Outer membrane phospholipase A’s roles in Helicobacter pylori acid adaptation

Our bioinformatics work, combined with an extensive literature study, and backed up by experimental work on variants with non-functional OMPLA is summarized in Fig. 1. The remainder of this article discusses all aspects of this model.

The H. pylori acid regulation pathway has been well studied and often debated (see Additional file 1 for further details) [1], but all currently available functional models still require unknown OM components for urea influx and ammonium efflux. Our in silico results indicate that H. pylori OMPLA can be an OM urea and ammonium channel, while AmCI and AmCII, that are likely co-regulated with OMPLA, can be involved in ammonium efflux from the cytosol. The hypothesis presented in this paper, depicted in Fig. 1, is mainly based on in silico models combined with current literature findings on the urea pathway and H. pylori acid tolerance.

Unlike E. coli OMPLA, H. pylori OMPLA is continuously breaking down membrane phospholipids to lysophospholipids when cultivated at physiological pH [17]. The optimal pH for OMPLA’s enzymatic activity is around 7.0, and this activity is abolished at pH 5.0 or lower, yet this protein is required for in vitro survival at very low pH [12].

The OMPLA structure is composed of a 12-stranded transmembrane β-barrel with short periplasmic turns and long extracellular loops. This is consistent with the structure of other outer membrane proteins (OMPs; including that of porins, the largest OMP subfamily) that diffuse molecules through the membrane [16]. The H. pylori OMPLA barrel is large enough to have pore activity [16]. Figure 5 supports the idea that there is enough space for urea to pass the OM through OMPLA. The pores of the E. coli and H. pylori OMPLA vary in shape and amino acid composition, as shown in Fig. 6, supporting our hypothesis that they differ in function. We therefore suggest that H. pylori OMPLA is a multifunctional protein with one function being a phospholipase and a second function being acid protection.

Standard sequence alignment methods are reliable for closely related sequences, but often fail when diverse sequences-like OMPLAs- are analysed [18, 19]. The alignment of the H. pylori OMPLA sequence with the sequence of the E. coli OMPLA template structure (1QD5 PDB) is complicated. To get this alignment we collected as much information as possible about homologous OMPLA sequences and generated a MSA using an iterative profile alignment process. Unlike standard sequence alignment methods, iterative profile alignments can use both structure and function information. The H. pylori and E. coli OMPLA protein sequences are highly different (see Additional file 2) and their sequence identity is just above the threshold for homology modelling [20] when the not-modelled insert (Fig. 2) is not taken into account.

EV analyses the evolutionary footprint left behind in a MSA and identifies regions in the protein in which conservation and variation relate to various aspects of thee protein’s function [16, 19]. The highly variable residues (coloured blue in Fig. 4) observed mainly at the outside facing loops likely are (a-specifically) involved in host evasion [16]. The two highly conserved loop tyrosines (Fig. 4) are located far from the trimer interface, yet they must be functionally important because they are conserved in the MSA.

Most gastric Helicobacters have a sequence insert that is lacking in the enterohepatic Helicobacters (and other species). Table 2 shows two gastric Helicobacters that are very different and that lack this sequence insert: Helicobacter mustelae and Helicobacter himalayensis. In phylogenetic analyses, H. mustelae cluster with enterohepatic Helicobacters [21]. The genome from H. mustelae type strain 12,198 [22] lacks pldA, the gene coding for OMPLA. H. mustelae possess a nickel-independent urease, called UreAB2 [23]. In the presence of urea this metalloprotein is by itself already sufficient for the bacteria to survive acid shock. It is activated by ferrous ions in the absence of auxiliary proteins [23]. H. mustelae has adapted a different mechanism to survive in the stomach of its host, the ferret. The other gastric species that differs, H. himalayensis lacks urease genes, which indicates that it lacks the whole urease pathway. Generally, gastric and enterohepatic Helicobacter OMPLAs are quite different; enterohepatic Helicobacters lack the insert found in most gastric Helicobacters, except the OMPLAs from H. mustelae and H. himalayensis, that seem to have evolved totally different systems to cope with low pH.

The not-modelled insert is similar to that found in the sequence alignment of a subset of sequences presented by Istvan et al. [11]. This not-modelled insert (see Table 2; Figs. 2, 7 for more details) can be of importance for pH-gating as observed in other proteins such as H. pylori UreI [24] and E. coli OmpG [25]. pH sensitivity has also been detected in the E. coli OmpF constriction loop [26].

We do not know the function of the residues in the predicted extracellular loop (the not-modelled insert highlighted in Fig. 2b), but literature has shown that mutating a charged extracellular loop residue can have deleterious effects on acid survival [27, 28]. Our laboratory results (the missense mutations P157S and S235R, see Additional file 5 for residue numbering) revealed the importance of these two OMPLA residues. Ser235 is located in the not-modelled loop insert. Arginine is a positively charged residue and its introduction will lead to electrostatic changes, perhaps resulting in a functionally closed pore. P157 is one of the highly-conserved residues. It is located at the trimer interface (see Fig. 4). This mutation likely destabilizes the β-barrel structure [16, 29, 30], or might disable acidic protection by disturbing the trimer interface that is required for pore-activity [16]. The conserved cysteines, found in the gastric Helicobacter OMPLA lie in this extracellular loop. Cys residues are seldom found in OMPs [31] and they are seldom found facing the surface; however, conserved cysteines have a wide range of functions and are usually of great importance to the protein [32‐34]. Although protein stabilization would be a likely function for this conserved disulfide bond, further laboratory work is needed to confirm its function. Inspection of the model suggests that Y240 stabilizes the extracellular loops; especially the interactions between the first and third extracellular loop. We have previously predicted that all porins function as a trimer [16], and provided evidence that OMPLA forms a trimeric structure too. The Y233 is located near the putative trimeric hole. This suggests a regulatory role for this residue, but what that role might be remains unclear.

OMPLA’s operon structure sheds light on other relevant proteins in the urea pathway (see Fig. 1). The gastric and enterohepatic Helicobacter species have different operon organizations, as illustrated in Fig. 8. Genes located in the same operon normally show co-expression and tend to be regulated by the same promoter. Price et al. found that the life-cycle of an operon is under strong selection [35] and genes found in the same operon are likely involved in the same process [36, 37]. Gastric Helicobacters have a different gene expression compared to enterohepatic Helicobacters, including a higher level of urease expression [23]. The hosts of the gastric Helicobacter group are more diverse than the hosts of the enterohepatic group, but their pldA operons nevertheless are more similar to each other; some consisting of five consecutive genes (2 SLC6sbd_Tyt1-Like genes, pldA, dnaN, and gyrB). The two transporter (channel) genes upstream of the pldA gene are generally predicted to lie in the same operon as pldA in gastric Helicobacters, as shown in Table 3. The gene encoding AmCI is also predicted to be phase variable (see Table 3). The two gastric bacteria H. himalayensis (lacking the urease genes) and H. mustelae (lacking the pldA gene) and enterohepatic Helicobacters, have different organization and they have probably evolved different mechanisms for acid adaptation.

Homology searches in clusters of orthologous groups (COGs) database [37] revealed that AmCI and AmCII both belong to COG0733 (Na⁺-dependent transporters (channels) of the SNF family). Since genes in an operon are regulated together, the two Na⁺-dependent transporters (channels) of the SNF family (AmCI and AmCII) are also likely implicated in acid survival and we suggest that they are involved in ammonium/ammonia transport or diffusion from the cytosol to the periplasm. We do not know why two very similar channels are needed in this process, but visual inspection of their 3D structure models suggests the possibility that one of the channels can be ammonium specific while the other is specific for ammonia. While the OMPLA family (COG2829) is mainly found in Proteobacteria, Na⁺-dependent COG0733 channels are widespread throughout bacteria and eukaryotes. In order to better understand the possible role of these proteins in the urea pathway (being co-regulated with OMPLA), 3D model structures were constructed for AmCI and AmCII (see Fig. 9). The AmCI and AmCII transporters have closest sequence similarity with Bacillus halodurans MhsT (a BLAST search against the PDB resulted in 31–35% sequence identity for the H. pylori HP0497 and HP0498 sequences). MhsT is a Na⁺-dependent neurotransmitter/sodium symporter and belong to the SLC6 family of Na⁺/Cl⁻-dependent neurotransmitter transporters. These proteins transport small substances, e.g. amino acids or similar structures [38, 39]. The substrates found among SLC6 transporters include glycine, serotonin, dopamine, and norepinephrine [40]. They have a so-called 5 + 5 core helix motif that is embedded in the membrane, albeit that we cannot exclude the presence of more helices; MhsT, for example, has 11 helices, while LeuT has 12 helices [38]. Evolutionary, these transporters adjust quickly to changes; some of them, for example, are voltage-gated channels under certain conditions [41]. We believe that the two putative ammonium channels, AmCI and AmCII, are located in the inner membrane, because helical proteins are seldom found in the outer membrane of Gram-negative bacteria [16]. Likely functions include pH sensing, or solute transport of small substances that are involved in buffering the environment [42‐46]. We hypothesize that since they are co-regulated with OMPLA, these transporters are involved in the urea pathway, as shown in Fig. 1. Since there are currently no NH₄ ⁺ channels known, we believe the two COG0733 are better candidates than the suggested UreI. UreI have an important role in urea influx, but no experiments show a role for UreI in ammonium efflux [42].

OMPs have multiple functions that aid adaptation in rapidly changing environments [16]. Current literature does not explain which H. pylori OMP is involved in maintaining higher periplasmic pH level compared to the acidic outside environment. In silico protein structure modelling of H. pylori OMPLA indicates a transmembrane pore of 4 Å, and H. pylori OMPLA could participate in acid protection through this transmembrane pore. Our group has previously linked intact H. pylori OMPLA to an increased risk for ulcer disease [47]. A possible explanation for this could be its role in the urea pathway that concludes with the efflux of ammonium that is implicated in ulcer formation.

We propose that urea passes through OMPLA into the periplasm, while ammonium might exit to increase the cytoplasmic and periplasmic pH. Since acid tolerance is observed only when OMPLA is intact, we suggest that OMPLA’s role is to maintain an optimal periplasmic pH as modelled in Fig. 9. We hypothesize that AmCI and AmCII, which are co-regulated with OMPLA, are also implicated in acid protection allowing NH₃/NH₄ ⁺ to move from the cytoplasm to the periplasm. We propose that OMPLA also is involved in the secretion of NH₄ ⁺ from the cell. That gives OMPLA two transport roles -urea influx and, ammonium efflux- which is possible because it is likely constitutively expressed (as observed in other pathogenic bacteria [33]) which suggests that it probably is abundantly present in the outer membrane.