Background

In 2001, Laribacter hongkongensis, a novel genus and species that belongs to the Neisseriaceae family of β-subclass of the Proteobacteria, was discovered from the blood and empyema pus of a patient with underlying alcoholic cirrhosis [1]. Subsequently, it was observed that L. hongkongensis was associated with freshwater fish borne community-acquired gastroenteritis and traveler's diarrhea in human [2–7]. The clinical syndrome of associated gastroenteritis is similar to those of Salmonella or Campylobacter gastroenteritis. About 80% and 20% of the patients have watery and bloody diarrhea respectively, one third of them have systemic symptoms and another one third have vomiting [4]. Pulsed-field gel electrophoresis of Spe I digested chromosomal DNA and multilocus sequence typing using seven housekeeping gene loci independently showed that the L. hongkongensis isolates recovered from freshwater fish and patients fell into separate clusters. These suggested that some L. hongkongensis clones could be more virulent or adapted to human than others [8, 9].

For a gastrointestinal tract pathogen to cause infection, after transmission through the oral route, the bacterium has to be able to survive the hostile acidic environment of the stomach, resist the action of bile in the small intestine, colonize the gastrointestinal tract epithelium through binding of adhesins of the bacterium to receptors on epithelial cells, evade host immune defense mechanisms before causing diarrhea and/or invading the gastrointestinal tract and cause systemic infections, as in the case of bacteremia and empyema thoracis [1]. Moreover, the possession of drug resistance determinants and phages also enhance the potential capability of the bacterium to resist to killing by antimicrobials and causing diseases. In this article, we present an overview of the genes and gene cassettes of the L. hongkongensis genome related to these various steps in the infective process, as well as drug resistance and phages. The phylogeny of these genes, most of them were thought to be acquired through horizontal gene transfer, was also analyzed.

Results and discussion

Resistance to acid

Urease

Similar to other gastrointestinal tract pathogens, L. hongkongensis has to face the highly hostile and acidic environment of the stomach before reaching the intestine. L. hongkongensis possesses a urease, that is able to hydrolyze the limited amount of urea available in the stomach to generate carbon dioxide and ammonia, which increases the pH. In the L. hongkongensis genome, a complete urease cassette, that occupies a 7,556 bp region, is observed. The cassette includes eight CDSs, which encodes three urease structural proteins (UreA, UreB and UreC) and five accessory proteins (UreE, UreF, UreG, UreD and UreI) [10]. Similar to the urease of other bacteria, the urease of L. hongkongensis is presumably a nickel containing enzyme [11]. The histidine residues at the carboxyl terminal of UreE are supposed to bind to the nickel ions that are transported into L. hongkongensis through a nickel transporter, and donate the nickel ions to UreC during urease activation. Most of the eight genes in the urease cassette of L. hongkongensis are most closely related to their homologues in bacteria of α- and γ-proteobacteria, rather than those in other bacteria of β-proteobacteria [12–16].

Arginine deiminase

Two arc gene clusters were encoded in the L. hongkongensis genome. Each cluster consists of four genes, arcA, arcB, arcC and arcD. arcA, arcB and arcC encode the three enzymes, arginine deiminase, ornithine carbamoyltransferase and carbamate kinase, of the arginine deiminase pathway, whereas arcD encodes a membrane bound arginine-ornithine antiporter. The arginine deiminase pathway converts L-arginine to carbon dioxide, ATP, and ammonia, which increases the pH. It has been shown in various bacteria, such as Streptococcus sanguis, Streptococcus rattus, Streptococcus suis, Streptococcus pyogenes, Enterococcus faecium and Pseudomonas aeruginosa that this gene cluster is useful for bacterial survival in acidic environment [17–19]. In S. pyogenes, it has also been shown that this pathway facilitates cell invasion and inhibits proliferation of human peripheral blood mononuclear cells [20, 21]. Phylogenetically, these four genes of the arc gene cluster in L. hongkongensis are most closely related to the corresponding homologues in Chromobacterium violaceum (Figure 1, 2, 3, and 4), whereas the gene cluster is absent in Neiserria meningitidis and Neisseria gonorrhoeae. Among all bacteria with complete genomes sequenced, L. hongkongensis is the only one that contains two adjacent arc gene clusters (Figure 5).

Figure 1
figure 1

Phylogenetic analysis of ArcA encoded in the arc gene cluster in L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Four hundred and nine and 409 amino acid positions in ArcA1 and ArcA2, respectively, were included in the analysis. The scale bars indicate the estimated number of substitutions per 10 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 2
figure 2

Phylogenetic analysis of ArcB proteins encoded in the arc gene cluster in L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Three hundred and thirty-four and 335 amino acid positions in ArcB1 and ArcB2, respectively, were included in the analysis. The scale bar indicates the estimated number of substitutions per 20 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 3
figure 3

Phylogenetic analysis of ArcC proteins encoded in the arc gene cluster in L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Two hundred and ninety-one and 314 amino acid positions in ArcC1 and ArcC2, respectively, were included in the analysis. The scale bars indicate the estimated number of substitutions per 10 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 4
figure 4

Phylogenetic analysis of ArcD encoded in the arc gene cluster in L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Four hundred and ninety-two, 478 and 478 amino acid positions in ArcD1, ArcD2 and ArcD3, respectively, were included in the analysis. The scale bars indicate the estimated number of substitutions per 10 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 5
figure 5

Genetic organization of ADI clusters in L. hongkongensis and other representative microbial genomes. The arrow boxes represent the CDSs. The relative positions of each gene are assigned as predicted by nucleotide sequence analysis.

Full size image

Bile resistance

Efflux pumps

Efflux of bile salts from bacteria is mediated through a number of efflux systems. These efflux systems pump a variety of compounds, including antibiotics, oxidative stress agents, organic solvents and bile salts, out of the bacterial cytoplasm. Among these efflux systems, the best studied one is encoded by acrAB-tolC of the resistance nodulation division (RND) family. This system has been shown to be present in the genomes of a variety of pathogenic and non-pathogenic bacteria of the human gastrointestinal tract, such as Escherichia coli and Salmonella Typhimurium [22, 23]. In the L. hongkongensis genome, three complete copies of acrAB-tolC, of which AcrB is located in the inner membrane and contains the conserved ACR_tran domain, AcrA is located in the periplasmic space and contains the conserved HlyD domain and TolC as the outer membrane channel protein, are present. A recent bioinformatics analysis on bile resistance mechanisms in Campylobacterales also found that one complete copy of acrAB-tolC is present in the C. jejuni genome [24]. In addition to efflux pumps encoded by acrAB-tolC, the genome of L. hongkongensis also contains two copies of emrAB-tolC of the major facilitator superfamily, one copy of acrAD-tolC of the RND family (AcrD is also an inner membrane protein and contains the conserved ACR_tran domain similar to AcrB), one copy of mdtABC-tolC of the RND family and one copy of ydgFE/mdtJI of the small multidrug resistance family. These four gene cassettes were also found to be encoding efflux pumps related to bile resistance in E. coli[22, 25–27]. In addition, acrAD-tolC and mdtABC-tolC have been documented to be related to bile salt resistance in S. Typhimurium [28].

Lipopolysaccharide (LPS) and Tol proteins

In addition to the efflux pumps, the integrity of the outer membrane is also important in resistance against bile. The O-antigen has been shown to be related to bile resistance in S. Typhimurium [29, 30]. Tol proteins, which are cytoplasmic and periplasmic proteins encoded by a gene cluster that consists of five genes (tolQ, tolR, tolA, tolB and pal), are also important in maintaining the integrity of the outer membrane and bile resistance, as shown in E. coli, S. Typhimurium and Erwinia chrysanthemi[31–33]. In the genomes of L. hongkongensis and C. violaceum, tolQ was not clustered with tolR, tolA, tolB and pal, although all five genes are present in their genomes.

Colonization of intestinal mucosa

The first step of infection is adhesion to host cells. In the L. hongkongensis genome, a putative adhesin, with 27-30% amino acid identity to the adhesins of diffusely adherent E. coli (DAEC) [34–36] and enterotoxigenic E. coli (ETEC) [37–40], encoded by aidA and tibA respectively, was observed (Figure 6). It has been shown that aidA deletion mutants of DAEC lost the ability to adhere to HeLa cells and tibA deletion mutants of ETEC lost the ability to adhere to human intestine epithelial cells [37, 41, 42]; and E. coli HB101 transformed with tib loci was able to adhere to HCT 8 cells [37, 42]. aidA and tibA encode proteins of the autotransporter family, type V protein secretion system of Gram-negative bacteria [43]. Proteins of this family possess three domains, an N-terminal signal sequence, a passenger or α-domain and a translocation or β-domain, which enable the proteins to transport themselves to cell surfaces. These three domains are all present in the putative adhesin in L. hongkongensis. Amino acid residues 1-36 is the putative signal sequence (predicted by SignalP). As in the passenger domains of other autotransporters, no cysteine residues, which were thought to interfere with transport of the proteins to cell surfaces because of formation of disulphide bonds, were present in the putative passenger domain (amino acid residues 37-756) of the putative adhesin in L. hongkongensis[41]. In the passenger domains of AIDA in DAEC, multiple copies of the consensus sequence VXNSGG, acceptor sites for heptose, addition of which catalyzed by AAH heptosyltransferase, encoded by aah located upstream to aidA, are present [44]. The addition of heptose was shown to be essential for the adhesion properties in the tibA adhesin in ETEC [45]. In the putative passenger domain of the putative adhesin in L. hongkongensis, nine copies of VXSGG, but not VXNSGG, were present; and a putative heptosyltransferase, with 52% amino acid identity to the TibC heptosyltransferase of ETEC, was present upstream to the putative adhesin gene in the L. hongkongensis genome. Interestingly, in the putative passenger domain of tibA adhesin in ETEC, 11 copies of VXSGG, but not VXNSGG, were present, but whether VXSGG is the acceptor sites for heptose has not been documented. In addition to their roles for adhesion, the passenger domains may also possess virulence functions, such as autoaggregation, biofilm formation, invasion and cytotoxicity. In the putative translocation domain, the consensus motif (Y/V/I/F/W)-X-(F/W) at the extreme carboxyl terminus of other autotransporter proteins, predicted to play a role in outer membrane localization and/or stability of these proteins, was present [41].

Figure 6
figure 6

Phylogenetic analysis of the putative adhesin of L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Six hundred and eight amino acid positions of the passenger domain were included in the analysis. The scale bar indicates the estimated number of substitutions per 20 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image

Evasion of host defense

To protect from the active oxygen species (superoxide and hydrogen peroxide) released from phagocytic cells, the genome of L. hongkongensis encodes superoxide dismutase and catalases, in line with its catalase-positive phenotype. The putative superoxide dismutase of L. hongkongensis, which decomposes superoxide to hydrogen peroxide and oxygen, is most closely related to those of C. violaceum, N. meningitidis and N. gonorrhoeae. There are three putative catalases in the L. hongkongensis genome, encoded by a katE (encoding hydroperoxidase II) and two katG (encoding hydroperoxidase I with catalase-peroxidase activity). These decompose hydrogen peroxide to water and oxygen. katE in L. hongkongensis is most closely related to the homologues in Ralstonia eutropha, whereas the two katG were most closely related to those in Shewanella amazonensis and Vibrio cholerae respectively. In addition to protection against the active oxygen species, some efflux pumps may export host-derived antimicrobial agents in addition to antibiotics, bile and other substances, hence protecting from such naturally produced molecules of the host.

Virulence factors

Lipopolysaccharide

LPS consists of three parts: lipid A, core oligosaccharide, and polysaccharide side chains. In E. coli, the minimal LPS required for growth include lipid A and two keto-deoxyoctulonate (Kdo) units of the core oligosaccharide. The LPS of wild type strains of E. coli consist of additional core sugars and polysaccharide side chains. The polysaccharide side chains are also known as the O-antigen, which varies among different species of Gram-negative bacteria and different strains of the same species. These sugars enhance survival during environmental stress, and help the bacteria evade the host immune system by modification of the structure. Lipid A, also known as the endotoxin, is the hydrophobic anchor of LPS. It is a glucosamine based phospholipid inserted into the outer membranes of most Gram-negative bacteria. Most Gram-negative bacteria synthesize lipid A by pathways similar to the one in E. coli. Through binding to Toll-like receptor 4 and CD14, lipid A of Gram-negative bacteria trigger the synthesis and secretion of pro-inflammatory cytokines. The actions of these cytokines lead to local and systemic inflammatory responses, which result in various clinical manifestations, and even deaths, of patients.

The same set of genes that encode enzymes in the biosynthetic pathways of lipid A, the two Kdo units and the heptose units are present in the L. hongkongensis, C. violaceum, N. meningitidis, N. gonorrhoeae and E. coli genomes. In contrast to E. coli, the lipid A of C. violaceum, N. meningitidis and N. gonorrhoeae had a symmetrical acylation pattern [46]. Both the reducing and terminal N-acetyl-glucosamine residues in these bacteria carry three acyl groups. The sequential addition of the last 12-carbon acyl group to the reducing and terminal N-acetyl-glucosamine residues are catalyzed by enzymes encoded by the htrB and msbB genes, respectively. It was found that msbB deletion mutants of N. meningitidis and N. gonorrhoeae had lower abilities to activate human macrophages to produce pro-inflammatory cytokines [47–49]. Phylogenetic analysis of the experimentally confirmed htrB and msbB genes in N. meningitidis and N. gonorrhoeae and the putative htrB and msbB genes in L. hongkongensis and C. violaceum showed that the four htrB genes and the four msbB genes fell into two separate clusters, with very high bootstrap values (Figure 7). Therefore, we speculate that the htrB and msbB genes in L. hongkongensis and C. violaceum serve similar functions as those in N. meningitidis and N. gonorrhoeae and that the lipid A of L. hongkongensis also had a symmetrical acylation pattern.

Figure 7
figure 7

Phylogenetic analysis of confirmed/putative HtrB and MsbB of L. hongkongensis , C. violaceum , N. meningitidis and N. gonorrhoeae. The tree was constructed by neighbor joining method using Kimura's correction and was rooted using HtrB of Rickettsia typhi (YP_067645). Two hundred and eighty-four amino acid positions were included in the analysis. The scale bar indicates the estimated number of substitutions per 20 amino acids. Numbers at nodes indicated levels of bootstrap support calculated from 1000 trees. All names and accession numbers are given as cited in the GenBank database.

Full size image

The genes that are responsible for the synthesis of α-chain L1, α-chain L2, β-chain and γ-chain in the core oligosaccharide in N. meningitidis and N. gonorrhoeae (lgtA, lgtB, lgtC, lgtD, lgtE, lgtF, lgtG, rfaK) and those for the addition of sialic acids to these chains (lst) are absent in the genomes of L. hongkongensis and C. violaceum[50]. On the other hand, nine genes which encode putative enzymes for biosynthesis of the polysaccharide side chains are present in the L. hongkongensis genome. Four of these genes (rfbA, rfbB, rfbC and rfbD) are also present in the genomes of C. violaceum, N. meningitidis and N. gonorrhoeae. The enzymes encoded by these four genes catalyzed reactions for the synthesis of dTDP-rhamnose, although mutations of them in N. meningitidis and N. gonorrhoeae did not result in any change in their phenotypes [51, 52]. The other five genes (wbmF, wbmG, wbmH, wbmI and wbmK), which encode putative nucleotide sugar epimerases/dehydratases and amidotransferase, are not present in the C. violaceum, N. meningitidis and N. gonorrhoeae genomes, but are most closely related to the corresponding genes for the biosynthesis of the O-antigens in Bordetella parapertussis and Bordetella bronchoseptica[53]. Although the structures of the LPS of L. hongkongensis and C. violaceum remain to be determined, these imply that the structures of the LPS of L. hongkongensis and C. violaceum are probably quite different from those of the lipooligosaccharides of N. meningitidis and N. gonorrhoeae.

Recently, a number of genes that encode proteins for the assembly and transport of LPS in E. coli have been discovered [54]. All these genes were also present in the genomes of L. hongkongensis, C. violaceum, N. meningitidis and N. gonorrhoeae (Table 1). The exact functions of these proteins have not been fully elucidated.

Table 1 Genes for assembly and transport of LPS
Full size table

Cytotoxins

The L. hongkongensis genome contains a number of CDSs that encode putative cytotoxins. These include cell surface acting cytotoxins, such as RTX toxin and hemolysins; and intracellular cytotoxins such as patatin-like proteins.

RTX toxins

RTX toxins, originally discovered in E. coli (α-hemolysin) [55, 56], are most commonly found in bacteria of the Pasteurellaceae family. Most RTX toxins are hemolysins or leukotoxins [57, 58]. The L. hongkongensis genome contains an RTX gene cluster (tolC-rtxA1-rtxD-rtxB) and an isolated rtxA2 gene. In the RTX gene cluster (Figure 8), tolC encodes the outer membrane component of the type I secretion apparatus, rtxA1 encodes the structural toxin, rtxD encodes the adaptor protein anchored to the inner membrane and rtxB encodes the inner membrane ATPase. TolC, RtxD and RtxB form the secretion apparatus for exporting RtxA. Similar to RtxA of other bacteria, RtxA1 and RtxA2 of L. hongkongensis possess tandem arrays of glycine-rich nonapeptide repeats (GGXGXDX[L/I/V/W/Y/F]X, where X is any amino acid) for binding of calcium ions (Figure 8). There are five nonapeptide repeats in RtxA1 and nine nonapeptide repeats in RtxA2. Unlike most other bacteria which contain rtxC genes, the RTX gene cluster of L. hongkongensis does not possess this gene. Instead, it contains a gene of putative adhesive function, located between rtxA1 and rtxD. Domain search using InterProScan showed that this gene contains nine repeats of 22 amino acids (TDNGTVTNVTLSSVTNGQTVAE) with parallel beta-helix structures. Each repeat is separated from the adjacent one by 82 amino acids (Figure 8). Although the genomes of L. hongkongensis, C. violaceum and N. meningitidis all contain RTX toxin, RtxA1 and RtxA2 of L. hongkongensis do not show clustering with the homologues in C. violaceum and N. meningitidis. This is in contrast to the other genes (tolC, rtxD and rtxB) in the RTX gene cluster, which are all most closely related to the corresponding homologues in C. violaceum and other species of β-proteobacteria [59, 60] (Figure 9, 10, 11, and 12). Moreover, the amino acid identities between TolC, RtxD and RtxB and their homologues in C. violaceum are much higher than those between RtxA1 or RtxA2 and their homologues in any other bacteria (Figure 9, 10, 11, and 12). These suggest that rtxA1 and rtxA2 have evolved much faster than tolC, rtxD and rtxB, so that the toxins can bind to their corresponding host cells more efficiently. Interestingly, similar to rtxA2 of L. hongkongensis, the structural toxin genes (frpC and frpA) in N. meningitidis are not linked to genes of the type I secretion system. However, it has been shown that FrpC and FrpA can be secreted by E. coli harboring hlyBD genes, indicating that they are probably secreted by secretion systems unlinked to their corresponding genes [61].

Figure 8
figure 8

Genetic organization of the RTX gene cluster ( tolC-rtxA1-rtxD-rtxB ) in L. hongkongensis. The boxes represent the CDSs. The number of amino acid residues of each gene is indicated above the boxes. The basic functional activities of the corresponding gene products are given on the top. Five copies of glycine-rich nonapeptide repeats (GGXGXDX[L/I/V/W/Y/F]X, where X is any amino acid) of rtxA1 are underlined. An CDS of unknown function, located between rtxA1 and rtxD, are also depicted, where nine repeats of 22 amino acids are highlighted. The relative positions of each gene are assigned as predicted by nucleotide sequence analysis.

Full size image
Figure 9
figure 9

Phylogenetic analysis of TolC in the RTX gene cluster of L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Four hundred and forty-two amino acid positions were included in the analysis. The scale bars indicate the estimated number of substitutions per 20 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 10
figure 10

Phylogenetic analysis of RtxA1 in the RTX gene cluster of L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. One thousand and eighty-seven amino acid positions were included in the analysis. The scale bars indicate the estimated number of substitutions per 20 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 11
figure 11

Phylogenetic analysis of RtxD in the RTX gene cluster of L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Four hundred and fifty-two amino acid positions were included in the analysis. The scale bars indicate the estimated number of substitutions per 20 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Figure 12
figure 12

Phylogenetic analysis of RtxB in the RTX gene cluster of L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Seven hundred and twenty amino acid positions were included in the analysis. The scale bars indicate the estimated number of substitutions per 20 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image
Hemolysins

In the L. hongkongensis genome, there are two gene loci that encode putative hemolysins. The first putative hemolysin contains three domains, the first one of the DUF21 superfamily, the second one of the CBS_pair superfamily and the third one of the CorC_HlyC superfamily. Among the five most closely related protein sequences, three of them were putative hemolysins of three different Yersinia species, and the other two were hypothetical proteins. The second putative hemolysin belongs to the HlyIII superfamily, which contains seven transmembrane domains with conserved amino acid residues present. It is most closely related to the hemolysin III of C. violaceum.

Patatin-like protein

Patatin, originally described in plants such as potatoes, has diverse functions such as storage glycoproteins [62], signal transduction [63] and defense against parasites [64]. In 2003, it was found that toxin ExoU of P. aeruginosa, delivered to eukaryotic cells via a type III secretion system, possessed the catalytic domains of patatin, iPLA(2) and cPLA(2) [65]. Direct injection of ExoU in mammalian cells resulted in irreversible damage to cellular membranes and rapid necrotic death [66]. Similar to patatin, ExoU of P. aeruginosa possessed phopholipase A2 activity. P. aeruginosa mutants with mutations at the active sites of the patatin-like protein were less virulent than wild type P. aeruginosa in a mouse model [67]. Subsequently, genes that encode putative patatin-like proteins were observed in many bacterial genomes, although none of them was characterized phenotypically [68]. It was also observed that the average copy number of genes that encode patatin-like proteins is higher in plant/animal bacterial pathogens than in non-pathogens [68]. In some pathogens, up to eight copies of genes that encode putative patatin-like proteins can be found. Similar to P. aeruginosa, the genome of L. hongkongensis also contains three copies of genes that encode putative patatin-like proteins. The lengths of the genes that encode putative patatin-like proteins in the genomes of L. hongkongensis, C. violaceum (7 copies), N. meningitidis (1 copy) and N. gonorrhoeae (1 copy) varied from 894 to 2,337 bp. The three copies in the L. hongkongensis genome are 951, 963 and 2,232 bp respectively. All three copies contain all the four domains that can be found in bacterial patatin-like proteins, including a putative oxyanion hole, a serine hydrolase G-X-S-X-G domain, a potential serine-containing phosphorylation site and an aspartate-containing active site domain (Figure 13). The serine in the hydrolase domain and the aspartate made up a patatin-specific catalytic dyad that has not been described in any other known proteins [68].

Figure 13
figure 13

Multiple alignments of the four conserved domains in the putative patatin-like proteins in the genomes of L. hongkongensis , C. violaceum , N. meningitidis and N. gonorrhoeae. The two arrows indicate the Ser-Asp catalytic dyad. Conserved amino acids in the four domains are in bold. ω, number of amino acids before and after the conserved domains.

Full size image

Enzymes

Outer membrane phospholipase A

It has been shown that outer membrane phospholipase A (OMPLA) is a virulence factor in a number of bacteria, including Helicobacter pylori and C. coli. Located on the outer membrane of bacteria, OMPLA lyses the outer membrane, leading to release of other virulence factors, such as urease and VacA in H. pylori. In the L. hongkongensis genome, a gene that encodes a putative OMPLA is observed. This OMPLA possesses a complete and highly specific consensus sequence motif (YTQ-Xn-G-X2-H-X-SNG) found in OMPLA of other bacteria. Phylogenetically, it is most closely related to the OMPLA of Methylibium petroleiphilum, a methyl tert-butyl ether-degrading methylotroph of β-proteobacteria (Figure 14) [69].

Figure 14
figure 14

Phylogenetic analysis of outer membrane phospholipase A of L. hongkongensis. The tree was constructed by neighbor joining method using Kimura's correction and bootstrap values calculated from 1000 trees. Three hundred and seventy-seven amino acid positions were included in the analysis. The scale bar indicates the estimated number of substitutions per 10 amino acids. All names and accession numbers are given as cited in the GenBank database.

Full size image

Drug resistance

A genome-wide analysis using similarity searches revealed the presence of a large number of antibiotic resistance-related genes in L. hongkongensis strain HLHK9. They are related to β-lactam (Table 2), multidrug efflux (Table 3) and other resistance genes (Table 4).

Table 2 CDSs related to beta-lactam antibiotics in L. hongkongensis
Full size table
Table 3 CDSs related to multidrug resistance in L. hongkongensis
Full size table
Table 4 Miscellaneous resistance genes in L. hongkongensis
Full size table

β-lactam resistance-related genes

A total of 10 CDSs related to β-lactam resistance were identified in the L. hongkongensis genome. Genes that exhibit similarity to penicillin-binding proteins (PBPs) (6 CDSs) of other bacterial species were found (Table 2). The PBPs identified in L. hongkongensis include PBP1a, PBP2, PBP3, PBP4a, PBP6a, and PBP7, which are essential proteins that are involved in biosynthesis of murein and peptidoglycan, and are targets for inhibition by β-lactams [70, 71]. Although the presence of PBPs per se does not confer resistance, chromosomal mutations in PBPs may render the bacteria resistant to β-lactams [72–75].

Apart from the ampC gene (LHK_03028) that encodes the previously characterized class C β-lactamase [76], there are two other putative β-lactamases (LHK_00876 and LHK_00878) observed in the L. hongkongensis genome. They are both putative metallo-β-lactamases containing a metallo-β-lactamase superfamily domain which included two zinc ligand-binding sites essential for its hydrolytic function on the β-lactam ring (Figure 15) [77–79]. However, these zinc ligand-binding sites were also present in most proteins of the metallo-β-lactamase superfamily, the function of which is not limited to β-lactam hydrolysis [79–81]. Therefore, in vitro experiments are required to confirm the actual function of these two putative metallo-β-lactamases.

Figure 15
figure 15

Multiple alignment of the partial amino acid sequences of the two putative metallo-β-lactamases in L. hongkongensis and those of known metallo-β-lactamases showing the conserved zinc-ligand binding sites. Amino acid residues high-lighted in yellow and blue representing two independent putative zinc-ligand binding sites of class B3 metallo-β-lactamase His116-His118-His196 and Asp120-His121-His263. Numbers in parentheses indicate the corresponding positions in the amino acid sequences. L1, Stenotrophomonas maltophilia IID1275 (accession no. CAA52968); FEZ-1, Legionella gormanii ATCC33297 (accession no. CAB96921); CAU-1, Caulobacter vibrioides DSM 4727 (accession no. CAC87665); BJP-1, Bradyrhizobium japonicum USDA 110 (accession no. NP_772870)

Full size image

Multidrug resistance genes

A total of 54 CDSs related to multidrug efflux were identified in L. hongkongensis genome (Table 3). The five major families of drug extrusion translocases were all present, including the Major Facilitator Superfamily (MFS) (7 CDSs), Small Multidrug Resistance (SMR) family (2 CDSs), RND family (7 CDSs), Multidrug and Toxic compound Extrusion (MATE) family (2 CDSs), and ATP-Binding Cassette (ABC) superfamily (6 CDSs).

Resistance-Nodulation-cell Division (RND) family proteins

For Gram-negative bacteria, the efflux pumps that are associated with most clinically significant resistance to antibiotics are those of the RND family. In this family, three gene loci homologous to acrRAB-tolC (LHK_00138, LHK_00140-00142; LHK_02129-02132; LHK_02825-02828) and one gene locus homologous to acrAD-tolC (LHK_02929-02931) of Escherichia coli were identified in the genome of L. hongkongensis. These three AcrRAB-TolC and the AcrAD-TolC multidrug efflux systems shared typical tripartite structure with other multidrug efflux systems in the RND family [82]. AcrB and AcrD are membrane transporter proteins, AcrA is membrane fusion protein and TolC is outer membrane channel protein. acrR is a transcription regulator gene located upstream of the acrAB-tolC loci. As a multidrug efflux system with broad-substrate spectrum, AcrAB-TolC confers resistance to chloramphenicol, tetracyclines, erythromycin, trimethoprim, β-lactams, and other organic and inorganic antiseptic agents in E. coli[83, 84]. AcrAD-TolC is less commonly reported compared to AcrAB-TolC system, where AcrD is a close homolog of AcrB. AcrAD-TolC multidrug efflux system is capable of exporting antibiotics of the aminoglycoside class including amikacin, gentamicin, neomycin, kanamycin, tobramycin, and streptomycin in E. coli[85, 86]. Another putative multidrug efflux system of the RND family identified in the genome of L. hongkongensis is homologous to MdtABC-TolC system (LHK_01285, LHK_01286, LHK_01288, LHK_01289). MdtABC-TolC system in E. coli confers at least novobiocin and bile salt resistance in the bacterium. A uniqueness of this system is that MdtB and MdtC will form a heterodimer as a membrane efflux component in cooperation with membrane fusion protein MdtA and outer membrane channel protein TolC. [27, 87] Moreover, one RND family multidrug efflux system with homology to hydrophobe/amphiphile efflux-1 subfamily was also discovered (LHK_01424-01426).

Major Facilitator Superfamily (MFS)

Two loci (LHK_01373-01376; LHK_03132-03134) homologous to emrAB-tolC system of E. coli belonging to MFS were found in the genome of L. hongkongensis. One of them had an additional transcription regulator emrR gene (LHK_01376) in its upstream sequence. EmrAB-TolC system in E. coli confers nalidixic acid and other toxic novobiocin substances resistance to bacterium [88]. Moreover, mutation of the emrR gene has been shown to lead to over-expression of the EmrAB pump and increased resistance to antimicrobial agents [89]. However, the substrate specificity of these EmrAB-TolC homologs identified in the genome of L. hongkongensis is yet to be investigated. There are five other multidrug efflux proteins belonging to MFS (LHK_00743; LHK_01870; LHK_02173; LHK_02539; LHK_02975) in the L. hongkongensis genome. One of them (LHK_00743) is a homolog to mdfA gene while another (LHK_02975) has high identities to bcr gene. mdfA encodes an MF-related protein, MdfA, which results in resistance to a diverse group of cationic and zwitterionic lipophilic compounds and antibiotics such as chloramphenicol and erythromycin when over-expressed in E. coli[90]. bcr gene codes for an efflux protein which is associated with bicyclomycin resistance in E. coli[91].

Small Multidrug Resistance (SMR) family

Two adjacently located multidrug efflux genes (LHK_01384 and LHK_01385) of the SMR family were identified in the genome of L. hongkongensis. They are homologous to mdtJI (also named ydgEF) genes in E. coli which confers resistance to spermidine and, deoxycholate and sodium dodecyl sulfate at low level [92, 93]. mdtJI have to be co-expressed for functionality and it is suggested that MdtJI may function as a heterodimer or heterooligomer [92–94].

Multidrug and Toxic compound Extrusion (MATE) family

Two multidrug efflux genes of the MATE family (LHK_00466 and LHK_02533) were also discovered in the genome of L. hongkongensis. One of them (LHK_02533) is a homolog of multidrug efflux protein NorA from Staphylococcus aureus, which confers resistance to antibiotics of the quinolone class and various organic compounds [95, 96]. Mutation of the norA gene in S. aureus has resulted in 5- to 30-fold increase in susceptibility to norfloxacin [96].

ATP-Binding Cassette (ABC) superfamily

Six CDSs of the ABC transporter family related to multidrug resistance were identified in the L. hongkongensis genome. A tripartite multidrug efflux system of the ABC transporter family composed of membrane transporter (LHK_02239), MFP (LHK_02240), and OMP (LHK_02241) was identified in the genome of L. hongkongensis. This system of proteins probably functions as a complex with composition resembling to that of RND family. Five other standalone putative ABC transporter genes (LHK_00222; LHK_01967; LHK_02051; LHK_02238; LHK_02949) coding for multidrug efflux proteins were scattered over the L. hongkongensis genome. One (LHK_02949) of them possessed homology to msbA from E. coli, which is responsible for mediating the transport of the lipid A core of LPS to the outer membrane [97, 98]. Interestingly, expression of E. coli MsbA in Lactococcus lactis which lacks LPS has been shown to significantly increase resistance to erythromycin [98].

In addition to these five major families, the L. hongkongensis genome also encodes a number of other possible multidrug resistance-related genes. Among these, there are five marC-like genes (LHK_01214; LHK_01383; LHK_01934; LHK_02292; LHK_02783), the expression of which was once believed to be associated with multidrug efflux system MarRAB in E. coli[99]. However, a recent report has shown that mutation in marC did not increase antibiotic susceptibility on E. coli[100]. Therefore, the actual function of MarC is still not identified yet. One CDS (LHK_02235) coding for a protein with 75% amino acid identities to putative integral membrane efflux protein of Yersinia pestis and possessing an AbgT family domain was also identified in the genome of L. hongkongensis. AbgT protein family includes two transporter members, AbgT protein of E. coli and MtrF of N. gonorrhoeae[101, 102]. MtrF, as an inner membrane protein, which enhances the activity of multidrug efflux system MtrCDE of the RND family, conferring higher level of resistance to hydrophobic antibiotics such as penicillin and erythromycin etc. [102, 103]. Since no mtrCDE gene homologs were found in the genome of L. hongkongensis, the role and function of the AbgT family protein in L. hongkongensis remains to be elucidated.

Miscellaneous resistance genes

Six other CDSs with homologies to other drug resistance genes were identified in the L. hongkongensis genome (Table 4). A putative dimethyladenosine transferase, encoded by ksgA gene (LHK_00025) was found. Kasugamycin and streptomycin resistance as a result of mutations in ksgA have been documented [104–106]. A bacA gene (LHK_02940) encoding putative bacitracin resistance protein BacA was also identified. BacA protein confers bacitracin resistance to E. coli by catalyzing the dephosphorylation of undecaprenyl diphosphate (C55-PP) into C55-P, which is important in peptidoglycan synthesis. The conversion of C55-PP into C55-P is normally catalyzed by a specific phosphatase which is inhibited by bacitracin leading to halted peptidoglycan synthesis [107]. The other four CDSs encode putative arsenical-resistance protein (LHK_00913), two camphor resistance proteins CrcB (LHK_01038 and LHK_01039), and chloramphenicol sensitive protein RarD (LHK_01350). Overexpression of CrcB in E. coli has been shown to protect the bacteria against chromosome decondensation by camphor [108]. The presence of two crcB genes in L. hongkongensis genome, but only one copy in the closely related bacterium, C. violaceum, and none in N. gonorrheae or N. meningitidis genomes suggested that this is an important defense mechanism in L. hongkongensis. Since the L. hongkongensis strain, HLHK9, used for genome sequencing is susceptible to tetracycline (MIC = 0.5 μg/ml), the tetA gene previously identified in L. hongkongensis strains resistant to tetracycline is not found in the present genome [109]. Recently, class 1 integrons carrying multiple antimicrobial resistance genes were identified in 6.5% of L. hongkongensis isolates from aquatic products in Guangzhou city, China [110]. However, such integron is not present in the genome of strain HLHK9.

Bacteriophages

The L. hongkongensis genome (genome size 3.16 Mbp) contains a total of eight putative prophages named LhP1 to LhP8, the positions of which are shown in Figure 16 and Table 5. This high number of prophages, compared to 3 prophages in C. violaceum (genome size 4.75 Mbp) (GenBank accession no. AE016825), 1 to 3 in N. meningitidis (genome size 2.14 to 2.27 Mbp) (GenBank accession no. CP000381, FM999788, AM421808, AE002098, AL157959, AM889136, CP001561) and 6 in N. gonorrhoeae (genome size 2.15 to 2.23 Mbp) (GenBank accession no. AE004969, CP001050) using the same parameters for prophage prediction by Prophage Finder, suggested that this is an important mechanism for acquisition and exchange of genetic materials in L. hongkongensis. While N. meningitides and C. violaceum cause mainly meningitis and invasive infections respectively that can lead to fatal septicemia, N. gonorrheae and L. hongkongensis were mainly isolated from human genital and gastrointestinal tract respectively. Interestingly, the presence of apparently high number of prophages also in N. gonorrhoeae is in line with our previous observation that horizontal gene transfer was particularly frequent among bacteria residing in human gastrointestinal and probably genital tract [111], suggesting that these anatomical sites may be an excellent incubator for bacterial gene transfer.

Figure 16
figure 16

Position of the LhP prophages and the CDSs coding for transposases in the L. hongkongensis genome. LhP1 to LhP8: L. hongkongensis prophages 1 to 8.

Full size image
Table 5 Prophages of L. hongkongensis HLHK9
Full size table

LhP1

Bacteriophage LhP1 is composed of 47 CDSs, accounting for 31,318 bp with G+C content 63.07%, close to the G+C content of the L. hongkongensis genome. LhP1 contains 34 phage-related CDSs. Analysis of these CDSs indicated that LhP1 is likely a P2-like phage, as 29 of its 34 phage-related CDSs were most similar to CDSs in P2-like prophages (Figure 17). A P2-like phage typically possesses an icosahedral head with a diameter of about 60 nm, containing a linear double-stranded DNA molecule of about 30-35 kb with cohesive ends and a straight tail with a contractile sheath [112]. Based on their morphology, P2-like phages are classified as members of the Myoviridae family (phages with contractile tails) in the order Caudovirales (tailed phages) [113]. Other CDSs exhibit similarity to other genes of phages such as Mu-like phages and unclassified phages under Myoviridae and Siphoviridae (phages with long non-contractile tails).

Figure 17
figure 17

Dot-plot analysis for LhP1, LhP8 and E. coli phage P2. (A) Dot-plot alignment of LhP8 sequences (vertical axis) versus LhP1 sequences (horizontal axis). (B) Dot-plot alignment of LhP1 sequences (vertical axis) versus Enterobacteria phage P2 sequences (horizontal axis). (C) Dot-plot alignment of LhP8 sequences (vertical axis) versus Enterobacteria phage P2 sequences (horizontal axis).

Full size image

LhP2

Bacteriophage LhP2 is composed of 32 CDSs, accounting for 26,141 bp with G+C content 64.81%. Analysis of its CDSs indicated that LhP2 is likely a Mu-like phage, with 10 of the 25 phage-related CDSs most similar to CDSs in Mu-like phages of C. violaceum (CvP1), Haemophilus influenzae and N. meningitidis. There are also other CDSs similar to other phage genes of lambda- and P2-like phages.

LhP3

Bacteriophage LhP3 is the smallest prophage in the L. hongkongensis genome. It is composed of 19 CDSs, accounting for 11,169 bp with G+C content 58.70%, lower than that of the host genome (62.35%), reflecting its heterologous origin. Of the 19 CDSs, 14 were phage-related CDSs with similarity to genes of BPP-1-, lambda- and epsilon15-like phages and other unclassified phages, indicating its genetic complexity. Further studies are required if this relatively small prophage is a functional tailed phage.

LhP4

Bacteriophage LhP4 is composed of 36 CDSs, accounting for 34,375 bp with G+C content 58.78%, also lower than that of the host genome, indicating its heterologous origin. Of the 23 phage-related CDSs, 14 possessed similarity to genes of Bordetella phage BPP-1. Other phage related genes resemble those of P4-, P2-, P22- and episolon15-like phages and unclassified phages of Siphoviridae and Myoviridae.

LhP5

Bacteriophage LhP5 is the largest prophage identified in the L. hongkongensis genome. Composed of 64 CDSs, it accounts for 43,997 bp with G+C content 59%, lower than that of the host genome. Of the 32 phage-related CDSs, 9 possessed homologies to genes of Mu-like phages, 7 even possessed homologies to genes of lambda-like phages. The other phage-related CDSs are most closely related to those of various phages including those belonging to Podoviridae (phages with short tails), Myoriviridae and Siphoviridae.

LhP6

Bacteriophage LhP6 is composed of 31 CDSs, accounting for 21,918 bp with G+C content of 62.04%. The 25 phage-related CDSs exhibit similarity to phage genes of Bordetella bronchiseptica and Bordetella avium. Of these 25 CDSs, 12 possessed homologies to genes of unclassified phages belonging to Siphoviridae and 5 to lambda-like phages.

LhP7

Bacteriophage LhP7 is composed of 31 CDSs, accounting for 19,992 bp with the lowest G+C content of 55.59% among the eight prophages, suggesting a heterologous origin. Of the 18 phage-related CDSs, 4 exhibits similarity to phage genes of N. meningitidis, Burkholderia, and C. violaceum genes of Mu-like phages, and others to those of unclassified phages, lambda-, P22-, and BPP-1-like phages.

LhP8

Similar to LhP1, bacteriophage LhP8 is also a P2-like phage (Figure 17). It is composed of 48 CDSs, accounting for 33,791 bp with G+C content of 63.87%, similar to that of the host genome. It contains the highest number of phage-related CDSs (n = 37) among the eight phages. Of the 37 phage-related CDSs, 30 were most similar to genes of P2-like phages and others to phages of Myoviridae, Siphoviridae and Mu-like phages. In fact, LhP1 and LhP8 are highly similar with the exception of a few CDSs, with most of their CDSs exhibiting similarity to phage proteins found in other gram-negative bacteria including Salmonella, Burkholderia, Yersinia, and Shigella species. Their gene organizations are also highly similar to P2 phage (Table 6) (Figure 17).

Table 6 CDSs of LhP1 and LhP8 from the L. hongkongensis HLHK9, and comparison of genome structures of LhP1 (reverse complement), LhP8 and E. coli P2 phage.
Full size table

Remnant phages

Among the eight putative prophages, LhP1 and LhP8 are most likely to represent intact prophages, while the remaining six prophages encode a diversity of prophage elements of phage-related structural and non-structural proteins. In addition to these putative prophages, 17 other phage-related CDSs were found scattered in the L. hongkongensis genome. However, these CDSs are either not flanked by other phage-related genes or that the region of these phage-related gene clusters was too short for confident prediction as prophages. Further studies are required to ascertain if the present putative prophages and phage-related gene clusters are intact or remnant phages.

Transposases and insertion sequences

There are 26 CDSs coding for transposases in the L. hongkongensis genome (Table 7). Fourteen of these 26 transposases possessed homologies to transposases of IS3 family, nine to those of IS5 family and three to those of IS481 family. The presence of transposases of IS481 family is unique in L. hongkongensis, as they are absent in other members of the Neisseriaceae family such as the pathogenic Neisseria species and C. violaceum[114]. The transposases of L. hongkongensis are most closely related to those of other members of β-proteobacteria, especially of the order Burkholderiales, with seven most closely related to those of Comamonas testosteroni, seven to those of Janthinobacterium sp., and four to those of Polaromonas sp. However, only two pairs of these transposases carry short imperfect inverted repeats at their ends that form insertion sequences most closely related to the IS3 family. Other transposases are likely remnant insertion sequences and lack associated inverted repeats. The first insertion sequence, of 1,183 bp, contains two ORFs, LHK_01280 (ORFb) and LHK_01281 (ORFa), with 38-bp inverted repeats with six mismatches. The second insertion sequence is relatively short in length, with 603 bp containing two ORFs, LHK_02311 and LHK_02312 (ORFa) and 50-bp inverted repeats with ten mismatches. The G+C content of both putative insertion sequences are lower (57.4% and 54.89% respectively) than that of the L. hongkongensis genome, suggestive of heterologous origin.

Table 7 Transposases identified in the genome of L. hongkongensis HLHK9
Full size table

Conclusions

The L. hongkongensis genome possessed genes and gene cassettes for acid and bile resistance, colonization of the intestinal mucosa, evasion of host defense and cytotoxicity and invasion. In addition, a broad variety of antibiotic resistance or multidrug resistance genes, a high number of prophages, together with other phage-related CDSs and CDSs coding for transposases, were also identified.

Methods

CDSs identified in the L. hongkongensis genome were annotated as described in our previous publication and classified functionally according to the Clusters of Orthologous Groups (COG) methodology [10]. CDSs belonging to COG clusters potentially associated with virulence (such as intracellular trafficking, secretion and vesicular transport) were selected for further examination, whereas those associated with housekeeping functions (such as chromatin structure and dynamics) were removed. The CDSs were then examined by comparison with the latest release of the reference Virulence Factor Database (VFDB) [115] and keyword searching using the following words and their variants: virulence, toxin, hemolysin/hemolysis, pathogenicity, adherence, invasion, secretion, phagocytosis, phase variation, stress, iron uptake, siderophore, resistance, efflux pump, damaging and regulation. For drug resistance, CDSs that were classified to COG V (defense mechanisms), COG Q (secondary metabolites biosynthesis, transport and catabolism), and COQ M (cell wall/membrane/envelope biogenesis) were manually annotated for identification of antibiotic resistance-related genes. CDSs from other COGs were searched for additional genes using keywords: resistance antibiotic, efflux, multi etc. Prophages were identified by Prophage finder http://bioinformatics.uwp.edu/~phage/ searches [116]. The genome was run under the parameters with an e-value of 0.01, hits per prophage of 7, and hit spacing of 5000. Transposases were identified by performing BlastP analyses for all CDSs identified in the genome of L. hongkongensis HLHK9 against the ISfinder database http://www-is.biotoul.fr/is.html[117] and inverted repeats by einverted (EMBOSS package) [118]. Manual confirmation of the assigned function was performed by sequence similarity search using BLAST against the NCBI nr database, and assisted by conserved domain search (CD-search), identification of signature sequence motifs and sequence analysis using InterProScan. Localization patterns of putative virulence factors were predicted using PSORTb where appropriate [119].