Multidrug Efflux Systems in Helicobacter cinaedi

Helicobacter cinaedi causes infections, such as bacteremia, diarrhea and cellulitis in mainly immunocompromised patients. This pathogen is often problematic to analyze, and insufficient information is available, because it grows slowly and poorly in subculture under a microaerobic atmosphere. The first-choice therapy to eradicate H. cinaedi is antimicrobial chemotherapy; however, its use is linked to the development of resistance. Although we need to understand the antimicrobial resistance mechanisms of H. cinaedi, unfortunately, sufficient genetic tools for H. cinaedi have not yet been developed. In July 2012, the complete sequence of H. cinaedi strain PAGU 611, isolated from a case of human bacteremia, was announced. This strain possesses multidrug efflux systems, intrinsic antimicrobial resistance mechanisms and typical mutations in gyrA and the 23S rRNA gene, which are involved in acquired resistance to fluoroquinolones and macrolides, respectively. Here, we compare the organization and properties of the efflux systems of H. cinaedi with the multidrug efflux systems identified in other bacteria.


Introduction
Helicobacter cinaedi is a motile, Gram-negative, spiral bacterium belonging to the enterohepatic group of Helicobacter species of genus Helicobacter (the other group consists of gastric Helicobacter species, whose most well-known representative is the infamous H. pylori) [1]. During the last two OPEN ACCESS decades, this bacterium has increasingly been recognized as a human pathogen that causes infections such as bacteremia, diarrhea and cellulitis in mainly immune-compromised patients and occasionally in immunocompetent ones with a high potential for recurrence [2,3]. A possible association between H. cinaedi and atrial arrhythmias and atherosclerosis was also suggested [4]. This pathogen grows slowly over several days on blood agar, even at its optimal conditions, such as a wet microaerobic atmosphere at 37 °C , and often appears as a swarming thin film that is difficult to observe [1,5]. Therefore, it is often problematic to isolate, detect and sub-culture [5,6]. Antimicrobial chemotherapy has been used successfully to treat such infections, but prolonged courses of multiple antimicrobials for at least 2-3 weeks may be required [1]. Recently, molecular epidemiological analysis in Japan showed that all H. cinaedi isolates since 2000 had acquired resistance to clarithromycin (macrolides) and ciprofloxacin (quinolones), for which the MIC 90 (μg/mL) was >128 and 128, respectively, and contained typical mutations in gyrA and the 23S rRNA gene, respectively [7,8]. Unlike H. pylori, enteric Helicobacter species, such as H. cinaedi, are intrinsically resistant to amoxicillin (penicillin) [9]. High-level resistance and intrinsic resistance often require the presence of endogenous multidrug efflux pumps [10,11], which have not yet been analyzed in H. cinaedi.

Antibacterial Resistance Revealed by the Complete Genome of H. cinaedi
Very recently, we announced the complete genome sequence of H. cinaedi PAGU 611 isolated from a case of human bacteremia in Japan [22]. The clinical microbiological aspect of this strain was described as H. cinaedi-case 1; strain 923 [3]. Three months after our original report, another group published the sequence of the strain ATCC BAA-847, which was isolated in the 1980s in the USA [23]. The genome sequence of H. cinaedi CCUG 18818, although just a whole genome assembly and not complete, is also available from the Human Microbiome Project [24]. H. cinaedi PAGU 611 had a threonine to isoleucine mutation at position 84 of GyrA and adenine to guanine at position 2060 in PAGU 611 and ATCC BAA-847 (position 2018 in CCUG 18818) in the 23S rRNA gene, both of which are the same mutations identified in recent ciprofloxacin-and clarithromycin-resistant H. cinaedi isolates in Japan [7,8]. In addition to the slow, poor and, sometimes, failed growth described above, genetic tools for H. cinaedi are not sufficiently developed to take full advantage of the wealth of information generated by genome sequencing and to elucidate the function of unknown genes identified through sequencing. Fortunately, gene replacement via homologous replacement in H. cinaedi is possible by electroporation; however, no complementation system, e.g., a plasmid vector, is currently available for this organism [25]. We identified 10 putative drug transporter genes (2 RND, 1 MF, 2 MATE, 1 ABC, 4 SMR) in the genome of H. cinaedi PAGU 611 [22] (Figure 1). All transporters have homologues in H. hepaticus ATCC 51449, while only two-fifths are in H. pylori 26695 (Table 1). Interestingly C. jejuni subsp. jejuni NCTC 11168 has, rather, the most homologues (Table 1). Here, we compare the organization and properties of the multidrug efflux systems of H. cinaedi with the characterized and uncharacterized pumps available in the database.

RND Efflux Gene Operons of H. cinaedi
We identified two open reading frames (ORFs) belonging to the hydrophobe/amphiphile efflux-1 (HAE1) sub-family [26] of the RND family (locus-tags HCN_0595 and HCN_1563) encoded in the 2.08 Mbp chromosome of H. cinaedi PAGU 611 ( Figure 1). One consists of three genes (HCN_0593-HCN_0594-HCN_0595) that encode OMP, MFP and RND, respectively, and the other consists of two genes (HCN_1564-HCN_1563) that encode MFP and RND, respectively. The ORFs were obtained from the chromosomes of ATCC BAA-847 and CCUG 18818. Both a three-gene operon (MFP, RND, and OMF) and a two-gene operon (MFP and RND) are genetically common as a multidrug efflux operon, while the latter is functionally associated with an OMF component that is encoded by a separate gene that is physically unattached to the other two members on the chromosome. For example, in P. aeruginosa PAO1, mexAB-oprM and mexXY encode two multidrug efflux pumps (MexAB-OprM and MexXY-OprM, respectively) and contribute to natural antimicrobial resistance [27]. However, three-gene RND-type multidrug efflux operons (e.g., mexAB-oprM of P. aeruginosa [28] and cmeABC of C. jejuni [29]) are usually in the order MFP-RND-OMF, unlike H. cinaedi, H. pylori and H. hepaticus [9,30]. Chromosomal positions of drug efflux genes coding for putative inner membrane efflux transporters (red), outer membrane proteins (green), membrane fusion proteins (orange), and cytoplasmic proteins (light blue) are indicated by the kb (kilobase pair) in the H. cinaedi PAGU 611 genome [22]. Arrows correspond to the lengths and directions of the genes.
COBALT analysis [45] of representative RND pumps in Gram-negative bacteria, including all RND pumps from P. aeruginosa PAO1 and E. coli K12, characterized their relationships, and we focused on two branches containing the two RND pumps of H. cinaedi (Figure 2). The branch belonging to the HCN_0595 pump only includes HefC of H. pylori and CmeF of C. jejuni, while the branch containing the HCN_1563 pump includes not only CmeB of C. jejuni, but also the BepE/G pumps of B. suis and TtgB of P. putida ( Figure 2).
Taken together, we assume that the HCN_0595 pump of H. cinaedi plays a similar role to HefC of H. pylori and CmeF of C. jejuni, while the HCN_1563 pump has a similar role as CmeB of C. jejuni. In addition, the two pumps of H. cinaedi must play very similar roles to those of H. hepaticus, which is not surprising, according to their biological and genomic similarities. Recently, HefA (HH0224), the OMF component of HefABC of H. hepaticus ATCC 51449, was shown to be involved in resistance to amoxicillin and some antimicrobials, as well as bile acids [9]. As the authors failed to isolate a mutant RND pump (HH0222 (HCN_0595 orthologue) and HH0174 (HCN_1563 orthologue) in Figure 2), we do not know if the resistance to amoxicillin and bile acids is caused by HH0222 or HH0174, because the HH0174 gene is a two-gene operon, like the HCN_1563 gene [9]. It is noteworthy that the HefC pump of H. pylori played a role in cholesterol-dependent resistance in the bile salt-rich enterohepatic environment [30]. Cholesterol enhanced H. pylori resistance to various antibiotics, such as clarithromycin, amoxicillin and ciprofloxacin, as well as bile salts (e.g., deoxycholate) [30,46]. It is intriguing to determine if H. cinaedi resistance is enhanced by cholesterol and if the RND pumps of H. cinaedi play a role in cholesterol-dependent resistance. Actually, hefABC of H. hepaticus and H. pylori and cmeABC of C. jejuni were inducible by bile acids [9,30,47]. It is of note that CmeABC of C. jejuni plays a critical role in colonization in vivo [48].

A Possible Regulator Gene of Multidrug Efflux Systems in H. cinaedi
Although cognate regulators (e.g., repressors, activators, or two-component systems) located upstream of the RND efflux genes often exist, no cognate regulator was found upstream or downstream of the RND efflux operons of H. cinaedi, H. hepaticus and H. pylori. In C. jejuni, cmeR, which is a transcriptional repressor located immediately upstream of the cmeABC operon, encodes a 210 amino-acid protein that shares sequence and structural similarities with the members of the TetR family of transcriptional repressors [49]. BLAST analysis did not identify a homologue of CmeR in the genomes of Helicobacter species. Actually, H. cinaedi possesses only a small set of genes encoding transcriptional regulators, very similar to H. hepaticus [22,38].
Very recently, CosR, an oxidative stress responsive global regulator essential for viability [50], was shown to regulate the cmeABC operon negatively by binding directly upstream of cmeABC in C. jejuni NCTC 11168 [51]. CosR homologues are found mostly in -proteobacteria [51]. BLAST analysis showed that a quite similar CosR homologue (HCN_1079, YP_006235418.1) exists in H. cinaedi PAGU 611 (74% (86%) identity (positive)) and the strains ATCC BAA-847 and CCUG 18818. This homologue might also be involved in the expression of an efflux gene in H. cinaedi. In Gram-negative bacteria, oxidative stress responses are linked to the development of antimicrobial resistance, resulting from the activation of a resistance mechanism in which the RND multidrug efflux system is an important component [52]. For example, exposure to reactive oxygen species, such as peroxide, leads to MexXY-dependent aminoglycoside resistance in P. aeruginosa [52,53]. We point out that the putative start codon of all CosR homologues (HCN_1079, HCBAA847_0895, and HCCG_01220) of the H. cinaedi strains is TTG, which is a minor start codon [54], and found that an ATG codon located 3 codons before this TTG is also a possible start codon that is preceded by ribosome binding site-like sequences [55].

Other Probable Drug Efflux Systems in H. cinaedi
Finally, we discuss other probable drug efflux systems found in H. cinaedi, although the clinical significance and natural function of their homologues in other characterized bacteria remain unknown.

B
One ABC family efflux system was found in H. cinaedi PAGU 611 (Figure 1). It consists of four genes (HCN_0962-HCN_0963-HCN_0964-HCN_0965) encoding an inner membrane transporter, ATP binding protein, MFP and OMF, respectively, which means that it is an ABC transporter that spans the entirety of the Gram-negative cell envelope. The same efflux system was observed in the two other H. cinaedi strains. BLAST analysis showed that HH1856 (NP_861387) of H. hepaticus ATCC 51449 was a strong homologue of HCN_0962 (87% (93%) identity (positive)) ( Table 1). BLAST analysis with E. coli K12 and P. aeruginosa PAO1 suggested that HCN_0962 was significantly similar to the inner membrane domains of both MacB (NP_415400; 34% (56%) identity (positive)) of E. coli K12 and PvdT (33% (54%)) of P. aeruginosa PAO1. MacB and PvdT are inner membrane components of the macrolide-specific ABC transporter MacAB of E. coli [19] and of the de novo synthesized pyoverdine secretion system PvdRT-OpmQ of P. aeruginosa, respectively [65].

Future Perspective
The genome of H. cinaedi possesses probable uncharacterized drug efflux systems consisting of two RND pumps, one MF pump, two MATE pumps, two SMR pumps and one ABC pump, all of which are very similar to those of H. hepaticus. Because multidrug efflux pumps have roles in not only bacterial drug resistance, but also in other systems, including virulence and the stress response [52,63], characterizing the multidrug efflux pumps of H. cinaedi should lead to the understanding of various physiological aspects of this organism and, ultimately, conquering H. cinaedi infections. To do so, it is necessary to develop genetic tools and improve the culture method for this organism, while we can also use multiplex technologies, such as real-time PCR, DNA microarrays, proteomics and metagenomics. In the meantime, each pump can be cloned and characterized in organisms that lack a homologue, such as E. coli, C. jejuni and H. pylori, but some uncertainties will remain. Interestingly, H. cinaedi PAGU 611, but not ATCC BAA-847, possesses one plasmid, pHci1 (~23 kbp, 29 predicted coding sequences, of which 27 are hypothetical proteins) [22,23]. As such, it may represent a diamond in the rough that can be developed into a stable shuttle vector, although no replication protein or origin of replication have yet been found in this plasmid.