Distribution of genes related to Type 6 secretion system and lipooligosaccharide that induced ganglioside mimicry among Campylobacter jejuni isolated from human diarrhea in Thailand

Background Campylobacter jejuni (C. jejuni) is one of the most common bacteria responsible for human gastroenteritis worldwide. The mode of human transmission is foodborne infections due to consumption of contaminated food, especially poultry. Type 6 secretion systems (T6SS) were described recently as Campylobacter virulence mechanisms. Furthermore, infection sequelae associated with neurological disorders like Guillain–Barré (GBS) and Miller Fisher (MF) syndromes can become serious health problems in some patients after Campylobacter gastroenteritis. Our objective was to determine the distribution of these virulence genes among C. jejuni isolated from stool of human diarrhea. Methods A total of 524 C. jejuni strains from travelers and pediatric cases of acute diarrhea in Thailand were selected for this study. All isolates belonged to one of 20 known capsule types and all were assayed by PCR for T6SS, a hemolysin co-regulated protein (hcp) gene, and GBS-associated genes (cgtA, cgtB, cstIIHS19 and cstIIHS2) which are involved in sialic acid production in the lipooligosaccharide (LOS) cores of C. jejuni. The distribution of these genes are summarized and discussed. Results Of all isolates with these 20 capsule types identified, 328 (62.6%) were positive for hcp, ranging from 29.2 to 100% among 10 capsule types. The GBS-associated LOS genes were detected among 14 capsule type isolates with 24.4% and 23.3% of C. jejuni isolates possessed either cstIIHS19 or all three genes (cgtA, cgtB and cstIIHS19), which were classified as LOS classes A and B whereas 9.2% of C. jejuni isolates possessing cstIIHS2 were classified as LOS class C. The C. jejuni isolates of LOS A, B, and C together accounted for 56.9% of the isolates among 14 different capsule types while 31.1% of all C. jejuni isolates did not possess any GBS-associated genes. No significant difference was detected from C. jejuni isolates possessing GBS-associated LOS genes among travelers and children, but changes between those with hcp were significant (p < 0.05). Conclusions Our results suggested a high diversity of hcp and GBS-associated LOS genes among capsule types of C. jejuni isolated from Thailand.


Introduction
Campylobacter jejuni is recognized as a major pathogen of gastroenteritis worldwide [1]. As illustrated by the latest epidemiological report, Campylobacter is the main cause of foodborne illness in Europe and the United States [1]. Campylobacteriosis cases range from 14 to 57 per 100,000 per year (US and Europe, respectively). In endemic regions, particularly in Southeast Asia, campylobacteriosis cases are estimated to be 10 times higher than in Europe and the USA. This high incidence of Campylobacter is well documented in children, travelers, and foreign military personnel from Thailand [2][3][4]. C. jejuni is considered a zoonotic disease. The major source of contamination is through consumption of improperly prepared or stored foods containing poultry, but other sources like unpasteurized milk or water have been documented. The infectious dose can be as low as 500 to 1000 bacteria [5]. Campylobacteriosis symptoms range from mild abdominal pain and mild to no diarrhea, to severe abdominal cramping, sometimes accompanied with fever, headache, myalgia, and large volumes of mucous and bloody diarrhea that can last for several days [6].
Despite a worldwide prevalence, there is a paucity of data regarding C. jejuni virulence factors. Nevertheless, flagella, cytolethal distending toxin (cdt), fibronectin binding protein (cadF), lipoprotein (ceuE) [7][8][9], and proteins involved in adherence and invasion acquired by plasmids (pVir) [10] contribute to virulence. The Campylobacter polysaccharide capsule (CPS) is the best characterized virulence factor. Mutants deficient in CPS production demonstrated lower adherence and binding in vitro, decreased serum resistance, and reduced colonization of ferret and chicken animal models [7,11]. Phenotypic assays can be routinely applied for identification and classification of putative C. jejuni samples in laboratory settings [12]. Penner's serotyping, developed in 1980s, is a well-known serotyping scheme for C. jejuni and C. coli that is based on capsular polysaccharide [13]. Molecular assays have replaced this method for routine C. jejuni and other Campylobacter spp. identification. Analyses of different CPS sequences were utilized to develop multiplex PCR assays to distinguish more capsule types among C. jejuni isolates [14][15][16][17]. This capsule typing scheme correlates well to Penner serotypes because the capsule structure is the major serodeterminant [14].
In addition to the virulence factors aforementioned, additional C. jejuni isolates were characterized and found to possess the recently described T6SS [18,19]. The T6SS apparatus is composed of proteins which structurally and functionally are related to contractile components of bacteriophages [20]. The hemolysin co-regulated protein (Hcp) in Pseudomonas aeruginosa T6SS resembles the major component of the T4 phage tail base plate [21]. The T6SS contributes to bacterial pathogenesis in Pseudomonas aeruginosa [22], Vibrio cholerae [23], Salmonella enterica [24], Helicobacter hepaticus [25], Edwardsiella spp. [26], Burkoderia mallei [27], as well as in C. jejuni [19]. Campylobacter T6SS plays an important roles in host cell adhesion, invasion, and persistent colonization in vivo [19]. T6SS was reported in C. jejuni isolated from clinical, poultry, and water sources, and its distribution varied in human and animal sources from different countries [28][29][30]. However, the distribution and role of T6SS pathogenesis in C. jejuni from human gastroenteritis has yet to be further determined.
A myriad of post infectious sequelae were linked to C. jejuni infections. These include autoimmune mediated Guillain-Barré (GBS) and Miller Fisher (MF) syndromes [31]. These autoimmune diseases are provoked by mimicry of C. jejuni lipooligosaccharides (LOS) structures that contain N-acetyl neuraminic (sialic acid), organized into structures that resemble human gangliosides [31]. Most C. jejuni strains contain genes for the endogenous synthesis of N-acetyl neuraminic acid, and there are specific genes involved in the biosynthesis of these ganglioside mimics [32]. The cgtA (β-1,4-N-acetylgalactosyl transferase) and cgtB (β-1,3-galactosyltransferase) and cstII (α-2,3 sialyltransferase) are required for molecular mimicry and are associated with human GBS. These genes were used as markers to screen for GBS-related strains of C. jejuni [33][34][35]. Multiple variants of cstII transferase were characterized in the C. jejuni LOS locus, including cstII HS19 and cstII HS2 [33]. The gene cstII has mono and bi-functional activities (α-2,3 and/or α-2,3/α-2,8-sialyltranferase), and both represent different specificity of enzymes involved in the transfer of the sialic acid residue to the LOS structure [32,36]. Genes in LOS biosynthesis loci were studied from different origins of C. jejuni, and the LOS classes A, B, and C were classified for involvement in human ganglioside mimicry. Recently, LOS classes M and R were reported to possess genes Conclusions: Our results suggested a high diversity of hcp and GBS-associated LOS genes among capsule types of C. jejuni isolated from Thailand.
The aim of this study was to determine the distribution of genes associated with T6SS and LOS implicated in GBS among different capsule types in clinical C. jejuni isolated from stool samples of travelers and children suffering diarrhea in Thailand.  [17]. The studies used archived frozen C. jejuni isolates with appropriate consent for sample donation and future research use. All links or keys to subject name, number, and personal identifiable information were destroyed. All selected C. jejuni isolates were stored frozen at − 70 °C at AFRIMS.

Growth and genomic DNA purification of C. jejuni isolates
Campylobacter jejuni were subcultured on sheep blood agar plates, using aliquots of the frozen isolates. The plates were incubated under microaerobic conditions (85% N 2 , 10% CO 2 , and 5% O 2 ) at 42 °C overnight. Genomic DNA extraction from C. jejuni isolates were performed using DNeasy tissue extraction kits (Qiagen, MD, USA). The quantity and quality of DNA was estimated using a NanoDrop (Thermo scientific, MA, USA) and agarose gel electrophoresis. Gels images were visualized using ethidium bromide and a gel documentation system (Syngene, Cambridge, UK).

PCR for C. jejuni capsule types
Primer sets specific for C. jejuni capsule types were used according to previously published articles [16,17]. In brief, four primer sets (35 capsule primer pairs) recognized a total of 47 Penner serotypes and complex types. All 524 C. jejuni genomic DNA were analyzed with four multiplex PCR sets (Alpha, Beta, Gamma, and Delta). Multiplex PCR and step cycles followed previously reported standardized procedures [16,17]. Primer pairs for the 331 bp lpxA gene (involved in lipid A biosynthesis) were used as an internal control for C. jejuni. All PCR amplification products were evaluated using a 2.0% agarose gel (Thermo Fisher Scientific -Invitrogen, MA, USA) in 0.5× Tris-Borate EDTA buffer at 150 volts for 60 min. The DNA bands were visualized and photographed under UV light after the gel was stained with ethidium bromide. The amplification products were interpreted accordingly as described in [16].

PCR detection for T6SS gene
An hcp sequence from C. jejuni T6SS was obtained from GenBank, accession number JX436640. Primers specific to the hcp gene were designed in this study using Primer3 software [38]. The hcp forward (hcp_1) and reverse (hcp_2) primer sequences are 5′-CAA ATG CGC AAG AGT CAA GT-3′ and 5′-TAA GCT TTG CCC TCT CTC CA-3′, respectively. Primers specific for lpxA was included in these PCR as an internal control. The DNA extracts of C. jejuni ATCC 43431 and clinical isolates BH-01-142 were used as positive controls for hcp [19]. The negative control strains were C. jejuni NCTC 11168 and 81-176 [18,28]. Multiplex PCR was performed on a Nexus GX2 thermal cycler (Eppendorf, NY, USA) using AmpliTaq Gold DNA Polymerase (Thermo Fisher Scientific, MA, USA). The PCR step cycles for hcp were 1 cycle for 94 °C for 5 min, 28 cycles for 94 °C for 1 min, 52 °C 1 min, 72 °C for 1 min, and a final cycle at 72 °C for 10 min. A 100 bp DNA ladder (Thermo Fisher Scientific, MA, USA) was applied as a molecular marker. All PCR amplification products were evaluated on a 2.0% agarose gel (Invitrogen) at 150 volts for 60 min. The DNA bands were visualized and photographed under UV light after the gel was stained with ethidium bromide. The estimated size of the hcp amplified products was 133 bp.

PCR detection for GBS-associated LOS genes
Four primer pairs specific for the LOS genes implicated in ganglioside mimicry: cgtA (β-1,4-N-acetylgalactosyltransferase), cgtB (β-1,3galactosyltransferase), cstII (α-2,3 sialyltransferase, and/or cstII (α-2,3/α-2,8 sialyltransferase), were used, as previously described [35]. Multiplex PCR was performed on a thermal cycler using AmpliTaq Gold DNA Polymerase. The PCR step cycles were: 94 °C for 5 min, 94 °C for 1 min, 53 °C (cgtA and cgtB, cstII HS19 and cstI-I HS2 ) for 1 min, and 72 °C for 1 min for 28 cycles, with a final 10 min extension at 72 °C. Positive controls for the GBS-associated LOS genes in PCR were genomic DNA from C. jejuni ATCC 43432 (LOS-A1), ATCC 43438 (LOS-A2), ATCC 43456 (LOS-B1), ATCC 43439 (LOS-B2), NCTC 11168, ATCC 43429 (LOS-C), LIO 87 (LOS-D), and ATCC 43434 (LOS-E) to determine LOS classes A-E in C. jejuni as previously described [32,39]. The internal PCR control was C. jejuni specific lpxA. All PCR amplification products were applied and evaluated on a 3.0% agarose-1000 gel (Invitrogen) at 150 volts for 100 min. A 100 bp DNA ladder was used as a molecular marker. The DNA bands were visualized and photographed under UV light after the gel was stained with ethidium bromide. The sizes of amplified products from cgtA, cgtB cstII HS19 , and cstII HS2 were 527, 502, 400, and 417 bp, respectively. Primers of the four genes, and one internal control gene herein distinguished control isolates as follows: ATCC 43432 and ATCC 43456, representing LOS classes A1 and B1, were positive for all three genes cgtA, cgtB, and cstII HS19 whereas the control strains ATCC 43438 and ATCC 43449 representing LOS classes A2 and B2 were only positive for gene cstII HS19 . Two control strains, NCTC 11168 and ATCC 43429 both representing LOS class C, were positive for only cstII HS2 . Two control strains LIO 87 and ATCC 43434 representing LOS classes D and E, respectively were negative for all four genes studied. Thus, C. jejuni isolates possessing cstII HS19 or all three genes (cgtA, cgtB, and cstII HS19 ) were characterized as either LOS class A or class B, whereas isolates positive for only cstII HS2 were classified as LOS class C. Isolates negative for all genes were classified as non-LOS classes A, B, or C.

Sequencing of PCR amplified products and analysis
Amplification products of selected clinical C. jejuni isolates positive for hcp, cgtA, cgtB, cstII HS19 , and cstII HS2 were purified for sequencing using the Wizard SV gel and PCR clean-up system (Promega, WI, USA). After purification, the amplified products for hcp and the four GBSassociated LOS genes were submitted for sequencing (Macrogen, South Korea). The amplified products of hcp, cgtA, cgtB, cstII HS19 , and cstII HS2 were edited and assembled by Sequencher version 5.0 (Gene Codes Corporation, MI, USA). The nucleotide sequences were analyzed with BLAST (http://www.ncbi.nlm.nih.gov/BLAST ).

The association of C. jejuni capsule types, hcp, and GBS-associated LOS genes
The distribution of hcp and GBS-associated LOS genes detected in C. jejuni isolated from travelers and child diarrhea cases were tested for statistical differences using Fishers exact test from GraphPad software [40]. A p value < 0.05 was considered statistically significant.

Association of capsule types of C. jejuni isolates with hcp and GBS-associated LOS genes
In the line graph display, three C. jejuni isolates capsule types: HS2, HS8/17, and HS15 possessed only cstII HS2 , representing LOS class C as 32.3%, 37.5%, and 27.8%, respectively. These C. jejuni isolates from the three capsule types also harbored hcp, whereas C. jejuni isolates from HS2, HS3, HS1/44, HS53, and HS15 possessed both LOS classes A, B, or C.

Discussion
In this study, HS4 and HS2 were the most common C. jejuni capsule types from traveler and pediatric diarrhea cases, respectively. These two capsule types were the two most common Penner serotypes in global human C. jejuni isolates, confirming previous reports [41]. Other capsule types exhibited varied distribution among isolates from travelers and children. The study of traveler's diarrhea isolates included C. jejuni, of which > 80% were obtained from stool samples of military personnel attending annual exercises from different locations in Thailand. A point-source of infection of an identical C. jejuni clone isolated from the same type of food consumed might attribute a distribution of one capsule type over the others (e.g., HS53, HS23/36). The different capsule types seen in travelers may be affected by temporal variations during the exercises of military personnel, and by the contaminated food consumed by travelers at the visiting sites (e.g., resort or hotel), but this was not seen in isolates from children in which samples were collected over a 2 year period. Although C. jejuni isolates from these two studies were from more than 5 years apart, the five most common capsule types from traveler isolates (HS4, HS2, HS5/31, HS8/17, and HS3) and Thai children isolates (HS2, HS 8/17, HS4, HS5/31, and HS3) were comparable to capsule serotypes of global C. jejuni isolates reported previously [41].
T6SS are bacteriophage-like structures which secrete effector molecules into eukaryotic cells or other bacteria    [20,42]. The presence of a secreted effector, the hcp needle protein, in culture supernatants suggested contribution of this protein to virulence of C. jejuni strains [18,19]. The hcp gene was used as a surrogate for the presence of T6SS in C. jejuni. Using this screening method on different C. jejuni isolates, heterogeneous results were reported. Harrison et al. [28] demonstrated the prevalence of hcp in 60.6% and 33.3% from clinical C. jejuni isolates in Vietnam and Thailand, respectively. However, only three human C. jejuni isolates from Thailand were studied. None of the T6SS genes were detected in 34 C. jejuni isolated from human diarrheal cases in Pakistan [29]. In our study, hcp was found in 62.6% of the 524 cases, and this proportion was similar to studied isolates in Vietnam (60.6%) and Egypt (57.6%) [28,43].
The hcp gene was associated with 10 capsule types of C. jejuni isolates whereas a variable hcp prevalence was reported by others. Sainato et al. [43] reported a hcp gene prevalence of 33.3% in HS2 isolates, but up to 86.2% was noted in this study. Moreover, numbers of hcp-C. jejuni isolates were more significant in children than travelers (22.2%, 86.7%; p = 0.0001), especially in HS8/17 isolates. Different distributions of the hcp gene were also observed among HS4cpx isolates (28.5% of HS4 cpxB , 92.5% HS4 cpxA , and 100% of HS4 cpxAB ). Our study was limited in that we tested for only one gene marker in T6SS, while the T6SS locus composed of at least 13 ORFs [18]. A fully-functional T6SS was only reported in two C. jejuni strains in published articles [18,19]. Further investigation on more conserved genes in the T6SS locus and expression of their encoded proteins may implicate the contribution of T6SS to Campylobacter virulence.
Hcp was previously reported to be associated with bloody diarrhea [28]. We found no significant association between hcp and C. jejuni isolates of any specific capsule type. This is similar to the finding by Agnetti et al. [44], where no significant difference in the clinical manifestations and the course of disease between patient with hcppositive and -negative C. jejuni isolates. Similarly, the presence of hcp gene in C. jejuni and C. coli from stool did not have any significant association to clinical outcome in Egyptian children [43].
Campylobacter jejuni and their LOS classes could be distinguished by different LOS biosynthesis locus, as referenced by others [32,33,39]. Parker et al, reported orf locus 7ab and 7c to encode for cstII and cstIII, respectively. In our study, the primers for four genes, cgtA, cgtB, cstII HS19 and cstII HS2 , were used following Nachamkin, et al. [35]. The amplified products representing clinical C. jejuni of different capsule types, shown in Fig. 4, were sequenced and analyzed with BLASTn to identify each gene. The alignment matched within locus of cgtA, cgtB, cstII, and cstIII, respectively. In this study, we focused on LOS classes A to C that can cause sialylated LOS synthesis resulting in ganglioside mimicry, which are crucial for ganglioside mimicry from various C. jejuni strains, as previous reported [32,39,[45][46][47]. The GBS genes for LOS classes A, B are predominately amongst C. jejuni HS19 isolates [47]. All 524 C. jejuni isolates were associated with gastroenteritis (GI). CstII HS19 was previously reported in GBS-HS19 isolates more than in GBS-non HS19 isolates [35]. In this study, almost 50% (23.3 and 24.4% in Table 2) of the GI-isolates were classified as LOS classes A or B, but only one isolate was HS19 serotype. GBS was commonly associated with HS1/44-, HS2-, and HS4-C. jejuni antecedent infections in the Netherlands, whereas HS19-, HS 23/36-, and HS41-C. jejuni were associated with GBS in Bangladesh, Japan, and South Africa [47]. Our study demonstrated a high prevalence of HS1/44, HS2, HS23/36, and HS4 cpx , but the HS41 serotype was not detected from the GI-isolates. Distribution of LOS classes A, B, and C in C. jejuni has been previously demonstrated, as 17%, 32%, and 19%, respectively, but there are data linked to only 5 Penner serotypes from these C. jejuni isolates [47]. Parker et al. [39] reported that 64% of non-GBS associated isolates were in LOS classes (A, B, or C), and this is comparable to our study that 56.9% of C. jejuni isolated were in LOS classes A, B, and C. A limitation of our study was that the primers used for LOS associated GBS genes could not amplify gene targets to differentiate LOS subclasses A1, A2, B1 and B2, and other LOS loci diversity [37,39,48,49]. More primers specific to other LOS classes should be included to characterize these C. jejuni isolates in more depth, especially in 63 C. jejuni isolates with unclassified LOS locus.

Conclusion
This data provides information on the diversity of hcp and LOS genes implicated with GBS-associated C. jejuni capsule types in travelers and children, as compared to other studies. A high prevalence of T6SS and GBS-associated LOS genes was found in certain C. jejuni capsule types: HS4 cpxAB , HS4 cpxA , HS2, HS1/44, and HS10, which may suggest an increase fitness of these Campylobacter since most isolates contained hcp and LOS classes A, B, or C genes, which are crucial for survival in the host and promote the synthesis of sialylated LOS. However, variations in clonal strains, different biological LOS functions, as well as host and environment factors, may be involved in triggering an autoimmune response and need further study. The understanding of global capsule types, antigenic epitopes, and virulence determinants may contribute towards future development of targeted therapeutics against C. jejuni diarrhea.