Characterization of the omlA gene from different serotypes of Actinobacillus pleuropneumoniae: A new insight into an old approach

The OmlA protein is a virulence factor of Actinobacillus pleuropneumoniae, an important pathogen in pigs. The polymorphisms present in the omlA gene sequence of 15 reference serotypes of A. pleuropneumoniae and non-serotypable isolates were assessed to determine the possible evolutionary relationship among them and to validate the importance of this gene as a molecular marker for the characterization of this bacterium. Divergence among the 15 serotypes of A. pleuropneumoniae probably resulted initially from two major evolutionary events that led to subsequent differentiation into nine groups. This differentiation makes it possible to characterize most of the serotypes by using bionformatics, thereby avoiding problems with immunological cross-reactivity. A conserved α-helix common to all the serotypes was most likely involved in connecting the protein to the outer membrane and acting as a signal peptide. A previously unknown gene duplication was also identified and could contribute to the genetic variability that makes it difficult to serotype some isolates. Our data support the importance of the omlA gene in the biology of A. pleuropneumoniae and provide a new area of research into the OmlA protein.


Introduction
Swine pleuropneumonia (SPP) is a significant respiratory disease and has been reported in all countries where pig farming is intensively practiced. The etiological agent of SPP is the Gram-negative coccobacillus Actinobacillus pleuropneumoniae, currently divided in 15 serotypes that are defined based on the antigenic properties of capsule polysaccharides. Although all of the serotypes are capable of causing SPP, differences in virulence make the serotyping of field isolates of A. pleuropneumoniae a key factor in the epidemiological study and control of this disease (Schuchert et al., 2004).
Numerous assays have been developed for the serological characterization of A. pleuropneumoniae isolates. Although immunological assays are relatively fast, their main limitation is that they commonly show crossreactivity (Jessing et al., 2003). Various molecular techniques have also been used to study the molecular epidemiology of bacterial pathogens and are extremely important in monitoring the characteristics of a given population (Ashis et al., 2012). Current approaches seek to develop molecular markers that can complement the sometimes inconclusive information obtained using serological techniques. Molecular phylogenies based on gene polymorphisms have been used to characterize and distinguish serotypes or isolates of microorganisms (Nightingale et al., 2005;Gonzalez-Escalona et al., 2008).
The objective of this study was to analyze polymorphisms of the omlA gene, which codes for an outer membrane protein, in isolates from different serotypes of A. pleuropneumoniae. This is the first in-depth study of the polymorphisms and phylogeny of the omlA gene in A. pleuropneumoniae and provides new insights on the structure and organization of this gene. This work provides additional molecular tools for genotyping A. pleuropneumoniae.
Genomic DNA from A. pleuropneumoniae strains was obtained using the Wizard Genome DNA purification 244 Rossi et al. (5'-ATTAAAAAGTAAAAAAGCTATCCC-3') (Gram and Ahrens, 1998), was used to amplify the omlA gene (the amplicon had an expected size of approximately 1270 bp). The PCR was done using 1.25 U of GoTaq DNA polymerase (Promega) in a final volume of 50 mL of enzyme buffer containing 1.5 mM MgCl 2-, 0.2 mM of each dNTP, 0.2 mM of each oligodeoxynucleotide and 50 ng of DNA in a C1000 TM thermal cycler (BioRad, Richmond, CA, USA). The DNA was initially denatured at 94°C for 3 min, followed by 35 cycles of denaturation at 94°C for 1 min, annealing at 50°C for 45 s, and an extension step at 72°C for 1.5 min, followed by a final extension step at 72°C for 10 min. The reaction products were analyzed by electrophoresis in 1.0% agarose gels, purified using a Wizard SV gel and PCR clean-up system (Promega) and sequenced using the Sanger sequencing method.

Nucleotide sequences
The omlA gene nucleotide sequences used in this study were from A. pleuropneumoniae isolates from different serotypes and origins. The NCBI GenBank database accession numbers and the serotypes of their respective isolates are listed in Table 1. The nucleotide sequences of isolates from serotypes 13, 14 and 15 that were previously unavailable in the databases were obtained in the present study and deposited under accession numbers JF311904, JF304624 and JF304622, respectively. In addition, omlA genes from isolates with serotypes that could not be defined by immunological methods because of cross-reactivity were sequenced (accession numbers JF304619, JF304621 and JF304623).

Structural analysis of the omlA gene
The +1 point of translation and the termination codon of the omlA gene were predicted using the analysis tool ORF finder (Rombel et al., 2002). Sequences corresponding to the promoter region (-10 and -35) of the omlA gene were predicted using the Bacterial Promoter Prediction Program BPROM, which was also used to predict possible cis elements for the recognition of transcription factors.

Organization of the omlA gene in A. pleuropneumoniae isolates
The copy number and organization of the omlA gene in the genomes of the different isolates was studied using Southern blotting (Sambrook et al., 1989). We selected 11 A. pleuropneumoniae clinical isolates (Table 1) that were obtained between 2003 and 2010 from six farms located in southeastern Brazil and included four non-serotypable isolates.
The primers omlAhF (5'-CGGTTTAGTCGCAGGTTTAGT-3') and omlAhR (5'-TCCTTAACCCCTAATTCCTTAAGA-3') were used to synthesize a 372-bp probe for hybridization ( Figure 1D). The probe was labeled using a PCR DIG Probe Synthesis kit (Roche, Mannheim, Germany), according to the manufacturer's instructions. Five micrograms of total DNA from the isolates was digested for 16 h with the restriction enzymes XbaI and BglII to generate an expected fragment of 943 bp. The fragments were separated by electrophoresis in a 0.8% agarose gel and transferred to a nylon membrane (Amersham Hybond N + ; GE Healthcare, Chalfont St. Giles, UK). The hybridization was done under high stringency conditions using the DIG High Prime DNA Labeling and Detection Starter kit II TM (Roche) to ensure the high specificity of hybridization. The results were visualized and documented using L-Pix Chemi photodocumentation system (Loccus, São Paulo, SP, Brazil).

Structural analysis of the OmlA protein
The amino acid sequences coded by the omlA genes were predicted using the bacterial genetic code in the program Mega 5.03 (Tamura et al., 2007). The putative secondary structure was predicted using the Phyre algorithm (Kelley and Sternberg, 2009) and the Jnet algorithm of Jpred3 (Cole et al., 2008). The resulting structures were compared and only motifs with probability scores > 80% were used to construct the consensus structure model of the OmlA protein. The membrane protein topology prediction method TMHMM, based on the Markov model, was used to predict transmembrane helixes (Krogh et al., 2001). Conserved domains were located using the PROSITE database (Sigrist et al., 2010) and the Conserved Domain Database (NCBI).

Phylogenetic molecular analysis
The 26 nucleotide sequences were initially aligned using Clustal W (Larkin et al., 2007). Phylogenetic trees were built using the maximum parsimony (MP), maximum likelihood (ML) and neighbor-joining (NJ) methods. The neighbor-joining tree was inferred using the program MEGA 5.03 (Tamura et al., 2007) and the remaining two were inferred using the program PAUP version 4.0b10 (Swofford, 2003). The ModelTest program version 3.7 (Posada and Crandall, 1998) was used to test 56 models and establish a DNA evolution model that would best fit the data for ML analysis. Subsequently, a heuristic search was done that was initiated using a NJ tree and the treebisection-reconnection (TBR) algorithm. The robustness of each internal knot of the trees was assessed statistically by using a combination of bootstrapping (1000 replications for the NJ and MP methods and 100 replications for the ML method) and the PAUP program. Because bootstrap values tend to be conservative, a Bayesian analysis was done using Mr. Bayes 3.0 (Huelsenbeck and Ronquist, 2001). The best model for the analysis was inferred using the program Mr. Modeltest 2.3 (Nylander, 2004).

Sequence analyses
The sequences used in this study contained 1092-1125 base pairs and 506 variable sites (~43%) in the aligned positions. There were 450 parsimony-informative sites (39%) and the high variability in the nucleotide sequences resulted in 216 base substitutions, 77 (36%) in the first base of the codon, 84 (39%) in the second and 55 (25%) in the third (data not shown). Since most of the substitutions occurred in the first and second bases, there was a large number of variable sites in the amino acidic sequences deduced by Mega 5.03. Among the 383 aligned amino acids, 260 sites were variable (~68% of the total). The extensive number of alterations in the primary sequence of the OmlA protein resulted from transitions and transversions. Typically, transition rates are approximately two times higher than transversion rates (Zhang and Gerstein, 2003) since the latter are usually rapidly detected by DNA repair mechanisms. However, for the omlA gene, the transition/transversion ratio was~0.99 (data not shown). Thus, the high transversion rate reflected the high genetic variability acquired during the evolution of different A. pleuropneumoniae serotypes.

Structural analysis of the omlA gene
The open reading frame (ORF) of omlA was flanked by the codons ATG and TAA as the initiation and termination codons, respectively, and the alignment of these sequences showed that the initial region of the gene was very conserved, as also pointed by Gram and Ahrens (1998). Between the +1 point of translation and point +160, 87.5% of the nucleotides were identical in all the serotypes and differences were observed in only a few isolates. The BPROM program identified the probable positions of the -10 and -35 regions of the promoters in the omlA genes ( Figure 1A). The sequence TATTTATT was prevalent in the promoter region of the gene and the BPROM program suggested that this might be a binding site for the regulatory protein Lrp, a global transcription regulator and member of a widely distributed family among Bacteria and Archaea (Brinkman et al., 2003). The putative Shine-Dalgarno region (ribosome binding site) was inferred based on the composition and position of the consensus sequence. Figure 1B shows the schematic organization of the promoter region of the omlA gene based on analyses of all available sequences.

Structural analysis of the OmlA protein
The deduced sequence of the OmlA protein was 362-375 amino acid residues long and 80% of the first 58 residues were identical in all of the sequences analyzed. The primary sequence of the A. pleuropneumoniae OmlA protein was similar to OmlA in Actinobacillus ureae and Haemophilus parasuis, both belonging to the family Pasteurellaceae. Alignment of the primary sequences of this protein from these three species ( Figure 1C) revealed conservation of the OmlA structure, including the N-terminal region, which reinforces the importance of this portion of the molecule. The secondary structure model of OmlA based on the profiles generated by Jpred3 and Phyre revealed that a large part of the protein is organized in b sheets ( Figure 1E), the positions of which vary only slightly according to the serotype; there was only one a-helix and this was located between the 5 th and 20 th amino acids and was conserved in all serotypes. In most serotypes, this region contained the amino acid sequence KLIAGLVAGLVLTAC, with variations only in serotypes 1, 8, 9 and 11. In these cases, the third (isoleucine), fifth (glycine), ninth (glycine) and tenth (leucine) amino acids were replaced by methionine, serine (two) and valine, respectively, to yield the sequence KLMASLVASVVLTAC. This sequence variation probably does not significantly affect the formation of the a-helix since the substituted amino acids are from the same charge groups as the original residues. The TMHMM model showed that the a-helix did not have a transmembrane insertion, which suggested involvement in another function. In the PROSITE database, the first 20 amino acids of the sequence matched a lipid-binding site (profile PS51257) and a signal peptide. Hence, the a-helix of the OmlA protein may serve to anchor the protein to the external membrane, in addition to its function in directing the transport of the protein to its extracellular location. Analysis using the Conserved Domain Database indicated that the OmlA protein belonged to the lipoprotein 5 superfamily, which contains proteins that bind to transferrin and is distinctly related to other protein families that bind to solutes.
The bootstrap values were equally high for all of the methods used to construct the phylogenetic trees, as were the posteriori probability values obtained in Bayesian analysis (close to or equal to 100%).
Whereas the initial regions of the genes were widely conserved, there was marked genetic variation in the internal and terminal regions among the serotypes. However, these differences in sequence homology largely disappeared when some of the groups in the phylogenetic trees were observed and analysed separated from the others. In group 1, the number of variable amino acid sites and variable nucleotide sites was 4 of 440 (0.9%) and 5 of 1320 (0.4%), respectively, which prevented their partition even if the group was analyzed separately from the others, given their almost identical omlA sequences. For group 2, the corresponding values for variable sites were 1.5% and 0.9%, and for groups 3 and 7, these values were 2.5% and 3.6%, respectively. Serotypes 13, 14 and 15 were distantly related and the clinical isolates grouped with serotype 15. The radial topology ( Figure 2B) of the tree shown in Figure 2A strongly suggested that the differentiation of the 15 A. pleuropneumoniae serotypes may have involved two distinct evolutionary events (highlighted by arrows in Figure 2B). This radial analysis separated the isolates into two major clusters: the first comprising serotypes 1,2,8,9,11,12 and 14 and the second, serotypes 3, 4, 5, 6, 7, 10, 13 and 15. These cluster profiles were very similar to those obtained when the reconstruction was done using the predicted amino acid sequences (data not shown), the only difference being that in the latter case, the serotype 4 organisms had distanced themselves from serotypes 3 and 6. This divergence most likely resulted from the duplication of a sequence present between positions +144 and +176 (GenBank accession number AB007575) and added 11 amino acids to the OmlA protein.

Organization of the omlA gene in clinical isolates of A. pleuropneumoniae
Since the nucleotide sequence of the omlA gene varies significantly among serotypes, the primer pair omlAhF/omlAhR was synthesized based on the conserved region located in the initial extremity of the omlA gene; the resulting probe contained 372 bp. The organization of the omlA gene was investigated in clinical isolates of A. pleuropneumoniae obtained from seven farms in southeastern Brazil. Southern blotting revealed that the organization of the omlA gene was conserved in most of the isolates from the farms, with an expected 943-bp fragment after the cleavage of genomic DNA (Figure 3). Two of the isolates diverged from the expected results. Isolate MV452 contained two copies of the gene and isolate MV 653 had a different organization from the other isolates because of the presence of a similar fragment at a different position. This isolate apparently contained only one copy of the gene, but 248 Rossi et al. since the fragment size was > 10 kb the genome of this isolate most likely did not have the same recognition sequences for cleavage by XbaI and BglII present in the other isolates.

Discussion
The OmlA protein belongs to a family of small, poorly characterized bacterial lipoproteins widely distributed in b and g proteobacteria (Vanini et al., 2008). OmlA protein is a virulence factor of A. pleuropneumoniae and has an important role in binding to transferrin to facilitate the acquisition of iron from the host (Baltes et al., 2002).
The genetic diversity of the omlA gene was first observed when a 970-bp amplicon was digested with restriction enzymes and the resulting fragments then used to classify 12 serotypes into five groups (Osaki et al., 1997). Subsequent analyses took advantage of the variability of the internal region of the omlA gene and used different techniques to try to distinguish the then known 12 serotypes, but could still only separate them into four or five groups (Gram and Ahrens, 1998;Gram et al., 2000a). In the present study, part of the work by Gram and Ahrens (1998) was reassessed and the polymorphisms of the omlA gene in the 15 currently known serotypes were identified and used to build a phylogenetic tree based on more recent methods and their respective best fit models. This approach allowed us to separate the A. pleuropneumoniae serotypes into nine (when using the nucleotide sequences) or ten (when using the amino acid sequences) groups and also to infer possible evolutionary relationships between them. Additionally, the omlA genes from clinical isolates of A. pleuropneumoniae were sequenced and analyzed.
The existence of various serotypes and the range of antigenic differences among them has made effective vaccination against A. pleuropneumoniae difficult since only serotype-specific immunity is generally observed (Nielsen, 1984). This situation reinforces the importance of the pre-cise characterization of isolates present in a given area. In addition, many A. pleuropneumoniae isolates are nontypable by currently used techniques but are nevertheless capable of causing disease (Fenwick, 2002). As shown here, the polymorphisms present in the nucleotide sequence of the omlA gene can be used to characterize isolates considered nontypable by conventional methods. In addition, the phylogenetic reconstruction described here for the omlA gene reinforced the characteristics noted elsewhere and confirmed the usefulness of this gene in distinguishing among A. pleuropneumoniae serotypes. For example, the clustering of serotypes 1, 9 and 11 agreed with the crossreactivity observed amongst these serotypes (Paradis et al., 1999). A similar conclusion is applicable to serotypes 3, 6 and 8 which also show similar tube agglutination, coagglutination and indirect hemagglutination (Mittal et al., 1988). In this case, however, the molecular phylogeny separated serotype 8 from the other two. Cross-reactivity between isolates of serotypes 4 and 7 has also been observed (Mittal and Bourdon, 1991) and suggests evolutionary proximity between them. This conclusion was validated by their close clustering in the phylogenetic tree, although they were placed in monophyletic branches.
Together, these results indicate that the polymorphisms present in the nucleotide sequence of the omlA gene and in the amino acid sequence of the OmlA protein can be used as markers to distinguish among the serotypes of many isolates. This is a useful approach for understanding the characteristics and origin of isolates in a delimited region.
All of the clinical isolates examined here were grouped with serotype 15, even those designated by other molecular techniques as serotype 8, i.e., some serotype 8 isolates of A. pleuropneumoniae in Brazil can show variation in the omlA gene that is actually closer to serotype 15. This finding agrees with Gram et al. (2000b) who identified some serotype 8 isolates in which the omlA gene was similar to serotypes 3, 4, 6 and 7. We also believe that although the isolate MV235 was not serotypable by commonly used methods, it was almost certainly a variation of serotype 8 because of its cluster position in the phylogenetic tree.
Although the expression of the omlA gene is constitutive in other organisms (Ochsner et al., 1999), there is not much information on the expression of this gene in A. pleuropneumoniae. The structural model of the promoter region of the omlA gene suggests that there are possible binding sites for a transcription factor, the leucine-responsive regulatory protein (Lrp), that may control gene expression, particularly under stress (Wagner and Mulks, 2007), as has been observed in iron-restricted conditions (Deslandes et al., 2007). Additionally, the predicted secondary structure of the OmlA protein indicates the existence of a conserved region that may be involved in the binding of this protein to the lipid region of the outer membrane of A. pleuropneumoniae and have a role as a signal Phylogeny of the omlA gene 249 peptide. The tertiary structure of this protein cannot be predicted because of a lack of homologous proteins in the databases.
In contrast to the conserved organization of the omlA gene previously reported for other genomes (Gerlach et al., 1993), we have shown that there are important variations in the organization of this gene in different A. pleuropneumoniae isolates obtained in Brazil, including a surprising duplication. Such variability has not been observed before and the duplication event may confer some advantage to this microorganism, e.g., in adapting to new environments or even making it more virulent.
Although some aspects of the structural characterization of the omlA gene and the corresponding protein described here require additional experiments to confirm their functional relevance, we nevertheless believe that further detailed analysis of the genetic variability of this gene can yield important information on its role in the 15 serotypes of A. pleuropneumoniae identified in this work. The resulting information will improve our understanding of infection by A. pleuropneumoniae.