Comparative Genomic Analyses of Attenuated Strains of Mycoplasma gallisepticum

Abstract

Mycoplasma gallisepticum is a significant respiratory and reproductive pathogen of domestic poultry. While the complete genomic sequence of the virulent, low-passage M. gallisepticum strain R (R_low) has been reported, genomic determinants responsible for differences in virulence and host range remain to be completely identified. Here, we utilize genome sequencing and microarray-based comparative genomic data to identify these genomic determinants of virulence and to elucidate genomic variability among strains of M. gallisepticum. Analysis of the high-passage, attenuated derivative of R_low, R_high, indicated that relatively few total genomic changes (64 loci) occurred, yet they are potentially responsible for the observed attenuation of this strain. In addition to previously characterized mutations in cytadherence-related proteins, changes included those in coding sequences of genes involved in sugar metabolism. Analyses of the genome of the M. gallisepticum vaccine strain F revealed numerous differences relative to strain R, including a highly divergent complement of vlhA surface lipoprotein genes, and at least 16 genes absent or significantly fragmented relative to strain R. Notably, an R_low isogenic mutant in one of these genes (MGA_1107) caused significantly fewer severe tracheal lesions in the natural host compared to virulent M. gallisepticum R_low. Comparative genomic hybridizations indicated few genetic loci commonly affected in F and vaccine strains ts-11 and 6/85, which would correlate with proteins affecting strain R virulence. Together, these data provide novel insights into inter- and intrastrain M. gallisepticum genomic variability and the genetic basis of M. gallisepticum virulence.

Mycoplasma gallisepticum is an avian respiratory and reproductive tract pathogen which has a significant economic impact on the poultry industry in the United States and worldwide. Elucidation of the pathogenic mechanisms by which M. gallisepticum exerts its effects on poultry is critical for the rational pursuit of improved vaccines and control strategies. To date, few virulence determinants or virulence-associated determinants have been identified. Attachment to the respiratory epithelium is essential to host colonization and is mediated by the primary and accessory cytadhesins GapA and CrmA, respectively (16, 46, 51, 53, 54). PlpA (44) and the OsmC-like protein (27) have been shown to bind host extracellular matrix molecules fibronectin and heparin, respectively, potentially aiding in attachment of M. gallisepticum to eukaryotic cells during infection. The OsmC-like protein also confers organic hydroperoxide resistance for M. gallisepticum in the extracellular milieu and may be essential for survival and pathogenesis in the host (28). The numerous lipoproteins encoded by the vlhA gene family have been shown to mediate phase variation of the bacterial surface architecture and are believed to be involved in evasion of the host immune system (5, 15, 18, 42). Metabolic pathways have also been shown to contribute to M. gallisepticum virulence. Dihydrolipoamide dehydrogenase (Lpd), a component of the pyruvate dehydrogenase complex, has been shown to be required for in vivo growth and survival in the host. A mutation in this gene resulted in the generation of the strain designated MG7, which exhibited reduced ability to cause tracheal lesions relative to the virulent progenitor strain R_low (25).

The need to control the spread of M. gallisepticum on multiage layer farms prompted the development of the live attenuated vaccine (LAV) strains F, ts-11, and 6/85. The F strain is a naturally occurring field isolate originally isolated in the 1950s in the United States (74). It is attenuated in older chickens yet retains virulence in young chickens and turkeys (2, 38). The ts-11 strain is a laboratory-generated, temperature-sensitive mutant produced via chemical mutagenesis of a virulent Australian field strain (73). Strain 6/85 was generated by serial passage of a virulent strain of M. gallisepticum (10). Although all three of these LAV strains are generally considered safe, the efficacy, transmissibility, and residual virulence of these vaccine strains vary considerably (2). Currently, the genetic basis for reduced virulence of these strains is poorly understood.

The virulent, low-passage M. gallisepticum strain R (R_low) has been serially passaged in broth culture to generate an attenuated high-passage strain (R_high) (34). Comparison of these two strains by SDS-PAGE has led to the identification of proteins specifically absent in R_high, including GapA, CrmA, PlpA, HatA, and Hlp3 (44, 54). This analytical approach, however, is limited in its ability to detect the full complement of genes/proteins that are missing or disrupted in R_high. The complete genome sequence of R_low was published in 2003 and has proved invaluable in studies of M. gallisepticum pathogenesis (52). Paralogous redundancy was observed in types of genes where plasticity is likely important for survival in the host environment, including 43 vlhA genes arranged in five multigene loci and 24 ABC transporter component genes. Approximately one-third of the predicted M. gallisepticum genes are of unknown function and/or are unique to M. gallisepticum, and these likely include novel genes important for virulence in the host. A comprehensive analysis of which genes are important for M. gallisepticum virulence is lacking.

In order to better understand the genetic basis of virulence in M. gallisepticum, we have utilized genomic sequencing and comparative genomic hybridizations (CGH) to identify genomic differences between virulent and avirulent strains that may relate to virulence. Here, we present complete genome sequence analysis of attenuated M. gallisepticum strains R_high and F and compare them to the virulent strain R_low as well as CGH data identifying features of the R_low genome absent from the genomes of strains ts-11 and 6/85. These specific genomic differences provide insights into determinants responsible for the avirulent nature of these strains and may provide targets for mutagenesis in the pursuit of development of a more efficacious vaccine.

MATERIALS AND METHODS

Strains, growth conditions, and DNA extraction.

M. gallisepticum strains R_low and R_high were obtained from Sharon Levisohn (The Hebrew University Hadassah Medical School, Jerusalem, Israel). Strains ts-11 and 6/85 were obtained from David Ley (College of Veterinary Medicine, North Carolina State University, Raleigh, NC). The F strain was purchased from Schering-Plough Animal Health (Kenilworth, NJ). All strains were grown in complete Hayflick's medium (23) at 37°C. Clonal isolates (three isolates) of strains R_high and F were selected for sequencing. Total genomic DNA was prepared from mycoplasma cultures using an Easy DNA Kit (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. An MGA_1107 isogenic mutant was constructed through random transposon mutagenesis of strain R_low as previously described (25) and selected using MGA_1107-specific PCR screening of mutant clones.

Library construction.

Random DNA fragments were generated by incomplete digestion of genomic DNA with the restriction endonuclease Tsp509I (New England BioLabs, Beverly, MA). DNA fragments (1.0 to 5.0 kb and 5.0 to 10.0 kb) were size selected using a Zymoclean Gel DNA Recovery Kit (Zymo Research, Orange, CA) and cloned into EcoRI-digested and dephosphorylated pUC19 plasmid (New England BioLabs, Ipswich, MA) to generate two libraries each for both the R_high and F strains. Plasmids were transformed into Escherichia coli One Shot Max Efficiency DH5α T1 chemically competent cells (Invitrogen) and subsequently extracted using a PerfectPrep Plasmid 96 Vac, Direct Bind DNA prep protocol according to the manufacturer's instructions (Eppendorf, Westbury, NY).

Sequencing and assembly.

DNA sequencing reactions were performed in 96-well format using M13 forward and reverse primers and Applied Biosystems (ABI) BigDye Terminator, version 3.1, Cycle Sequencing Kits (Applied Biosystems, Foster City, CA) and cleaned via ethanol precipitation. Reactions were run on an ABI 3730xl DNA Analyzer (ABI), and data were collected, bases were determined, and sequences assembled using ABI Data Sequencing Analysis Software (version 5.2), Phred (11), Phrap (http://www.phrap.org), CAP3 (24), and MIRA2 (8), with manual editing using Consed (17). Finishing reactions were performed using custom primers for PCR amplification to close gaps in the assembly. Final, circularized R_high and strain F sequences were derived from assemblies containing approximately 10.3-fold and 10.7-fold sequence coverage, respectively.

Reannotation and correction of the R_low genome.

The genome sequence of the virulent R_low strain was published in 2003, but our comparative analyses indicated that reannotation was necessary. This included updating gene predictions utilizing the three translational start sites believed to be used by mycoplasmas (ATG, GTG, and TTG) (58) rather than the eight included in NCBI Codon Table 4. The annotation was also updated to include previously undetected functional assignments for unique and conserved hypothetical proteins and new pseudogene information. Loci discrepant between R_low and R_high consensus sequences were manually examined for support at the trace data level. Sequences ambiguous or erroneous in the original R_low consensus were PCR amplified from the shotgun-sequenced clonal isolate (R_low clone 2), cloned, sequenced, and corrected if necessary.

TABLE 4.

R_low genes absent or divergent in vaccine strains as indicated by CGH^a

Locus	Product	Absent or divergentb
MGA_0581	ATP synthase subunit beta fragment	D
MGA_1328	Deoxyribose-phosphate aldolase fragment	−
MGA_0604	tRNA modification GTPase MnmE	D
MGA_0801	Subtilisin-like serine protease fragment	−
MGA_0802	Subtilisin-like serine protease fragment	−
MGA_0836	Putative Holliday junction resolvase	D
MGA_1108	Putative transposase fragment	−

^aGenes generating negative hybridization signals and verified by PCR.

^bD, gene present but divergent; −, gene absent.

Sequence analyses.

General nucleotide analyses and comparisons, single nucleotide polymorphism (SNP) analysis, and repeat analysis were conducted with DOTTER and programs in the MUMmer, EMBOSS, and REPuter packages (32, 59, 66). Open reading frames (ORFs), potential coding DNA sequences (CDSs), rRNA, and tRNA were identified using EMBOSS, Glimmer (9), and Prodigal (D. Hyatt, L. Hauser, G.-L. Chen, P. Locascio, F. Larimer, and M. Land [http://compbio.ornl.gov/prodigal/]) using a modified genetic code 4 and programs implemented in tRNAscan-SE and RNAmmer (33, 40), respectively. Screening for unusual coding differences in the R_high and F genomes (stops and frameshifts) relative to predicted proteins of strain R_low were conducted using FASTA program packages (55, 56) and BLAST (3). ORFs were used in local searches against nonredundant (NCBI), domain and family profile protein databases, as done previously (52), and against the HAMAP database (37), using BLAST, rpsblast (http://www.ncbi.nlm.nih.gov/BLAST/), and HMMer (http://hmmer.wustl.edu/). ORF graphical visualization and manual annotation were aided using Artemis, release 9 (The Sanger Institute, Genome Research Unlimited, United Kingdom). In addition to laboratory computational resources, resources available at the University of Connecticut Biotechnology-Bioservices Center Bioinformatics Facility were used.

CGH.

Oligonucleotide-based microarrays, designed to represent 756 ORFs using R_low genomic information and used to perform CGH analysis, have been previously described (7). DNA was extracted from mid-log-phase cultures of the respective strains of M. gallisepticum using the manufacturer's protocol 3 of the Easy DNA kit (Invitrogen, Carlsbad, CA). DNA concentrations were determined from readings of the optical density at 260 nm (OD₂₆₀), and purity was measured by the OD_260/280 ratio using a Thermo Spectronic Biomate 3 spectrophotometer (Thermo Scientific, Waltham, MA). Four micrograms of genomic DNA was labeled using a BioPrime Array CGH Genomic Labeling System (Invitrogen, Carlsbad, CA), according to the manufacturer's instructions (36), and tagged with Cy3 fluorescent dye (Amersham Biosciences, Piscataway, NJ). Microarrays were hybridized with a single DNA sample overnight at 42°C and washed in decreasing concentrations of SSC buffer (2× SSC plus 0.1% SDS, 1× SSC, 0.1× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]). Image acquisition was performed using a GenePix 4000B microarray scanner (Axon Instruments, Union City, CA) at a photomultiplier tube (PMT) gain of 700. Hybridizations were performed in triplicate.

Analysis of CGH data.

CGH data were analyzed in Microsoft Excel using log₂ ratio transformations of the median signal intensity of the test strain compared to the reference strain R_low [M = log₂ μ_test/μ_{reference R(low)}]. Genes which generated an M value of less than or equal to −2 (4-fold signal reduction in the test strain) were considered to be divergent or absent.

Confirmation of absent or divergent genes.

Genes suspected of being absent or divergent in the attenuated vaccine strains by CGH were confirmed by PCR amplification, cloning, and sequencing. The PCRs were performed with an AmpliTaq DNA Polymerase kit (ABI) using standard procedures and cycling conditions. Initial primers were designed (MWG Biotech, High Point, NC) to be within the ORF and flanking the microarray probe. If a PCR was negative, then a second primer set was designed flanking the ORF. Amplicons were cloned into the Topo TA vector (Invitrogen), and plasmids were extracted using a Qiagen Plasmid Mini kit according to the manufacturer's protocol (Qiagen Inc., Santa Clarita, CA). Sequencing reactions were carried out by MWG Biotech using an ABI capillary DNA sequencer, and contiguous sequences were assembled and analyzed using Sequencher, version 4.2 (Gene Codes Corporation, Ann Arbor, MI) and Artemis, release 9.

Animal challenge experiments.

Challenge experiments were conducted as previously described (12). Briefly, groups of six, 4-week-old, female, specific-pathogen-free, White Leghorn chickens (SPAFAS, North Franklin, CT) were challenged in high-efficiency particulate air (HEPA) isolator units, per IACUC protocol (A07-001). Cultures of R_low and the MGA_1107 mutant were grown from frozen stocks at 37°C with shaking at 130 rpm in fresh Hayflick's medium for 5 h prior to animal inoculation and quantified using the OD₆₂₀ and color-changing unit (CCU) serial dilutions. Chickens were challenged intratracheally on days 0 and 2 with 200 μl (1 × 10⁸ CFU) of culture or Hayflick's medium alone and humanely sacrificed on day 14 postinoculation.

Chickens were necropsied, and gross pathological findings were noted. Annular segments from cranial, middle, and caudal thirds of the trachea were sampled for mycoplasma isolation and histopathologic assessment. Tissue samples taken for histopathologic study were fixed by immersion in 10% neutral buffered formalin, routinely processed for paraffin embedment, sectioned at 4 μm, and stained using hematoxylin and eosin. Histopathological assessments were conducted in a blinded fashion by a veterinary pathologist, with tracheal scoring adapted from Nunoya et al. (50) and implemented as in Gates et al. (12): 0, no significant findings; 0.5, minimal multifocal lymphocytic or lymphofollicular infiltrates of one to three discrete foci; 1, mild mucosal thickening resulting from (i) multifocal lymphocytic or lymphofollicular infiltrates of four or more discrete foci or (ii) a mild diffuse lymphocytic infiltrate; 2, moderate mucosal thickening resulting from multifocal to diffuse lymphocytic and histiocytic infiltrates with or without lymphofollicular infiltrates or intraepithelial and lamina proprial infiltrates of heterophils; and 3, severe mucosal thickening resulting from diffuse infiltrates of lymphocytes, histiocytes, and heterophils with necrosis or squamous metaplasia of epithelium and luminal exudates. Tracheal sections that varied in severity along segments between two scoring values were assigned mid-range scores, e.g., 1.5 or 2.5. In addition to histopathological lesion scores, tracheal mucosal widths were measured as described previously (25). Using a light microscope and micrometer, the width of the tracheal mucosa was measured from the base of the lamina propria to the top of the epithelium at the base of the cilia. Measurements were made at four equidistant points around the circumference of each tracheal segment. Lesion scores and mucosal widths were analyzed using a nonparametric Kruskal-Wallis one-way analysis of variance on ranks test, with pairwise multiple comparison procedures performed using the Student Newman-Keuls method as implemented in SigmaPlot, version 11 (Jandel Scientific, San Rafael, CA), with a P value of <0.05 used to determine statistical significance.

Tracheal samples taken for mycoplasma isolation were cultured in 5 ml of Hayflick's medium for 4 h, vortexed, and filtered through 0.45-μm-pore-size Millipore filters (Millipore, Billerica, MA) to eliminate nonmycoplasmal contaminants. Cultures that turned acidic (yellow) after being filtered were adjusted to an approximate pH of 7.4 by the addition of 10 N NaOH. Serial 10-fold dilutions prepared to 10⁵ were incubated at 37°C for 28 days to assess CCU.

Accession numbers.

M. gallisepticum strain R_high and F genome sequences have been deposited in GenBank under accession numbers CP001872 and CP001873, respectively. CGH array data were deposited in the NCBI Gene Expression Omnibus database under accession number GSM489674.

RESULTS AND DISCUSSION

Reannotation and correction of the R_low genome.

De novo sequencing of the R_high genome and comparative analyses ultimately allowed for updates to be made to the previously deposited sequence of the progenitor strain R_low. These comprised changes at 52 loci and included 19 single nucleotide indels and 21 substitutions, with several leading to incorrect disruption of eight genes (see Table S1 in the supplemental material). The changes corrected misassembly in the clustered regularly interspaced short palindromic repeats (CRISPR) region, a noncoding region characterized by 36-bp exact repeats interspersed by 30-bp unique regions (reviewed in reference 67). Added to the R_low genome sequence was approximately 2.2 kb of sequence originally excluded from within, and adjacent to, an approximately 1.7-kbp direct repeat within the CRISPR. Also added in the vhlA4 locus region of the R_low genome was an approximately 14.2-kbp tandem repeat sequence containing six vlhA genes (MGA_1336 to MGA_1342) (Fig. 1, vlhA4.7.1 to vlhA4.7.6). These changes resulted in a revised genome size of 1,012,800 bp for R_low. Use of standard bacterial start codons altered translational start positions of 419 (58%) previously annotated coding sequences (CDSs), changing their predicted coding and upstream sequences. Finally, reanalysis altered the predicted genetic content of R_low, with 40 RNA genes, 682 intact CDS gene regions, and 62 potential or likely pseudogenes (784 total genes) now predicted.

FIG. 1.

Major genomic differences between F and R strain genomes. (A) Dot plot of F and R_low genome sequences. vlhA loci are indicated with dotted lines and expanded to demonstrate gene structure at vlhA5 loci (B), vlhA1 loci (C), and vlhA4/3 loci (D). vlhA genes are indicated with gray arrows and also noted with vlhA numbers in R strain; genes present only in F strain adjacent to the vlhA5 locus are indicated with black arrows. Genes similar and potentially orthologous between the two genomes are indicated with blocks or diagonals.

Genome comparisons of R_high and R_low.

The 156-passage R_high genome was 1,012,027 bp in length and highly similar to that of its R_low parent, exhibiting no large-scale insertions, deletions, or genomic rearrangements. Genomic differences were observed at a total of 64 loci, with the majority comprised of small-scale changes, including 55 indels of which 53 were located in small repetitive tracts (Table 1). This highlights the variable nature of repeats in M. gallisepticum and is consistent with differences in tandem repeat elements observed between strains of Mycoplasma hyopneumoniae (70). Twenty-three indels were located in predicted CDSs, and 32 were in noncoding regions, including 29 located in variable GAA repeat sequences involved in phase-variable expression of vlhA genes (14). Noted were only nine substitutions (five transitions and four transversions), six of which were found in ORFs and were nonsynonymous. Thus, of the 29 differences observed in potential coding regions, all introduced amino acid substitutions or indels, translational stops, or frameshifts that might indicate functional protein differences between R_high and R_low (Table 1). Though some of these genomic changes have been previously described, specifically in cytadherence-related genes gapA (54), crmA (51), plpA (52), and hlp3 (44), mutations in additional genes accumulated over 156 in vitro passages likely contribute to the attenuated R_high phenotype.

TABLE 1.

Nucleotide and coding differences identified between genomes of R_low and R_high

Rlow		Nucleotide difference (Rlow/Rhigh)d	Rhigh		Locus and/or protein/region	Effect on Rhigh codingb
Position (nt)a	Lengthc		Position (nt)a	Length
15684	1	G/T	15684	1	MGA_0644, glycerol kinase	Stop; 157-aa truncation
46081	1	C/T	46081	1	MGA_0677, glycerol-3-phosphate transport protein	T417I substitution
65904	1	A/C	65904	1	MGA_0704, conserved membrane protein	Y64D substitution
218873	54	Repeat/—	218872	0	MGA_0928, adherence-associated protein Hlp3	18-aa deletion
223599	0	—/A	223546	1	MGA_0934, cytadhesin GapA	Frameshift out 995 aa
255751	6	GAA repeat	255697	0	Intergenic, VlhA4.03 promoter region
255971	12	Repeat/—	255911	0	MGA_0968, VlhA4.03a N-terminal fragment	TPNP repeat deletion
260507	3	GAA repeat	260435	0	Intergenic, VlhA4.05 promoter region
262821	9	GAA repeat	262746	0	Intergenic, VlhA4.06 promoter region
265255	3	GAA repeat	265171	0	Intergenic, VlhA4.07 promoter region
269969	3	GAA repeat	269882	0	Intergenic, VlhA4.07.2a promoter region
270185	0	—/Repeat	270097	12	MGA_1336, VlhA4.07.2.a	TP repeat insertion
272304	0	GAA repeat	272228	3	Intergenic, VlhA4.07.3 promoter region
274692	6	GAA repeat	274617	0	Intergenic, VlhA4.07.4 promoter region
277009	9	GAA repeat	276928	0	Intergenic, VlhA4.07.5 promoter region
279442	0	GAA repeat	279354	15	Intergenic, VlhA4.07.6 promoter region
282061	6	Repeat/—	281986	0	MGA_0979, VlhA4.08	TP repeat deletion
284335	1	T/C	284255	1	Intergenic, VlhA4.09
284341	1	G/A	284261		Intergenic, VlhA4.09
284396	6	Repeat/-	284315	0	MGA_0981, VlhA4.09	TP repeat deletion
291716	6	GAA repeat	291629	0	Intergenic, VlhA4.12 promoter region
307991	3	ATG/—	307898	0	MGA_1000, RNA polymerase subunit beta	D repeat deletion
358895	18	Repeat/—	358799	0	MGA_1071, conserved hypothetical protein	TPTPTP repeat deletion
389142	1	C/G	389029	1	Intergenic, asparaginyl-tRNA synthetase
393735	0	—/G	393623	1	MGA_1117/9, CrmB-like protein	ORF fusion
421612	1	G/A	421500	1	MGA_1161, conserved hypothetical protein	S323N substitution
448139	0	—/Repeat	448028	17	MGA_1199, fibronectin binding protein PlpA	Frameshift out 166 aa
480523	0	GAA repeat	480429	3	Intergenic, VlhA5.03 promoter region
483041	0	GAA repeat	482950	6	Intergenic, VlhA5.04 promoter region
487897	9	GAA repeat	487810	0	Intergenic, VlhA5.06 promoter region
490341	0	GAA repeat	490247	33	Intergenic, VlhA5.07 promoter region
492725	0	GAA repeat	492664	3	Intergenic, VlhA5.08 promoter region
495034	3	GAA repeat	494974	0	Intergenic, VlhA5.09 promoter region
495312	13	13 bp/7 bp	495250	7	MGA_1251, VlhA5.09	TP repeat deletion
497434	48	GAA repeat	497365	0	Intergenic, VlhA5.10 promoter region
499881	18	GAA repeat	499764	0	Intergenic, VlhA5.11 promoter region
594921	3	GAA repeat	594786	0	Intergenic, VlhA1.02 promoter region
597248	12	GAA repeat	597110	0	Intergenic, VlhA1.03 promoter region
599667	6	GAA repeat	599517	0	Intergenic, VlhA1.04 promoter region
606824	0	GAA repeat	606670	12	Intergenic, VlhA1.06 promoter region
609365	12	GAA repeat	609221	0	Intergenic, VlhA1.07 promoter region
611661	3	GAA repeat	611505	0	Intergenic, VlhA1.08 promoter region
657054	1	A/C	656896	1	MGA_0141, nucleotide phosphodiesterase	N56K substitution
670334	6	Repeat/—	670175	0	MGA_0162, dihydrolipoamide acetyltransferase	GG repeat deletion
672071	0	—/CTT	671908	3	MGA_0165, pyruvate dehydrogenase E1 subunit	E repeat insertion
677821	0	—/TGTATTTG (repeat)	677661	8	MGA_0173, TlyC-like protein	Frameshift out 67 aa
711767	0	—/GA (repeat)	711615	2	MGA_0221, ABC peptide transport permease	Frameshift out 99 aa
713340	20	Repeat/—	713188	0	MGA_0223/4, ABC peptide transport permease	ORF fusion
723519	1	T/C	723348	1	MGA_0237, ABC peptide transport solute BP	D467G substitution
736772	0	—/T (repeat)	736602	1	MGA_0250, unique hypothetical	61-aa N-terminal extension
740191	37	Repeat/—	740020	0	MGA_0256/8, phase variable protein PvpA	ORF fusion
757659	12	Repeat/—	757451	0	MGA_0287, amino acid permease	4-aa C-terminal deletion
815905	3	GAT/—	815685	0	MGA_0368, ABC transport permease	I repeat deletion
821389	0	GAA repeat	821168	21	Intergenic, VlhA3.01 promoter region
822227	0	—/ATAT (repeat)	822027	4	MGA_0379, VlhA3.02	Frameshift out 218 aa
823701	0	GAA repeat	823505	3	Intergenic, VlhA3.02 promoter region
828562	12	GAA repeat	828367	0	Intergenic, VlhA3.05 promoter region
833126	0	—/TGGAGT	832921	6	MGA_0388, VlhA3.06	TP repeat insertion
835662	0	GAA repeat	835463	12	Intergenic, VlhA3.07 promoter region
840488	3	GAA repeat	840299	0	Intergenic, VlhA3.09 promoter region
910723	581	581 bp/—	910531	0	MGA_0508, fructose-specific PTS component	N-terminal deletion
929813	66	Repeat/—	929040	0	CRISPR	Repeat deletion
930766	0	—/Repeat	929929	66	CRISPR	Repeat insertion
938625	1	A/— (repeat)	937852	0	Intergenic, MGA_0536 ATPase and MGA_0537 hsdM

^ant, nucleotide.

^baa, amino acids.

^cLength in nucleotides.

^dSpecific nucleotides, length of nucleotide sequences, and/or the presence of repeated sequences noted as indels or substitutions between R_low and R_high are given.

Among the genes truncated in R_high, two are involved with sugar metabolism, the regulation of which is thought to play a major role in the survival and virulence of mycoplasmas in vivo (reviewed in reference 21). The glycerol kinase gene glpK (MGA_0644) and the fructose/mannose-specific IIABC component gene fruA (MGA_0508) (60) of the phosphoenolpyruvate:fructose phosphotransferase system (PTS) are likely inactive, as both lack significant portions of coding sequence (157 and 179 amino acids, respectively). The fruA lesion (a deletion of 581 bp) is the largest indel observed between R_low and R_high. These genes are predicted to be necessary for transport/utilization of what is believed to be the three primary alternative carbon sources (glycerol, fructose, and mannose) used by M. gallisepticum in glycolysis and subsequent ATP production. Glycerol catabolism has also been implicated in mycoplasmal virulence as it yields very high concentrations of H₂O₂ per mole of O₂ and glycerol consumed (68). In Mycoplasma mycoides subsp. mycoides biotype small colony, glycerol uptake and utilization are strongly correlated with virulence. It is critical for the production of cytotoxic hydrogen peroxide by bacterial l-alpha-glycerophosphate oxidase (GlpO), which catalyzes the oxidation of glycerol-3-phosphate and release of H₂O₂ (6, 57, 71). M. gallisepticum encodes a homolog (MGA_0646) of the Mycoplasma pneumoniae GlpO, encoded by a glycerol dehydrogenase-like gene (glpD; MPN051), which was recently shown to mediate H₂O₂ production, thereby affecting host-cell cytotoxicity (22). Thus, glycerol uptake and conversion to H₂O₂ are potentially significant in M. gallisepticum virulence. The loss of GlpK function could conceivably result in reduced levels of glycerol-3-phosphate available for GlpO-mediated H₂O₂ production. Notably, one of the observed amino acid substitutions in R_high occurs within the ATP binding region of a putative ABC-type, sn-glycerol-3-phosphate transport system ATP-binding protein (MGA_0677), making a functional change in glycerol-3-phosphate transport also a possibility. PTS components have also been implicated as virulence factors. The plant pathogen Spiroplasma citri requires fructose PTS activity for virulence (13), and PTS functions and carbohydrate metabolism have been linked with in vivo survival in extraintestinal pathogenic E. coli (61). The fruA PTS component may also have a conceivable role in M. gallisepticum virulence.

Disrupted in R_high are other genes with limited functional commonality. Two disrupted genes had highly similar paralogs elsewhere in the genome and encoded a potential ABC-type peptide/nickel transporter permease (MGA_0221) and a member of the variable lipoprotein family, vlhA3.02 (MGA_0379). These genes have coding sequences that are frameshifted relative to orthologs in R_low, with R_high MGA_0221 lacking two transmembrane helices and a potential substrate binding domain and R_high vlhA3.02 lacking the C-terminal 218 amino acids present in R_low. Paralogs of these genes conceivably provide equivalent functions and thus directly complement MGA_0221 and MGA_0379 mutations. Conversely, disruption in these genes may indicate loss of, and/or functional shift away from, very specific transport or attachment functions affecting virulence, with paralogs conceivably complementing functions with different specificity.

Also disrupted in R_high was a gene with no obvious function, the conserved hypothetical gene MGA_0173. Identified in MGA_0173 was a tlyC domain which consists of a duf21 (domain of unknown function) transmembrane domain and a CBS domain potentially involved in ligand binding. MGA_0173 was also similar to M. pneumoniae tlyC, a gene proposed to encode a hemolysin but noted by other investigators as lacking a hemolysin domain (GenBank accession NP_109847). Despite the lack of obvious function for this gene, its disruption in R_high makes it an obvious target for investigation as a potential virulence determinant.

Genomic comparison of strains R_low and F. (i) General comparison.

The F strain genome was similar to that of R strain, sharing 747 predicted gene orthologs in largely syntenic genomic regions. The F strain was predicted to contain 781 total genes, including 688 intact CDS genes and 53 potential or likely pseudogenes. Despite sharing similar overall genomic content, however, the F strain demonstrated features indicating that it is clearly distinct from the R strain. Overall, F strain genes were 2.7% divergent at the amino acid level relative to orthologs in strain R, indicating significant genetic distance between these two strains of the species M. gallisepticum. The F strain genome was 977,612 bp in length, approximately 35 kbp shorter than that of R strain, largely due to a unique and reduced vlhA gene complement. Notable large-scale genomic differences included a genomic inversion of approximately 500 kbp between the vlhA3 locus and the vlhA4 locus (Fig. 1A) and approximately 17 kbp of novel sequence present in the F strain and absent in the R strain (Fig. 1B). Notable in the F strain genome was an insertion of a tandem duplication of approximately 25.7 kbp (duplicated within positions 305536 to 356916) which contained 17 genes corresponding to R strain MGA_0271 to MGA_0312. Other larger-scale indels in non-vlhA and non-CRISPR regions were limited (18 indels of 100 bp or more), with 14 of these present in regions containing only transposons and/or potential pseudogenes in both strains but with others affecting intact genes (Table 2). A total of 126 genes exhibited length differences between F and R strains, but most of these resulted in minor internal or terminal changes in affected proteins (52 genes), amino-terminal changes which likely reflected incorrect start codon prediction (17 genes), or changes in genes predicted to be fragmented in both strains (22 genes). Changes did, however, account for at least 25 F strain genes potentially intact or disrupted relative to strain R orthologs (Tables 2 and 3). Differences in coding potential were examined for clues as to the genetic basis of M. gallisepticum virulence. Of particular interest were differences in genes absent, disrupted, or highly divergent in strain F relative to strain R (Fig. 1 and Table 2; see also Table S2 in the supplemental material).

TABLE 2.

Genes fragmented or absent in strain F relative to strain R^a

F strain locus	Gene change(s)b	R strain locus	Predicted gene product
MGF_0026f	Fr	MGA_0626	ABC-type multidrug/protein/lipid (MdlB-like) transport system component
MGF_0725	Fr, I	MGA_0798	Subtilisin-like serine protease
MGF_0748f	Fr	MGA_0805	Putative ABC-type transport system protein
MGF_1439	Fr	MGA_0957	Conserved hypothetical protein
MGF_1750f	Fr	MGA_0323	Conserved hypothetical membrane protein
MGF_1948	Fr	MGA_0271	Unique hypothetical protein
MGF_2099f	Fr	MGA_1361	Unique hypothetical protein
MGF_3286	N, Fr	MGA_1305	MaoC-like dehydratase
MGF_3677f	Fr	MGA_1220	Arginine deiminase
MGF_4156	N	MGA_1100	Asparaginyl-tRNA synthetase
	I	MGA_1107	Conserved hypothetical RmuC-domain protein
MGF_4199	Fr, I	MGA_1083	HAD superfamily hydrolase
MGF_4219f	Fr	MGA_1081	Putative transposase
MGF_4466	N	MGA_1027	Conserved hypothetical protein
MGF_4633	Fr	MGA_1343	Unique hypothetical protein
MGF_5191f	Fr	MGA_0516	Unique hypothetical protein
MGF_5319f	Fr	MGA_0541	Type I site-specific restriction-modification system restriction (R) subunit
MGF_5453	B	MGA_0566	Unique hypothetical protein
MGF_5461/3	B	MGA_0567	Unique hypothetical protein

^aGenes shorter by at least 10% in length by deletion, frameshift, or premature stop relative to intact ortholog, and not likely the result of incorrect start codon prediction.

^bFr, gene fragmented relative to R strain; N, gene encoding protein lacking amino-terminal region relative to R strain; I, gene affected by indel event (>100 bp) relative to strain R; B, gene potentially fragmented in both R and F strains.

TABLE 3.

Genes intact or present in strain F relative to strain R^a

F strain locus	Gene changeb	R strain locus	Predicted gene product
MGF_0017	F	MGA_1322d/625	ABC-type multidrug/protein/lipid (MdlB-like) transport system component
MGF_0872	F	MGA_0829	conserved hypothetical lipoprotein
MGF_1196	F	MGA_0908/11	unique hypothetical membrane protein
MGF_2103	I		putative transposase
MGF_2868	I		putative transposase
MGF_3370	F	MGA_1284/5	ABC transporter ATP-binding protein
MGF_3373	I		unique hypothetical protein
MGF_3468	F	MGA_1350	hypothetical protein fragment
MGF_3470	Iu	MGA_1349	conserved hypothetical protein
MGF_3484	Iu		conserved hypothetical protein
MGF_3486	Iu		ABC-type maltose/maltodextrin transporter permease MalC
MGF_3501	Iu		ABC-type maltose/maltodextrin transporter permease MalG
MGF_3505	Iu		ABC-type maltose/maltodextrin transporter ATP-binding protein MalK
MGF_3514	Iu		unique hypothetical protein
MGF_3519f	Iu		alpha amylase superfamily protein (fragmented)
MGF_3529	Iu		GntR-family transcriptional regulator
MGF_3537	Iu		pullulanase
MGF_3555	Iu		putative alpha,alpha-phosphotrehalase
MGF_3567	Iu	MGA_1265	maltose phosphorylase
MGF_3987	F	MGA_1364	unique hypothetical protein
MGF_4139	I		putative transposase
MGF_4207f	I	MGA_1083	PTS lichenan-specific IIA component (fragmented)
MGF_4515	F	MGA_1014	unique hypothetical protein
MGF_5077	F	MGA_0487	conserved hypothetical protein
MGF_5456	B	MGA_1358	unique hypothetical protein
MGF_5547	F	MGA_0583/4	conserved hypothetical protein

^aGenes longer by at least 10% in length by deletion, frameshift, or premature stop relative to intact ortholog, and not likely the result of incorrect start codon prediction.

^bF, F strain gene of full-length relative to R strain; I, gene encoded in sequence present in F strain but absent in R strain; I^u, gene encoded in 17-kbp unique region; B, gene potentially fragmented in both R and F strains.

(ii) vlhA gene regions.

The F strain genomic regions most variable relative to strain R contained vhlA genes. vlhA loci displayed divergent and nonsyntenic gene complements suggestive of local and inter-vlhA rearrangements and paralogous gene gain/loss, with only the two-gene vlhA2 locus highly conserved between F and R strains (Fig. 1B to D). Indeed, the F strain contained 23 vlhA genes (20 intact), 28 fewer than the 51 present (44 intact) in R_low. Intact VlhA ORFs demonstrated lower average amino acid identity (88%, ranging from 61 to 99%) to R strain homologs than did non-VlhA homologs (98%). One intact F strain vlhA ORF (MGF_4735; predicted 75-kDa protein) was notably divergent from VlhA ORFs in the R strain (59% amino acid identity to VlhA4.11) yet more similar (73%) to the strain S6 ORF pMGA 1.4 (43). A 75-kDa immunodominant protein specific to the F strain has been previously described (30, 69); however, whether MGF_4735 encodes this protein remains to be shown. The nonsyntenic nature of many interstrain VlhA best-matches and the presence of a large genomic inversion, gene duplication, and indels suggested recombination in and around vlhA loci (Fig. 1). This included potential vlhA3-vlhA4 locus rearrangement bounding the genomic inversion, where genes present in the 3′ region of each R strain locus appear to have recombined and switched positions, resulting in the vlhA3/4 locus and the vlhA4/3 locus in the F strain (Fig. 1D). Additional complexity at these and other vlhA loci makes elucidation of discrete rearrangement events speculative. In addition, the unique 17-kbp sequence adjacent to the vlhA5 locus in the F strain is essentially in the same locus as most vlhA5 locus genes present in the R strain (Fig. 1B).

vlhA genes encode immunodominant lipoproteins and hemagglutinins that undergo phase-variable expression both in vitro and in vivo (5, 15, 41), and they are thought to be virulence determinants which facilitate establishment of chronic infection through immune evasion. Overall, the divergence between F and R strain loci was extreme relative to the single vlhA gene disruption observed between R_high and R_low and, regardless of mechanism, was consistent with phase-variable gene locus variation observed between strains of other mycoplasma species (62). Extreme variation in vlhA complement, be it phase-variable expression or interstrain genetic heterogeneity, likely reflects significant disruptive selective pressure exerted on these genes as they encode major immune targets, and it likely results in elicitation of different serological specificities in the host. Whether frameshifted vlhA genes are expressed directly or though recombination with other vlhA genes, as seen in Mycoplasma synoviae (49), is unknown; however, these, too, could conceivably contribute to antigenic variation in the host. Similarly, a recombinatorial effect on gene order and ultimately phase variation is not known in M. gallisepticum, nor is such a mechanism obvious, given the data here. Notably and despite this variation in vlhA complement, F strain continues to induce immune responses that are generally protective against distinct strains.

(iii) Transposons.

Transposons are mobile genetic elements encoding transposases and are capable of random genomic integration, disruption of coding sequences, and/or mediating movement of nontransposon sequence within or between organisms. Such transposon-mediated changes likely occurred between F and R strains as their transposase gene loci demonstrated variability. These included two distinct F strain transposase insertions (MGF_2103 and MGF_2868) into intergenic regions upstream of potential lipoproteins (MGF_2102 and MGF_2118) and a ribosomal protein. A third F strain transposase gene (MGF_4139) was intact, and it essentially replaced a 2,441-bp locus which contained both transposase gene fragments (MGA_1108/9) and a conserved protein gene (MGA_1107) present in strain R. MGF_4139 was similar to MGA_0910 transposase, which at a different locus disrupts the MGA_0908/0911 hypothetical transmembrane protein (multigene family) gene in strain R but is absent in strain F, leaving an intact MGA_0908/0911 ortholog (MGF_1196) (Table 3). Although a direct transposition between these two loci is possible, similarity to other M. gallisepticum transposases and remnants of MGF_4139 in R strain sequence leave this unclear. Overall, strain F contains 14 transposase genes (3 intact) relative to the 16 genes (2 intact) present in strain R. In addition, transposons may affect indel events in adjacent genes, as six genes with coding potential affected by larger indels in F strain relative to the R strain were adjacent to transposon loci.

(iv) MGA_1107 and virulence assessment.

Transposon-mediated genomic changes ultimately may act to alter virulence, host range, or tissue tropism. Notably, this may be the case for the R strain MGA_1107 gene, which again is adjacent to transposase genes and is absent in the F strain. MGA_1107 contains a domain shared among proteins involved in DNA metabolism, including proteins similar to the putative nuclease RmuC, thought to affect DNA recombination of short inverted repeat sequences in E. coli (31, 65). MGA_1107 also shares a relatively high level of amino acid identity (92%) with M. synoviae MS53_0172, indicating that the MGA_1107 gene, and perhaps flanking transposon sequences, have involved or mediated a horizontal gene transfer (HGT) event, consistent with previous observation of a likely vlhA HGT between M. synoviae and M. gallisepticum (48, 70). MGA_1107 genes were identical between R_low and R_high (Table 1). Based on these genomic data and on previous data indicating that MGA_1107 is transcriptionally upregulated in R_low upon exposure to cells to which M. gallisepticum binds (7), an isogenic mutant of MGA_1107 was assessed in a chicken challenge system for virulence in vivo. Though lung lesions and minor airsacculitis were induced by the MGA_1107 mutant, lung and air sac lesions have been previously reported to be significantly variable in this experimental system and thus are precluded from being used for quantitative purposes (54). Tracheal mucosal thickness was reduced compared to that in the wild type, and histopathological lesion scores were similar to those of negative-control birds (P < 0.05, analysis of variance [ANOVA] on ranks and posthoc pairwise comparison) (Fig. 2). This mutant was recovered from lung and air sac tissues (albeit to a lesser degree than R_low) but was unrecoverable from the trachea. These data indicate that MGA_1107 contributes in a yet uncharacterized manner to the generation of tracheal lesions typical of virulent M. gallisepticum infection in vivo and that the loss of this ORF may be a factor in the attenuated phenotype of the F strain. This experimental evidence illustrates the utility of the approach of genomic comparison of virulent and attenuated strains in identifying genetic factors that influence survival in the host or the production of lesions.

FIG. 2.

Attenuation of MGA_1107 mutant of R strain in chickens. (A) Histopathological lesion scores in tracheas of chickens infected with Hayflick's medium (Medium), mutant (MGA_1107) organism, and virulent organism (R_low). Horizontal bars indicate 25th percentile (bottom), median, and 75th percentile (top). Lowercase letters indicate statistically similar groups. (B) Tracheal mucosal thickness. Bar and letters are as described for panel A. (C) Microbiological recovery from tracheas of chickens. Dots indicate titers of samples recovered from individual birds.

(v) Subtilisin-like proteases.

Indels occurred within subtilisin-like protease genes, of which there are five paralogs (three intact) encoded in the R strain. F strain lacks 1,693 bp encoding the majority of the MGA_0798 subtilisin-like gene intact in the R strain. A similarly sized deletion (1,686 bp) removed the majority of the nearby MGA_0801 subtilisin-like locus; however, MGA_0801 was a likely pseudogene and thus was predicted to be nonfunctional in both strains. Orthologous, but fragmented, MGF_5102F and MGA_0517/8 subtilisin-like loci also demonstrated coding variation, with two frameshifts restoring all but the likely N terminus to MGF_5102F. Subtilisins are a ubiquitous family of proteases with a range of functions, including roles in bacterial virulence. All five genes in the R strain belong to the D-H-S subgroup of subtilases encoded in other pathogenic bacteria, including Bacillus anthracis (reviewed in reference 63). Other essential subtilases conferring microbial virulence include dentilisin and SufA. Dentilisin is used by the oral spirochete Treponema denticola to degrade host chemokines, cytokines, and fibrinogen (4, 45) and to rearrange the bacterial outer sheath (26). SufA, encoded by the Gram-positive bacterium Finegoldia magna (an opportunistic pathogen of humans), inactivates antimicrobial peptides and chemokines and is believed to aid in bacterial survival in the host (29).

(vi) HAD-like proteins.

Also affected by indels were genes encoding potential hydrolases of the haloacid dehalogenase (HAD) superfamily, of which strain R contains five paralogs. MGF_4199 lacked the PTS-like N-terminal domain of the R strain ortholog (MGA_1083), a fusion reflecting deletion of additional PTS lichenan-specific IIA component gene sequences (present as fragments in F strain MGF_4207f) from strain R. Two HAD hydrolase genes were duplicated in the 24-kbp tandem repeat, yielding seven HAD hydrolase loci in F strain. Though functions of mycoplasmal subtilases and HAD hydrolases are unknown, the presence of, and variability among, multiple copies in M. gallisepticum suggest a role in host interaction and pathogenesis.

(vii) hsd genes.

Disrupted or variable in the F strain relative to the R strain were host specificity of DNA (hsd) genes adjacent to a fragmented transposase (MGF_5343f). These encode protein subunits of a type I restriction-modification system (R-M) complex which mediates methylation (modification subunit, hsdM), sequence-specific recognition of methylation state (specificity, hsdS), and restriction enzyme activity (hsdR). The F strain hsdR gene (MGF_5319f) is prematurely terminated at nucleotide position 2307 (of 3,198), possibly resulting in a loss of restriction enzyme function. HsdS often contains N- and C-terminal domains of similar structures, but each has discrete sequence specificities (or target recognition domains). Both the R and F strains, however, encode two separate single-domain HsdS units, akin to a single N-terminal domain and similar to the single-domain ORFs present in M. pneumoniae and other bacteria. This is consistent with data indicating that single-domain dimerization confers proper HsdS function in E. coli (1). Notably, while the first copy of HsdS is identical between R and F strains (MGF_5309/MGA_0539), the central domain of the second copy (MGF_5313/MGA_0540) is highly divergent, with little recognizable nucleotide similarity.

Hsd systems primarily protect bacteria from large fragments of foreign DNA such as those encountered during bacteriophage infection and which may interfere with normal cellular processes (reviewed in reference 47). Mycoplasma pulmonis encodes a unique hsd system that undergoes phase-variable gene expression and generates hsdS sequence variation (and likely target sequence specificity) through sequence-specific recombination between two distinct hsd loci (64). In addition, M. pulmonis hsd expression has been associated with bacterial tissue tropism as expression becomes active in the lower, but not upper, respiratory tissues of rodent hosts in vivo (19). While the extreme sequence divergence observed here between orthologous MGF_5313 and MGA_0540 genes could conceivably be generated through a recombination process, lack of both a second hsd locus and obvious inverted repeats bounding the divergent hsdS domain makes this unlikely. Though a role for the hsd system in M. gallisepticum host range and/or virulence is speculative, hsd genes do appear to be under different selective pressures in the R and F strains.

(viii) Transport proteins.

Though none appeared to involve glucose metabolism, multiple solute transporter-like genes were variable in F strain relative to R strain. These included two genes fragmented and two genes intact in F strain relative to R strain. Fragmented in F strain were MGF_0748f, a protein with weak similarity to ABC-transport proteins, and MGF_0026f, a prematurely terminated ortholog of the intact MGA_0626 mdlB-like gene in R strain. MGF_0026f contained a stop site located between the ABC transporter transmembrane domain and the ATP binding domain, likely affecting this protein, which is similar to proteins involved in multidrug efflux and transport of lipids and proteins. Conversely, intact in F strain relative to R strain were MGF_0017, a second mdlB-like gene adjacent to the fragmented MGF_0026f paralog, and MGF_3370, an intact ATPase component of an ABC transport protein of unknown specificity. The MGF_3378f locus contained a fragmented ortholog of the MGA_1283 mtlA-like gene in R strain; however, MGA_1283 itself is similar only to the C-terminal EIIB domain of MtlA. MtlA is a PTS transporter for mannitol and, although all genes required for mannitol utilization are present in the members of the pneumoniae clade, this system does not appear to work in M. pneumoniae (20), and mannitol transport has been reported to be absent in M. gallisepticum strain NCTC 10115 (68). The variability between F and R strains in multiple solute transport proteins indicates that, again, they may affect growth and survival in different hosts or host tissues, likely affecting virulence.

(ix) Other disrupted strain F genes.

Other genes disrupted in F strain relative to R strain have metabolic functions or are of unknown function (Table 2). MGF_3677f is a highly fragmented arcA gene encoding arginine deaminase and is intact in R strain (MGA_1220). MGF_4156 encodes a protein orthologous to the MGA_1100 asparaginyl-tRNA synthetase in strain R, including the AspRS/AsnRS core domain; however, it lacks an N-terminal domain conserved in other species. While these changes potentially affect F strain metabolism, paralogs (MGF_2849 and MGF_4297) conceivably provide compensatory functions. Four conserved and six unique hypothetical proteins are fragmented in F strain relative to R strain. One genomic region containing four small hypothetical proteins (MGF_5453 through MGF_5463) is highly variable between F and R strains, suggesting that, although no intact homolog has been observed, these may represent fragments of a novel gene.

(x) Highly divergent genes.

Many genes demonstrated above-average amino acid divergence between F and R strains (see Table S2 in the supplemental material). While 173 intact F ORFs were identical to R strain homologs at the protein level, the rest differed on average by 2.4%. Among the most divergent protein homologs were those similar to known or putative cytadhesins and cytadherence-related proteins (Table S2). Indeed, of the 37 intact ORFs differing between strains by 5% or more, nine (24%) are putatively involved in cytadherence or tip structure formation, including GapA, CrmA, CrmB, MGC2, and several ORFs with similarity to HMW cytadhesin-related proteins. Also divergent is PvpA, a protein that is localized to the terminal tip structure, undergoes phase variation under antibody pressure both in vitro and in vivo, and is potentially involved in antigenic variation and immune evasion in the host (35, 75). Genomic sequences presented here confirm the loss of about 230 nucleotides previously reported in the direct repeat 1 (DR1) and DR2 regions in the F strain relative to the R strain (39).

(xi) Genes present or functionally intact in strain F.

In addition to genes absent or disrupted, the F strain genome contains several genes that are absent in the R strain genome or that are intact relative to homologs in strain R (Table 3). The most striking of these included the 11 genes present in the approximately 17-kbp sequence adjacent to the F strain vlhA5 locus (Fig. 1). This “17-kbp locus” includes genes involved in acquisition, transport, and metabolism of maltose/maltodextrin and other sugars (Table 3). The ORFs at the 17-kbp locus shared homology with syntenically conserved or semiconserved loci in other mycoplasmas, in particular, members of the hominis group (72). This includes similarity to Mycoplasma fermentans, in which the entire locus was conserved in content and was highly similar at the amino acid level (27% to 71%), and to M. synoviae, another poultry pathogen which contains three genes of the locus which also are highly similar to those in the F strain (59% to 62% amino acid identity). Notably, homologs of genes in this locus were not obvious in other species of the pneumoniae group to which M. gallisepticum belongs. Thus, the 17-kbp locus may have been the product HGT from a species from the hominis group, conceivably a common ancestor of M. fermentans and M. synoviae. In addition, two genes bounding the 17-kbp locus in the F strain (MGF_3470 and MGF_3567) are present as remnants in the R strain (MGA_1349 and MGA_1265, respectively), indicating that the locus was lost from the R strain subsequent to its introduction to M. gallisepticum.

Other genes in the F strain represent genes intact relative to fragments present in strain R (Table 3). MGF_5077 encodes a conserved hypothetical protein fragmented in the R strain (MGA_0485 and MGA_0487) and is located at the 5′ end of the F_oF₁ ATP synthase operon, where it conceivably plays a part in energy production. The MGF_4515 unique hypothetical protein contains an additional C-terminal 35 amino acids relative to the R strain MGA_1014, which contains a frameshift. Similarly, the MGF_0872 conserved hypothetical lipoprotein gene contains an additional C-terminal 141 amino acids relative to MGA_0829, potentially affecting a surface-exposed domain and the ability to interact with the host. How these and other intact genes affect the phenotype of strain F relative to strain R is unknown, but they conceivably mediate virulence or host range functions.

Notably, sequence analyses using R_high and F strains enabled identification of differences in the genome sequence of R_low that were likely specific to the particular clone selected for sequencing (R_low clone 2) (52). These included the 37-bp frameshifting insertion noted in pvpA (52) and mutations in the MGA_0223/4 ABC-transporter permease, MGA_0250 unique hypothetical protein, and MGA_1117/9 cytadherence-related molecule B (crmB)-like protein genes that disrupted coding potential in R_low clone 2. While the effect of these changes on the phenotype of R_low clone 2 relative to the R_low wild-type population is unknown, the comparative genomics approach proved a powerful means to discern them.

CGH of vaccine strains.

Comparative genomic hybridizations of the three commercially available vaccine strains (F, ts-11, and 6/85) were performed, and the fold difference relative to results obtained with R_low were determined. In an attempt to identify gene divergence/loss associated with M. gallisepticum attenuation, focus was given to features absent in all vaccine strains relative to R_low. Only seven non-vlhA and non-CRISPR region probes hybridized 4-fold or less in all the vaccine strains compared to R_low, and these genomic lesions were further probed with PCR and sequencing (Table 4).

Of the seven gene features absent in all vaccine strains, five are located in likely gene fragments and verified subtilisin and transposase gene loci affected by indels in F strain genome analysis (Table 4). Although divergence or loss of sequence within pseudogenes might be expected, these data verified their absence in vaccine strains. Sequences for a GTPase similar to the tRNA modification protein MnmE (MGA_0604) and a putative Holliday junction resolvase (MGA_0836) gene were confirmed to be present in vaccine strains by sequencing, with SNPs within sequence spanning the 50-mer probe responsible for the lack of hybridization signal. Whether divergence in these enzymes might contribute to an attenuated phenotype remains to be proven. In addition to genes that are missing in all three vaccine strains, multiple features were observed to be divergent or absent in only one or two strains (see Table S3 in the supplemental material). Strain-specific phenotypes may be associated with these mutations, supporting the conclusion that M. gallisepticum virulence is complex and multigenic.

Conclusions.

In this study, we used comparative genomic analyses of virulent and attenuated M. gallisepticum strains to identify determinants involved in pathogenesis and survival in the host. Genomes of the attenuated high-passage derivative, R_high, and heterologous vaccine strain F were sequenced and compared to the known genome sequence of the virulent, low-passage strain R_low, revealing mutations in numerous genes and indicating a range of protein functions potentially involved in virulence. While these included suspected or known cytadherence-related functions, which are of primary importance for M. gallisepticum virulence, other novel virulence determinants were indicated. Relative to other genomic changes, those associated with the vlhA major variable lipoprotein genes were highly represented—as promoter region variability in strain R_high and as a highly divergent gene complement in strain F—supporting the notion that vlhA gene expression and VlhA phenotypic diversity are important for persistence of M. gallisepticum in the host. Notably, changes in sugar metabolism and solute transport functions were apparent in both attenuated strains, indicating that metabolic substrate utilization may be a significant mechanism by which strains exhibit phenotypic differences in the host. While genes involving metabolism, proteolysis, and restriction-modification were predicted to be compromised in attenuated strains, other potential virulence determinants included genes with no or nonspecific functional prediction and thus of interest for further characterization. We proceeded to characterize one such gene by demonstrating reduced tracheal lesions in chickens infected with an isogenic mutant of the MGA_1107 gene. CGH analysis identified few common genes missing or divergent in vaccine strains F, ts-11, and 6/85, indicating that no single gene was likely responsible for their attenuation. This supports the notion that M. gallisepticum pathogenesis is complex, multifaceted, and multigenic, consistent with these and previous results indicating that independent genes may be essential for virulence. The comparative genomic analysis presented here point to additional factors potentially critical for colonization and virulence in the host, and it provides the framework essential for the rational design of future vaccines.