fapH (hp0838) is conserved in Helicobacter species and linked to flgP

H. pylori HP0838 (FapH) is a predicted lipoprotein encoded by a gene immediately downstream of flgP, which encodes a flagellar motor accessory protein that forms the basal disk in Campylobacter jejuni and Aliivibrio fischeri [13]. In the genomes of other Helicobacter species that we examined, a fapH homolog is located immediately downstream of a flgP homolog (Fig 1 and S1 Table). Further analysis of FapH using the Position-Specific Iterative (PSI)-BLAST tool identified FapH homologs in other members of the phylum Campylobacterota (S1 Table). In contrast to the synteny of fapH and flgP in Helicobacter species, however, fapH and flgP only display synteny in Wolinella succinogenes for the other Campylobacterota that we examined (Fig 1 and S1 Table). Notably, the W. succinogenes fapH homolog is in an operon with six other genes known or predicted to be involved in flagellum assembly (Fig 1), suggesting a role for fapH in flagellar function. As with H. pylori FapH, all of the FapH homologs that were examined have predicted lipoprotein signal peptides (S1 Table). Given the synteny between flgP and fapH in Helicobacter species and W. succinogenes, close association of fapH with several flagellar genes in W. succinogenes, and the predicted periplasmic location of FapH as a lipoprotein, we hypothesized that FapH is a flagellar motor accessory protein.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Organization of operons containing fapH in various members of Campylobacterota and the predicted lipoprotein signal peptides for the FapH homologs.

(A) In many genomes, fapH is located immediately downstream of flgP. Approximate locations of predicted promoters for the operon containing fapH in H. pylori B128 are indicated (green arrows), which is based on a transcriptional start site database determined from the H. pylori 26695 transcriptome [63]. P-p: primary promoter, P-s: secondary promoter, P-i: internal promoter. (B) Predicted lipoprotein signal peptides of the FapH homologs. Amino acids sequences of the FapH homologs were analyzed for signal peptides and their cleavage sites using the SignalP-6.0 server (https://services.healthtech.dtu.dk/service.php?SignalP) [64]. Locations of the cleavage sites in the FapH homologs are indicated by the arrow.

https://doi.org/10.1371/journal.ppat.1012860.g001

In examining the FlgP homologs of various Helicobacter species, we noted that many of the FlgP homologs possess predicted lipoprotein signal peptides, while others have signal peptides that are cleaved by signal peptidase I and are not predicted lipoproteins (S1 Table). This differentiation followed a morphological distinction as Helicobacter species that have sheathed flagella (FS⁺ species) typically possessed a FlgP homolog predicted to be a lipoprotein, while the FlgP homologs in Helicobacter species with sheath-less flagella (FS^- species) lacked a predicted lipoprotein signal peptide. Two FlgP homologs in FS⁺ Helicobacter species that we examined were not predicted to have signal peptides, but this may be due to incorrect assignments for the translational start sites as selecting alternative upstream translational start sites resulted in protein sequences with predicted lipoprotein signal peptides. The relationship between the type of signal peptide and sheathed/unsheathed flagella does not apply to all members of Campylobacterota as some species (e.g., C. jejuni and Nautilia profundicola) have FlgP homologs that are predicted lipoproteins but possess unsheathed flagella (S1 Table).

FapH is not essential for motility in H. pylori

To determine if FapH plays a role in flagellar biosynthesis or function, fapH was deleted in H. pylori B128 and the flagellation and motility of the resulting mutant was assessed. Examination of the ΔfapH mutant by transmission electron microscopy (TEM) and cryo-ET revealed the cells were indistinguishable from wild type with regard to the flagellation pattern (Fig 2). Examples of cryo-ET images showing the flagellar sheath are presented in Fig 2C and 2D. The flagellar sheaths of the ΔfapH mutant appeared to be intact as evidenced by the uniform thickness of the flagella of the cells used for flagellar counts (n = 100), suggesting that FapH is not required for sheath biosynthesis. The H. pylori B128 ΔfapH mutant was impaired in its motility in soft agar medium as it generally produced no discernible swim halo following a 7-d incubation (Fig 3A). Migration of bacterial cells from the point of inoculation in soft agar medium involves both the ability to swim and chemotaxis. In our hands, a chemotaxis deficient H. pylori B128 ΔcheA mutant forms a small, but readily discernible, swim halo following a 7-d incubation [24]. Thus, the failure of the ΔfapH mutant to form a swim halo was not due to loss of chemotaxis, but instead resulted from the inability of the mutant to swim in the soft agar medium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Electron micrographs of H. pylori B128 wild type and ΔfapH pflA^T strain.

TEM images of representative cells of H. pylori B128 wild type (A) and ΔfapH pflA^T mutant (B). Section of cryo-ET images of H. pylori B128 wild type (C) and ΔfapH pflA^T mutant cells (D). In panels C and D, flagellar basal bodies are indicated by red arrows, flagellar sheath is indicated by the blue arrows, and flagellar filaments are indicated by the green arrows. OM–outer membrane; IM–inner membrane.

https://doi.org/10.1371/journal.ppat.1012860.g002

Small swim halos were sometimes observed on soft agar medium plates inoculated with the ΔfapH mutant, suggesting that motile suppressors of the ΔfapH mutant were arising. To enrich for motile variants of the ΔfapH mutant, cells from the edge of the small swim halos were used to inoculate fresh soft agar plates. Over the course of several passages in soft agar medium the average swim halo diameter of the ΔfapH mutant increased from ~5 mm to ~35 mm, and clonal isolates from the cells at the edges of these swim halos were obtained.

Whole genome sequencing (WGS) of the original ΔfapH mutant revealed an insertion of a guanosine at position 1429 in pflA (paralyzed flagellar protein A) (S2 Table). PflA is required for motility in C. jejuni and H. pylori [25,26]. In-situ structures of the C. jejuni motor generated by cryo-ET and single particle cryo-electron microscopy (cryo-EM) suggest PflA forms a set of spoke-like structures that are closely associated with the stator units [13,27]. Stator units are not visible in the in-situ structure for the motor of the C. jejuni ΔpflA mutant, suggesting that PflA is required for the recruitment and/or retention of the stator units within the motor [13]. The insertion in pflA shifted the reading frame and introduced a stop codon four codons downstream of the insertion, which is predicted to result in expression of a truncated PflA that is shortened by 325 amino acid residues. We hereafter refer to the original, non-motile ΔfapH mutant as the ΔfapH pflA^T strain. Introducing pflA on a shuttle vector into the ΔfapH pflA^T strain (hereafter designated as ΔfapH pflA^T/ppflA) restored motility in soft agar medium to near wild-type levels (Fig 3A), indicating that the frameshift mutation in pflA was responsible for the loss of motility in the ΔfapH pflA^T strain. Moreover, these data indicated that FapH is not essential for motility in soft agar medium.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Motility of H. pylori B128 wild type and ΔfapH mutants and their sensitivity to bacitracin.

(A) Motilities of wild type, ΔfapH pflA^T, ΔfapH pflA*,ΔfapH pflA^T/ppflA, and ΔfapH pflA*/pfapH were assessed by stab inoculating the strains in soft agar medium and then measuring the diameters of the resulting swim halos following a 7-d incubation period. Bars indicate mean values for the diameters of the swim halos and the error bars indicate the standard error of the mean (SEM). At least 4 replicates were used in motility assays. (B) Efficiency of plating assays were done with H. pylori strains (indicated in A) on TSA (-bac) or TSA supplemented with 200 μg/ml bacitracin (+bac). Cells from freshly grown cultures of the strains were resuspended in tryptic soy broth to the same cell densities (OD₆₀₀). Ten-fold serial dilutions of the resuspensions (10⁰ to 10⁻⁵) were then spotted onto the media, and the cultures were incubated for 7 d. The ΔpflA mutant was included to demonstrate that disrupting pflA was not responsible for the increased sensitivity to bacitracin. (C) Growth of H. pylori strains in soft agar medium in the absence (-bac) and presence of 200 μg/ml bacitracin (+bac). (D) H. pylori B128 wild type, ΔfapH pflA*, and ΔfapH pflA*/pfapH were stab inoculated into soft agar medium that contained 0, 100, or 200 μg/ml bacitracin and the diameters of the resulting swim halos were measured following a 7-d incubation period. Bars indicate mean values for swim halo diameters and the error bars indicate the SEM. The asterisk indicates statistically significant differences in swim halo diameters as determined using a two-sample t test (p-value <0.0001). At least 3 replicates were done for each sample.

https://doi.org/10.1371/journal.ppat.1012860.g003

WGS analysis of a ΔfapH motile variant that produced a robust swim halo revealed a deletion of a cytosine at nucleotide position 1393 in pflA (S2 Table), which restored the reading frame following the insertion of the guanosine at position 1429 and allowed for the expression of a full-length PflA variant in which the amino acid sequence from Leu-465 to Ile-476 (L⁴⁶⁵FSMEGNTQEKI⁴⁷⁶) is altered to F⁴⁶⁵FPWRGTRKKKS⁴⁷⁶. PflA monomer structure predicted by AlphaFold 2 [28] possesses a β-sandwich domain near the N-terminus and a tetratricopeptide repeat (TPR) domain that comprises most of the remainder of the protein (S1A Fig). The insertion in ΔfapH pflA^T creates a truncation in the tetratricopeptide repeat region of PflA (S1B Fig). The altered sequence in PflA is primarily within a loop between α-helices 16 and 17 in the TPR domain, and the predicted tertiary structure of the PflA variant (hereafter referred to as PflA*) is essentially identical to that of wild-type PflA (S1A and S1C Fig). Given that the isolate expressing PflA* was highly motile and the amino acid substitutions did not impact the predicted tertiary structure of PflA, we proceeded to analyze further the motile isolate, hereafter referred to as the ΔfapH pflA* strain.

The ΔfapH pflA* strain has reduced fitness

Although our data indicated that fapH is not required for H. pylori motility, we wished to determine if loss of fapH resulted in other phenotypic traits. Examination of the ΔfapH pflA* strain by TEM indicated that it was indistinguishable from wild type and the parental ΔfapH pflA^T strain in the number of flagella per cell (S2A Fig). Unexpectedly, the mean length of the flagellar filaments in the ΔfapH pflA* strain was slightly, but significantly, longer than those of the ΔfapH pflA^T strain and wild type (S2B Fig).

To determine if loss of FapH impacted the growth rate of H. pylori, we compared the growth rates of wild type, the non-motile ΔfapH pflA^T strain, and the motile ΔfapH pflA* strain. Growth rates for cultures of wild type and ΔfapH pflA^T strain were similar (mean doubling time was ~6 h; S2C Fig). By contrast, the growth rate of the ΔfapH pflA* strain was significantly slower with a mean doubling time of ~13 h (S2C Fig). To determine if deletion of fapH was responsible for the slower growth rate of the ΔfapH pflA* strain, fapH was introduced into the strain on the shuttle vector pHel3 to generate the strain that we hereafter refer to as ΔfapH pflA*/pfapH. Strain ΔfapH pflA*/pfapH had a growth rate similar to that of wild type and the ΔfapH pflA^T strain (S2C Fig). Taken together, these data suggest the loss of FapH results in a motility-dependent reduction of fitness in H. pylori.

Strain ΔfapH pflA*, but not ΔfapH pflA^T, has increased sensitivity to bacitracin

The LPS in the outer leaflet of the OM forms an effective barrier to hydrophobic antibiotics and other noxious chemicals, and increased sensitivity to these compounds is often indicative of the intrusion of GPLs into the outer leaflet of the OM [17]. Bacitracin is a cyclic polypeptide antibiotic that interferes with peptidoglycan biosynthesis by binding undecaprenyl-pyrophosphate to disrupt the undecaprenyl-phosphate cycle and thereby inhibit synthesis of the peptidoglycan precursor molecule lipid II [29]. In an efficiency of plating assay, the motile ΔfapH pflA* strain displayed increased sensitivity to bacitracin compared to wild type, the non-motile ΔfapH pflA^T strain, the motile ΔfapH pflA*/pfapH strain, and a ΔpflA mutant (Fig 3B). These data demonstrate that loss of FapH in H. pylori results in increased sensitivity to bacitracin and the antibiotic sensitivity is dependent on a functional flagellum. Moreover, these data show that disrupting pflA has no discernible effect on bacitracin sensitivity.

Sensitivity of the H. pylori strains to bacitracin was examined further under conditions where motility was assessed simultaneously by inoculating the strains in soft agar medium that contained bacitracin. H. pylori strains were inoculated into soft agar medium that contained 0, 100, or 200 μg/ml bacitracin and the subsequent swim halos were measured. Consistent with the results from the efficiency of plating assays, the ΔfapH plfA* strain displayed increased sensitivity to bacitracin as the swim halo diameter of the strain was markedly decreased in the presence of the antibiotic at a concentration of 200 μg/ml, while swim halo formation by wild type was unaffected (Fig 3C and 3D). Although the ΔfapH pflA^T strain did not form a swim halo since its flagella are paralyzed, colony growth of the strain in the soft agar medium containing bacitracin was dense indicating that the strain was resistant to bacitracin under the assay conditions (Fig 3C). As observed in the efficiency of plating assay (Fig 3B), the fapH pflA*/pfapH strain was resistant to bacitracin in the motility-based assay indicating that the loss of FapH was responsible for the antibiotic sensitivity of the ΔfapH pflA* strain (Fig 3C and 3D).

To address the possibility that PflA* is required for the antibiotic sensitivity of the ΔfapH pflA* strain, we generated additional ΔfapH mutants in H. pylori B128 and examined the sensitivity of the resulting strains to bacitracin. The resulting ΔfapH mutants were motile and sensitive to bacitracin (S3 Fig). WGS of the strains indicated that they did not have secondary mutations in any known flagellar gene, including pflA (S2 Table). These findings indicate that the sensitivity of H. pylori to bacitracin upon disruption of fapH is not dependent on the PflA variant PflA*.

FapH forms a flagellar motor accessory associated with the OM

Given the sensitivity of the ΔfapH pflA* and ΔfapH strains to bacitracin, which is normally excluded by the OM, we hypothesized that FapH is required to form a motor accessory that helps to preserve OM barrier function during flagellar rotation. To address this hypothesis, we compared in-situ structures of the motors of wild type H. pylori B128, ΔfapH plfA^T, and ΔfapH pflA*, which were determined by cryo-electron-ET and subsequent subtomogram averaging. The motor of ΔfapH pflA^T contained many of the core components and accessories that were present in the wild-type motor, but lacked many of the periplasmic components of the motor (S4A and S4B Fig). Since the periplasmic components are missing in the in-situ structure for the motor of the C. jejuni ΔpflA mutant [13], we infer that the truncation of PflA in ΔfapH pflA^T was responsible for the motor assembly and motility defects in the strain.

Comparing the in-situ structures of the motors of wild type and ΔfapH pflA* revealed that the motors of the two strains were identical with the exception of a globular density near the junction of the OM and flagellar sheath that was missing in the motor of the ΔfapH pflA* strain (Fig 4A–4C). Perpendicular cross-sectional imaging revealed that the globular densities form a ring-like structure with 18-fold symmetry that is absent in the ΔfapH pflA* motor (Fig 4D and 4E), and we designate the structure as the FapH ring. The FapH ring subunits are anchored on a motor accessory that we refer to as the outer disk and extend towards the OM (Fig 4F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Architecture of the FapH ring in the H. pylori motor.

(A and B) In-situ structures of the flagellar motors from ΔfapH pflA* and wild-type cells were determined using subtomogram averaging, respectively. Medial slices through the motor structures indicate the ΔfapH pflA* motor is composed of all motor components, except globular densities in the outer membrane region. (C) A model showing the location of the FapH ring and outer disk structures. (D) Cross section through the globular densities in the ΔfapH pflA* motor. (E) Cross section through the globular densities in the wild-type motor. (F) 3D model of the FapH ring adjacent to the outer membrane. FR–FapH ring, OD–outer disk, BD–basal disk, OM–outer membrane, IM–inner membrane.

https://doi.org/10.1371/journal.ppat.1012860.g004

The tertiary structure of FapH predicted by Alphafold 2 [28] indicates α-helices near the N-terminus and at the C-terminus that are separated by a domain comprised of two β-sheets linked by three short α-helices (S5A Fig). The C-terminal α-helix of FapH is dominated by acidic amino acid residues, while the N-terminal α-helix and surface of the β-sheet domain contain several basic amino acid residues (S5B Fig). The predicted tertiary structure of FapH fits well into the density of the FapH ring subunits (S5C Fig), suggesting that each subunit represents a FapH monomer. From the modeling of the FapH tertiary structure on the FapH ring subunits, the β-sheet domain of FapH contacts the outer disk, while the N-terminal and C-terminal α-helices are distal to the outer disk and located at the top of the FapH ring subunits. Consistent with the proposed orientation of FapH, the N-terminus is closely associated with the OM, which would allow the acyl side chains of the FapH lipoprotein to insert into the OM.

Identification of FapH interaction partners

We sought to identify proteins that interact with FapH to gain a better understanding of how FapH protects the barrier function of the OM from flagellum-mediated damage. Moreover, given the close association of the FapH ring and outer disk, we reasoned that identified FapH interaction partners would be good candidates for outer disk components. FapH was reported to interact with HP1409/HP0426 (these proteins are duplicated in H. pylori 26695) and the basal disk protein FlgP in a high-throughput yeast two-hybrid screen [30]. HP1409 and HP0426 are predicted cytoplasmic proteins and are unlikely to interact with FapH within the context of the flagellar basal body. To identify additional FapH interaction partners, we expressed a c-Myc-tagged FapH protein in ΔfapH pflA* and performed co-immunoprecipitation (co-IP) assays with cell extracts of the strain. Examination of the immunoprecipitated proteins by SDS-PAGE followed by Coomassie blue staining revealed a few prominent protein bands in the sample that contained the FapH-Myc fusion protein (Fig 5A, lanes 4 and 5) that were not apparent in the sample prepared with cell extracts from wild type H. pylori B128, which served as a negative control (Fig 5A, lanes 2 and 3). Proteins in three of these bands were identified by peptide mass fingerprinting. FapH was the only identified protein in the ~26-kDa band, which was the most intense band on the gel. Two proteins, HopD and FtsH, were identified in the ~70-kDa band. HopD is one of five β-barrel OM proteins (HopA, HopB, HopC, HopD, and HopE) in H. pylori that are identified as porins and are structurally homologous with the E. coli OmpF porin [31]. FtsH is an integral membrane, ATP-dependent zinc metallopeptidase that functions in quality control of integral membrane proteins [32]. HP1456 (also referred to as Lpp20), a predicted lipoprotein of unknown function, was the only identified protein in the ~15-kDa band. A crystal structure for H. pylori HP1456 indicates that the protein is elongated and bent, consisting of a three-stranded anti-parallel β-sheet flanked by three α-helices on one side and a short α-helix on the other side [33]. Drobnič and co-workers reported that the predicted tertiary structures of C. jejuni FlgP and Vibrio alginolyticus FlgT are very similar to that of H. pylori HP1456 [27]. Like FlgP, FlgT forms a disk-like motor accessory [34].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Identification of FapH interaction partners.

(A) SDS-PAGE gel of proteins isolated from co-IP assay using anti-myc antibody. Lane 1 contained the molecular weight standards. Sizes of the protein standards are indicated to the left of the gel. Lanes 2 and 3 are negative controls and are samples from the co-IP assay with cellular extracts prepared from wild type H. pylori B128 and correspond to 5 μl and 10 μl of loaded sample, respectively. Lanes 3 and 4 are samples from the co-IP assay with cellular extracts prepared H. pylori B128 strain expressing the FapH-myc fusion protein and correspond to 5 μl and 10 μl loaded sample, respectively. Lane 6 shows a c-myc tagged E. coli protein (indicated by **) that served as a positive control for the co-IP assay and was supplied with the Pierce c-Myc Tag Magnetic IP/co-IP kit. Protein bands in lane 5 that were excised and analyzed to identify proteins present are indicated by an asterisk (*). HopD and FtsH were identified in the top band, the middle band corresponded to the FapH-myc protein, and HP1456 was identified in the lower band. (B) Results from β-galactosidase assays for strains with select combinations of adenylate cyclase T15 and T25 fragments fused to FapH, FlgP, HP1454, and HP1456. The cartoons shown on the y-axis illustrate the orientation of the adenylate cyclase fragments and H. pylori proteins in the fusion proteins relative to each other. In the cartoons, FapH is indicated in blue, FlgP is indicated in yellow, HP1456 is indicated in red, and HP1454 is indicated in green. pKNT25 and pKT25 indicate the adenylate cyclase T25 fragment at the N-terminus or C-terminus of the fusion protein, respectively. pUT18 and pUT18C indicate the adenylate cyclase T18 fragment at the N-terminus or C-terminus of the fusion protein, respectively. An E. coli strain containing the BACTH vectors pKT25 and pUT18 served as a negative control, and an E. coli strain bearing plasmids that expressed the T25-zip andT18-zip proteins served as a positive control. The bars indicate the average β-galactosidase activity for each strain and the error bars indicate the SEM. At least 3 replicates were done for each sample. A single asterisk (*) indicates a p-value of < 0.0001, two asterisks (**) indicate a p-value of < 0.001, and three asterisks (***) indicated a p-value of < 0.005 as determined using a two-sample t test. Significance of interaction between pKNT25-FlgP and FlgP-pUT18C is compared with pKT25 and pUT18. For the FapH-FlgP interactions, 7 of the 8 possible plasmid combinations produced significantly elevated levels of β-galactosidase activity, but only 3 of the 7 strains are shown in the figure. (C) Protein-protein interaction map constructed from co-IP and BACTH experiments. Solid black lines indicate interactions observed in co-IP assays involving the FapH-myc protein and dashed red lines indicate interactions observed in the BACTH assay.

https://doi.org/10.1371/journal.ppat.1012860.g005

Additional proteins that were pulled-down in the co-IP assays used for the fingerprinting analysis, as well as co-IP samples from two additional replicates were identified by analyzing the total protein content of the samples by mass spectroscopy (S3 Table). Notably, HP1456 was identified in all three replicates of the co-IP assay and FlgP was identified in one of the three replicates. HP1454, which is encoded in the same operon as HP1456, was identified in two of the replicates of the co-IP assay. HP1454 is a secreted protein composed of three distinct domains [35], the first of which is structurally similar to HP1456 [33].

Since FlgP was identified in one of the replicates of the co-IP assay with the FapH-Myc fusion protein, we used a co-IP assay to determine if FapH was pulled-down with FlgP and to also identify other potential FlgP partners. For these co-IP assays, we used cell extracts prepared from a H. pylori B128 ΔflgP mutant that expressed FlgP-myc fusion protein. A caveat of the experiment was the ΔflgP mutant was aflagellated. WGS of the ΔflgP mutant revealed that the strain had a single nucleotide deletion within a homopolymeric tract of eight guanosines in fliP, which encodes a component of the flagellar protein export apparatus. Slipped-strand mispairing-mediated mutagenesis within this poly(G)-tract in fliP was reported previously for H. pylori 26695, and was suggested as a mechanism for switching between “on” and “off” phases for flagellation and motility [36]. Despite the mutation in fliP, we proceeded with co-IP assays with the expectation that FlgP can associate with at least some of its interaction partners in the absence of the axial components of the motor. Mass spectrometry analysis of the proteins pulled-down in two biological replicates of the co-IP assay with the FlgP-myc protein identified several potential FlgP interaction partners (S3 Table). Notably, FapH was identified in both replicates of the co-IP assay. Other proteins of interest identified from the co-IP assay with the FlgP-myc protein included HP1456 and HP1454.

To examine the validity of the protein interactions predicted from the co-IP assays, we examined the interactions of FapH with selected proteins using the bacterial adenylate cyclase two-hybrid (BACTH) assay [37]. The BACTH assay is based on the reconstitution of adenylate cyclase (CyaA) activity where two complementary fragments from the catalytic domain of the Bordetella pertussis CyaA toxin are brought into proximity by interacting proteins that are fused to the CyaA fragments. Reconstitution of CyaA activity leads to production of cAMP, which is assessed by monitoring the expression of cAMP-dependent genes in an E. coli cyaA mutant. We constructed vectors that expressed FapH, HP1454, HP1456, or FlgP fused to the N- or C-termini of the B. pertussis CyaA fragments T18 and T25. In constructing the expression vectors, the sequences encoding the signal peptides of the H. pylori proteins were omitted to avoid secretion of the fusion proteins into the periplasm.

All possible combinations of expression vectors in the E. coli reporter strain BTH101 were examined by culturing the cells on McConkey agar medium that contained 1% maltose (S6 Fig), and β-galactosidase activities were measured for strains with plasmid combinations that produced pink or red colonies on McConkey-maltose agar medium (Fig 5B). Consistent with the co-IP data that suggested interactions between FapH and HP1456, five of the eight combinations of FapH-CyaA and HP1456-CyaA fusion proteins reconstituted active CyaA as demonstrated by the significantly higher β-galactosidase activities in strains harboring these vector combinations versus the negative control (Fig 5B). In addition, three of the eight combinations of FapH-CyaA and HP1454-CyaA fusion proteins reconstituted active CyaA (Fig 5B). Consistent with the results from the co-IP assays with both the FapH-myc and FlgP-myc proteins as well as the reported interactions between FlgP and FapH in a yeast two-hybrid assay [30], seven of the eight combinations of FapH-CyaA and FlgP-CyaA fusion proteins reconstituted active CyaA (Fig 5B). FlgP also interacted with HP1454 and HP1456 in the BACTH system, as well as itself (Fig 5B). The strong interaction of FlgP with itself is consistent with the assembly of FlgP monomers to form the basal disk. Additional results from the BACTH assays suggest that HP1454 interacts weakly to moderately with itself but does not interact with HP1456 (Fig 5B). In addition to interacting with FlgP and FapH, HP1456 interacted weakly with itself (Fig 5B). Results of the protein-protein interaction studies are summarized in Fig 5C.

Given the interactions between FapH and HP1456 in the co-IP and BACTH assays (Fig 5) as well as the structural similarity between HP1456 and the disk-forming motor accessory proteins FlgP and FlgT [27], HP1456 seemed a good candidate for an outer disk protein. To address the hypothesis that HP1456 forms the outer disk of the H. pylori motor, we generated an unmarked deletion of hp1456 in H. pylori B128 and characterized the resulting mutant. The number of flagella produced by the Δhp1456 mutant and the motility of the mutant in soft agar medium were the same as observed for wild type (S7A-S7C Fig). In addition, the Δhp1456 mutant displayed wild-type resistance to bacitracin (S7D Fig), indicating that HP1456 is not required by FapH to protect the barrier function of the OM from flagellum-mediated damage. The in-situ structure of the Δhp1456 mutant motor was indistinguishable from that of wild type (S7E and S7F Fig), indicating that HP1456 is not required for assembly of the outer disk. Thus, despite being a promising candidate, HP1456 is not an outer disk component.

Bacterial strains and culture conditions

E. coli Turbo cells (New England Biolabs, Ipswich, MA, USA, which were used for cloning procedures, were grown on LB medium supplemented with ampicillin (100 μg/ml) or kanamycin (30 μg/ml) as required. H. pylori strains used in the study were derived from H. pylori B128 (kindly provided by Richard M. Peek, Jr.). and are listed along with plasmids used in the study in S4 Table. Liquid cultures of H. pylori were grown at 37°C with shaking in brain heart infusion (BHI) supplemented with 5% heat-inactivated horse serum (Gibco; Thermo Fisher Scientific, Waltham, MA, USA) in serum bottles that contained an atmosphere consisting of 5% CO₂, 10% H₂, 10% O₂, and 75% N₂. H. pylori strains were grown on solid medium under an atmosphere consisting of 10% CO₂, 6% O₂, and 84% N₂ at 37°C on tryptic soy agar supplemented with 5% heat-inactivated horse serum (TSA-HS). Growth medium for H. pylori was supplemented with kanamycin (30 μg/ml), bacitracin (100 or 200 μg/ml), or 5% sucrose (w/v) as needed.

PCR methods

Genomic DNA (gDNA) from H. pylori B128 was purified using the Wizard genomic DNA purification kit (Promega, Madison, WI, USA) and used as the template for PCR using PrimeSTAR DNA polymerase (Takara Bio, San Jose, CA, USA) or Phusion DNA polymerase (New England Biolabs). To facilitate cloning into the pGEM-T Easy vector (Promega), amplicons were incubated with Taq polymerase (Promega) at 72° C for 10 min to add A overhangs at the 3’-ends.

Construction of H. pylori B128 ΔfapH, ΔflgP, and Δhp1456 mutants

A schematic illustrating the process for deleting fapH in the H. pylori B128 chromosome is shown in S8 Fig, and the same general approach was used to delete flgP and hp1456. Regions flanking fapH were amplified from H. pylori B128 gDNA using primer pair 59 and 60 (for upstream region) and primer pair 61 and 62 (for downstream region; S5 Table). Primers 60 and 61 were complementary at their 5’-ends and introduced XhoI and NheI restriction sites for subsequent introduction of a kan^R-sacB cassette. The amplicons corresponding to the regions upstream and downstream of fapH were joined together by overlapping PCR, A overhangs were added to the 3’-ends of the resulting amplicon, which was then ligated into pGEM-T Easy (Promega) to generate plasmid pKR10. Plasmid pKR3 is a derivative of pJC038 [48] that carries a kan^R-sacB cassette where sacB is under control of the H. pylori ureA promoter. In pKR3, the H. pylori ureI promoter was introduced at the end of the cassette in an outward orientation to compensate for potential polar effects on genes that are downstream of the kan^R-sacB insertion. The kan^R-sacB cassette was excised from plasmid pKR3 by digesting with NheI and XhoI (New England Biolabs) and then introduced into the NheI and XhoI sites in plasmid pKR10 to generate plasmid pKR11, which was introduced by natural transformation into H. pylori B128. Since plasmid pKR11 does not replicate in H. pylori, transformants in which fapH had been replaced with the kan^R-sacB cassette by homologous recombination were enriched by selecting for kanamycin resistance. Replacement of fapH with the kan^R-sacB cassette was confirmed by PCR using primer pair 59 and 62. One of the kanamycin-resistant isolates in which fapH was replaced with the kan^R-sacB cassette, which was designated strain H6, was transformed with the suicide vector pKR10. Transformants in which the kan^R-sacB cassette was replaced with the unmarked deletion of fapH resulting from homologous recombination between plasmid pKR10 and the chromosome were counter-selected on TSA-HS supplemented with 5% sucrose as described [49]. Sucrose-resistant isolates were screened for kanamycin-sensitivity on TSA-HS supplemented with kanamycin, and deletion of fapH in kanamycin sensitive isolates was confirmed by PCR using primer pair 59 and 62 and DNA sequencing of the resulting amplicon (Eton Biosciences, Research Triangle, NC, USA). A H. pylori B128 strain in which fapH was deleted was designated as strain H16.

An unmarked deletion of flgP in H. pylori B128 was constructed following the protocol used for constructing the ΔfapH mutant. Regions flanking flgP were amplified from H. pylori B128 gDNA using primer pair 39 and 40, and primer pair 41 and 42. The resulting amplicons were joined together by overlapping PCR and cloned into pGEM-T Easy to generate plasmid pKR4. The kan^R-sacB cassette was introduced into plasmid pKR4 to generate the suicide vector pKR9, which was introduced into H. pylori B128 to replace flgP with the kan^R-sacB cassette through homologous recombination to yield strain H5. Replacement of flgP with the kan^R-sacB cassette was confirmed by PCR using primer pair 39 and 42. The suicide vector pKR4 was introduced into H5 to replace the kan^R-sacB cassette in the flgP locus with the unmarked deletion of flgP in plasmid pKR4 using the sucrose counter-selection as described above. Sucrose-resistant isolates were screened for kanamycin-sensitivity on TSA-HS supplemented with kanamycin, and deletion of flgP in kanamycin sensitive isolates was confirmed by PCR using primer pair 39 and 42 and DNA sequencing of the resulting amplicon. A H. pylori B128 strain in which flgP was deleted was designated as strain H6. gDNA from strain H6 was prepared and provided to the SeqCenter (Pittsburgh, PA, USA) for whole genome Illumina sequencing. The resulting genome sequence data were analyzed using the breseq computational pipeline [50].

An unmarked deletion of hp1456 in H. pylori B128 was constructed following the protocol used for constructing the ΔfapH mutant. Regions flanking hp1456 were amplified from H. pylori B128 gDNA using primer pair 195 and 196, and primer pair 197 and 198. The resulting amplicons were joined together by overlapping PCR and cloned into pGEM-T Easy to generate plasmid pKR67. The kan^R-sacB cassette was introduced into plasmid pKR67 to generate the suicide vector pKR77, which was introduced into H. pylori B128 to replace hp1456 with the kan^R-sacB cassette through homologous recombination to yield strain H143. Replacement of hp1456 with the kan^R-sacB cassette was confirmed by PCR using primer pair 195 and 198. The suicide vector pKR67 was introduced into H143 to replace the kan^R-sacB cassette in the hp1456 locus with the unmarked deletion of hp1456 in plasmid pKR67 using the sucrose counter-selection as described above. Sucrose-resistant isolates were screened for kanamycin-sensitivity on TSA-HS supplemented with kanamycin, and deletion of hp1456 in kanamycin sensitive isolates was confirmed by PCR using primer pair 195 and 198 and DNA sequencing of the resulting amplicon. A H. pylori B128 strain in which hp1456 was deleted was designated as strain H145. gDNA from strain H145 was prepared and provided to the SeqCenter (Pittsburgh, PA, USA) for whole genome Illumina sequencing. The resulting genome sequence data were analyzed using the breseq computational pipeline [50].

An unmarked deletion of pflA in H. pylori B128 was constructed following the protocol used for constructing the ΔfapH mutant. Regions flanking pflA were amplified from H. pylori B128 gDNA using primer pair 43 and 44, and primer pair 45 and 46. The resulting amplicons were joined together by overlapping PCR and cloned into pGEM-T Easy to generate plasmid pKR5. The kan^R-sacB cassette was introduced into plasmid pKR5 to generate the suicide vector pKR7, which was introduced into H. pylori B128 to replace pflA with the kan^R-sacB cassette through homologous recombination to yield strain H3. Replacement of pflA with the kan^R-sacB cassette was confirmed by PCR using primer pair 43 and 46. The suicide vector pKR7 was introduced into H3 to replace the kan^R-sacB cassette in the pflA locus with the unmarked deletion of pflA in plasmid pKR5 using the sucrose counter-selection as described above. Sucrose-resistant isolates were screened for kanamycin-sensitivity on TSA-HS supplemented with kanamycin, and deletion of pflA in kanamycin sensitive isolates was confirmed by PCR using primer pair 43 and 46 and DNA sequencing of the resulting amplicon. A H. pylori B128 strain in which pflA was deleted was designated as strain H7. gDNA from strain H7 was prepared and provided to the SeqCenter (Pittsburgh, PA, USA) for whole genome Illumina sequencing and the resulting genome sequence data were analyzed using the breseq computational pipeline.

Complementation of mutants

To facilitate complementation assays in H. pylori, we modified the shuttle vector pHel3 [51] for Golden Gate cloning that employed tandem sites for the Type IIS restriction enzyme BspQ1 (S9A Fig). The modified plasmid, which we designated pHel3-GG, contains the fliF promoter upstream of the tandem BspQ1 sites where the cloned gene is introduced. The fliF promoter is flanked by unique BamHI and NheI sites that can be used to replace the fliF promoter with other promoters if desired.

Primers 169 and 170 were used to amplify fapH from H. pylori B128 gDNA (S5 Table). The primers introduced BspQ1 sites immediately upstream and downstream of the start and stop codons of fapH, respectively. The resulting amplicon and pHel3-GG were digested together with BspQ1 (New England Biolabs) for 1 h at 50°C, after which the amplicon and vector were ligated using Fast-Link DNA ligase (Biosearch Technologies, Hoddeson, UK) for 30 min at room temperature. The reaction mix was incubated with BspQ1 to digest any uncut pHel3-GG vector, and the reaction mix was then used for transformation of E. coli. A plasmid containing the expected insert, designated as pKR54, was verified by restriction enzyme digestion and sequencing of the inserted DNA.

Primers 159 and 160 were used to amplify pflA from H. pylori B128 gDNA. The primers introduced BspQ1 sites immediately upstream and downstream of the start and stop codons of pflA, respectively. The resulting amplicon was cloned into pHel3-GG as described above. A plasmid containing the expected insert, designated as pKR50, was verified by restriction enzyme digestion and sequencing of the inserted DNA. Plasmids pKR54 and pKR50 were introduced by natural transformation into various H. pylori strains for complementation assays.

Construction of c-Myc tagged FapH and FlgP proteins

To facilitate construction of a c-Myc-tagged FapH in H. pylori, we modified the shuttle vector pHel3 [51] to allow for the introduction of the c-Myc epitope at the C-terminus of the protein (S9B Fig). The modified plasmid, which we designated pHel3-Myc, contains the fliF promoter upstream of the tandem BspQ1 sites for Golden Gate cloning followed by a sequence encoding a flexible linker, c-Myc epitope, and DDDDK epitope. Primers 162 and 163 were used to amplify fapH from H. pylori B128 gDNA (S5 Table). The primers introduced BspQ1 sites immediately upstream and downstream of the start and stop codons of fapH, respectively. The resulting amplicon and plasmid pHel3-Myc were digested together with BspQ1 and ligated using Fast-Link DNA ligase as described above. The reaction mix was incubated with BspQ1 to digest uncut pHel3-Myc vector and the reaction mix was then used for transformation of E. coli. A plasmid containing the expected insert, designated as pKR55, was verified by restriction enzyme digestion and sequencing of the inserted DNA.

To construct a c-Myc-tagged FlgP, primers 78 and 79 were used to amplify flgP from H. pylori B128 gDNA. The resulting amplicon and plasmid pHel3-Myc were digested with BspQ1 and ligated together to generate plasmid pKR23. The flgP-myc chimeric gene in plasmid pKR23 was confirmed by DNA sequencing.

Motility assay

Motility was evaluated using a semisolid medium containing Mueller-Hinton broth (MHB), 10% heat-inactivated horse serum, 20 mM MES (2-(4-morpholino)-ethane sulfonic acid) (pH 6.0), and 0.4%-0.6% Noble agar. A minimum of three technical replicates were used to assess the motility of each strain. H. pylori strains grown on TSA-HS for 2 d were stab-inoculated into the motility agar and incubated at 37° C under an atmospheric condition consisting of 10% CO₂, 6% O₂, and 84% N₂. The diameters of the resulting swim halos were measured 7 d post-inoculation and a two-sample t-test was used to determine statistical significance. For motility-base antibiotic resistance assays, bacitracin was included in the soft agar medium at the indicated concentrations.

Efficiency plating assay

H. pylori strains were grown on TSA-HS, then resuspended in MHB. The densities of each cell suspension were adjusted to an OD₆₀₀ of ~0.50. Cell suspensions were diluted serially, 3 μl of each dilution were plated onto TSA-HS and TSA-HS supplemented with bacitracin, and the cultures were incubated for 5 d under a microaerobic atmosphere. A minimum of two replicants were done for each strain and tested condition.

Co-immunoprecipitation procedure

Wild-type H. pylori B128 and the H. pylori strains expressing either the FapH-Myc or FlgP-Myc fusion proteins were grown on TSA-HS plates, harvested, and resuspended in 7 ml phosphate-buffer saline (PBS) with Pierce Protease Inhibitor (Thermo Fisher Scientific). Cells were lysed at 18,000 psi with a French press and the resulting cell lysate was centrifuged at 7,700 x g for 10 min to remove unlysed cells and cell debris. The resulting supernatant was centrifuged at 100,000 x g to pellet membrane vesicles, which were subsequently resuspended in PBS buffer containing 50 mM n-dodecyl-β-D-maltoside (DDM) (Chem-Impex International, Wood Dale, IL, USA). The solution was diluted to 20 mM DDM and the insoluble material was pelleted by centrifugation at 10,000 x g for 10 min and the resulting supernatant was used for the co-IP procedure using the Pierce c-Myc Tag Magnetic IP/Co-IP kit (Thermo Fisher Scientific) as described by the supplier. Samples were incubated with the magnetic beads either overnight at 4°C or at room temperature for 30 min. Magnetic beads with bound proteins were washed at least 10 times with a 1:20 solution of Buffer 2 (supplied with the kit) containing 20 mM DDM. Proteins were eluted from the magnetic beads using 100 μl of 1x non-reducing sample buffer (supplied with the kit) followed by incubation at 95–100°C for 5 to 10 min. Samples were loaded on a 12% SDS-PAGE gel and proteins were visualized by staining with Coomassie Brilliant Blue R-250. Protein bands were excised from the gel and submitted to the University of Georgia Proteomics and Mass Spectrometry Facility for protein identification by peptide mass fingerprinting analysis. Proteomic analysis of total protein from the co-IP assays was done by loading the samples on a 12% SDS-PAGE gel and allowing the samples to enter the top of the resolving gel. The protein band was excised from the gel and submitted to UGA PAMS for analysis by in-gel trypsin digestion followed by LC-MS/MS on the Orbitrap mass spectrometer coupled with nano-HPLC using a 90-minute elution gradient.

Transmission electron microscopy

H. pylori strains were grown to late-log phase (OD₆₀₀ ~1.0) in MHB supplemented with 5% heat-inactivated horse serum. Cells from 1 mL of culture were pelleted by centrifugation (550 x g) then resuspended in 125 μL of PBS. Cells were fixed by adding 50 μL of 16% EM grade formaldehyde and 25 μL of 8% EM grade glutaraldehyde to the cell resuspension. Following incubation at room temperature for 5 min, 10 μL of the cell suspension were applied to a 300 mesh, formvar-coated copper grid and incubated at room temperature for 5 min. The cell suspension was wicked off the grids using a filter paper, and the grids were washed 3 times with 10 μL of water. Cells were stained by applying 10 μL of 1% uranyl acetate to the grids for 30 s. After removing the stain with filter paper, the grids were washed three times with 10 μL of water and then air-dried. Cells were visualized using a JEOL JEM 1011 transmission electron microscope. The number of flagella per cell were determined for at least 100 cells for each strain. Flagellar lengths were determined using Fiji ImageJ.

Cryo-ET sample preparation

H. pylori strains were grown on Columbia agar plates supplemented with 5% horse red blood cells at 37°C under microaerobic conditions. Bacteria from the agar medium were resuspended in PBS and mixed with 10 nm of BSA gold tracers (Aurion, Wageningen, NL). The mixtures were deposited on freshly grow-discharged cryo-EM grids (Quantifoil R2/1a Cu 200 mesh) and then frozen into liquid ethane by using a manual plunger as described previously [15,48].

Cryo-ET data collection and processing

Frozen-hydrated specimens were visualized in a 300kV Titan Krios G2 transmission electron microscope (Thermo Fisher Scientific) equipped with a K3 summit direct detection camera and a BioQuantum energy filter (Gatan). Tilt series images were acquired at 42,000x magnification (corresponding to a pixel size of 2.148 Å at the specimen level) and ~4.5 μm defocus using SerialEM [52] and FastTomo script [19] based on a dose-symmetric scheme from -48° to +48° in 3° increments [53]. The total accumulative dose for each tilt series was ~60 e^-/Ų. MotionCor2 [54] was used to correct the beam induced motion. IMOD [55] was used to create image stacks and align all images in each tilt series by tracing the gold tracer beads. Gctf [56] was used for the defocus estimation for all images in the tilt series, and then the ctfphaseflip function in IMOD was used for contrast transfer function (CTF) correction [57].

Subtomogram averaging

Tomo3D [58] was used to generate 3D reconstructions with the simultaneous iterative reconstruction technique (SIRT), and a total of 268, 225, and 146 tomograms were reconstructed from wild type, ΔfapH pflA^T, and ΔfapH pflA*, respectively. 6 x binned tomograms were used to select 824, 640, and 534 flagellar motors from wild type, ΔfapH pflA^T, and ΔfapH pflA*, respectively. After the particle picking, tomograms with weighted back projection (WBP) were used for the initial subtomogram averaging with the i3 suite [59,60]. For further structural analysis, 4 x binned subtomograms were used to refine the intact motor structures.

FapH modeling in the intact flagellar motor

Alphafold 2 was used to generate a structural prediction for FapH [28,61]. The predicted structure deemed the best model was used to fit into the intact motor structure derived from subtomogram averaging by using the “fit in map” option in UCSF ChimeraX software [62]. All models examined had the signal peptides removed.

Whole genome sequencing and analysis

gDNA from the H. pylori strains was purified using the Wizard genomic DNA purification kit (Promega) and submitted to the SeqCenter (Pittsburgh, PA, USA) for genomic library preparation and Illumina sequencing. Reads for H. pylori gDNA sequence were mapped using the breseq computational pipeline [50] with the published NCBI genome for H. pylori B128 (Accession no.: NZ_CP024951.1).

Construction of plasmids for BACTH assays

The Bacterial Adenylate Cyclase Two-Hybrid (BACTH) System Kit (Euromedex, Souffelweyersheim, France) was used to detect and characterize protein-protein interactions in vivo. Primer pairs 208/209, 2010/211, 212/213, and 214/215 (S5 Table) were used to amplify hp1454, hp1456, fapH, and flgP, respectively, from H. pylori B128 gDNA. The forward primers corresponded to sequences immediately following the regions that encoded the predicted signal peptides to prevent secretion of the recombinant proteins. Following the addition of A’s to the 3’-ends, the resulting amplicons were ligated into pGEM-T Easy. The resulting plasmids were designated pKR69, pKR70, pKR71, and pRK72 (S4 Table), Tand the inserted DNA sequences were confirmed by sequencing. BamHI-KpnI fragments from these plasmids were cloned into BamHI and KpnI sites in the BACTH vectors pKT25, pKNT25, pUT18, andpUT18C) to generate plasmid pRK96 though pRK112. DNA fragments that were cloned into the BACTH vectors were sequenced to confirm that the plasmid constructs were correct. E. coli reporter strain BTH101 (Euromedex) was used for co-transformation of BACTH plasmids. Eight combinations of plasmids were co-transformed and screened in BTH101 for each of the following protein-protein interaction pairs: FapH-FlgP, FapH-HP1454, FapH-HP1456, FlgP-HP1454, FlgP-HP1456, HP1454-HP1456, FlgP-FlgP, HP1454-HP1454, and HP1456-HP1456, for a total of 72 different combinations. Transformants were plated onto McConkey agar supplemented with 1% (w/v) maltose, 100 μg/ml ampicillin, 50 μg/ml kanamycin, and 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG).

β-galactosidase assay

Cultures of E. coli BTH101 harboring various combinations of BACTH plasmids were grown at 30°C to mid- to late-log phase in LB supplemented with ampicillin (50 μg/ml), kanamycin (25 μg/ml), and 0.5 mM IPTG. Cells were diluted five-fold in M63 medium (15 mM (NH₄)₂SO₄,100 mM KH₂PO₄, 1.8 μM FeSO₄, 1 mM MgSO₄, and 4 μM thiamine) containing 0.2% maltose as the carbon source in a final volume of 2.5 ml and OD₆₀₀ readings were recorded for each culture. Cells were permeabilized by adding 30 μl of toluene and 35 μl of a 0.1% sodium dodecyl sulfate solution then incubating for 30–40 min at 37°C with shaking. The permeabilized cells were diluted 10-fold in PM2 buffer (70 mM Na₂HPO₄, 30 mM NaH₂PO₄, 1 mM MgSO₄, 0.2 mM MnSO₄, 100 mM β-mercaptoethanol, pH 7.0) to a final volume of 1 ml and placed in 5 ml glass tubes. Cells were incubated for 5 min in a water bath at 28°C, and the enzymatic reaction was started with the addition of 0.25 ml of 13 mM o-nitrophenol-β-galactoside (ONPG) in PM2 buffer. Enzyme reactions were stopped at various recorded times by the addition of 0.5 mL of 1 M Na₂CO₃, and then the OD₄₂₀ values were recorded for each sample. The enzymatic activity (A; units (in nmol ONPG hydrolyzed min^-1)/ml) was calculated according to the instructions provided by the BACTH kit supplier using the formula A = 200 x (OD₄₂₀ / min of incubation) x dilution factor. The factor 200 in the formula is the inverse of the OD₄₂₀ for 1 μM o-nitrophenol using a 1 cm light path, which is 0.0045 at pH 11.0. The dilution factor for each was 10. Specific activities (SA) expressed in units per mg dry weight were calculated by considering 1 ml of culture at OD₆₀₀ = 1 corresponds to 300 μg dry weight bacteria and using the formula SA = A / (OD₆₀₀ x 0.3 mg dry weight per OD₆₀₀). Each strain was assayed with at least three replicates, and a two-sample t-test was used to determine statistical significance.