Keywords

1 Introduction

Bacterial colonization of the stomach is fraught with danger. The constant production of hydrochloric acid and the resulting low pH, churning of the stomach, and the rapid turnover of epithelial cells (on the order of every 2–3 days) makes the stomach an inhospitable place (Belanger and Leblond 1946; Lee 1985; Schreiber et al. 2004). However, H. pylori has evolved mechanisms to neutralize the acid (Langenberg et al. 1984; Mobley 1996) and to move towards the epithelial cell surface through the mucosal barrier (Beier et al. 1997; Croxen et al. 2006; Keilberg and Ottemann 2016) that protects the stomach lining, where the nearly neutral pH provides a much more tolerable environment. Once encountering the gastric epithelial cell layer, H. pylori must anchor themselves to host cell plasma membranes in order to prevent moving back into the lumen and expulsion from the stomach. Bacterial survival depends on these mechanisms of host cell adherence.

Once H. pylori bacteria have adhered to host cells, they will remain within the stomach and the host will remain asymptomatic in approximately 80% of infected individuals (Blaser et al. 1995; Israel et al. 2001; Parsonnet et al. 1997). However, the remaining 20% will go on to develop gastritis, peptic ulcer disease (PUD), mucosa-associated lymphoid tissue (MALT) lymphoma and/or gastric cancer (GC) during their lifetimes (Blaser et al. 1995; Israel et al. 2001; Parsonnet et al. 1997). The more virulent strains of H. pylori that can translocate the CagA protein are strongly associated with these diseases (Censini et al. 1996; Parsonnet et al. 1997). CagA is an oncoprotein that interferes with cell signaling pathways through its interactions with host factors such as E-cadherin, CRK, CSK, PAR-1, SHP-2, GRB-2 and ASPP-2 (Buti et al. 2011; Lu et al. 2008; Mimuro et al. 2002; Murata-Kamiya et al. 2007; Segal et al. 1999; Selbach et al. 2009; Tegtmeyer et al. 2011; Tsutsumi et al. 2003; Zhang et al. 2015), thereby perturbing cytoskeletal organization, motility, proliferation, cell-cell contact, mitogenic gene expression and apoptosis (compare Chap. 3 of this book). CagA is encoded by the cytotoxic-associated gene pathogenicity island (cagPAI), a region of the H. pylori genome that contains ~30 genes which encode for a T4SS that delivers CagA into host cells (Censini et al. 1996; Backert et al. 2015). CagL, another member of the cagPAI, forms part of the T4SS injection pilus and aids in adherence of H. pylori to host cells and the successful translocation of CagA (Posselt et al. 2013).

To achieve adherence, the H. pylori genome contains over 60 outer membrane protein (OMP) genes which can be divided into five paralogous gene families (Alm et al. 2000). The largest family consists of the Hop (H. pylori OMP) and Hor (Hop-related) genes, which encode 33 proteins. The second family is Hof (Helicobacter related) with eight genes, whilst the third, Hom (Helicobacter outer membrane) is the smallest with four genes (Alm et al. 2000). The remaining OMPs are contained within the iron-regulated and efflux pump OMP families. Driven by gene recombination and duplication, increase in mutational rate and the exchange of DNA between different strains (Didelot et al. 2013; Kennemann et al. 2011; Morelli et al. 2010), every genome of each strain of H. pylori differs in the OMPs that it possesses (Alm et al. 1999; Tomb et al. 1997). In addition to the genetic differences in the OMPs between strains, each strain also regulates expression of OMPs through several different mechanisms, including phase and allelic variation, gene conversion, gene duplication, and regulation in response to pH and salt. The combination of different OMPs and their regulation allows H. pylori to respond to the local environment in the stomach and host immune mechanisms in order to establish and maintain colonization (Kang and Blaser 2006; Odenbreit et al. 2009).

This review summarizes our current understanding of the bacterial adhesins, from both the OMP families and the cagPAI, that have been demonstrated to be involved in adherence of H. pylori as well as their pathogenic roles in the promotion of disease. An overview of adhesins and their known ligands is shown in Fig. 1.

Fig. 1
figure 1figure 1

Graphical representation of H. pylori adhesin interactions with host cell proteins. HopQ (yellow) interacts with CEACAM1, -3, -5 and -6 (IgV domain, brown, IgC2 domains, blue). CEACAM1 and CEACAM3 contain an immunoreceptor tyrosine-based inhibition (−) and activation (+) motifs, respectively. CagL (grey) is attached to the pilus tip (light grey circles) of the T4SS, where it can interact with the α5β1, αVβ3, αVβ5 and αVβ6 integrins. Mucins (green) are heavily glycosylated proteins and can be decorated with the blood group antigens (Leb, A-Leb and B-LeB), both sialyated and asialyated Lex, and lcadiNAc, which bind to BabA (cyan), SabA (blue) and HopD (purple), respectively. SabA also binds both sialyated and asialyated Lex attached to Laminin. AlpA (pink) also binds Laminin. HorB, HomA/B/C/D, OipA and HopZ are outer membrane proteins of H. pylori. The host cell ligands of these proteins are currently unknown

Full size image

2 OMP Domain Organization

To date, all H. pylori OMPs established as bona fide adhesins belong to the Hop, Hor and Hom families. An analysis of the genomic sequences of strains J99 and 26.695 grouped the OMPs into five separate families (Alm et al. 2000). Furthermore, comparison of the Hop and Hor families indicates that they share a common domain organization, including an N-terminal signal peptide of ~20–25 residues, a C-terminal β-barrel to anchor the protein to the outer membrane and a central domain that confers host protein specificity (Fig. 2a). The domain organization is reminiscent of autotransporters. Recently, however, several C-terminal β-barrels of the Hop proteins have been shown to be split, with a single strand found at the N-terminus, with the remaining strands found at the C-terminus (Coppens et al. 2018). Thus, the large extracellular domains are found to be inserted in the extracellular loops of the β-barrel (Fig. 2a).

Fig. 2
figure 2figure 2

Structures of H. pylori adhesins. (a) Domain organization of both the Hop and Hor proteins. Top – The original domain organization of Hop and Hor proteins consists of a N-terminal signal peptide (SP, orange), an extracellular domain (grey) and a C-terminal beta-barrel (blue), reminiscent of autotransporters. Bottom – The new domain organization of Hop and Hor proteins which show that part of the beta-barrel is formed from several residues of the N-terminus. (b) Left – Structure of BabA (PDB entry 5F9D, grey) interacting with the Lewis b blood group B heptasaccharide ligand (cyan sticks) through its insertion domain (red). Right – Closer inspection of the interaction. (c) Structure of SabA (PDB entry 4O5J, grey) with proposed s-Lex binding site highlighted in red. (d) Left – Structure of the Type I HopQ-CEACAM1 complex (PDB entry 6AW2, grey and cyan, respectively). CEACAM1 does not contact the smaller insertion domain (red). (e) Conformational changes in CagL (grey) in response to pH exposes the RGD motif (cyan spheres). Right – At low pH, CagL exists in an extended state, with the α1 helix burying the arginine of the RGD motif (PDB entry 4X5U). Left – At high pH, CagL compacts and α1 moves exposing the RGD motif and allows recruitment of integrins (PDB entry 3ZCI)

Full size image

3 Adhesins

Protein adhesins may recognize small host cell molecules (e.g., sugars) or larger ligands (e.g., proteins). Below, the known H. pylori adhesin proteins are discussed in detail, grouped according to the host cell binding partner to which they bind.

3.1 Adhesins with Known Small Ligands

3.1.1 BabA/BabB/BabC

These adhesins are also known as HopS/HopT/HopU, respectively. The blood group antigen-binding adhesins (Bab) are approximately 80 kDa in size. The majority of research conducted on these proteins has been focused on BabA. BabA recognizes and can bind the mono- (H) or di-fucosylated (Lewis b, Leb) blood group antigens from the O blood group (Boren et al. 1993; Ilver et al. 1998) and the A and B blood group antigens (A-Leb and B-Leb) (Aspholm-Hurtig et al. 2004), all of which are found on the surfaces of gastric epithelial cells and certain secreted mucins, including MUC1 and MUC5AC of the stomach (Linden et al. 2004; Sakamoto et al. 1989) and MUC5B in the salivary glands (Bosch et al. 2000; Veerman et al. 1997). Analysis of BabA binding to salivary proteins by matrix-assisted laser desorption ionization-mass spectrometry (MALDI-MS) identified two proteins, which could bind BabA – proline-rich glycoprotein and gp-340 (Walz et al. 2009). BabB displays no Leb binding activity, whilst the binding specificity of BabC is unknown (Kim et al. 2015; Saberi et al. 2016). BabA exists in two allele forms, babA2, which encodes for the full length protein, and babA1, which contains a 10-bp deletion in the signal peptide that results in a frame shift (Backstrom et al. 2004; Ilver et al. 1998). The location within the chromosome is highly variable. BabA and BabB are typically found at either locus A or B, with BabC commonly found in locus C (Armitano et al. 2013; Colbeck et al. 2006; Hennig et al. 2006). Expression of BabA is regulated by phase variation in the signal peptide through a cytosine-thymine (CT) dinucleotide repeat (Ilver et al. 1998). The CT repeats can allow slipped strand mispairing (SSM) during replication and, thus, induce a reading frame shift (Colbeck et al. 2006; Ilver et al. 1998; Solnick et al. 2004; Styer et al. 2010). Regulation can also occur in response to host mucin expression (Skoog et al. 2012) and through gene conversion of babA with babB, creating BabA/B chimeras that have varying abilities to bind Leb (Backstrom et al. 2004; Colbeck et al. 2006; Matteo et al. 2011; Pride and Blaser 2002; Solnick et al. 2004). In models of acute H. pylori infections in mice, gerbils and rhesus monkeys, BabA expression was found to be lost regularly during the early stages of infection due to phase variation or gene conversion (Hansen et al. 2017; Ohno et al. 2011; Solnick et al. 2004; Styer et al. 2010). This, however, appears to be a rare event in humans (Nell et al. 2014). However, these strains were isolated from chronically infected humans, for which mutation rates are approximately ten times slower compared to strains commonly found in acute infections (Nell et al. 2014). This surge of mutations observed in acute infections allows these H. pylori strains to respond quickly and adapt to the host environment (Linz et al. 2014).

The affinity of the extracellular domain of BabA for the blood group antigens is rather weak. Measured affinities include: Leb antigen hexasaccharide, KD = 250 μM; Leb antigen pentasaccharide, KD = 80 μM; H antigen pentasaccharide, KD = 620 μM; B-Leb septasaccharide, KD = 40 μM; and A type 1 hexasaccharide (A6–1), KD = 150 μM (Hage et al. 2015; Moonens et al. 2016). These are much weaker than Leb antigen interaction with full length BabA as measured by surface plasmon resonance (SPR) analyses and cell binding assays, which exhibits an affinity of 390 pM (Aspholm-Hurtig et al. 2004; Ilver et al. 1998; Imberty et al. 2005; Moonens et al. 2016). Cross-linking of full length BabA indicates that it oligomerizes, primarily as trimers in H. pylori outer membranes (Moonens et al. 2016). The structure of the extracellular domain of BabA is a 4 + 3 helical bundle fold (Moonens et al. 2016) (Fig. 2b). An 80-residue insertion domain comprised of a four-stranded sheet with a helical loop crowning the beta-strands is located between helices 4 and 5. Several structures of BabA in complex with various blood group antigens have been determined by X-ray crystallography (Hage et al. 2015; Moonens et al. 2016). In these structures, all of the glycans bind to the insertion domain, specifically to two loops (Loop 1 and Loop 2) that connect the strands and the crowning helical loop (Moonens et al. 2016). The crowning helical loop is constrained by a disulfide that, upon reduction, prevents glycan binding (Moonens et al. 2016). DL1 and DL2 differ in sequence considerably across H. pylori strains and, consequently, H. pylori isolates exhibit distinct ABO preferences and Leb affinities (Aspholm-Hurtig et al. 2004). Most strains produce BabA generalist adhesins that promote binding to each blood group glycan; however, several strains are Leb-only specialists. These strains are found to have a shorter Loop 1, thereby preventing binding of the N-acetylgalactosamine or galactose sugar moieties in the larger A-Leb and B-Leb antigens, respectively (Aspholm-Hurtig et al. 2004).

As H. pylori resides in the stomach, it experiences a large pH gradient. The adherence of H. pylori in the gastric mucosa mediated by BabA-Leb interactions display similar affinities between pH 4.0 and 6.0 (Bugaytsova et al. 2017). Further lowering the pH results in a 1000-fold reduction in adherence. Reconditioning of the H. pylori to a higher pH results in the recovery of binding to gastric mucosa, demonstrating that BabA displays a reversible pH sensitivity, which has been localized to residue 199 of the crowning helical loop (Bugaytsova et al. 2017). This residue resides in yet another region of BabA that is hypervariable in length and sequence amongst H. pylori strains. The pH50 of Leb binding (the pH value at which BabA retains 50% of its Leb binding) was determined for tens of strains and found to vary ~2.5 pH units, from 2.3 to 4.9. Deletion of residues 199–200 of BabA in strain 17.875 resulted in an increase of the pH50 from 3.3 to 3.9, mimicking strains where these residues are naturally missing (Bugaytsova et al. 2017). The presence of BabA can aid in the adherence of H. pylori, though it is not essential as many strains exist that lack the babA gene (Odenbreit et al. 2009) and strains that do contain babA have been observed to cause PUD or GC (Gerhard et al. 1999; Yamaoka et al. 2002c). This is particularly true if the cagPAI is also present (Azevedo et al. 2008).

3.1.2 SabA/SabB

These adhesins are also known as HopP and HopO, respectively. The sialic acid binding (Sab) proteins are slightly smaller than BabA with a molecular weight of ~70 kDa (Alm et al. 2000). SabA recognizes and binds sialylated glycans, whilst SabB does not (Mahdavi et al. 2002). The most studied ligand is the sialyl-Lewis x (s-Lex) sugar found attached to O-glycans. Sialyated glycans are typically found in low concentrations in normal gastric musoca (Kobayashi et al. 2009). However, SabA can bind two minor gangliosides in the stomach, Neu5Acα3-neolactohexaosylceramide and Neu5Acα3-neolactooctaosylceramide, which can promote initial infection (Benktander et al. 2018). H. pylori most often cause gastric inflammation by triggering IL-8 induction in host cells responding to the infection. IL-8 activates FUT3 and B3GNT5, two genes involved in the biosynthesis of s-Lex (Magalhaes et al. 2015), leading to dramatic alterations in ganglioside sialyation profiles of the gastric mucosa and an enrichment of s-Lex (Benktander et al. 2018; Magalhaes et al. 2015). SabA has been found to interact with several host cell glycoproteins that are sialyated including MUC5B, MUC7, laminin, carbonic anhydrase VI, zinc α2-glycoprotein, parotid secretory protein and the heavy chain of secretory IgA1 (Aspholm et al. 2006; Walz et al. 2005, 2009). SabA can also bind sialylated proteins on erythrocytes, which leads to hemagglutination (Unemo et al. 2005).

Regulation of SabA expression is complex. Phase variation is observed through two SSM mechanisms: one is observed in the CT dinucleotide repeat of the signal peptide as for babA and the second found within a poly-thymine repeat in the promoter region (Harvey et al. 2014; Kao et al. 2012; Yamaoka et al. 2002b, 2006). This can affect transcription of sabA through either altering regulatory protein interactions and/or RNA polymerase. The high sequence similarities between sabA and sabB also allow gene conversion between the two (Talarico et al. 2012). SabA expression is also regulated by the external pH, through the acid-responsive ArsRS two-component signal transduction system (Goodwin et al. 2008). At low pH (pH <5.0), SabA and SabB expression are repressed, whilst at higher pH they are upregulated. The type of mucins present in the mucosa can also regulate the expression of SabA (Skoog et al. 2012). Tumor mucosa from several patients was found to consist of different mucins and glycosylation patterns. These differences were found to have an effect on SabA expression (Skoog et al. 2012). High salt concentrations can upregulate SabA expression (Loh et al. 2018); several studies have revealed a link between high salt intake and an increase GC risk in humans.

The affinity of the extracellular domain of SabA for s-Lex is slightly tighter than BabA is for Leb, with an affinity of 20 μM as determined by SPR (Pang et al. 2014). The same study indicated that SabA can also bind non-sialyated Lewis x (Lex), albeit weaker with an affinity of 50 μM (Pang et al. 2014). No binding was observed between SabA and Lewis A, Leb or Lewis Y glycans. Structurally, SabA is similar to BabA, as they share the 4 + 3 helical bundle fold (Fig. 2c). However, the insertion domain of SabA differs in sequence, is 50–70 residues shorter and some of the residues appear to be conformationally dynamic, as they are not resolved in the X-ray crystal structure. Although no high-resolution structure of a SabA-s-Lex exists, an alanine scan of conserved residues, from a multiple sequence alignment of BabA and SabA sequences and a ligand binding site prediction program, identified a potential ligand binding pocket on the surface of SabA (Pang et al. 2014) (Fig. 2c). Two mutations (Y148A and K152A) that had no effect on s-Lex binding were found to weaken binding to Lex. The Q159A mutation inhibited SabA binding to both s-Lex and Lex, whereas the Q162A mutation only inhibited binding to Lex (Pang et al. 2014). This binding pocket is distinct from the insertion domain of BabA which binds blood group antigens and the carcinoembryonic antigen-related cell adhesion molecules (CEACAM) binding loop of HopQ (see below).

SabA + strains appear to be associated with GC as observed in a diverse cohort of patients (Yamaoka et al. 2006). However, another study restricted to Taiwanese patients failed to identify any significant differences with patients infected with sabA+ and sabA− strains and the prevalence of gastric atrophy (Sheu et al. 2006). A similar observation is seen in a Japanese study restricted to Japanese patients (Yanai et al. 2007), suggesting that there may be geographical and environmental factors confounding the link between SabA and disease incidence and severity.

3.1.3 HopD

This GalNAcβ1-4GlcNAc glycan motif (N,N′-diacetyllactosediamine or lacdiNAc) -binding adhesin (LabA) is a protein with a molecular weight of 77 kDa (Alm et al. 2000). LacdiNAc is only observed as an O-linked glycan on MUC5AC expressed on the superficial and foveolar epithelium of the stomach (Rossez et al. 2014). The lacdiNAc motifs comprise ~7% of human adult gastric mucin O-glycans (Kenny et al. 2012; Rossez et al. 2014). Several different strains of H. pylori were shown to adhere to lacdiNAc, with strain 26.695 showing the strongest adherence. H. pylori lysate from strain 26.695 was incubated with gastric mucins in the presence and absence of soluble lacdiNAc as a competitive binder. The supernatants were compared by SDS-PAGE analysis and revealed a prominent band in the competition experiment. Proteomic analysis identified this protein as HopD (Rossez et al. 2014). No structure has been reported for HopD, though it probably has a structure grossly similar to BabA and SabA.

3.2 Adhesins with Known Proteins

3.2.1 HopQ

HopQ is a 68 kDa protein (Alm et al. 2000) found in two allelic forms, Type I and Type II (Cao and Cover 2002). These two forms share approximately 70% sequence identity at the protein level. Both HopQ types have a significant association with GC and gastritis, with Type I HopQ also found to be associated with an increased risk of PUD (Leylabadlo et al. 2016; Ohno et al. 2009). Furthermore, HopQ Type I is found significantly more often in cagPAI+ versus cagPAI− strains (Loh et al. 2008). hopQ was the first gene identified that is located outside of the cagPAI and is also essential for the translocation of the CagA oncoprotein through the T4SS (Belogolova et al. 2013; Jimenez-Soto et al. 2013). HopQ expression is regulated by salt concentrations like SabA, with higher amounts of salt leading to HopQ upregulation (Loh et al. 2018).

HopQ binds CEACAM receptors on host cell surfaces (Javaheri et al. 2016; Königer et al. 2016). Twelve CEACAMs are found in humans (Tchoupa et al. 2014) and display distinct expression patterns, with certain CEACAMs only expressed in specific tissue types (Hammarstrom 1999; Zebhauser et al. 2005). Various CEACAM members are found to possess a similar domain architecture: they are comprised of a single N-terminal IgV domain, which predominately homodimerizes, though a few can also heterodimerize; followed by a variable number of IgC2 domains and a C-terminal transmembrane helix or a glycosylphosphatidylinositol anchor (Gray-Owen and Blumberg 2006; Tchoupa et al. 2014). HopQ is found to bind the N-terminal dimerization domains of only CEACAM1, -3, -5 and -6 as determined by flow cytometry with CEACAM6 binding the weakest (Javaheri et al. 2016; Königer et al. 2016). A humanized mouse model that expresses CEACAM5 results in a significant difference in gastritis activity upon H. pylori infection compared to control mice (Königer et al. 2016). Isolation of H. pylori from these humanized mice after 6 weeks of infection display a more active T4SS as observed by the amount of CagA translocated in AGS cells in vitro. This is followed by a less active T4SS and lower IL-8 induction in CEACAM5 mice after 3 and 12 months of infection compared to wild-type mice (Königer et al. 2016). It is thought that to prevent clearance of the infection, H. pylori responds to the inflammation by lowering the activity of T4SS (i.e., a rheostat model) (Barrozo et al. 2013; Königer et al. 2016). HopQ binds one CEACAM monomer (Bonsor et al. 2018; Javaheri et al. 2016; Königer et al. 2016; Moonens et al. 2018; Tegtmeyer et al. 2019), with CEACAM1 and CEACAM3 binding affinities of ~200 and ~400 nM, respectively (Bonsor et al. 2018). CEACAM1 binding to the Type II HopQ allele is ~6-fold tighter (Moonens et al. 2018). The structure of HopQ, like those of BabA and SabA, is comprised of the common 4 + 3 helical bundle, with an insertion domain that is longer and better resolved than that of SabA, but shorter than that of BabA (Javaheri et al. 2016) (Fig. 2d). Crystal structures of the Type I HopQ in complex with CEACAM1 and CEACAM3 clearly show that CEACAMs interact with a disordered loop (in the unbound HopQ structure) of 13 residues that folds across the CEACAM dimerization interface and extends the CEACAM beta-sheet, such that Type I HopQ recognizes the monomeric form of CEACAMs and disrupts their dimerization (Bonsor et al. 2018; Moonens et al. 2018). Alanine mutagenesis of the Type I HopQ loop or the CEACAM dimerization interface failed to identify a critical residue that was important for binding (Bonsor et al. 2018), though larger substitutions in Type I HopQ such as L150N and V156N resulted in significant reduction in H. pylori binding to CEACAM1 expressing MKN28 cells (Moonens et al. 2018). Shortening of the loop weakened binding, whereas swapping the loop with that of BabA inactivates binding of CEACAMs and impairs translocation of CagA (Bonsor et al. 2018). The crystal structure of the Type II HopQ CEACAM1 complex is very similar to the Type I HopQ structure with two major differences. First, the disordered loop is shorter in Type II HopQ and as such does not extend the CEACAM beta-sheet. Second, the loop is more hydrophobic and results in less hydrogen bonds across the dimerization interface (Moonens et al. 2018).

The role of pH, disulfide bonds and glycans on the HopQ-CEACAM interaction have also been investigated. BabA binding to Leb was both pH sensitive and reversible (Bugaytsova et al. 2017), whereas HopQ could still bind CEACAM1 at pH 4.0, but at lower pH values binding was neither detectable nor reversible (Bonsor et al. 2018). The CEACAM binding loop of HopQ, unlike the Leb binding site on BabA, is not constrained by a disulfide bond, however, a disulfide exists preceding the loop and other loops proximal to the CEACAM binding site contain disulfides. Reduction of these disulfide bonds had little impact on the affinity of HopQ for CEACAM1 (Bonsor et al. 2018). CEACAMs are heavily glycosylated proteins. CEACAM1 decorated with high mannose type glycans or no glycans bind with a similar affinity to HopQ compared with the aglycosylated N-terminal domain of CEACAM1. However, CEACAM1 glycosylated with complex type glycans are found to bind eightfold tighter to HopQ, suggesting that glycans may have a role in HopQ binding through some unknown mechanism (Bonsor et al. 2018).

3.2.2 AlpA/AlpB

Also known as HopC and HopB, respectively, AlpA and AlpB share 45% sequence identity with a molecular weight of ~56–57 kDa (Alm et al. 2000). AlpA and AlpB appear to be co-expressed in all clinical strains of H. pylori, suggesting an essential or important function for these proteins (Odenbreit et al. 2009), in contrast with other OMPs. AlpA expression is upregulated in response to oxidative stress (Huang and Chiou 2011). Deletion of these genes results in the failed infection of guinea pigs and gerbils, the inability to adhere to human gastric tissue sections (de Jonge et al. 2004a; Senkovich et al. 2011; Sugimoto et al. 2011), and failed to stimulate secretion of IL-8 and IL-6, suggesting that these proteins are pro-inflammatory (Lu et al. 2007; Selbach et al. 2002). Stimulation of IL-8 expression may be specific to the geographic location of the strain, as deletion of AlpA and AlpB only reduced IL-8 secretion with East Asian strains (Lu et al. 2007). All strains that possess AlpA and AlpB can perturb host cell signaling pathways such as ERKs, c-Fos and cAMP-responsive element binding protein, whereas activation of Jun N-terminal Kinase, c-Jun and NF-κB signaling were specific to East Asian strains (Lu et al. 2007).

A solution of semi-purified human extracellular matrix causes aggregation of H. pylori (Williams et al. 2008), which is significantly reduced in a ΔalpA/B strain (Senkovich et al. 2011). This effect was also observed in Matrigel, a mixture of predominately collagen IV and laminin from Engelbreth-Holm-Swarm mouse sarcoma cells (Senkovich et al. 2011). Experiments with purified mouse laminin show clear binding of AlpA and AlpB by flow cytometry (Senkovich et al. 2011). Expression of AlpA and AlpB in Escherichia coli causes a gain of function, allowing these bacteria to adhere to mouse laminin (Senkovich et al. 2011).

3.2.3 CagL

CagL is a 25 kDa protein that forms part of the H. pylori T4SS, responsible for translocation of CagA into host gastric epithelial cells (Fischer et al. 2001; Kwok et al. 2007). CagL is thus strongly associated with an increased risk of GC. CagL is expressed and attached to the surfaces of the T4SS pili that form when H. pylori physically contacts gastric epithelial cells (Shaffer et al. 2011). CagL is essential for the translocation of CagA as deletion of CagL causes the failure of the formation of the pili (Fischer et al. 2001; Kwok et al. 2007; Shaffer et al. 2011). CagL contains an Arg-Gly-Asp (RGD) motif, a known integrin binding ligand. Various studies have demonstrated that CagL can bind to α5β1, αVβ3, αvβ5 and αvβ6 through the RGD motif (Barden and Niemann 2015; Conradi et al. 2012a, b; Kwok et al. 2007; Wiedemann et al. 2012), although three studies have shown that CagA can still be translocated in an RGD-independent manner – deletion of CagL in the P12 strain had no effect on CagA translocation nor IL-8 secretion (Jimenez-Soto et al. 2009), deletion of the RGD motif in the SU2 strain resulted in a weakened interaction between CagL and the α5β1 and αvβ5 integrins and no observed binding to αvβ6 (Bonig et al. 2016) and the CagLRGA mutation only weakened the interaction with αvβ5 (Wiedemann et al. 2012). CagL binds to integrins with an affinity of ~100–200 nM (Koelblen et al. 2017; Kwok et al. 2007; Wiedemann et al. 2012), independent of whether the integrin is in an extended closed or open state (Koelblen et al. 2017). This interaction is pH sensitive, with maximal binding occurring at pH 6.5 (Bonsor et al. 2015).

Several X-ray crystal structures of CagL have been determined. The first two structures revealed a four helix bundle (Barden et al. 2013; Choi et al. 2015). The RGD motif is located on a helix, which is currently the only known RGD motif not found in a loop (Fig. 2e). The following two structures identified that CagL could undergo a large conformational change, where two of the antiparallel helices become fused to form a single longer helix (Fig. 2e) and then dimerized through a domain swapped dimer mechanism (Barden et al. 2014; Bonsor et al. 2015). This conformational change is a result of low pH in both crystals and solution, though it is found not to be important for adhesion, IL-8 secretion nor CagA translocation (Bonsor et al. 2015). However, subtle conformational changes are apparent in the first helix (α1), which packs against the RGD motif (Barden et al. 2013; Bonsor et al. 2015). At low pH, α1 buries the arginine of the RGD preventing adhesion to host cells, whereas at higher pH α1 undergoes a registry shift, exposing the RGD motif and thus allows adherence, cell spreading, focal adhesion formation and heparin-binding epidermal growth factor activation (Bonsor et al. 2015; Saha et al. 2010; Tegtmeyer et al. 2010).

Several polymorphisms exist in the α1-α2 loop (residues 58–62), which may affect disease outcome in H. pylori infections. Worldwide, E59/I60 polymorphisms are associated significantly with GC-associated H. pylori isolates (Gorrell et al. 2016). Several studies indicated specific polymorphisms in local populations: in Iranian patients, the D58 polymorphism is typically observed in PUD, whereas the N58 polymorphism is associated with GC (Cherati et al. 2017); whilst in Taiwanese patients, the Y58/E59 mutations were over-represented in GC patients (Yeh et al. 2011). The role of Y58/E59 mutations is, however, debated as it has produced conflicting data. Replacement of the Y58/E59 mutation with D58/K59 resulted in a strain with a less active T4SS, lower CagA translocation and IL-8 secretion (Yeh et al. 2013). However, mutation of CagL in strain 26,695 (N58Y/E59E) inhibited CagA translocation in another study (Tegtmeyer et al. 2014). A final report investigated a larger group of polymorphisms (Y58/E59, D58/K59, D58/E59, N58/E59 and N58/K59) in both the P12 and 26.695 strains of H. pylori and found no significant changes in CagA translocation or IL-8 secretion (Tafreshi et al. 2015).

3.3 Adhesins with Unknown Ligands or Proteins

3.3.1 HorB

HorB or HP0127 is a 30 kDa protein. While very little is known about this protein, it is predicted to have a C-terminal β barrel domain and an N-terminal signal peptide, as well as a domain architecture similar to the Hop proteins (Snelling et al. 2007). Deletion of the gene results in a H. pylori strain that has a twofold reduction in adhesion, a lower production of LPS O-chains and thus the Lewis X and Y antigens attached to it. In mouse infection assays, colonization was reduced for ΔhorB strains (Snelling et al. 2007).

3.3.2 HomA/HomB/HomC/HomD

These four proteins form the smallest OMP family and are each approximately 75 kDa in size (Oleastro et al. 2008). HomA and HomB share 90% sequence identity, and are 50% identical to HomC and HomD (Alm et al. 2000). HomC exists in three allelic forms (Kim et al. 2016). HomA and HomB occupy two defined loci within the bacterial chromosome (Alm et al. 2000). HomB can exchange positions with HomA and may have resulted from gene duplication (Alm et al. 1999, 2000; Oh et al. 2006; Tomb et al. 1997). Deletion of HomB results in H. pylori strains that exhibit less adherence to gastric epithelial cells and cause lower IL-8 secretion (Oleastro et al. 2008). One strain tested had two copies of HomB. Sequential deletion of the genes led to a further decrease in adherence (Oleastro et al. 2008). All four proteins are predicted to be 24-stranded β barrels (Servetas et al. 2018). The sequence variation between HomA and HomB is predicted to occur within the extracellular loops. Similar variance is observed in HomC (Servetas et al. 2018).

HomB is strongly associated with PUD (Oleastro et al. 2008). It also correlates with the presence of the cagA, babA, hopQ and oipA genes (Oleastro et al. 2008). Correlation is also observed with one of the homC alleles and the presence of babA at locus A (see BabA), suggesting that HomC may play a similar role in disease as BabA (Kim et al. 2016).

3.3.3 OipA

Outer inflammatory protein A, or HopH, is relatively small compared to the other abovementioned OMP adhesins, with a molecular weight of ~34 kDa (Alm et al. 2000). No structural data exists for this protein, but like the other Hop proteins is predicted to have a C-terminal β barrel and an N-terminal signal peptide (Alm et al. 2000). Therefore, the extracellular domain would be small, consisting of approximately 75 residues. OipA regulation is achieved through phase variation in the CT nucleotide repeat of the signal peptide and is found either in an oipA-off or oipA-on state (Saunders et al. 1998; Yamaoka et al. 2000). OipA is believed to stimulate IL-8 secretion from host cells and cause inflammation, hence its name, though the evidence for this is conflicting. Several studies showed that oipA mutants did not alter IL-8 secretion in vitro or inflammation in gerbils (Dossumbekova et al. 2006; Franco et al. 2008). However, other studies clearly show that OipA does in fact cause IL-8 secretion and promotes inflammation (Yamaoka et al. 2000, 2002a). Furthermore, it has been shown that IL-8 secretion by OipA is regulated through PI3K/Akt activation and inactivation of FoxO1/3a (Tabassam et al. 2012). This conflict is thought to be compounded by the fact that OipA is strongly associated with the cagPAI, which stimulates IL-8 secretion. Indeed, two studies have shown that greater than 95% of cagPAI+ strains contain an oipA-on allele, whilst no cagPAI− strain has been found with an oipA-on allele (Ando et al. 2002; Farzi et al. 2018; Odenbreit et al. 2009). Mutation of the H. pylori J68 strain (cagPAI−/ oipA-off) to produce an oipA-on strain, results in a bacterium with increased adherence, but did not alter IL-8 secretions of host cells (Horridge et al. 2017). Mutation of the 26.695 strain (cagPAI+/oipA-on) to produce an oipA-off strain failed to stimulate IL-8 secretion, demonstrating that OipA is essential, but insufficient for IL-8 secretion (Horridge et al. 2017). The 26.695 cagPAI+/oipA-off mutant strain was also found incapable of CagA translocation (Horridge et al. 2017). This is the second adhesin gene that does not reside with the cagPAI locus (HopQ was the first, see above), which has been shown to be essential for a functional T4SS. As the oipA-on allele correlates so strongly with cagPAI+ strains, OipA is found to be associated with PUD, GC and MALT lymphoma (Dabiri et al. 2009). Purified OipA is found to trigger apoptosis of gastric epithelial cell lines through increasing Bax and cleaved Caspase 3 concentrations and lowering of Bcl-2 (Teymournejad et al. 2017).

3.3.4 HopZ

HopZ is a ~74 kDa protein (Alm et al. 2000; Peck et al. 1999). Like hopQ, its closest homologue (Oleastro and Menard 2013), hopZ exists as two alleles, which can undergo recombination (Kennemann et al. 2011; Peck et al. 1999). HopZ expression is regulated through SSM within the CT dinucleotide repeats in the signal peptide. The switch from off-to-on is influenced by colonization density and during infection (Kennemann et al. 2011, 2012; Peck et al. 1999; Yamaoka et al. 2002b). Deletion of hopZ results in lower adherence of H. pylori to gastric epithelial cell lines (Yamaoka et al. 2002b), while mutation of the gene does not affect colonization of guinea pig stomachs (de Jonge et al. 2004a). However, in germ-free transgenic mice, hopZ deletion reduces H. pylori survival in their stomachs (Giannakis et al. 2009). HopZ does not appear to be associated with gastric disease, except for MALT lymphoma where lower expression levels of HopZ are found (Chiarini et al. 2009; de Jonge et al. 2004b; Kennemann et al. 2012). Indeed, the hopZ-off state is found to be associated with MALT lymphoma (Lehours et al. 2004).

4 Conclusions

H. pylori has successfully adapted to life in the human stomach. The ever changing local environment (churning of the stomach, acidic environment, constant shedding of epithelial cells) in the stomach has placed substantial selective pressure on H. pylori to escape the acidic lumen on the stomach and swim towards the gastric epithelial lining, where it can use a repertoire of over 60 outer membrane proteins to achieve adherence to host cells. This large selection of adhesion molecules allows H. pylori to rapidly change the proteins that it presents on its cell surface to adhere the host proteins and/or ligands that are present in the local environment. Furthermore, altering its outer membrane proteins aids in lessening of the immune response by the host and, thus, its elimination. However, some of the proteins are associated with a more severe clinical outcome. This review has discussed several important adhesion-host protein/ligand interactions which are important in H. pylori colonization, survival and disease. These data could provide druggable or vaccine targets for the eradication of H. pylori.