in proteins involved in complex carbohydrate degradation [35], while sequences rich in both proline and glutamine are not uncommon and can form a conformation known as a polyproline II helix [36, 37]. Such proline-rich sequences are often involved in binding processes and are highly immunogenic [37]. However, because the function of the P146 protein remains unknown, correlations with virulence or adhesion are speculative and need further investigation.

P159 is a proteolytically processed surface adhesin of M. hyopneumoniae[17]. Three proteins with apparent molecular masses of 27 (P27), 52 (P52), and 110 (P110) kDa were identified through proteomic analysis of M. hyopneumoniae lysates [17], with each representing a different region spanning P159. These cleavage fragments are located on the cell surface and present at all growth stages. In 168-L, MHP168L_504 (P159) has a missense mutation resulting in a G240A replacement in the (S)(S)G(G)S repeat region of P159. Although this (S)(S)G(G)S repeat region has been reported, its biological function is unknown.

P216 (MHP168L_503/MHP168_503) is a proteolytically processed cilium and heparin binding protein of M. hyopneumoniae[18]. This surface protein is post-translationally processed to generate N-terminal P120 and C-terminal P85 fragments, both of which can bind cilia [18]. The 168-L P216 gene has an in-frame four amino acid deletion in a poly Q motif near the C-terminus. Previous studies have suggested that poly Q and KEKE motifs may play a role in maintaining P85 on the cell surface [18, 34]. Collectively, the deletion mutations affecting P216 may affect its cilium adhesion and may be associated with virulence attenuation in 168-L.

The MHP168L_424 gene and its gene product LppT were analyzed in detail because they showed approximately 22% identity to the LppT protein from M. conjunctivae. LppT is the second gene in a two gene operon with LppS, which was reported to be an adhesin in M. conjunctivae[16]. The LppT gene lacked a promoter and is likely to be co-transcribed with LppS, thus suggesting a functional relationship between LppS and LppT [16]. In M. hyopneumoniae, LppT encoded a protein of 954 aa with a calculated molecular mass 108 kDa. The gene product encoded by LppT is also a membrane protein, with a signal sequence of 34 aa at the amino-terminal end and a transmembrane structure. Notably, one of the amino acid substitutions in 168-L occurs near the C-terminus of LppT, resulting in a L814F replacement.

Mutations altering the cell envelope and genes encoding secreted proteins

Cell envelope proteins and secreted proteins are involved in virulence, host cell interaction, and immune responses [14, 38, 39]. Outer membrane protein-P95 is a cell envelope protein in M. hyopneumoniae. In 168-L, MHP168L_103 (P95) is truncated by a nonsense mutation compared to 168, resulting in an E965* termination near the C-terminus of P95. Significant alteration in this outer membrane protein could conceivably cause a truncation in coding region, and in turn alter the function of P95 outer membrane protein.

M. hyopneumoniae contains an abbreviated membrane protein secretory system [1]. The pathway consists of secA (MHP168L_088), secY (MHP168L_128), secD (MHP168L_259), prsA (MHP168L_664), dnaK (MHP168L_069), trigger factor (MHP168L_154), and lepA (MHP168L_076). It has recently been demonstrated that some pathogenic bacteria use a type IV secretion system, composed of subunits related to the conjugation machinery, to deliver effector molecules to host cells [40], and that this system may be involved in pathogenesis [41]. We found no pathogenic mutations in the protein secretory system, except a synonymous substitution (P407P) in MHP168L_069, which encodes chaperone protein DnaK.

Mutations affecting antigens

Studies on the antigenic properties of M. hyopneumoniae revealed several immunodominant proteins, including the P36 cytosolic protein [12, 42], the P46, P65, and P74 membranous proteins [43–46], the elongation factor Tu [47], the chaperone protein DnaK[47], the pyruvate dehydrogenase E1-beta subunit [47], and the P97 adhesin [4]. The functions of these proteins have not been well elucidated, but specific reactants may eventually be useful tools to diagnose M. hyopneumoniae[48].

The cytosolic P36 protein is a lactate dehydrogenase [49] that induced an early immune response in pigs that are experimentally and naturally infected by M. hyopneumoniae[50]. Comparative studies with other Mycoplasmas commonly found in pigs demonstrated that the P36 proteins carry highly conserved species-specific antigenic determinants for M. hyopneumoniae[42]. Hyperimmune sera produced against recombinant P36 protein showed no reactivity against other porcine Mycoplasmas, including M. flocculare, M. hyorhinis, and Acholeplasma laidlawii[12]. Notably, one of the observed amino acid substitutions in 168-L occurs near the C-terminus of P36 (MHP168L_167), resulting in a N204D replacement.

P65 is an immunodominant surface lipoprotein of M. hyopneumoniae that is specifically recognized during disease [14]. Analysis of the translated amino acid sequence of the gene encoding p65 revealed similarity to the GDSL family of lipolytic enzymes [14]. The monospecific antibodies against heat shock protein-like P42 antigen, part of P65, can block the growth of M. hyopneumoniae[51]. In 168-L, MHP168L_668 (P65) has a missense mutation resulting in a T138A replacement.

Mutations affecting transport proteins

As Mycoplasmas are dependent on the exogenous supply of many nutrients, it has been predicted that they may need many transport systems [3]. Motif analysis revealed a family of proteins with a phosphotransferase (PTS) motif. The open reading frames (ORFs) included sgaA (MHP168L_422), sgaB (MHP168L_421), sgaT (MHP168L_563), mtlF (MHP168L_739), mtlA (MHP168L_561), nagE (MHP168L_582), and licA (MHP168L_041) [7]. However, no mutations were identified in this PTS transporter family. There are approximately 30 genes with ABC transporter family signatures in the genome of M. hyopneumoniae 168-L (Additional file 7: Table S7), and five missense mutations and one synonymous substitution were identified in this group. These included a cobalt import ATP-binding protein (MHP168L_284, K64E), an ABC transporter permease protein (MHP168L_394, A492G), a xylose ABC transporter ATP-binding protein (MHP168L_523, S8S), and three ABC transporter ATP-binding proteins (MHP168L_413, D118Y; MHP168L_462, V219L; MHP168L_631, R60W). Interestingly, the expression of MHP168L_394 and MHP168L_413 was reported to be up-regulated in vivo during disease relative to in vitro-grown [11]. The variability between strains 168 and 168-L in multi-transport proteins indicates that they may affect growth and survival in different hosts or host tissues.

Mutations affecting genes directly related to metabolism and in vivo growth

M. hyopneumoniae strain 168 encodes 695 genes, approximately one quarter of which are involved in metabolism and in vivo growth. Of particular interest were the mutations observed in the genes involved in various metabolic pathways (Figure 3), including glycolysis/gluconeogenesis (MHP168_167, MHP168_186), purine metabolism (MHP168_289, MHP168_639), pyrimidine metabolism (MHP168_086), glycerophospholipid metabolism (MHP168_596), oxidative phosphorylation (MHP168_085), aminoacyl-tRNA biosynthesis (MHP168_058), and the pentose phosphate pathway (MHP168_142, MHP168_152). An in-frame insertion of two amino acids (TG), a missense mutation (E393G) and a nonsense mutation (N456*) were identified in 168-L near the C-terminus of MHP168L_085, which encodes a NADH oxidase involved in oxidative phosphorylation. Mycoplasma genomes are deficient in genes coding for components of intermediary and energy metabolism [3]. Thus, Mycoplasmas depend mostly on glycolysis to synthesize ATP [3]. In the glycolysis pathway, missense mutations in both L-lactate and pyruvate dehydrogenase were observed, resulting in N204D and S194G replacements, respectively. Iron deprivation, is a prominent feature of the host innate immune response, and most certainly impacts growth of Mycoplasmas in vivo[52]. Through transcriptome analysis, MHP168_639 was identified to be down-regulated during iron limiting conditions [52]. This suggests that MHP168_639 may play a role in M. hyopneumoniae’s response to iron stress. In 168-L, MHP168L_639 has a missense mutation resulting in an E411K replacement. Mutations in these metabolism-related genes accumulated over 300 in vitro passages likely affect growth and survival within host cells.

In addition, approximately 41% of mutations affected genes coding for hypothetical proteins. Despite the lack of functional annotations for these genes, their disruption in 168-L makes them obvious targets for investigation as potential virulence factors. Further molecular genetics and in vivo studies are required to confirm and assess the relative importance of these genes in the attenuation of virulence in 168-L.

Bacterial strains, growth conditions, and DNA extraction

Clonal isolates of M. hyopneumoniae strain 168 and 168-L were selected for sequencing. Both of the strains were grown in KM2 cell-free medium at 37°C. The culture was harvested from 100 mL KM2 cell-free medium by centrifugation at 1,200×g for 30 min, and then total genomic DNA was extracted from mycoplasma cultures using a TIANamp Bacteria DNA Kit (Tiangen, Beijing, China) according to the manufacturer’s instructions.

Genome sequencing and assembly

Genomic libraries containing 8 kb inserts were constructed according to the manufacturer’s protocols. Whole-genome sequencing of strain 168-L was performed by combining GS FLX and Solexa paired-end sequencing technologies. A total of 242,507 reads (67.4% paired ends) were produced with the GS FLX system, giving 44.5-fold coverage of the genome. Eighty-eight percent (215,346) of reads were assembled into one large scaffold using Newbler (454 Life Sciences, Branford, CT, USA). A total of 1,971,358 reads were generated with an Illumina Solexa genome analyzer IIx (Illumina, San Diego, CA, USA) and were mapped to the scaffold with the Burrows-Wheeler Alignment (BWA) tool [53]. Gaps were filled by local assembly of the Solexa/Roche 454 reads or sequencing PCR products using a Prism 3730 capillary sequencer (Applied Biosystems, Foster City, CA, USA). All repeated DNA regions and low-quality regions were verified by PCR and sequencing of the product amplified from genomic DNA.

Annotation and sequence analyses

Open reading frames containing more than 30 amino acid residues were predicted using Glimmer 3.0 [54] with modified genetic code 4 and verified manually using the strain 168 annotation. Loci discrepancies between the 168 and 168-L consensus sequences were manually examined for support at the trace data level. Transfer RNA (tRNA) and ribosomal RNA (rRNA) genes were predicted using the tRNAscan-SE program [55] or by observing similarities with the M. hyopneumoniae strain 232 and strain J rRNA genes. Artemis (release 12) [56] was used to collate and annotate data. Functional predictions were based on BLASTP similarity searches against the UniProtKB [57], GenBank [58], Swiss-Prot protein [59], and COG [60] databases. EC numbers were assigned using the Kyoto Encyclopedia of Genes and Genomes (KEGG) [67] and metabolic pathways were mapped and analyzed using KEGG Pathway Database (http://www.genome.jp/kegg/pathway.html). Pseudogenes were detected by BLASTN analysis, comparing the genome sequences of 168-L with those of 232 and J, and then the annotation was revised manually.

Single nucleotide polymorphism (SNP) analysis

Nucleotide comparisons and single nucleotide polymorphism (SNP) analysis for strains 168 and 168-L were performed using the Artemis Comparison Tool (ACT) [61] and Mauve 2.3.1 genome alignment software [62]. ORF graphical visualization and manual annotation were carried out using Artemis, release 12 [56]. Screening for unusual coding differences between the 168 and 168-L genomes (stops and frame shifts) was conducted using FASTA program packages [63, 64] and BLAST [65]. The coding differences between the 168 and 168-L genomes were checked manually.

Accession numbers

M. hyopneumoniae strains 168 and 168-L genome sequences have been deposited in GenBank under accession numbers CP002274.1 and CP003131, respectively.