Genome Biology of Actinobacillus pleuropneumoniae JL03, an Isolate of Serotype 3 Prevalent in China

Actinobacillus pleuropneumoniae is the etiologic agent of porcine contagious pleuropneumonia, a cause of considerable world wide economic losses in the swine industry. We sequenced the complete genome of A. pleuropneumoniae, JL03, an isolate of serotype 3 prevalent in China. Its genome is a single chromosome of 2,242,062 base pairs containing 2,097 predicted protein-coding sequences, six ribosomal rRNA operons, and 63 tRNA genes. Preliminary analysis of the genomic sequence and the functions of the encoded proteins not only confirmed the present physiological and pathological knowledge but also offered new insights into the metabolic and virulence characteristics of this important pathogen. We identified a full spectrum of genes related to its characteristic chemoheterotrophic catabolism of fermentation and respiration with an incomplete TCA system for anabolism. In addition to confirming the lack of ApxI toxin, identification of a nonsense mutation in apxIVA and a 5′-proximal truncation of the flp operon deleting both its promoter and the flp1flp2tadV genes have provided convincing scenarios for the low virulence property of JL03. Comparative genomic analysis using the available sequences of other serotypes, probable strain (serotype)-specific genomic islands related to capsular polysaccharides and lipopolysaccharide O-antigen biosyntheses were identified in JL03, which provides a foundation for future research into the mechanisms of serotypic diversity of A. pleuropneumoniae.


INTRODUCTION
Actinobacillus pleuropneumoniae is a Gram-negative, facultatively anaerobic, non-motile, rod-shaped bacillus in the family of Pasteurellaceae, which is chemoheterotrophic possessing both metabolic patterns of fermentation and respiration. Members of the Pasteurellaceae are obligate parasites, primarily of mammals and birds, while A. pleuropneumoniae is the etiologic agent of porcine contagious pleuropneumonia, an infectious respiratory disease of swine, which causes important world wide economic losses in the pig industry. The pathogen invades the porcine tonsil and upper respiratory tract, and can be isolated from nasal cavities, tonsils, the middle ear cavity and the lungs of infected animals [1,2]. Depending on the number of bacteria reaching the lung, the particular serotype of the infection and the immunological status of the host, the course of the disease can be divided into peracute, acute and chronic forms [1]. Peracute and acute cases usually show high mortality with pulmonary lesions characterized by severe oedema, inflammation, haemorrhage and necrosis, whereas the chronic form of disease is characterized by haemorrhagic, fibrinous and necrotic pleuritis, pericarditis and pneumonia [1].
The virulence of A. pleuropneumoniae is known to be associated with several factors, such as exotoxins, capsular polysaccharide (CPS), lipopolysaccharide (LPS), outer membrane proteins (OMPs), and iron uptake proteins [3]. In addition, some enzymes involved in anaerobic respiration also appear to play an important role in the virulence of A. pleuropneumoniae [4].
A. pleuropneumoniae has been classified into two nutritional biotypes: the biovar 1 is b-NAD-dependent while the less common biovar 2 is b-NAD-independent [3]. On the basis of their capsular and lipopolysaccharide antigens, 15 serotypes of A. pleuropneumoniae have been recognized, with variations in their virulence and regional distributions [5]. Serotypes 1, 5, and 7 are most commonly found in North America, whereas serotype 2 predominates in many European countries [3]. In China, the prevalent serotypes are 1, 3, 4, 5 and 7 [6].
To date, 8 complete genomic sequences are available within the family of Pasteurellaceae. Seven of them, i.e., A. pleuropneumoniae L20 (accession no. CP000569), Pasteurella multocida Pm70 (AE004439) [7], Haemophilus influenzae Rd KW20 (L42023) [8] and 86-028NP (CP000057) [9], H. ducreyi 35000HP (AE017143), Mannheimia succiniciproducens MBEL55E (AE016827) [10] and H. somnus 129PT (CP000436) [11], are in the GenBank. While the complete sequence of A. actinomycetemcomitans is available from the web site of University of Oklahoma's Advanced Center for Genome Technol-ogy (http://www.genome.ou.edu). Among them, A. pleuropneumoniae L20 (serotype 5b) genomic sequence is the only one available in the Actinobacillus genus. In this study, we sequenced and analyzed the genome of A. pleuropneumoniae strain JL03, a Chinese field isolate of serotype 3. Together with the genomic sequence of L20, this information provides a firm foundation for future research into the genetic basis of metabolism, pathogenesis, virulence and serotype/ biotype determination in A. pleuropneumoniae.

RESULTS AND DISCUSSION
General features of the genome The genome of A. pleuropneumoniae strain JL03 is composed of 2,242,062 base pairs (bps) with a single circular chromosome ( Figure 1A). Referring to genomic coordinates of strain L20, the dnaA gene, designated APJL0001, was selected as the first gene of the JL03 genome. The putative replication origin (oriC) of JL03 chromosome was identified between two genes, gidA (APJL1688) and cof (APJL1689), based on GC skew and the presence of DnaA protein recognition sequences (DnaA-boxes) [12,13] with typical gamma proteobacterium oriC features as what found in other genera of the Pasteurellaceae family ( Figure 1B and Table 1).
The JL03 genome is approximately 1.4% smaller than that of strain L20 (2,274,482 bps). The genomic comparison of the linear organization at the nucleotide level between strains JL03 and L20 is presented in Figure 2A. Notably, strain L20 possesses a strainspecific genomic island of 37.7 kb encoding a number of phagerelated proteins, which is absent in strain JL03.
There were eleven repetitive elements in the JL03 genome (designated JLRP1 to 11, hereafter) divided into several categories according to their coding sequences, i.e., transposase, adhesin, elongation factor Tu and unknown proteins ( Table 2). Among them, JLRP2, with its characteristic 25 bp inverted repeats in both ends, was presumed to be a novel insertion sequence element (IS) of the IS3 family. Submitted to the IS database (http://www-is. biotoul.fr), this sequence was designated ISAp12. In addition, a noncoding 2071 bp clustered regularly interspaced short palindromic repeats region (CRISPR) was identified in the vicinity of the cas1 gene (APJL0215) that has been found adjacent to CRISPR loci in different bacteria [14]. This CRISPR is composed of an array of 28 bp direct repeats (DR) individually separated by 34 unique spacers of 32 bp or 33 bp. On the other hand, nine spacers in JL03's CRISPR all bear high sequence similarities with the corresponding sequences of plasmids from related bacteria (A. actinomycetemcomitans, H. influenzae and H. ducreyi). The inheritable feature of CRISPR spacers has been interpreted as evolutionary remnants derived from other extrachromosomal elements [15], and the CRISPR loci were successfully applied to studies in evolution, typing, and comparative genomics [16].
Annotation of the JL03 genome is summarized in Table 3 and compared to those of strains L20 (A. pleuropneumoniae), Pm70 (P. multocida) and 35000HP (H. ducreyi). The entire JL03 genome has six ribosomal operons (16S-23S-5S rRNA) and an additional 5S rRNA. Sixty-three tRNA genes corresponding to the 20 common amino acids were identified in the JL03 genome. Four copies of tRNA-Ile and tRNA-Ala genes were located in the spacer regions between the 16S and 23S rRNA genes. A distinct selenocysteine tRNA gene containing the UCA anticodon was also identified. This tRNA gene is located adjacent to two genes (APJL1590, 1589) encoding L-seryl-tRNA selenium transferase (SelA) and selenocysteine-specific elongation factor (SelB), respectively. This kind of organization is the same as that found in H. influenzae strain 86-028NP [9].
The JL03 genome contained 2,097 potential CDSs with an average size of 941 bps, which in sum account for 88.1% of the whole chromosome. A graphical representation of CDSs by category and genetic characteristics of the JL03 genome are shown in Figure 1A. As shown in other completed microbial genomes, 18.3% of the CDSs were found to be similar to hypothetical proteins of unknown functions. The ortholog relationship between A. pleuropneumoniae and other species within the family Pasteurellaceae was consistent with their phylogenetic relationship based on the Circles are numbered from 1 (outer circle) to 10 (inner circle). The circles 1/2 shows predicted CDSs on the plus and minus strand in JL03 color-coded by COG categories. All genes are colored according to biological functions: gold for translation, ribosomal structure and biogenesis; orange for RNA processing and modification; light orange for transcription; dark orange for DNA replication, recombination and repair; antique white for cell division and chromosome partitioning; pink for defense mechanisms; tomato for signal transduction mechanisms; peach for cell envelope biogenesis and outer membrane; deep pink for intracellular trafficking, secretion and vesicular transport; pale green for posttranslational modification, protein turnover and chaperones; royal blue energy production and conversion; blue for carbohydrate transport and metabolism; dodger blue for amino acid transport and metabolism; sky blue for nucleotide transport and metabolism; light blue for coenzyme metabolism; cyan for lipid metabolism; medium purple for inorganic ion transport and metabolism; aquamarine for secondary metabolites biosynthesis, transport and catabolism; gray for function unknown. Circle 3/4, the putative horizontal transferred genes in deep pink identified by SIGI-HMM on the forward and reverse strand. Circle 5, repetitive elements in yellow, above 200nt and cutoff value 1e-10. Circle 6, transposases in green and potential prophage genes in dark orange. Circle 7, mean centered GC content of JL03 genes (red: above mean, blue-below mean). Circle 8, tRNA genes in orange. Circle 9, rRNA genes in red. Circle
Besides fermentation, A. pleuropneumoniae performs both aerobic and anaerobic respirations and the latter is an important factor for pathogenesis (see Table S1). The electron transport chains in A. pleuropneumoniae might be branched and modular depending on its growth conditions ( Figure 3). Cytochrome D ubiquinol oxidase encoded by cydAB (APJL0308, 0309) should be responsible for reducing the terminal electron acceptor oxygen to water in aerobic environments [21]. While, genes coding for various kinds of reductases specific for terminal electron acceptors of anaerobic respiration were also identified ( Figure 3). Besides the arsenate reductase encoded by arsC (APJL1105), the napFDAGHBC (APJL1463-1457) operon encodes a periplasmic nitrate reductase system (NAP) highly homologous to that in H. ducreyi [22], which, as the sole nitrate reductase in A. pleuropneumoniae, should be essential to support anaerobic growth in the presence of nitrate [23]. Furthermore, albeit less favorable than nitrate, identification of frdABCD (APJL1556-1553) encoding a fumarate reductase and dmsABC (APJL1705-1707) encoding an anaerobic DMSO reductase in.JL03 inferred that this strain may be able to utilizing fumarate or dimethyl sulfoxide (DMSO) as electron acceptors as well ( Figure 3) [4,24].
Three global transcription regulators Hlyx (APJL0646), ArcA (APJL0049) and NarP (APJL0059) are encoded in all known genomes of Pasteurellaceae, including A. pleuropneumoniae. Under anaerobic conditions, these transcription factors may activate genes for anaerobic respiration while repress genes for aerobic respiration and fermentation [25].
Complete sets of genes coding for enzymes of glycolysis and gluconeogenesis, as well as non-oxidative pentose phosphate pathways were confirmed in strain JL03 ( Figure 3). However, the tricarboxylic acid (TCA) cycle pathway in A. pleuropneumoniae was incomplete. Genes encoding three key enzymes of TCA cycle, i.e., citrate synthase, aconitase and isocitrate dehydrogenase were not found in the genome. This pattern of metabolism was the same as species of genus of Haemophilus, e.g., H. influenzae, H. ducreyi, and H. somnus [11]. In addition, genes encoding malate synthase and isocitrate lyase, essential for glyoxylate pathway were also missing in JL03. Nevertheless, in JL03, the provision of C4 metabolites is unaffected and C5 metabolic intermediates should be offered by the non-oxidative synthesis process of the pentose phosphate pathway ( Figure 3). In contrast to the bacteria in the genera Actinobacillus and Haemophilus, P. multocida and M. succiniciproducens, species of the genera Pasteurella and Mannheimia respectively, may perform their catabolism via an intact TCA cycle pathway [7].
The JL03 genome encodes almost all the enzymes involved in fatty acid metabolism, biosynthesis of glycerophospholipid, terpenoid, amino acid and purine/pyrimidine nucleotides ( Figure 3). Interestingly, a specific operon cysGHDNJI (APJL1886-1881) encoding several proteins involved in assimilatory sulfate reduction was identified in JL03, and this operon is incomplete in the genome of A. pleuropneumoniae strain L20 ( Figure 2B), and can not be found in the genomes of other members of the family Pasteurellaceae except for M. succiniciproducens MBEL55E. The assimilatory sulfate reduction has been extensively studied in Escherichia coli as a model for Gramnegative bacteria [26,27], and the related genes, organized in a single operon cysGHDNJI in JL03, are dispersed into three operons in the genome of E. coli K12 MG1655 (U00096) [26]. The biological significance of the different genotypes and genomic organization deserves to further study.
JL03 needs nicotinamide adenine dinucleotide (NAD) for in vitro growth, but NAD is not required by H. ducreyi or H. somnus [11]. Concordantly, H. ducreyi 35000HP genome contains a duplication of the intact gene nadV (HD1447, 1455, 495aa) while H. somnus 129PT has one (HS0002, 465aa), which encodes the nicotinamide phosphoribosyltransferase (NAmPRTase) [11]. However, JL03 only bears a mutated nadV' CDS (APJL0638, 203aa) encoding merely a truncated domain. This is, for the first time, that genetic evidence was presented to support the previous notion that A. pleuropneumoniae serotype 3 belongs to the NAD-dependent biotype I category [28]. In addition, pathways involved in ubiquinone biosynthesis as well as riboflavin and vitamin B6 metabolism were also complete in JL03, as they are in H. ducreyi and P. multocida.

Analysis of pathogenesis/virulence factors
Pathogenesis/virulence of A. pleuropneumoniae has been known to be related to many specific factors in addition to its metabolic features well adapted to in vivo growth and in vitro survive described above. The genomic characteristics of the specific pathogenesis/virulence factors are described in detail below:

APX exotoxins
Although the virulence of A. pleuropneumoniae is multifactorial, the major factor primarily responsible for the characteristic hemorrhagic lesions of the porcine contagious pleuropneumoniae are the pore-forming exotoxins belonging to the repeat in toxin (RTX) family [29,30]. Widely distributed among Gram-negative bacteria, RTX toxins share structural and functional properties, including a characteristic nonapepetide glycine-rich repeat motif, a particular mode of secretion with a signal sequence at the C-terminus, post-translational activation, and cell toxicity via pore-forming mechanism [31]. RTX toxins in A. pleuropneumoniae are called Apx toxins (for A. pleuropneumoniae RTX toxins): the strongly hemolytic and cytotoxic ApxI, the weakly hemolytic and moderately cytotoxic ApxII, the nonhemolytic but strongly cytotoxic ApxIII, and the weakly Figure 2. The schematic comparison of genetic organizations among three isolates of A. pleuropneumoniae. A co-linearity comparison diagram of the genomic organization at the nucleotide level between A. pleuropneumoniae strain JL03 and strain L20 (A). Color code stands for maximal length of those regions with highly homologous sequences between genomes: red, .10 kb; blue, 5-10 kb; cyan, 1-5 kb. The boxes in green represent phage-associated CDSs of L20. Besides the strain L20-specific prophage region illustrated below the linear genomic diagram as an enlarged drawing, four special genomic regions highlighted (B, C, D, E) were magnified in the corresponding panels. The genetic organizations of the cys operons (B), the CPS biosynthesis and export gene clusters (D), and the LPS O-antigen biosynthesis gene clusters (E) were compared among three isolates of A. pleuropneumoniae: JL03 (serotype 3), L20 (serotype 5b) and 4074 (serotype 1). Comparative genetic organization of the flp operons between JL03 and L20 is illustrated in panel C. Regions presented in gray represent highly homologous sequences. Blue arrows represent putative CDSs with either forward or reverse transcription directions. doi:10.1371/journal.pone.0001450.g002 hemolytic and cytotoxic ApxIV [1,29]. Different serotypes secrete different sets of Apx toxins, causing variations in both of their hemolytic and cytotoxic activities [32]. Apx toxins are encoded by apx operons that usually consist of four contiguous genes arranged in the order of apxCABD. The apxC encodes a product that directs the cytoplasmic conversion by an acylation reaction of the structural toxin encoded by apxA to the active form, exported by a transporter encoded by apxBD [31]. The high degree of conservation of the RTX-B and RTX-D secretion proteins is reflected by the functional exchangeability of these proteins [33].
The absence of apxICABD operon in JL03 genome confirmed that it bore the moderate toxicity property of serotype 3, in contrast to serotypes 1, 5, 9, 10, and 11, all of which secrete ApxI [33]. Two tightly linked gene clusters, apxIICAB' (APJL0968-0966) and apxIIICABD (APJL1347-1344) [30,34], were identified in the JL03 genome. The apxII operon is truncated in JL03, consisting of apxIICA but a partial apxIIB' without apxIID. It was evident that the secretion of ApxII may use the exporter encoded by apxIBD which are present in all serotypes except serotype 3, and that ApxII effect in serotype 3 is mainly cytoplasmic but barely hemolytic [29]. ApxIII has been known to be expressed and secreted by serotypes 2, 3, 4, 6, and 8 [29].
Remarkably, ApxIVA might be impaired in JL03 due to a Trp to nonsense mutation (tgGRtgA) in the coding region of this gene (APJL1015-1016). We further sequenced four independent isolates of serotype 3 and found that besides JL03, both S1421 and HB12 had the same TGA mutation. On the other hand, neither strain GDSB nor strain HV114 is mutated, bearing the prototype Trp codon (tgG). Concerning all of the genetic determinants of Apx encoded by JL03, it is worth mentioning that serotype 3 has very low virulence and secretes little ApxII, but normal amounts of ApxIII [29]. The absence of the most important operon apxI is likely to be an important factor leading to a decreased virulence of JL03.
Adherence As previously reported, a 14-gene flp (fimbrial low-molecular-weight protein) operon (flp1-flp2-tadV-rcpCAB-tadZABCDEFG) has been found in the genera of Haemophilus, Pasteurella, Pseudomonas, Yersinia, Caulobacter and others, which is essential for Flp-pilus production, rough colony morphology, autoaggregation, and biofilm formation [34]. However, although JL03 possesses a series of genes encoding proteins responsible for bacterial adherence to host cells and biofilm formation (Table 5), the flp operon of it is truncated, composed of only 11 genes (APJL0549-0539), where the 59-proximal flp1, flp2 and tadV genes found in strain L20 were absent in JL03 ( Figure 2C). In addition, the JL03 rcpC (APJL0549) is truncated with only a quarter of the C-terminal CDS maintained comparing to that of L20. The remaining JL03 flp operon genes were highly homologous to those from H. ducreyi 35000HP. For instance, the TadZABCDEFG CDSs have 66%, 88%, 71%, 75%, 70%, 54%, 52%, and 49% amino acid identity comparing to those of 35000HP, respectively [35]. However, RT-PCR experiments indicated that the tadZABCDEFG genes were not expressed in JL03 (data not shown) and it is likely due to the truncation of the promoter region of the JL03 flp operon, which was identified in strain L20. Because the Flp1 protein is the major structural component of Flp pili required for adherence-related phenotypes [34] and TadZABCDEFG is also known for tight adhesion [35], a truncated non-expression flp operon of strain JL03 should lead to failure of adherence. Polyglycolic acid (PGA), a linear polymer of N-acetyl-Dglucosamine residues in b-1,6 linkage, has been suggested to play a role in the intercellular adhesion and cellular detachment and dispersal in A. actinomycetemcomitans biofilm [36]. An operon consisted of pgaABCD genes (APJL1968-1971) encoding hexosamine-containing extracellular polysaccharide adhesin biosynthesis enzymes and another gene dspB (APJL1110) encoding N-acetyl-bhexosaminidase were identified in the JL03 genome. The presence of these genes is consistent with the hypothesis that biofilm formation may be relevant to the colonization, pathogenesis and transmission of A. pleuropneumoniae [36].
Capsular polysaccharides (CPS) Bacterial polysaccharides are extremely diverse and occur in multiple forms, with substantial variations within a species. They include CPS, exopolysaccharides (EPS) and O-antigens [37]. CPS is required for virulence of bacteria and variation in CPS content may contribute to the differences in virulence among A. pleuropneumoniae isolates [38]. A specific genomic island-like fragment, approximately 8.6 kb, encoding genes involved in CPS biosynthesis and export was identified in JL03 (Table 6). BLASTn searches revealed that the CPS biosynthetic enzymes encoded by the cps3ABCD (APJL1614-1611) operon were serotype 3-specific in strain JL03 ( Figure 2D). The putative proteins encoded by cps3A and cps3D both have CDP-glycerol glycerophosphotransferase motifs. Cps3B contained a cytidylyltransferase motif. This is a key regulatory enzyme for phosphatidylcholine biosynthesis. The putative Cps3D was 57% similar to the TagF protein of Campylobacter jejuni involved in teichoic acid biosynthesis. The proteins encoded by the genes of the cps operon showed low homology with those encoded by genes of different A. pleuropneumoniae serotypes [39].
Upstream of cps3A, transcribed in the opposite orientation, there is another operon of four genes cpxDCBA (APJL1615-1618) encoding proteins involved in the export of CPS. These genes showed a high degree of homology to the group II capsule export genes hexDCBA in P. multocida strain Pm70 and bexDCBA in H.  Table S2. doi:10.1371/journal.pone.0001450.g003 influenzae type b strain [23,39]. Surprisingly, nucleotide sequences of the cps genes had no sequence similarity or seemingly no DNA rearrangement between JL03 of serotype 3 and L20 of serotype 5b. On the other hand, the cps genes seemed to be identical between L20 of serotype 5b and 4074 of serotype 1. These data suggested that the serotype diversity of A. pleuropneumoniae might be associated with CPS biosynthesis, but not with CPS export.
Lipopolysaccharide (LPS) biosynthesis LPSs of Gramnegative bacteria are essential structural components of the cell membrane and are considered to be virulence determinants [40]. These substances are complex molecules composed of three welldefined biochemical moieties: the lipid A; the core oligosaccharide, which contains 3-deoxy-D-manno-oct-2-ulosonic acid (KDO) and is essential for optimal adhesion of A. pleuropneumoniae to the host [3]; and the O-antigen, a polysaccharide consisting of repeating sugar units [41].
We identified all the genes encoding enzymes for LPS biosynthesis in the JL03 genome ( Table 7). Majority of the genes encoding enzymes for lipid A and KDO core biosynthesis were highly conserved among different species within the family Pasteurellaceae [11,18]. These CDSs are scattered throughout the JL03 chromosome, like most Gram-negative prokaryotes, such as E. coli, Neisseria meningitidis, Yersinia pestis, P. aeruginosa and Fusobacterium nucleatum.
A cluster of genes coding for enzymes that catalyze the biosynthesis of O-antigen was identified in JL03 ranging from wzz (APJL1485) to rmlB (APJL1497), of which, only the dTDPglucose 4,6-dehydratase RmlB was conserved across a wide range of species. These CDSs could be divided into three groups [37]: nucleotide sugar biosynthesis related (APJL1496 and 1497); glycosyltransferases (APJL1486-1489 and 1493) involved in sugar transfer; and oligosaccharide repeat unit processing related, wzz, wzy and wzx (APJL1490 and 1491).
A bacterial sugar transferase (436aa) encoded by APJL1493 shares 55% identity with Orf9 (400aa) found in A. actinomycetemcomitans [42], but only 34% identity with a sugar transferase (472aa) encoded by APL1471 of strain L20. Two proteins (encoded by APJL1487 and 1488) among the four closely linked glycosyl transferases contain a Glycos_transf_1 domain (PF00534) in their Ctermini and a Glycos_transf_2 domain (PF00535) in their N-termini, respectively, both unique among Pasteurellaceae species. Gene wzz encodes a protein (370aa) bearing 45% identity with the O-antigen chain length determining protein (MHA1853, 375aa) found in M. haemolytica [18]. Although there are much sequence variabilities among the O-antigen-processing enzymes in different Gramnegative bacteria, structural conservation and stability of membrane spanning regions still indicate that they should perform similar function predicted by the numbers and loci of relevant transmembrane helices (TMHs) (Figure 4).
The G+C content of the gene cluster coding for enzymes responsible for O-antigen chain biosynthesis was much lower (31%) than that of the JL03 chromosome (41%). On the other hand, genomic comparison with strains L20 or 4074 revealed that these CDSs were more variable than those for the synthesis of LPS lipid A and core oligosaccharide between serotypes (Table 7).   As defined by the sequences of these serotype-specific CDSs, serotype 3 should be able to be distinguished from serotype 5b and 1 as another group ( Figure 2E). Therefore, comparison of the Oantigen regions, which were analogous with that of the cps operons, could also be used as one of the markers in classifying serotypes of A. pleuropneumoniae.
Virulence related enzymes Various enzymes, such as urease and proteases, are known to play important roles in the disease process of A. pleuropneumoniae. Quite a few respiratory tract pathogens produce urease, which catalyzes the hydrolysis of urea to produce ammonia and carbon dioxide [43]. As previously reported [44], the gene cluster (ureABC, APJL1651-1649) identified in the genome of JL03 encodes the structural subunits of urease while the closely linked genes ureEFGH (APJL1647-1644) encode the accessory subunits. All of these genes are orthologues of those in the ure operon of H. influenzae. Furthermore, a 6-gene cluster, upstream of the ure operon and transcribed in the same direction, was also identified. Five of the aforementioned six genes formed a cbi operon (APJL1657-1652) encoding a putative nickel and cobalt periplasmic permease system, which may affect the total urease activity in A. pleuropneumoniae [44].
The pepN (APJL1358) encoding aminopeptidase N was identified in JL03. It was characterized on the basis of a zinc binding motif (aa 294-303) found in metalloproteases. Expression of this protease was observed in lung tissue of pigs that had died from porcine pleuropneumonia [45].
Genes encoding the enzymes required for anaerobic respiration, i.e., periplasmic nitrate reductase and DMSO reductase were described above. These enzymes are probably accessory virulence factors in A. pleuropneumoniae pathogenesis [4].
Iron acquisition and utilization Iron is essential for bacterial growth and acts as an environmental signal that regulates the expression of many virulence factors [3]. Mammals have evolved a mechanism to reduce the availability of iron to potential bacterial pathogens by using of very-high-affinity ironchelating molecules, while host-adapted pathogens have accordingly evolved means to use these iron-bearing molecules as an iron source [46]. It is known that A. pleuropneumoniae can use porcine transferrin, hemoglobin and ferrichrome [3,46]. Approximately 2.6% (55 genes) of the JL03 genome are involved in iron uptake with additional 5 related pseudogenes likely impaired by mutations. Comparing with the genomes of other Pasteurellaceae members, large proportion of genes involved in iron metabolism seems common at least within the family (Table S3). These iron metabolism related proteins are highly conserved between JL03 and L20 except for TbpB1 (APJL1598) and FhuA (APJL2066). An analogous cell model of some iron-related protein complexes and other virulence factors is illustrated in Figure 5.
The TonB system plays a key role in iron acquisition by many Gram-negative bacteria. Two functional TonB systems were reported in detail for P. aeruginosa [47] and A. pleuropneumoniae [46]. Two sets of closely linked genes encoding the TonB1 and TonB2 systems, namely tonB1(246aa)-exbB1-exbD1 (APJL1601-1599) and exbB2-exbD2-tonB2 (244aa) (APJL0078-0076), were identified in the JL03 genome. The identity between tonb1/tonb2 was only 15%, suggesting two structurally independent TonB systems for iron uptake. Transferrin-binding protein (Tbp), a kind of iron receptors, has been found in many species of the families Pasteurellaceae and Neisseriaceae [3]. Two pairs of tbp genes were found in the JL03 genome, tbpB1-tbpA1 (APJL1598-1597) and tbpB2-tbpA29 (partial)-tbpA29 (partial) (APJL0250-0252). Moreover, the nucleotide sequence identity between the two sets of genes was 54.4%, indicating that they were likely to be duplicated copies. The tbpB1-tbpA1 operon was located immediately downstream of The fhu operon encodes CDSs homologous to proteins involved in the uptake of hydroxamate siderophore across the outer membrane of several bacteria [3]. The fhuCDBA (APJL2063-2066) genes encode four proteins with 28.5 kDa, 35.8 kDa, 69.4 kDa and 77.1 kDa in sequential order, in agreement with previous studies [48]. Orthologs of ccmABCDEF (APJL1390-1385) were found in H. ducreyi and are involved in post-translational attachment of heme and catalyze the reduction of disulfide bonds in the cytochrome c apoprotein [22,49]. Both ccmE and ccmF have multiple TMHs and probable signal peptide in their N-termini. A number of genes that encode putative iron-binding receptors were found in JL03 in addition to the tbp genes, such as two hbpA genes (APJL0866 and 2060) encoding the heme-binding protein A and an outer membrane iron-receptor protein (99 kDa, APJL1922). JL03 also has the hgbA gene (APJL1065) encoding a haemoglobinbinding protein located at the outer membrane, which is regulated by a highly conserved gene fur (APJL1231) encoding ferric uptake regulator Fur [50,51]. Upstream of hgbA, there is a CDS encoding a potential haemin-binding protein homologous to the HugZ in Plesiomonas shigelloides [50].
Thus, a remarkably large number of genes encoding putative iron uptake proteins were found in the genome of JL03. A. pleuropneumoniae appears well-equipped to overcome iron shortages during infection.
In summary, we have sequenced the complete genome of A. pleuropneumoniae strain JL03, a Chinese field isolate of serotype 3 and annotated the genome in comparison against other members of the Pasteurellaceae family. A complicated metabolic network with various kinds of oxidation-reduction enzymes for catabolism and anabolism was comprehensively illustrated at the genomics level for this genus, and, for the first time, genomic discoveries were made to account for assimilatory sulfate reduction (intact operon cysGHDNJI) and NAD-dependent biotype I character (truncated nadV) of the strain (Figure 3). Meanwhile, we identified a series of genes encoding proteins of Apx toxins, adhesins, iron-uptake systems as well as enzymes for the biosynthesis of CPS and LPS, which underlined the genetic basis related to the pathogenesis/ virulence of A. pleuropneumoniae. Furthermore, comparing to the genomes of strains L20 (serotype 5b) and 4074 (serotype 1) of A. pleuropneumoniae, probable strain (serotype)-specific genomic islands and genome reductions were identified in JL03. These data should provide a foundation for future research into the mechanisms of virulence and serotype diversity of A. pleuropneumoniae.

Bacterial strain
The A. pleuropneumoniae strain JL03 used for genomic sequencing was isolated from the lung of a pig from a Chinese commercial pig farm in 2003 [52] This isolate was identified as serotype 3 [53,54] and was deposited in China Center for Type Culture Collection (CCTCC, Wuhan) available upon request. It grows well at 37uC on Tryptic Soy Agar or in Tryptic Soy Broth, supplemented with 10 mg/ml nicotinamide adenine dinucleotide (NAD) and 5% bovine serum.

Genomic sequencing, assembly and analysis
A whole genome shotgun strategy was adopted. Two genomic DNA libraries of JL03 with 1.5-4 kb or 6-8 kb insertion fragments were constructed in pUC18 or pSmart-LC respectively. Total of 13,440 clones (10,080 from the 1.5-4 kb pUC18 clones and 3,360 from the 6-8 kb pSmart-LC clones) were sequenced from both ends by ABI 3700 DNA analyzer, and altogether, 25,650 sequencing reads (Phred value.Q20, of which 760 bp was the confirmed mean length of reads) gave an 8.6-fold coverage of the genome. Employing Phred [55] and the Staden software package [56], 170 contigs were assembled. Sequence and physical gaps of the unfinished genome were filled by primer walking with 226 effective PCRs. The final closure was confirmed by sequencing the PCR amplified corresponding contig-connecting fragments using the JL03 genomic DNA as the template. The finished complete genomic sequence was analyzed by conventional genomic annotation methods, which was described in detail in the Text S1 as Materials and Methods.