Preliminary comparative analysis of the genomes of selected field reisolates of the Mycoplasma synoviae vaccine strain MS-H reveals both stable and unstable mutations after passage in vivo

Genomic comparison of Mycoplasma synoviae vaccine strain MS-H and the MS-H parental strain 86,079/7NS established a preliminary profile of genes related to attenuation of MS-H. In this study we aimed to identify the stability of mutations found in MS-H after passage in experimental or field chickens, and to evaluate if any reverse mutation may be associated with changes in characteristics of MS-H in vitro or in vivo. Whole genome sequence analysis of 5 selected MS-H field reisolates revealed that out of 32 mutations reported previously in MS-H, 28 remained stable, while four found to be reversible to the wild-type. Each isolate possessed mutations in one to three of the genes obg, oppF1 and gap and/or a non-coding region. Examination of the 4 reversible mutations by protein modeling predicted that only two of them (in obg and oppF1 genes) could potentially restore the function of the respective protein to that of the wild-type. These results suggest that the majority of the MS-H mutations are stable after passage in vaccinated chickens. Characterisation of stable mutations found in MS-H could be utilised to develop rapid diagnostic techniques for differentiation of vaccine from field strains or ts- MS-H reisolates.

exactly known whether temperature sensitivity is the cause of attenuation or just a coincidental phenotype in these organisms [4,5].
The majority of MS clones recovered from vaccinated flocks display their ts + phenotype, but it has been suggested that MS-H proliferation in vaccinated birds generates a mixture of ts + and ts-clones in the farm [6,7].
Unlike the non-virulent MSH strain, ts-field reisolates cause only minor lesions in the tracheal mucosa of the experimentally infected birds, significantly lower than the vaccine parent strain [7]. These results suggest that factors other than ts + phenotype are involved in the attenuation of the MS-H vaccine.
While the genetic basis of the MS-H temperature sensitivity and attenuation is not fully known yet, a mutation detected in obg gene was proposed as a likely explanation for the MS-H ts + phenotype [8]. Also, further comparison of the MS-H genome with that of its wild-type parent strain 86,079/7NS has revealed a frameshift mutation in an oligopeptide permease transporter (opp) gene, oppF 1 [9]. OppF is essential in establishment of systemic infection by M. bovis and its persistence in lower respiratory tract of calves [10]. Also, oppD was found to be required for full expression of virulence of M. gallisepticum in chickens [11].
Partial sequence analysis of obg and oppF genes [8,12] in five MS-H isolates have found different combinations of obg and oppF genotypes. Of the five isolates, MS-H 3 , 101,564 and 101,731 had obg w (w = wild-type) and oppF v (v = vaccine-type), MS-H 4 had obg w and oppF w , and MS-H 5 had obg v and oppF w . In this study the MS-H reisolates MS-H 3 , 101,564, 101, 731 MS-H 4 , and MS-H 5 were subjected to a comparative genome analysis to establish if any other mutations previously reported for the MS-H [12] may be reversible to the wild-type and evaluate if they could potentially influence MS-H attenuation.

Results
Phylogenetically, all selected reisolates from vaccinated flocks were closely related to MS-H Illumina paired reads from MS-H field isolates (Gen-Bank accession number PRJNA649354) were De novo assembled successfully using SPAdes with an average 162 of contigs generated for each ranging from 125 to 131,331 bp per isolate. The vlhA pseudogene region, a~50 kb locus covering large number of highly repetitive sequences, as well as the repetitive and the highly similar IS failed to assemble. Otherwise, the SPAdes generated an average 790,468 bp, representing a complete genome with high identity (93%) to other MS sequences available in the Gene Bank [12][13][14][15][16]. Alignment of the draft genomes of MS-H field isolates with that of MS-H exhibited an overall high degree of sequence similarity (99.99%) with no large-scale chromosomal insertions, deletions, duplications or rearrangements except for vlhA locus.
The maximum likelihood and NJ analysis performed using platforms REALPHY and MEGA, respectively, on whole genome sequences of 7 MS strains/isolates generated highly comparable results, reflecting a close relationship between MS-H and its field isolates. Notably, MS-H 3 and MS-H 4 were respectively the most closely and distantly related to MS-H ( Fig. 1).    [12], however the other 21 were found only in the 5 reisolates. The MS-H 3 had an identical sequence to MS-H except for the coding DNA sequence (CDS) of oppF in which insertion of 'T' at position 468 resulted in the restoration of the full-length oppF sequence identical to that of 86,079/7NS. Three non-synonymous differences were found between genomes of MS-H and MS-H 5 . These included deletion of two nucleotides 'AT' in a tandem repeat in a non-coding region (positions 502,825) upstream of the cytosine-5-methyltransferase CDS, insertion of nucleotide 'A' (causing frame-shift mutation) in a gene (CDS 164) encoding a protein of unknown function, and nucleotide substitution 'A' to 'G' in obg gene resulting in restoration of the wild-type Obg (Arg123Gly). Also, a synonymous substitution ('C' to 'T') in Glu322 was found in the CDS 966 which codes for Desert Hedgehog Signalling Molecule.
Twelve genomic differences were found between MS-H 4 and MS-H, three of which had been described to exist between MS-H and 86,079/7NS and were reverted to wildtype sequence. These comprised of insertion of 'AT' at position 502,827 in a tandem repeat within a non-coding region, a SNP in obg gene (similar to that found in MS-H 5 ) and a frameshift mutation in the oppF gene (identical to that found in MS-H 3 ). The other 9 mutations comprised of 6 in genes coding for Cardiolipin, two hypothetical protein, TatD deoxyribonuclease, S1 RNA-binding domain, and Thymidine phosphorylase, 1 in a gene with unknown function, 2 in non-coding regions.
In the isolate 101,546, two genomic differences were found to cause reversion to wild-type sequence. These included 'A' to 'G' in gap gene (at CDS 554) which resulted in a conservative change (Ala185Val), and a frameshift mutation in oppF identical to that from MS-H 3 and MS-H 4 . Moreover, four non-synonymous substitutions were found in CDSs corresponding to Obg, YbhB/YbcL Raf kinase inhibitor, and two hypothetical proteins. These substitutions were due to 'C' to 'T' at CDS 629 in obg gene resulting in a conservative change (Ala210Val); 'C' to 'A' at CDS 909 causing a conservative change (Asp303Glu) in a gene encoding a hypothetical protein; 'G' to 'A' at CDS 220 resulting a conservative change (Val74Ile) in gene encoding YbhB/YbcL Raf kinase inhibitor, and 'C' to 'T' at CDS 3979 resulting in a non-conservative substitution (Ala1327Thr). Moreover, a synonymous substitution in Thr197 was found in CDS corresponding to LemA ('C' to 'T' at CDS 591).
Comparison of MS-H and 101,731 were found changes in obg and oppF genes consistent with those of 1,015, 465. Moreover, similar to MS-H 4 , 'AT' insertions at positions 502,827 was found resulting in reversion to wildtype sequence. In addition, a 'T' deletion at CDS 387 in a gene encoding a hypothetical protein, substituted Tryr135 to a premature stop-codon. Moreover, genes encode Cls and DNA-directed RNA polymerase subunit beta were found to have a 'G' to 'A' substitution which resulted in a conservative amino acid change (Ser263-Asn), and a 'T' to 'C' substitution which resulted in a non-conservative substitution (Glu1037Gly), respectively. Additionally, a synonymous substitution in Asx362 was found in CDS corresponding to a hypothetical protein ('G' to 'A' at CDS 1086).
Mutations found in genes coding for Obg, OppF, Cardiolipin, and YbhB/YbcL Raf kinase-inhibitor were computationally predicted to affect the proteins structure The final alignment between targeted proteins and templates using Phyre2 for eight proteins (OppF, Obg, Cls, TatD, S1 RNA-binding, NAD-dependent glyceraldehyde-3-phosphate dehydrogenase, YbhB/YbcL Raf kinase  a "Yes" indicates different secondary structure of respective protein between MS-H and its reisolate/s; "No" shows that MS-H and its reisolate/s have the same secondary structure of the respective protein; "_" indicates that only a domain of respective protein was modelled and as the result the percentage of α helix and β strand were not reliable b "High" indicates that respective mutation highly likely has functional/phenotypic effect; "Medium" shows that respective mutation moderately likely has functional/phenotypic effect; "Low" shows that respective mutation has low potential functional/phenotypic effect; "NA" indicates that position corresponding to respective mutation was not modelled inhibitor and DNA-directed RNA polymerase subunit beta) exhibited an average 76% of residues modelled at 100% confidence. For the other five proteins (hypothetical proteins) only an average 40% of residues modelled at > 50% confidence and none of residues modelled at > 90% confidence ( Table 2). The model template IDs, protein lengths, predicted secondary structures, and the degree (%) of mutation sensitivity at a given position in the respective proteins are detailed in Table 2. Consistent with findings from a previous study [8], homology modelling of Obg in MS-H showed that Arg123 is located in large pocket region, which are frequently the active sites [17], however the probability that a missense mutation at this position effecting function of the corresponding protein was predicted low. The percentages of predicted alpha helices in the Obg protein sequences of MS-H 4 or MS-H 5 (29%), was different from that of MS-H (26%), due to the Arg123Gly difference between these strains. By contrast, the Ala210Val difference in the Obg protein sequence of strains 101, 546 and 101,731 did not change the secondary structure of Obg compared to that of MS-H.
The frameshift mutation corresponding to oppF gene in MS-H 3 , MS-H 4 , 101,546 and 101,731 restored the full-length OppF (reversion to wild-type). Based on the protein homology analysis conducted as part of this study, the functional domain of OppF is identified at the C terminus. As a result, the secondary structure of OppF in above-mentioned isolates (which possessed 68% alpha helices and 7% beta strands) was significantly different to that of MS-H (31% alpha helices and 24% beta strands).
Residue 166 in Cls of MS-H is in a highly sensitive mutation region and therefore Ala166Thr in MS-H 4 was predicted to affect its function. Also, due to this amino acid change, the secondary structure of Cls in MS-H 4 (containing 51% alpha helices) was different to that of MS-H (containing 50% alpha helices). By contrast, amino acid at position 263 was in a low mutation sensitive region. Therefore, Ser263Asn was unlikely to affect the protein Cls function in isolate 101,731. Also, the secondary structure of Cls in 101,731 was modelled identical to that from MS-H.
The TatD deoxyribonuclease and S1 RNA-binding proteins were also modelled and compared between MS-H and MS-H 4 for their mutations at positions 143 and 57, respectively. In MS-H both these mutations were in low sensitive mutation regions and therefore were unlikely to affect the function of the respective proteins in MS-H 4 . However, the latter substitution resulted in a slight change in the percentage of alpha helices from 55% in MS-H to 56% in MS-H 4 , and this could potentially alter the secondary structure of the protein.
The Arg185 in NAD-dependent glyceraldehyde-3phosphate dehydrogenase in MS-H was in a low mutation sensitive region. The Arg185lys could potentially change the secondary structure of the protein in 101,546 compared to that of MS-H as the percentage of beta strands changed from 31% in MSH to 32% in 101,546.
The protein YbhB/YbcL Raf kinase inhibitor was also modelled in MS-H and 101,546. Residue Val74 was found in a highly sensitive mutation region and therefore Val74Ile could potentially affect the function of the respective protein in 101,546. The secondary structure of this protein was identical in MS-H and 101,546.
The effect of Glu1037Gly in DNA-directed RNA polymerase subunit beta in 101,731 was found neutral as Glu1037 was located in a low sensitive mutation site and the secondary structure of respective protein was identical in MS-H and 101,731.

Full-length OppF was detected in all MS-H reisolates
Amongst all mutations detected in MS-H reisolates, the frameshift mutation in the oppF gene appeared to have the most significant impact on the structure of its encoded protein and therefore was further investigated. The wild-type oppF was predicted to encode a polypeptide of 797 amino acids (approximately 94 kDa). Immunoblotting experiments with rabbit-anti-OppF-N antibodies detected the OppF protein of expected size (~94 kDa) in 86,079/7NS, MS-H 3 , MS-H 4 , 101,546 and 10,173,118, while did not detect any protein of similar size in MS-H and MS-H 5 cells (Fig. 2). The rabbit-anti-OppF-N antibodies also detected several presumably nonspecific bands of similar sizes in all MS strains/isolates lysates tested.

Discussion
This is the first study that investigates the stability of all mutations in a live attenuated mycoplasma vaccine after in vivo passage under field conditions. The initial swab cultures collected from MS-H vaccinated birds were passaged three times in vitro by selection of an individual colony from each step. It may be possible that in vitro passage of the clones may have incorporated selection pressure and bias into the expansion of a clonal population, however it is notable that the clones were compared against an in vitro propagated MS-H vaccine strain. Whole genome sequencing directly from clinical materials collected from vaccinated birds would be ideal to circumvent the potential of in vitro selection pressure, but currently available does not allow compilation of complete genome sequence reliable at a base pair level. Also, current techniques may run the risk of generating a chimeric genome generated from multiple clonal populations that may cohabit the bird's respiratory system. A recent study [12] described 32 mutations within the MS-H genome as compared to its parent strain 86,079/ 7NS. However, the stability of these mutations after passage in vivo had only been tested for those found in obg [8] and oppF [9].
Protein homology modelling found that four sequence variations between MS-H and reisolates from vaccinated flocks, located in genes coding for Obg, OppF, Cardiolipin, and YbhB/YbcL Raf kinase-inhibitor, were likely to affect the in vitro and/or in vivo characteristics of MS.
Given that oppF is involved in pathogenesis of M. bovis [10], and the wild-type OppF from MS shares 43% amino acid similarity with that of M. bovis, the genomes of five MS-H field isolates differing in the oppF gene with that from MS-H and 86,079/7NS was analysed in this study to reflect the possible role of oppF in temperature sensitivity/attenuation phenotype of MS-H. Mycoplasmas can survive in vivo due to complex interaction between the microorganism and the host environment [18]. A continuous source of a nutrient used by a gene that is essential for in vivo survival may be a vital factor in the capability of a pathogen to cause disease [10]. Several nutrients are gained from exogenous sources by mycoplasmas as a result of their limited synthesis pathways. Hence, the ability to integrate molecules over membrane-associated transport systems appears to be a substantial factor for in vivo survival of mycoplasmas. In M. bovis, two transporters (oligopeptide transporter oppABCDF and an uncharacterized transporter) were essential for colonization on the tracheal mucosa [10]. In M. mycoides subsp. mycoides, a glycerol transporter (gtsABC) has been specified as a virulence factor related with hydrogen peroxide production and induction of cytotoxicity [19][20][21]. The level of mRNA expression of oppD of M. hyopneumoniae was moderately upregulated throughout in vivo infection [22] and under iron-depletion conditions [23]. Therefore, all available studies on the role of OppF in several Mycoplasma species are highly suggestive that OppF has a major contribution to the attenuation of MS-H. It is notable that in Western Immunoblot analysis conducted as part of this study, the truncated OppF was not delectable in MS-H and MS-H 5 (Fig. 2). It is speculated that the truncated version of OppF does not react well with polyclonal antibody against N terminus of OppF. The repeat of this Western Immunoblot in this and our previous publication [24] has shown that truncated version of OppF has only minimal reaction against anti-OppF-N polyclonal antibody. It is postulated that most of epitopes of this antibody are probably conformational (as opposed to linear) and may require of the remaining OppF protein to fully react and provide a readily detectable band on Western Immunoblot.
In bacteria, the Cardiolipin levels have been found to elevate in the stationary growth phase due to upregulation of Cls activity in response to osmotic stress [25]. The importance of anionic phospholipids cl in the osmotic adaptation and in the membrane structure of Bacillus subtilis cultures was demonstrated by impairment of osmotolerance in a Cls mutant (clsA) of this organism. As well as the lack in cl synthesis, this mutant indicated other deficiencies in lipid and fatty acid content compared to the wild-type, signifying a crossregulation in membrane lipid pathways, critical for the conservation of membrane functionality and integrity [25]. Therefore, it appears that elucidation of the role of Cardiolipin in attenuation of MS-H needs further investigation.
Given that the amino acid substitution in Cls of MS-H 4 was predicted to change secondary structure of the respective protein compared to that of MS-H and mutation resides in a highly sensitive mutation region, it is likely that this mutation affects the function of the respective protein in MS-H 4 .
The two proteins YbhB and YbcL belong to Raf kinase family and play role in the regulation of protein phosphorylation by kinases in E. coli [26]. Phosphorylation and dephosphorylation of proteins play a fundamnetal role in signalling in bacteria [27,28]. Previous studies have confirmed the significance of the phosphorylation of threonine and histidine residues and serine/threonine kinases which were involved in pathogenicity and stress responses in several prokaryotes [29]. Although the Val74Ile substituion in YbhB/YbcL of isolate 101,546 was unlikey to influence the secondary structure of the respective protein, it was found in a highly sensitive mutation region and therefore could affect the function of this protein.
Earlier studies in our laboratory have revealed that GapA + M. gallisepticum (MG) ts-11 vaccine was more immunogenic and induced higher antibody response than the GapA − ts-11 population [30]. In MG, the GapA is determined as the primary cytadhesin molecule and is known to play role in prolonged colonization and survival of MG [31,32]. Interestingly, isolate 101,546 was recovered from a MS-H vaccinated flock with unusually high systemic antibody response to MS. This isolate had a mutation in NAD-dependent glyceraldehyde-3phosphate dehydrogenase, located in large pocket regions and found likely to change the secondary structure of the respective protein compared to that of MS-H.
Comparative analysis of the genomes of selected MS isolates from MS-H vaccinated flocks revealed that they were true reisolates of the MS-H vaccine as they had highly similar genome to that of MS-H as opposed to 86,079/7NS. Results of this study also demonstrated that out of 32 mutations found in MS-H genome [12], only four to be reversible (Table 1) after passage in field birds. Thus, the 28 other mutations appear to be stable in MS-H. Of the four unstable mutations, twos (found in the Obg and OppF), were predicted to have some effects on MS virulence.
The mutations which are prone to revert are those that provide advantages to the organism to grow faster or grow in different parts of the respiratory system. For example, reversion mutations in obg provide organism higher capacity to live in lower respiratory system or mutation in oppF provides organism utilising amino acids more efficiently. These are important to drive reversion to wild-type organism. The mutations that were not found to revert organism to wild-type state probably do not provide the vaccine a significant advantage in vivo.
Given that obtaining pure cultures of the MS-H reisolates characterised here had to undergo multiple steps of growing in liquid and solid media, it may be possible that some of the mutations detected were as a result of in vitro passage. Future studies should therefore target these mutations directly in clinical specimens collected from vaccination chickens.
The data generated in this study also set the foundation for future research aiming to develop strain identification tests that reliably distinguish MS-H from other MS strains that possess identical vlhA gene sequence. Furthermore, using a set of mutations found here, it may be possible to correlate results emerging from genotyping techniques to variations in characteristics of MS isolates.

Conclusion
Results of this study reveal that most of the MS-H mutations are stable after passage in vaccinated chickens. Characterisation of stable mutations observed in MS-H could be applied to develop rapid diagnostic techniques for differentiation of vaccine from field strains or ts-MS-H reisolates.

MS strains, growth conditions, and DNA extraction
All MS-H isolates used in this study (Table 3) were made from flocks vaccinated with MS-H at various times after vaccination. All initial swab cultures were cloned by selection of individual colonies three times. The MS-H isolates were grown in mycoplasma broth supplemented with 10% swine serum (Sigma-Australia) and 0.01% (w/v) of nicotinamide adenine dinucleotide (NAD) (Sigma-Australia) [33] at 37°C in a 50 mL final volume until late logarithmic phase (approximately pH 6.8). Cells were collected followed by extraction of genomic DNA as described previously [12]. The DNA concentration was measured using the optical density at 260 nm (OD 260 ) using a NanoDrop™ 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) and purity was evaluated by calculating the OD 260/280 ratio. The integrity of DNA was assessed using chromatography through 0.8% agarose gel and DNA products were stored at − 80°C until use.

Next-generation sequencing (NGS)
NGS of all MS-H field isolates was performed using Paired-end 125-bp reads by the Illumina MiSeq platform at the Australian Genome Research Facility Ltd. (AGRF, Melbourne, VIC, Australia).

De novo assembly and sequence analysis
SPAdes assembler version 3.10.0 (Geneious® version 11.1.3) was used to perform De Novo assembly of contiguous sequences. To visualize overall sequence similarity and identify genomic organisation between the MS-H and its field isolates, the contigs were aligned to MS-H genome (GenBank accession number CP021129) as reference using Mauve (Mauve Contig Mover (MCM)), Geneious®. The MCM aligns a draft genome to a reference sequence and orders the contigs in the draft genome according to their position along the reference genome [34,35].
The resulting contigs and Illumina short reads were mapped to the MS-H genome using Geneious as mapper in Geneious®. Subsequently the alignments were subjected to single nucleotide polymorphism (SNP) and insertion/deletion (indel) analysis. To detect SNPs and indels, 'Find Variations/SNPs' in Geneious® was used.
The genome sequence of strain 86,079/7NS (GenBank accession numbers NZ_CP012624) was also included as reference for analysis of SNPs.

Homology modelling of proteins vary between isolates
The Phyre2 (protein homology/analogy recognition engine V 2.0) web portal for protein modelling, prediction and analysis (http://www.sbg.bio.ic.ac.uk/~phyre2/html/ page.cgi?id=index) [38] was used for homology modelling of proteins deduced from genes harboured SNP and indel variants in MS-H field isolates. Intensive mode of modelling was selected which performs complete modelling of the entire protein using multiple templates and ab initio techniques. Furthermore, the resultant modelled protein was subjected to Phyre investigator for more in-depth analysis [39].

Detection of OppF expression in MS strains/isolates
One ml volumes of mycoplasma broth were inoculated with 1/10 dilution of MS strains/isolates (Table 3) and grown to late exponential phase (~pH 6.8). The cells were treated as described previously [24] and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by Immunoblotting with mono-specific rabbit sera raised against N terminus of OppF [24].