The bacterial strains and plasmids used in this study are described in Table 1. H. pylori cells were grown on CDC anaerobic blood agar (BBL, Microbiology Systems, Cockeysville, MD) or in Brucella broth containing 10% (v/v) horse serum (Gibco BRL, Life Technologies, Rockville, MD) at 37°C in microaerophilic conditions (5% O₂, 10% CO₂ and 85% N₂). E. coli was grown on Luria-Bertani (LB) (BD Biosciences, San Jose, CA) agar or in broth. Bacteria harboring antibiotic resistance determinants were grown in the presence of the appropriate antibiotics at the following concentrations: ampicillin (Amp, 100 μg ml⁻¹); chloramphenicol (Cm, 25 μg ml⁻¹ for E. coli, 10 μg ml⁻¹ for H. pylori); kanamycin (Km, 50 μg ml⁻¹ for E. coli, 10 μg ml⁻¹ for H. pylori). All strains were stored at −80°C in Brain-Heart Infusion (BHI) broth (H. pylori) or LB broth (E. coli) containing 20% (v/v) glycerol until testing.

The human gastric carcinoma cell line AGS (purchased from ATCC; American Type Culture Collection, Manassas, VA, USA) was grown in Ham's F-12 medium (Invitrogen Life Technologies, Rockville, MD) supplemented with 10% (v/v) fetal bovine serum (Gibco BRL) in an atmosphere consisting of 5% CO₂ at 37°C. The human gastric epithelial immortalized GES-1 cells (a gift from Prof. Wei-Lun Chang, National Cheng Kung University Hospital) were grown in RPMI 1640 medium (Invitrogen Life Technologies) supplemented with 10% (v/v) fetal bovine serum in an atmosphere consisting of 5% CO₂ at 37°C.

Mini Qiagen columns and a QiaAmp DNA extraction kit (Qiagen, Valencia, CA, USA) were used for plasmid and chromosomal DNA extraction. PCR was carried out according to the manufacturer's instruction using Taq polymerase (Promega, Madison, WI, USA).

H. pylori cells were grown on CDC plates in microaerophilic conditions for 36 h, then transferred to 100 ml Brucella broth containing 10% (v/v) horse serum at an optical density (OD) of 0.2 and incubated with shaking (150 rpm, to reduce cell aggregation) for 18 h in microaerophilic conditions. Mini Qiagen columns and a Qiagen RNAeasy mini kit (Qiagen, Valencia, CA) were used for RNA extraction. Ribosomal RNA was depleted according to the manufacturer's instruction using Bacteria Minus™ Transcriptome Isolation Kit (Invitrogen Life Technologies). The Applied Biosystems SOLiD™ Total RNA-Seq kit was used to generate the cDNA template library. The SOLiD™ EZ Bead system (Invitrogen Life Technologies) was used to perform emulsion clonal bead amplification to generate bead templates for SOLiD™ platform sequencing. Samples were sequenced on the 5500XL SOLiD™ platform. The 50-base short read sequences produced by the 5500XL SOLiD™ sequencer were first run through SOLiD Accuracy Enhancement Tool (SAET) to improve color call accuracy, then were mapped in color space using SOLiD™ LifeScope™ software version 2.5 using default parameters against the H. pylori J99 reference genome (NCBI accession number, NC_000921). The BAM file from LifeScope™ was performed the analysis of gene expression with Partek software package. The gene expression from each sample was then tested for statistical differences using one-way ANOVA at 5% confidence level. The complete set of RNA-seq files has been deposited in Gene expression omnibus (GEO), NCBI (https://www.ncbi.nlm.nih.gov/geo/, accession number GSE95006). The fold change of each gene was measured as the mean of three independent experiments. Gene expression of the csrA mutant compare to wild-type J99 with a fold change > 1.5-fold was selected and confirmed by RT-qPCR.

H. pylori cells were grown on CDC plates in microaerophilic conditions for 36 h, then transferred to 100 ml Brucella broth containing 10% (v/v) horse serum at an OD of 0.2 and incubated with shaking (150 rpm) for 18 h in microaerophilic conditions. RNA extraction and reverse transcription PCR were described previously (Kao et al., 2012b). Thirty microliters of culture media were centrifuged at 1,000 × g for 5 min at 4°C and then washed with ice-cold 0.2 M sodium acetate buffer (pH 5.5) twice. The bacterial pellet was then re-suspended in 600 μl acetate buffer (20 mM sodium acetate, 1 mM EDTA, and 0.5% (w/v) SDS), and 600 μl acid-phenol (pH 4.5) was added to isolate bacterial RNA. The sample was incubated at 65°C for 10 min, and centrifuged at 12,000 × g for 10 min to collect the supernatant. After isopropanol precipitation, the sample was treated with DNase I (Promega) at 37°C for 2 h. Finally, phenol/chloroform was used to extract total RNA, and the sample was dissolved in diethylpyrocarbonate (DEPC)-treated deionized water and stored at −80°C until used. The RNA was quantified at an absorbance of 260 nm. Random hexamers (Mission biotech, Taiwan) and MMLV reverse transcriptase (Promega) were used to generate cDNA from 1 μg of total RNA, and the cDNA was stored at −20°C until testing.

The primers used for RT-qPCR are listed in Table S1. RNA quantification was carried out by RT-qPCR with a KAPA PROBE FAST Universal 2 x qPCR Master Mix (KAPA Biosystems Inc., Woburn, MA) specifically adapted for one-step RT-qPCR in glass capillaries using a Light Cycler instrument (Roche Diagnostics, Indianapolis, IN). Cycling conditions were as follows: activation of the polymerase for 10 min at 95°C, followed by 40 cycles of denaturation at 95°C for 20 s, annealing at 60°C for 1 min, and elongation at 72°C for 15 s. Fluorescence was detected at the end of each extension step, and the Cp values were calculated by the LightCycler 1.5 software.

The primers used in this study are listed in Table S1. In order to construct a revertant strain from rpoN mutant J99 (SW837), the plasmid containing the rpoN fragment, pMW758 (Kao et al., 2014), was transformed into SW837 to generate a revertant strain containing the wild-type rpoN. In brief, after natural transformation (Haas et al., 1993), H. pylori was grown on CDC anaerobic blood agar without antibiotic for 3 days. Colonies were picked and subcultured on CDC anaerobic blood agar without antibiotic and Brucella agar plate containing 10% (v/v) horse serum and Cm (10 μg ml⁻¹) at the same time and incubated at 37°C in microaerophilic conditions for 3 days. Colonies that lost the ability to grow on a Cm-containing plate (that only grew on CDC anaerobic blood agar) were considered as the revertant strain and were verified by PCR-sequencing and motility assay.

For flaB mutant construction, the 1,175 bp flaB fragment obtained from H. pylori J99 genomic DNA and PCR with flaB-Mut-1 and flaB-Mut-2 primers, was digested by PstI and KpnI and ligated to the plasmid pUC18 to generate plasmid pMW817. A chloramphenicol acetyltransferase cassette (cat cassette) containing 806 bp was obtained from plasmid vector 78, cut with HincII and inserted into plasmid pMW817 digested with HincII. Two plasmids were designated as plasmid pMW815 (flaB/cat in opposite transcriptional orientation) and pMW816 (flaB/cat in same transcriptional orientation), respectively, and transformed into J99. Cm (10 μg ml⁻¹) was used to select for the flaB mutants, SW859 and SW861, generated from chromosomal flaB double cross-over with plasmids pMW815 and pMW816, respectively. Plasmid pMW817 was then transformed into SW859 and SW861 to generate a revertant strains SW858 and SW860, respectively.

For jhp0106 mutant construction, the 1,004 bp jhp0106 fragment, obtained from J99 genomic DNA and PCR with jhp0106-Mut-1 and jhp0106-Mut-2 primers, was digested with EcoRI and ligated into the plasmid pUC18 to generate plasmid pMW801. A kanamycin resistance cassette (aph(3′)-III cassette) was obtained from plasmid vector pBHP489K, cut with ClaI and inserted into the plasmid pMW801 digested with BsmI. This plasmid was designated as plasmid pMW802 and transformed into J99. Km (10 μg ml⁻¹) was used to select for the jhp0106 mutant, SW863. In order to construct a jhp0106 revertant from SW863, the plasmid pMW801, was transformed into SW863 to generate SW862 containing wild-type jhp0106.

For flaA and flaA/flaB mutants construction, the 1,283 bp flaA fragment obtained from H. pylori J99 genomic DNA and PCR with FlaA-Mut-1 and FlaA-Mut-2 primers, was digested by PstI and KpnI and ligated to the plasmid pUC18 to generate plasmid pMW833. A aph(3′)-III cassette obtained from plasmid vector pBHP489K was inserted into plasmid pMW833 digested with AfeI. This plasmid was designated as plasmid pMW834 and transformed into J99 and SW861 (flaB mutant). Km (10 μg ml⁻¹) was used to select for the flaA mutant (SW866) and flaA/flaB double mutant (SW868).

The primers used for construction of the flaB promoter mutants and revertant are listed in Table S1. The 1,001 bp flaB promoter fragment obtained from H. pylori J99 genomic DNA and PCR with flaB-ProMut-1 and flaB-ProMut-2 primers, was digested by HincII and EcoRV and ligated to the plasmid pUC18 to generate plasmid pMW814. Primers flaB-ProMut-3 and flaB-ProMut-4 were used to amplify pMW814 to generate plasmid pMW813 containing a 46 bp partial deletion of the flaB promoter and carrying a NaeI site in the flaB promoter fragment. A cat cassette obtained from plasmid vector 78 was inserted into plasmid pMW813 digested with NaeI. This plasmid was designated as plasmid pMW812 and transformed into J99. Cm was used to select for the flaB promoter-cat mutant, SW856. In order to construct flaB promoter mutants with GG or CG/GG nucleotide mutations, primer pairs flaB-ProMut-7/flaB-ProMut-8 and flaB-ProMut-9/flaB-ProMut-10 were used to amplify pMW814 to generate plasmid pMW811 and pMW810, respectively. pMW814, pMW813, pMW811, and pMW810 were transformed into SW856 to generate SW857 (revertant), SW855 (deletion), SW853 (CG to AA/GG to AA), and SW854 (GG to AA), respectively.

The motility assay was described previously (Kao et al., 2014). Bacterial colonies were applied to one spot in the motility agar plate containing Brucella broth, 0.3% (w/v) Bacto agar, and supplemented with 10% (v/v) horse serum. The plates were incubated at 37°C under microaerophilic conditions for 7 days, and the motility was assessed by the diameter of migration of bacteria through the agar, from the inoculated center toward the periphery of the plate. The motility of each strain was measured as the mean of three independent experiments.

Complete amino acid sequence of H. pylori J99 Jhp0106 (Accession number: WP_001028953) was downloaded from NCBI database (http://www.ncbi.nlm.nih.gov/protein/). A three-dimensional structural modeling was carried out on the SWISS-MODEL Workspace server (http://swissmodel.expasy.org/) (Arnold et al., 2006). The structure representation figures are generated by the program PyMOL (http://www.pymol.org).

A grid covered with a carbon-coated parlodion film (300 mesh copper grid) was floated onto a 20 μl sample drop and left for 2 min for adsorption of the sample to the grid. The grid was then removed from the drop and floated on a drop of 1% (w/v) phosphotungstic acid (Sigma-Aldrich) and left for 1 min. Excess stain was removed by touching the edge of the grid to a piece of Whatman filter paper. All samples for electron microscopy were examined in a Hitachi H-7650 transmission electron microscope (Hitachi, Tokyo, Japan).

The assay was performed according to a previous study with modification (Kwok et al., 2002). AGS and GES-1 cells (1 × 10⁶/well) were grown overnight in 6-well culture dishes to approximately 80% confluence. H. pylori cells were added to the wells at a multiplicity of infection (MOI) of 100 without centrifugation and were incubated for either 30 min for the adhesion assay or 8 h for the cellular IL-8 production assay. For the adhesion assay, each dish with AGS-H. pylori coculture was washed three times with prewarmed phosphate-buffered saline (PBS) buffer to remove unbound bacteria. Adhered H. pylori were quantified by lysing the cells for 15 min with 0.1% (w/v) saponin-containing PBS buffer, followed by serial dilution and spreading on Brucella agar plates containing 10% (v/v) horse serum. The number of adhered bacteria was measured by the plate counts after 3 days incubation. For the cellular IL-8 production assay, the culture supernatants were collected and stored at −20°C until assayed. IL-8 concentration in the supernatant was determined by standard ELISA with commercially available assay kits according to the manufacturer's procedures (Arigo Biolaboratories Corp., Taiwan).

The Student's t-test and paired t-tests were applied as appropriate for the parametric differences. ANOVA was used for comparing groups of more than two strains. All tests of significance were two-tailed with a p value < 0.05 taken as significant.

To characterize the CsrA regulatory system, RNA-seq analysis on the wild-type J99 and its respective csrA mutant, SW835, was carried out. Three independent biological replicates were sequenced for each strain. The raw sequence output of the two strain transcriptomes included ~14 and ~15 million reads of the wild-type and csrA mutant, respectively. Approximately 80% of the reads were perfectly aligned to the J99 reference genome. Based on the genomic alignment, our analysis determined the expression of 1,559 genes in each strain.

In this study, genes found to be differentially expressed compared to the wild-type J99 (a > 1.5-fold change) by RNA-seq data were taken into consideration. In the csrA mutant, 53 chromosomally encoded genes were found to be differentially expressed compared to the wild-type J99 (Table 2). Of these genes, 94% (50 genes) were expressed at a lower level in the csrA mutant compared to the wild-type, with only 6% (3 genes) having higher expression in the csrA mutant (Table 2). These results indicate that CsrA acts primarily as a global positive regulator. RT-qPCR was further employed to validate the expression of 53 CsrA regulated genes in the wild-type and csrA mutant (Table 2). In agreement with the RNA-seq data, RT-qPCR confirmed the expression of genes differentially expressed in the csrA mutant compared to the wild-type J99 with a > 2-fold change (except jhp1296). Three genes, jhp1334, jhp1169 and jhp1132, originally identified with a >1.5-fold change by RNA-seq, showed inconsistent results between the RNA-seq and RT-qPCR methods (change <1.5-fold by RT-qPCR) (Table 2). These transcriptomic analyses also indicated that most of CsrA regulated genes (18 genes) were classified as genes encoding components involved in flagella formation, chemotaxis and motility, followed by genes with unknown function (15 genes) (Figure S1).

The CsrA-regulated jhp0106 gene was of interest, as it encodes a protein suggested to be involved in H. pylori strain 1016 flagellin glycosylation and motility (Schirm et al., 2003) (Table 3). The schematic diagram showing the pseudaminic acid biosynthetic pathway in H. pylori is shown in Figure S2. At present, the function of Jhp0106 in flagellin glycosylation is still unclear. We validated the transcription of pseB, pseC, pseH, pseG, pseI and jhp0106 in the wild-type, SW835 (csrA mutant) and SW836 (csrA revertant) strains by RT-qPCR. The results were consistent with the RNA-seq data. Only the expression of jhp0106 was dramatically reduced in SW835 compared to the wild-type J99 (Figure S3).

The genes close to jhp0106 in H. pylori J99 are shown in Figure 1A. jhp0106 was located immediately downstream of the flaB gene, which has been shown to be regulated by RpoN. The gene order and orientation of flaB-jhp0106 are conserved in H. pylori genomes. Therefore, we propose genes flaB and jhp0106 are in an operon controlled by RpoN binding to the flaB promoter. To validate this hypothesis, RT-qPCR was carried out to determine the expression of rpoN and jhp0106 in the wild-type, SW835 (csrA mutant), SW836 (csrA revertant), SW837 (rpoN mutant), and SW838 (rpoN revertant) strains. The results showed that the expression of rpoN was reduced in SW835 (Figure 1B), consistent with our previous study (Kao et al., 2014). Moreover, expression of jhp0106 mRNA was decreased to 24 and 8% in SW835 and SW837, respectively (p < 0.001). These results indicated that CsrA and RpoN positively regulate jhp0106 expression.

RT-PCR with different primer pairs was performed to confirm the co-transcription of flaB-jhp0106, and the results showed that 1.4- and 1.8-Kb transcripts were observed using primer pairs jhp0106-1/flaB-4 and jhp0106-3/flaB-4 for RT-PCR, respectively (Figure 2A). In addition, the expression of the flaB-jhp0106 transcript was reduced in SW837 (Figure 2A). The predicted RpoN binding site of the flaB promoter was examined by construction of flaB promoter mutants, as shown in Figure 2B. The mRNA levels of flaB and jhp0106 were dramatically reduced in SW837, SW853, SW854, SW855, and SW856, as determined by RT-qPCR (Figure 2C). We further investigated whether the RpoN binding site of the flaB promoter was conserved among different H. pylori strains, and the results showed that the RpoN binding sequence was identical among 17 examined strains (Figure S4). Our results indicated the transcription of jhp0106 is positively controlled by RpoN bound to the flaB promoter. However, these results could not exclude the possibility that the jhp0106 gene had its own promoter. Therefore, northern blotting was performed to determine the size and number of jhp0106 transcripts. Although several different fragments of the jhp0106 gene were used to serve as probes, the signal was still too weak to be detectable (data not shown).

In order to clarify the role of the flaB-jhp0106 operon in H. pylori J99 motility, soft-agar analysis was used to determine the motility of flaB promoter mutants (Figure 2D). SW837 was used as a non-motile negative control. The results showed that strains SW853, SW854, SW855, and SW856 exhibited deficient motility (Figure 2D). These results indicated that the flaB-jhp0106 operon was controlled by RpoN, and that it plays a critical role in H. pylori motility.

Josenhans et al. reported that when the flaB gene was disrupted in H. pylori strain N6, the motility decreased by 30 to 40% (Josenhans et al., 1995). To validate this observation in strain J99, we constructed flaB mutants SW861 and SW859 (Figure 3A). A cat cassette was inserted in the same (SW861) or opposite (SW859) direction to flaB transcription in wild-type J99 (Figure 3A). Strain SW861 showed decreased motility, and the quantified results revealed 83.8% motility ability compared with wild-type J99 (p < 0.05) (Figure 3B). In contrast, strain SW859 showed a non-motile phenotype, compared with the wild-type J99 (Figure 3B). To evaluate whether the deficient phenotype of SW859 was caused by the insertion of the cat cassette into the flaB gene, which interfered with downstream jhp0106 expression, RT-qPCR was performed to determine the expression of jhp0106 in the examined strains. The results showed that the transcription level of jhp0106 was decreased to 13% in SW859, but increased to 242% in SW861, compared to the wild-type (Figure 3C). The increase of jhp0106 expression in SW861 resulted from a cat cassette inserted into flaB in same transcriptional orientation. Thus triggering downstream jhp0106 transcription due to the leakage of the transcriptional terminator of the cat cassette. These results indicated that jhp0106 in flaB-jhp0106 operon, but not flaB, plays a critical role in H. pylori motility.

The NCBI protein database was used to search for proteins showing high similarity to Jhp0106, and the results are shown in Table 3. Among them, HMU07160 (62%) showed the highest identity to Jhp0106, followed by WS2199 (55%), NAMH_1610 (44%), Sdel_2228 (42%), and CMTB2_07872 (41%) (Table 3). The phylogenic tree based on the homologous full-length sequence of 16 proteins was also displayed in Figure S5. Previous studies reported that Maf4, PseD and PseE were involved in flagellin glycosylation in C. jejuni (McNally et al., 2006; van Alphen et al., 2008), and they also displayed high identity to Jhp0106 (Table 3).

The attempt to determine the crystal structure of Jhp0106 was unsuccessful, in spite of extensive efforts to crystallize the recombinant Jhp0106 protein from an E. coli expression system. Therefore, we performed computational modeling of the Jhp0106 protein to gain structural insights into Jhp0106 function. A 3D-model for Jhp0106 of 227 amino acid residues in length (from Asp231 to Phe457) predicted by the SWISS-MODEL server is shown in Figure 4A. The overall folding of the Jhp0106 structure is similar to an alpha-2,3/8-sialyltransferase CstII from C. jejuni in complex with a substrate analog, CMP-3FNeuAc (PDB: 1R07) (Figure 4B) (Chiu et al., 2004), suggesting Jhp0106 is a glycotransferase.

The protein interaction networks for Jhp0106 showed that the protein interacts with flagella structure proteins (FlaA, FlaB, FlaG, FlaL, FliD), pseudaminic acid synthase (NeuB), CagDelta protein (Jhp0417), hypothetical proteins (PdP, Jhp0578), and a septum formation inhibitor (MinC) (Figure S6). These results suggested that Jhp0106 functions as a glycosyltransferase and is involved in the transfer of pseudaminic acid to flagellin.

The roles of Jhp0106 in H. pylori flagella formation, adhesion and cellular cytokine induction were examined by constructing jhp0106 mutant (SW863) and revertant J99 (SW862) strains. The genetic loci of jhp0105 and flaB are close to jhp0106 in the J99 genome (Figure 1A). In order to rule out the possibility of polar effects in SW862 and SW863, the expression of jhp0105, jhp0106 and flaB were determined by RT-PCR (Figure S7). The data revealed that the mRNA expression levels of jhp0105 and flaB in strains SW863 and SW862 were similar to J99, and the transcription of jhp0106 was only disrupted in SW863 (Figure S7). In addition, the 72 h growth curves of SW863 and SW862 were similar to that of the wild-type J99 (Figure S8).

The deficient motility of SW863 is shown in Figure 5A. The bacterial shape and flagellar structure of J99, SW837, SW838, SW862, and SW863 were examined by TEM with negative staining. No flagellar structure was detected in SW837 and SW863, whereas the characteristic multiple polar, sheathed flagella were abundant on J99, SW838 and SW862 (Figure 5B). In addition, there was no dramatic difference in shape between the strains examined. These observations demonstrated that flagella formation was severely defective in the jhp0106 mutant.

To determine the role of flagella structure and Jhp0106 in H. pylori pathogenesis, the adhesion rate of the bacteria and the IL-8 production in H. pylori-infected AGS and GES-1 cells were determined (Figure 5C and Figure S9). The results showed that only SW868 (flaA/flaB double mutant) had lower adhesion to AGS cells, compared with J99 (p < 0.05) (Figure 5C). No difference in IL-8 production was observed between cells infected with the six examined strains, as determined by an ELISA assay (p > 0.05) (Figure S9).

Many studies suggest that genetic diversity in H. pylori virulence factors such as sabA, babA, cagA, and vacA genes is high among isolates from different geographic regions, and may be associated with different pathological outcomes (van Doorn et al., 1998; Yamaoka et al., 2006). As a result, we evaluated the prevalence of the jhp0106 gene in 95 isolates from patients with different diseases by PCR, including 38 gastritis strains, 21 duodenal ulcer strains, 17 gastric ulcer strains, 18 gastric cancer strains, and 1 MALToma strain. The results showed that all tested isolates contained the jhp0106 gene. To determine whether Jhp0106 is a critical factor in H. pylori motility among different strains, we constructed rpoN and jhp0106 mutants of 14 clinical isolates. The results of soft-agar motility assay showed that all rpoN mutants and jhp0106 mutants were non-motile (Figure 6).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer TC and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.