Introduction

Helicobacter pylori (H. pylori) is a bacteria that colonize gastric mucosa in humans and increase the risk of serious diseases such as gastrointestinal ulcers and some cancers [1]. Most of the case, H. pylori cause a minimal damage to the gastric system and the immune response es effective in the bacterial elimination. However, in individual with a genetic predisposition, these bacteria have the ability to form subpopulations capable of immunomodulating the immune response and favoring a more severe inflammatory process, favoring effective harvest of nutrients to feed the bacteria and suppressing the immune response, allowing it to persist in the gastric system [2, 3]. The role of H. pylori infection in the pathogenesis of several extra-gastric diseases has been also suggested [4, 5].

Different mechanisms explain the pathogenicity of H. pylori and the initiation of the inflammatory process with the release of cytokines and chemokines [6]. These molecules facilitate the alteration of endothelial morphology and the passage of different proteins of the microorganism to the general circulation. Therefore, the role of H. pylori in different systemic diseases has been studied, and different mechanisms proposed; the systematic inflammatory could induce an autoimmune response generated by molecular mimicry between the proteins of H. pylori and humans [4, 7]. Molecular mimicry allows different microorganisms to evade the human immune system due to the similarity of some human proteins and microorganisms. However, when lymphocytes recognize these antigens, autoantibodies and/or autoreactive T lymphocytes can be generated.

Pathological processes that lead to Autoimmune Gastritis (AG) and type 1 gastric Neuroendocrine Tumors (type 1 gNETs) are not understood but some evidence suggests at both are autoimmune diseases in which the gastric mucosa is damaged and the production of autoantibodies against the exposed intracytoplasmic proteins are spread. Autoimmune atrophic gastritis is frequently found in association with Hashimoto’s thyroiditis, Graves disease, type 1 diabetes mellitus, Addison’s disease, and chronic urticaria [8]. Infection by H. pylori is common in AG, with a prevalence of 33,3 to 57%, however, the contribution of H pylori to type 1 gNET oncogenesis has not been demonstrated [9–11]. In chronic urticaria, autoimmunity has been strongly associated with its causality and severity [12, 13]. As in GA and gNETs, several epidemiological studies show a possible relationship between H. pylori and urticaria [14, 15], however, the reason for this association has not been explored.

The repertory of components common between H. pylori and the host has been increasing. From the H. pylori proteome, it is possible to evaluate the identity of its proteins in humans. Our hypothesis was the following: Considering that during chronic inflammatory processes there is an increased expression of some hidden proteins, it is possible that autoantibodies are generated against these proteins. In this study, through in silico analysis, we evaluated this hypothesis by comparing the identity of H. pylori proteins with some proteins identified in urticaria inflammation and in type 1 gNET, two extra-gastrointestinal systemic diseases with growing evidence of their possible association with H. pylori.

Material and methods

Antigen analysis

After an exhaustive review of the literature, we selected proteins for in silico analysis involved in urticaria, autoimmune thyroiditis, AG and type 1 gNET. A total of thirty-two amino acid sequences of tumoral neuroendocrine and AG antigens and nine antigens related with urticaria, and thyroiditis were retrieved from Uniprot database (Table 1).

Download:

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

Table 1. Tumoral neuroendocrine antigens and proteins associated with urticaria used to compare with the H. pylori proteome.

https://doi.org/10.1371/journal.pone.0281485.t001

Amino acid sequences of each antigen was used as input in PSI-BLAST to find similar antigens from H. pylori and alignment using two algorithms, BLOSUM62 through PRALINE pairwise from the center for integrative bioinformatics IBIVU (https://www.ibi.vu.nl/programs/pralinewww/) and PAM250 using the EMBOSS Needle pairwise (https://www.ebi.ac.uk/Tools/psa/emboss_needle/), based on the assumption of both short and long evolutionary distances, respectively. For the progressive alignment strategy, gap penalty was used of 12 opens, 0.5 and 1 extension, with an iteration of 3. Search was limited to taxid 210, corresponding to the specific species H. pylori. Antigens with similarity lower than 30% were discard.

Results

H. pylori antigens selection

From the thirty-two tumoral neuroendocrine and AG human antigen sequences retrieved, eight share identities with ten H. pylori proteins: The Lewis antigens (3-galactosyl-N-acetylglucosaminide 4-alpha-L-fucosyltransferase and Galactoside alpha-(1,2)-fucosyltransferase 2), heat shock protein 60 (HSP 60), H/K ATPase, carbonic anhydrase, the heavy neurofilament protein and the neuro-specific alpha and gamma enolases. The H/K ATPase was the only one that present identity with more than one H. pylori protein (Table 2).

Download:

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

Table 2. Pairwise alignment results from human proteins and H. pylori proteins using Blosum 62 and PM250 algorithms.

https://doi.org/10.1371/journal.pone.0281485.t002

None of the nine proteins associated with urticaria and thyroiditis showed identity with the H. pylori proteome.

All the human proteins with significant identity, presented functional homology with the H. pylori proteins and have enzymatic functions: two fucosyltransferases, three ATPases, one carbonic anhydrase and two phosphopyruvate hydratase. There were not a high different in the identities between the BLOSUM62 and PAM250 algorithms. The highest conserved sequence was found in the HSP 60 antigen, with an identity of 54% and a cover of 92%, followed by the alpha and gamma enolases, both had an identity level of 48% and cover of 96%. The H/K ATPase showed high identity with two H. pylori proteins (35,21% with both HAD IC family P type ATPase and Cadmium-translocating P-type ATPase), but the cover were low (only 6%). The lowest conserved sequence was present with the third H/K ATPase pairwise, with an identity of 17% and cover of 11%.

Modelling and epitope prediction

From the eight tumoral neuroendocrine human antigens, five did not present a 3D experimental structure: HSP60, H/K ATPase (Chain A), heavy neurofilament protein, and the two Lewis antigen (FUT2 and FUT3) and were modelling based on homology with the Swiss Model server. The templated used were 60 kDa heat shock protein mitochondrial homology (100%) for HSP60, Potassium-transporting ATPase alpha chain 1 (96,71%) for H/K ATPase, Keratin type I cytoskeletal 10 (37,5%) for the heavy neurofilament protein and Alpha-(1,6)-fucosyltransferase (17,27%) and Alpha1,3-fucosyltransferase (21,95%) for the two Lewis antigens, FUT 2 and FUT 3, respectively. Models showed typical fold expected for their protein family. For those whose did not fit in range of 0–1 was refined in Deep view software.

By using Ellipro server, the lineal and discontinuous epitopes on the tumoral neuroendocrine protein were predicted. By the selection criteria, the server threw four lineal and four discontinuous for FUT 2, three both lineal and discontinuous epitopes for FUT 3, eight lineal and three discontinuous epitopes for HSP 60, three lineal epitopes for carbonic anhydrase and non-valid discontinuous epitopes, three lineal and one discontinuous epitopes for alpha enolase and gamma enolase, nine lineal and three discontinuous epitopes for H/K ATPase and none lineal but three discontinuous epitopes for the heavy neurofilament protein. For each of these proteins, some epitopes presented a possible antigenic patch among the alignment, in sequence with a high homology as of see in Table 3.

Download:

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

Table 3. Lineal and discontinuous epitopes from human HSP60, alpha enolase and gamma enolase, predicted by Ellipro server.

https://doi.org/10.1371/journal.pone.0281485.t003

In the case of HSP60, the second epitope (LE2: 276–299) was in the most conserved region compared to the other three epitopes. For the alpha enolase, the first epitope described (LE1: 49–106) was the largest one of the three antigenic patches, with a total of 57 residues and 26 (45,6%) of them being identical with the H. pylori protein. By last, gamma enolase presents one conservated epitope of 59 (LE2: 47–105), with an identity of 45,76% with the H. pylori protein (Fig 1). The 3D models of the human HSP60, alpha enolase and gamma enolase shown a similar arrangement with H. pylori proteins chaperonin GroEL and the two phosphopyruvate, respectively. The furthest RMSD was observed between HSP60 and chaperonin GroEL, with a value of 5.391, however, PDB validation of HSP60 are low and may have a low molecular spatial distribution. The overlay of the 3D models between the three human antigens and the H. pylori whit the highest identity values are shown in the Fig 2.

Download:

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

Fig 1.

3D structure of the human HSP60 (A), alpha enolase (B) and gamma enolase (C). Cartoon and surface models are showed to represent the position of the epitope predicted in the human antigens. All the epitopes are indicated in color on surface models: blue, red, green, yellow, cyan, magenta, orange. LE. Lineal epitope. DE: discontinuous epitope.

https://doi.org/10.1371/journal.pone.0281485.g001

Download:

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

Fig 2. Overlay of the 3-D models between type 1 gNET antigens and H. pylori proteins.

Samples: human HSP60 (gold) vs H. pylori chaperonin Groel (Magenta) are shown as the complete unit (A) and a subunit (B). (C) Human Alpha enolase (orange) vs H. pylori phosphopyruvate (cyan). (D) Gamma enolase (gray) vs phosphopyruvate (yellow). RMSD: Root-mean-square deviation are showed in Armstrong (Å).

https://doi.org/10.1371/journal.pone.0281485.g002

Discussion

Despite the multiple epidemiological studies that associate infection with H. pylori and various systemic diseases, it is not clear if this association is involved in the pathogenesis of these diseases [6, 18–22]. One of the most discussed mechanisms is the induction of autoimmunity by H. pylori through molecular mimicry [23, 24], but few human antigenic proteins have been associated with this mechanism and they vary depending on the type of disease [1, 4, 25]. Also, studies are lacking to show that these autoantibodies are functional.

In urticaria, different proteins have been described that can be recognized by IgG, IgM or IgE autoantibodies [12, 26, 27], some of these autoantibodies can induce the activation of basophils and mast cells, indicating that they can induce an inflammatory response [12, 28]. Some studies suggest that the frequency of H. pylori infection is higher in patients with urticaria [29]; case series have been published where elimination of H. pylori has been associated with remission of the disease [15]. Although these results are still controversial [30, 31] suggest that the presence of H. pylori may contribute to the development of urticaria. When we analyzed the nine human antigen proteins that have been recognized by autoantibodies with functional activity, none shared identity with the H. pylori proteins analyzed in this study. Although our results suggest that there is no relationship between H. pylori and urticaria through molecular mimicry, they also do not completely rule out this hypothesis; Schmetzer et al. [27], observed that more than 200 human proteins could be recognized by IgE autoantibodies in patients with urticaria, therefore H. pylori proteins could have molecular mimicry with other proteins that we did not evaluate; nevertheless, not all those autoantibodies are functional, and the 9 proteins included in this study are those that have antibodies with functional studies. There are also proposals for other mechanisms that could explain the relationship between H. pylori and urticaria not dependent on molecular mimicry [32].

In the case of type 1 gNET, there is also controversy about the role of H. pylori [33, 34] but its association with other types of cancer, specially gastric cancer is strong [7, 22, 24]; Following our hypothesis of a possible cross-reactivity between H. pylori proteins and human proteins associated with type 1 gNET, we found that HSP60 from H. pylori and human share identity and we identified possible antigenic patches.

HSP60 is a mitochondrial localized quality control protein responsible for maintaining mitochondrial function. HSP60 is considered both a tumor suppressor and promoter in different types of cancer; its role in the oncogenesis of type 1 gastric neuroendocrine tumor needs to be explored. HSPs participate as immunomodulators in both innate and in both innate and acquired immune responses [35, 36].

Enolase is a dimeric protein composed of three isoenzymes, alpha, beta, and gamma (α, β and γ). This enzyme catalyzes the phosphoenolpyruvate and 2-phosphoglycerate, one of the last steps in glycolysis. Enolase, specially α enolase, have been widely used as markers of NETs through immunostaining. Glycolysis is a fundamental metabolic pathway, and its enzymes are highly conserved and present in procaryotic and eukaryotic species, including pathogens, like Streptococcus aureus, Streptococcus pneumoniae, Candida albicans, and Leishmania mexicana, make it an excellent candidate to produce cross-reactive autoantibodies.

These results show the utility of in silico assays for clinical research; without putting patients at risk, in silico analysis allows us to explore research hypotheses and prioritize resources in those that are most promising. At the same time, it allows the integration of multiple disciplines, which generates a more holistic vision in the approach and study of patients. According to these results, subsequent investigations of the role of H. pylori in urticaria require a different hypothesis, while in the case of NETs the results suggest that some autoantigens could explain the relationship between H. pylori in a subgroup of patients.

HSP60 and enolase have previously been identified in other gastrointestinal disorders like autoimmune colitis and Chron’s disease [37, 38]. Therefore, our results support those autoantibodies against the human homonymous protein may have some role in this disease, but it is necessary to carry out functional studies to evaluate whether these antigens, when recognized by the immune system, generate an inflammatory response. It is also necessary to note that our study has some weaknesses that could explain the lack of cross-reactivity between the urticaria auto-antigens, and the H. pylori antigens evaluated; Due to the lack of knowledge about the various proteins involved in urticaria and gNET, some important molecules could not be included, and this could explain the lack of association between the pathogen and urticaria. In addition, it is necessary to explore other mechanisms such as epitope spreding, which may also be associated with this relationship between urticaria, type 1 gNET and H. pylori.

However, these autoantibodies can be useful as biomarkers with clinical utility. Molecular mimicry usually occurs by evolutionary conservation of proteins between different species; with the results of our study, we identified for the first-time shared epitopes between H. pylori proteins and human proteins related to the pathogenesis of type 1 gNET. This finding may have biological implications such as the potential formation of an autoimmune response mediated by antibodies secondary to the recognition of H. pylori proteins. Previous analyses, comparing in silico techniques versus functional techniques [39, 40], show that in silico analyzes allow to detected proteins that share identity with a 90% of precision, therefore they are quite useful for the development of new research hypotheses and they are cost/effective since they allow a better administration of research resources, especially those where there is little information with functional studies, as is the case of our study where we explored the relationship between H. pylori, urticaria and type 1 gNET; our results provide a rational basis for future research between H. pylori and type 1 gNET but advise against such research in urticaria where other mechanisms or proteins should be evaluated.

In conclusion, some human proteins associated with type 1 gNET like the HSP60 and enolases retain common epitopes with H. pylori proteins, suggesting that molecular mimicry could be a mechanism that explains the relationship between the microorganism and this disease. Our results allowed us to identify possible epitopes with molecular mimicry between H. pylori and type 1 gNET; these regions are the most likely to be associated with cross-reactivity; however, cross-reactivity requires in vitro studies to confirm. Additionally, our results suggest with a high level of certainty that the probability of molecular mimicry between H. pylori and the evaluated human urticaria-related proteins is low, so it is unlikely that they present cross-reactivity, which suggests that new proteins should be investigated, or other mechanisms explored to determine the relationship between H. pylori and chronic urticaria. However, studies evaluating the functional impact of this relationship are necessary.

Supporting information