Salivary proteome of aphthous stomatitis reveals the participation of vitamin metabolism, nutrients, and bacteria

There are currently no preventative options for recurrent aphthous stomatitis, and the only available treatments are palliative. This is partly due to a poor understanding of its etiopathogenesis. In this case–control study, we characterized the salivary proteome of patients with recurrent aphthous stomatitis in the presence and absence of lesions. Through mass spectrometry-based proteomics and bioinformatics tools, we identified that the presence of oral ulcers is associated with several specific biological processes, including the metabolic pathways of vitamin B9, B12, nitrogen, selenium, and the bacterium Neisseria meningitidis. These changes occurred only in the presence of clinically visible lesions, and there were no relevant differences between patients in anatomical regions unaffected by ulcers. Additionally, using western blot and ELISA assays, we verified that carbonic anhydrase 1 (CA1) and hemoglobin subunit beta (HBB) proteins are highly expressed during the ulcerative and remission phases of recurrent aphthous stomatitis. Our results cumulatively support saliva as an indicator of the pathophysiological changes, which occur during the clinical course of lesions. From a clinical perspective, we suggest that recurrent aphthous stomatitis is a condition triggered by temporary biological changes in people with lesions.

www.nature.com/scientificreports/ of ulcers) 14 . The ulcerative and remission phases are the stages able to be evaluated with greater objectivity in dental examinations 15 . Our research team wanted to try a different approach by applying state-of-the-art technology for the study of molecular processes: mass spectrometry-based proteomics. This technique analyzes a mixture of proteins, both qualitatively and quantitatively, using mass spectrometry. Additionally, the high feasibility and general sensitivity of this approach is particularly desirable for the identification and characterization of proteins 16 . Proteomic technologies have a great capacity to generate a holistic view of complex processes that control health and disease states. To our knowledge, there are no studies applying global characterization of the proteome in patients with recurrent aphthous stomatitis.
Unfortunately, for research purposes, incisional or excisional biopsy of recurrent aphthous stomatitis ulcers is recommended only in cases of uncertainty 17 . Saliva is an ideal source of biological information because it contains a set of biomolecules that cover oral structures, such as mucosa and teeth 18 , and is therefore in close contact with ulcers. Mass spectrometry-based proteomics has been used to discover biomarkers in various oral conditions using saliva as a liquid biopsy, such as caries and periodontal disease 19 , cleft palate 20 , Sjögren's syndrome 21 , burning mouth syndrome 22 and oral cancer 23 , but not in recurrent aphthous stomatitis. Mass spectrometry allows us to examine the salivary proteome in detail 24 . Since saliva collection is less invasive and has no clinical complications associated with blood collection, substituting saliva samples for blood in biomarker analysis is of great interest 25 .
Considering these antecedents, the objective of this study was to characterize the salivary proteome of patients with recurrent aphthous stomatitis using Mass spectrometry-based proteomics and assess its clinical usefulness in identifying the most representative biological and molecular processes during the course of lesions. To do this, we conducted a case-control study, analyzing the saliva of healthy controls and patients with recurrent aphthous stomatitis during a complete ulcerative cycle (active ulcers and absence of lesions). Using bottom-up proteomics, salivary proteins were sequenced with a timsTOF Pro mass spectrometer connected to nanoflow liquid chromatography (nLC-MS/MS). From these data, we were able to recognize biological processes related to vitamins, nutrients, interference of tissue destruction, and immune response against bacteria. We then identified bacterial proteins and the genera, species, and strains to which they belong. Finally, considering that recurrent aphthous stomatitis is tissue-specific, we evaluated the proteomic profiles of surfaces with and without ulcers. Applying a top-down proteomics approach, an Autoflex Speed MALDI-TOF mass spectrometer combined with machine learning analysis (MALDI-MS/ML) were used to obtain a specific and sensitive pattern of the salivary pellicle covering recurrent aphthous stomatitis ulcers and ulcer-free regions.

Methods
General design. Using Mass spectrometry-based proteomics, we evaluated the salivary proteome of subjects with recurrent aphthous stomatitis in the presence and absence of ulcers. We also compared these profiles with healthy subjects without a history of aphthous ulceration. All procedures were performed in accordance with the Declaration of Helsinki. The Ethics Committee of the University of Antofagasta (protocol #156/2018) approved this research.
Participants. We prospectively collected 118 salivary samples from 68 volunteers after obtaining informed consent. The subjects were divided into healthy controls (n = 31; people with no history of oral ulcers) and recurrent aphthous stomatitis (n = 36). The latter group was evaluated at study entry, during the ulcerative stage, and when lesions disappeared (remission stage, n = 36). Additionally, we collected samples from recurrences that occurred after the remission phase (n = 15). Subject demographics are provided in Supplementary Table S1. At the time of the examination, patients should have presented ulcers for no longer than 3 days. The exclusion criteria were the use of medications to treat ulcers during the previous 2 days and use of topical or systemic corticosteroids during the month before entering the study. Other exclusion criteria were the presence of other types of lesions of the oral mucosa, diseases with acute or chronic pain, smoking, excessive consumption of alcohol (more than three times weekly), and hematologic deficiency-related diseases. Patients with autoinflammatory syndromes, immunodeficiency states, gastrointestinal disorders, and hematinics deficits were also excluded (the complete list of diseases can be found in Supplementary Table S2). For their inclusion as controls, volunteers should never have presented oral ulcers, and except for the absence of lesions, this group shared the same inclusion and exclusion criteria. Salivary liquid biopsy. Subjects were requested not to eat or drink for at least 30 min before collection.
All patients rinsed their mouths with 10 mL of water for 30 s. After 10 min, whole unstimulated saliva collection began. Between 8:00 a.m. and 11:00 a.m., each participant expectorated continuously in a 50 mL sterile propylene tube for a period of 5 min. Additionally, we collected the salivary pellicle deposited on lesions and the contralateral anatomical equivalent of the unaffected oral mucosa using cotton swabs. Immediately after collection, samples were deposited on ice and a protease inhibitor cocktail (cOmplete Tablets EASYpack, Roche) 26 . We then centrifuged the samples at 14,000×g at 4 °C for 20 min to eliminate undissolved components and cellular detritus 27 . The total protein concentration in the supernatants was determined using the BCA Protein Assay (Thermo Scientific). The samples were stored at − 80 °C for further processing.
Protein identification and prioritization. When multiple subjects are being studied, pooling samples in proteomics experiments can help address resource constraints 28 . Saliva samples were randomized and pooled into 9 sets: healthy controls (n = 3, 11 samples per set), ulcerative (n = 3, 12 samples per set), and remission phase (n = 3, 12 samples per set Mass spectrometry data were acquired by analyzing 500 ng of peptides after tryptic digestion and peptide fractionation using a nanoElute LC system coupled to a timsTOF Pro mass spectrometer (Bruker Daltonics). A detailed description of the instrument is provided in Refs. 29 . Mass spectrometry raw files were processed using MsFragger through Fragpipe 30 . For all searches, a protein sequence database of reviewed human proteins (accessed 06/15/2020 from UniProt; 74,823 entries) and bacterial proteins of species assigned to the oral cavity (accessed 12/11/2020 from The Human Microbiome Database; 728 files of annotated sequences from genomes) were used. Label-free protein quantification was performed using the IonQuant algorithm, as described previously 30 . We used MSstats 31 to calculate protein abundance based on the ion abundances reported by MSFragger. We followed the general workflow analysis for data-dependent acquisition (DDA or shotgun) experiments 32 . The dataProcess function with log2 intensity transformation was used to calculate the protein abundance. We considered clinically significant salivary proteins with a log2 fold change value less than − 0.5, or greater than 0.5, with a p value ≤ 0.05, when comparing between groups (Student's t-test). See Supplementary file (Methods section) for further details and explanation.
Biological and molecular processes. To understand the set of clinically significant proteins, we performed an enrichment analysis. Pathway enrichment analysis helps to obtain a mechanistic view, identifying biological pathways that are enriched in a list of proteins more than would be expected by chance 33 . To that end, we imported the proteins into the g:GOSt profiling online tool of G:Profiler 34 and queried for functional enrichment of terms derived from Kyoto Encyclopedia and WikiPathways. We used the g:SCS algorithm to address multiple testing, and a p value ≤ 0.05 as a user-defined threshold for statistical significance. We also consulted the REACTOME database using the WebGestalt online tool 35 . We considered those pathways with an FDR of ≤ 0.05. To understand the individual functions of each protein, we consulted QuickGO. With QuickGO, it is possible to determine a biological process or molecular event, with a known beginning and end, in which a protein participates 36 .
Proteomic profiles comparison. We analyzed the affected and unaffected oral mucosa using cotton swab samples. Given its speed, simplicity, and low cost, as well as the availability of equipment and expertise in many hospital laboratories in developing countries, MALDI-MS/ML is a promising alternative 37 . In this experiment, we did not pool samples (each was analyzed individually) and we added new samples from ulcer recurrences (new lesions after the remission stage) experienced by some of the patients (n = 15). Samples were analyzed on an Autoflex Speed MALDI-TOF mass spectrometer (Bruker Daltonics) following the protocol recently published by our group 37 and detailed in Supplementary File (methods section). Proteomic profiles were obtained and then preprocessed to generate a matrix of peak intensities. The most relevant peaks were selected using the correlation-based feature subset selection method (CFS) 38 implemented in the Weka software 39 . To explore and compare spectra in multidimensional space, principal component analysis (PCA) was performed using the R packages FactoMineR 40 and factoextra 41 (data were scaled to unit variance).
Sample size calculation. The number of patients was estimated following guidelines for the use of Mass spectrometry-based proteomics in clinical studies 42,43 . We consider a two-sided hypothesis test with a confidence level of 99.9% (α = 0.001) and a power of 95% (1 − β = 0.95). Changes in protein abundance or effect size (relative intensity data) were considered as differences based on log2 (0.5). Biological variation (containing technical variation) was established with a magnitude of 0.5. We performed each analysis once. The number of participants was adjusted to a probable loss of 20%, establishing a minimum of 31 volunteers per group.

Results
Clinicopathological characteristics. The subjects were from the same population base. The mean age of the participants was 26 years for both groups, with the majority being women. The profile of our population agrees with that reported in the international literature, both in age and in the predisposition for the appearance of these lesions in women 44,45 . Figure 1 shows representative ulcers observed in this research. Most lesions occurred in the non-keratinized oral mucosa, where the lips were the most affected territory (Supplementary  Table S1). Ulcers mainly corresponded to single lesions classified as minor, with moderate pain according to the visual analog scale. The participants reported presenting more than seven episodes per year. Considering that recurrent aphthous stomatitis lasts for approximately two weeks, the large number of ulcer events highlights the clinical relevance of the condition.
The presence and absence of lesions are associated with pathways related to vitamins and nutrients. To understand the context in which the protein collective might participate, we performed an enrichment analysis using g-Profiler. For this, we included all differentially expressed proteins. Figure 3 shows the comparison between the healthy control group and the ulcerative stage patients (Fig. 3A) and the ulcerative stage versus the remission-stage patients (Fig. 3B). These comparisons revealed the metabolic pathways of macronutrients and micronutrients (nitrogen and selenium) along with the digestion and absorption of vitamins (B9 and B12). Notably, there was no enrichment when analyzing the list of differentially expressed proteins between healthy control samples and the remission stage. In other words, in terms of signaling, there were no differences between the groups. This result indicates that alterations in signaling pathways are only present during the clinical manifestation of the ulcer. It is necessary to note that the enrichment analysis placed a part of the identified proteins into context; therefore, to complete the picture, it is also necessary to describe the individual biological roles of each of them.

Proteins involved in the inhibition of proteolysis and responses against bacteria stand out in the presence of lesions.
Because the enrichment analysis did not provide information on all proteins, we list each differentially expressed protein in Table 1. To understand their individual roles, we examined the functional annotations in QuickGO. In contrast, information from patients with recurrent aphthous stomatitis during the ulcerative stage and remission stands out. Several upregulated proteins participate in the negative regulation of peptidase activity (SERPINA6, SERPING1, SERPINA3, and A2M) and the immune response against bacteria (IGHV1-8, IGHV1-69D, IGHV3-74, IGHV4-30-2, IGHV3-43, IGHV3-66, IGHV3-16). This result suggests that the ulcerative activity may be a response to the presence of a microorganism that is harmful to the oral mucosa or that the presence of lesions alters the microbiota and that a decrease in proteolysis is necessary for the regeneration of lesions.  in the analyzed saliva pools (Fig. 4). In this orthogonal validation, CA1 and HBB levels were elevated in recurrent aphthous stomatitis (ulcerative and remission stages, Fig. 4A,B upper panels, respectively). Additionally, we checked for the presence of vitamins B9 and B12 in saliva. In general, patients with recurrent aphthous stomatitis had lower vitamin levels compared to controls but without significant differences (Fig. 4C).
Neisseria meningitidis proteins were present at the ulcerative stage. Because defense against bacteria was one of the prominent pathways identified, we investigated whether we could detect bacterial proteins and species in salivary proteomes. For this purpose, we downloaded all the protein sequences of the bacterial species assigned to the oral cavity in the Human Oral Microbiome Database 46,47 . The resulting database was approximately 46 times larger than that of humans (1,174,712 KB vs. 26,318 KB). We identified 134 bacterial proteins. Figure 5 shows that two proteins, isocitrate dehydrogenase [NADP] (icd, UniProtKB Q9JZS1) and elongation factor Ts (tsf, UniProtKB P64051), belonging to Neisseria meningitidis strains were elevated only in the presence of lesions. The response indicated by the salivary proteome against bacteria could result from the activity of Neisseria meningitidis.
The oral surface epithelium without lesions did not show any differences between the groups. As recurrent aphthous stomatitis is a tissue-specific condition, we analyzed the salivary pellicle that covered the area affected by recurrent aphthous ulcers and the healthy contralateral anatomical region, to provide a local picture. In this analysis, we also included recurrent lesions (new ulcerative stages after remission). MALDI-MS/ML analysis showed that the proteomic profiles (mass spectra) were different when presence and absence of lesions were compared in a multidimensional space (PCA graphs, Fig. 6). However, the data from the healthy tissue regions could not be separated completely, both at the remission stage (Fig. 6B) and at all unaf- www.nature.com/scientificreports/ fected oral mucosa sites (Fig. 6D). This evidence shows that the salivary pellicle that covers the epithelium differs only during active lesions, but not during epithelial integrity.

Discussion
In this investigation, we determined the salivary proteome of patients with recurrent aphthous stomatitis during the presence and absence of lesions using mass spectrometry-based proteomics. Our analyses revealed that the ulcerative stage is related to several biological pathways, including vitamin and nutrient metabolism, inhibition of tissue destruction, and defense against bacteria, probably Neisseria meningitidis. Additionally, we established that the proteomic profiles of ulcer-free states do not differ from those of healthy individuals. Our evidence suggests that recurrent aphthous stomatitis corresponds to an oral manifestation of a temporary local or systemic condition. Because tissue biopsy is not indicated in recurrent aphthous stomatitis 15 , we used saliva as a "liquid biopsy" to obtain information on the health status of the participants. The lesions corresponded to small, solitary or multiple ulcers, with moderate pain, present almost exclusively in the non-keratinized oral mucosa. These characteristics are consistent with the classic profile of recurrent aphthous stomatitis 48,49 . In the analysis flow with nLC-MS/MS and bioinformatics, we used the fold change as a measure of contrast between the groups. From these values, we concentrated on the salivary proteins that presented a differential expression on which we explored the biological information. Fold change is a value that reports the extent to which a quantity changes from one state to another. It is used in proteomics to evaluate the differential expression of proteins 50 and to identify disease biomarkers 51 .
Through enrichment analysis with g:Profiler and WebGestalt, we found that ulcerative activity is related to the metabolism of vitamins (B9 and B12) and nutrients (selenium and nitrogen). Our evidence complements one of the main lines of inquiry for explaining the etiopathogenesis and severity of recurrent aphthous stomatitis with respect to nutritional deficiencies; low blood levels of iron 52 , vitamin D 53-55 , vitamin B9 56 and B12 57,58 have previously been reported to be associated with the presence of lesions. In our study, the levels of salivary vitamins B9 and B12 did not show significant differences. This may be due to the fact that there is a low correlation between serum and salivary levels, as has already been demonstrated in the geographic tongue 59 . Some of hematinic deficiencies already named can lead to anemia in patients with recurrent aphthous stomatitis 60 . Previous studies have shown that anemia causes atrophy of the oral epithelium, leading to a decrease in the mucosal barrier 61 . Selenium deficiency can also make the host susceptible to infection 62 . The presence of these molecules was not examined in our patients, as it was outside the objectives of our research; this may constitute a limitation in our analyses, and it would be advisable to measure these molecules in the future. However, it is interesting that saliva represents, at least indirectly, the nutritional states previously described in the literature in these patients. The high expression of salivary proteins associated with metabolic pathways may be attributed to  www.nature.com/scientificreports/ the lack of vitamins and nutrients and suggests that these proteins may not be involved in these pathways (i.e., they are freely available). In this sense, it has been shown that supplementation with vitamin B12 decreases the duration of ulcer episodes, number of injuries, and pain 63 . The finding of no differences between healthy subjects and those with recurrent aphthous stomatitis in the remission stage was particularly notable, as it indicates that these two clinical states are not significantly different; alterations that determine an ulcerative state are silenced in the absence of lesions. This evidence indicates that diet plays an important role in triggering lesions and that the aforementioned alterations may be temporary and susceptible to intervention. The enrichment analysis showed a high level of various proteins during the ulcerative stage. Interestingly, HBB and SERPINA3 have anti-inflammatory effects. Using ELISA, we verified that patients with recurrent aphthous stomatitis present a high expression of HBB. HBB can be cleaved into two chains, one of which is called Table 1. Biological processes of differentially expressed proteins in recurrent aphthous stomatitis. Fold change (log 2 ) is a measure that describes how much an amount changes when going from an initial value to a final value. Therefore, if the initial value is A and the final value is B, the fold change is (B − A)/A or equivalently B/A − 1. *We obtained biological processes from QuickGo (https:// www. ebi. ac. uk/ Quick GO/).    64 , and additionally promotes pain reduction by inhibiting the degradation of endogenous opioids 65 . SERPINA3 inhibits neutrophil cathepsin G protease, a pro-inflammatory enzyme released at sites of inflammation that contributes to the activation of inflammatory cytokines, pathogen degradation, and tissue remodeling 66 . Inhibition of cathepsin G likely contributes to reduction of tissue damage from proteases that are released at the site of epithelial injury 67 . Repeated topical application of SERPINA3 (2 mg/mL) on induced wounds in mice has been found to significantly accelerated their closure 67 .
Considering that the enrichment analysis did not provide information on all proteins, we consulted each of them individually in QuickGo. In this analysis, a protective response stood out again, and peptidase activity was negatively regulated (SERPINA6, SERPING1, SERPINA3, and A2M). In a recent study, Bao et al. used LC-MS/ MS and miRNA microarrays to evaluate the serum from patients suffering from recurrent aphthous ulcers 68 . The results showed SERPINA1 upregulation in patients with lesions compared to that in healthy controls. Serpins are protease inhibitor proteins that inhibit target enzymes by causing conformational changes. Many key proteolytic cascades, including mammalian coagulation processes, are regulated by these proteins. Together, these data indicate that after epithelial destruction, proteins are released to inhibit tissue destruction promoted by neutrophils. 69 .
In our experiments we verified a high expression of CA1 during the ulcerative and remission stages. CA1 is a member of the CA family of carbonic anhydrases, which reversibly catalyzes the hydration of CO 2 to generate HCO 3 − , which then rapidly binds to calcium ions to produce calcium carbonate 70 . Carbonic anhydrases are known to modulate immune cell activation. Evidence from experiments in mice shows that inhibition of CA1 possesses therapeutic potential that can be used to treat mast cell-mediated inflammation 71 . Pre-ulcerative lesion of recurrent aphthous stomatitis demonstrates subepithelial inflammatory mononuclear cells with abundant mast cells 72 . In addition, it has been reported that the presence of mast cells and their degranulation activity at the connective tissue of recurrent aphthous stomatitis lesions does not necessarily mean that they play a destructive role; mast cells also have the ability to mediate a protective role and might be involved in the repair process 73 . Whether in a destructive or protective sense, salivary CA1 could be related in some way to those activities. Subjects with ulcers therefore have active processes to destroy, or conversely, recover and preserve the integrity of the oral mucosa toward the remission stage. These processes have been induced by, or induce changes to, the salivary proteome.
Another relevant biological process identified was the immune response against bacteria, associated with several differentially expressed proteins (IGHV1-8, IGHV1-69D, IGHV3-74, IGHV4-30-2, IGHV3-43, IGHV3-66, IGHV3-16). To further investigate these findings, we analyzed over 700 species sequence files from the oral microbiome. Although the samples were centrifuged, we were able to identify that two strains of Neisseria meningitidis (53442 and MC58) were consistently present only in the ulcerative stage when compared to states without lesions. This is due to the identification of proteins belonging to this bacterium (isocitrate dehydrogenase [NADP] and elongation factor Ts). Isocitrate dehydrogenase [NADP] participates in the synthesis of  www.nature.com/scientificreports/ NADPH, which provides reducing power that drives numerous anabolic reactions, including those responsible for the biosynthesis of all major cellular components 74 . The elongation factor Ts participates in the elongation of the polypeptide chain during prokaryotic translation, allowing the synthesis of proteins 75 . This leads us to infer that Neisseria meningitidis can be found in an active phase of growth and synthesis. This result should be interpreted with caution because bacteria in the oral cavity may exhibit considerable sequence similarity 76 . Neisseria meningitidis is a transient commensal of the human oropharynx that causes life-threatening diseases, such as meningitis and bacteremia 77 . The asymptomatic oropharyngeal carriage of Neisseria meningitidis reaches its peak in adolescents and young adults 78 . This is consistent with the peak and frequency of recurrent aphthous stomatitis in the general population 48 84,85 . The use of mass spectrometry-based proteomics to identify proteins in saliva is reliable. A previous study showed that the percentage of tryptic peptides that share the same sequence between humans and bacteria is only 0.04% 86 . Our findings are limited to salivary identification and should subsequently be subsequently verified in microbiological cultures from salivary samples of patients with recurrent aphthous stomatitis in different phases of the disease to elucidate potential causality. Furthermore, all the biological knowledge consolidating our arguments has been sourced from public databases, whose information evolves over time.
Considering that the saliva was centrifuged, a good biological resource to exploit further is the pellet; from this, PCR or MALDI Biotyper could be used to explore the presence of Neisseria meningitidis and other bacteria for complete understanding of an eventual oral dysbiosis.
Our results show that, from a systemic point of view, the saliva of subjects who had never presented ulcers compared to subjects with recurrent aphthous stomatitis in the remission stage did not present differences in terms of biological processes. Considering that this condition is tissue specific, we examined the adhered salivary pellicle on the surfaces with and without lesions using MALDI-MS/ML. In this last experiment, we also www.nature.com/scientificreports/ added some samples of ulcer recurrence that patients made after the remission stage. MALDI-MS/ML analysis showed that the proteomic profiles of all lesion-free states did not differ, except when the lesion was present. This emphasizes that if recurrent aphthous stomatitis is considered a disease, it does not leave a mark on the epithelial salivary pellicle. Furthermore, when comparing the lesion-free states, no enriched biological processes are found, leading us to reflect on the true nature of recurrent aphthous stomatitis. The activity of recurrent aphthous stomatitis represents a pathological condition; according to the Medical Subject Headings, pathological conditions are abnormal anatomical or physiological conditions and objective or subjective manifestations of the disease, not classified as a disease or syndrome (MeSH Unique ID: D013568). There are various diagnoses and systemic states that occur with aphthous ulcerations, such as PFAPA syndrome, Behçet's syndrome, Crohn's disease, celiac disease, HIV infection, clinical neutropenia, among several others. There are many paths for the same lesion. Our data and results lead us to hypothesize that recurrent aphthous stomatitis is an oral manifestation on a local or systemic basis in susceptible subjects. This is likely due to nutritional deficiencies and bacteria-triggered lesions normally found in the oral cavity of adolescents and young adults. In this investigation, data obtained by mass spectrometry-based proteomics cumulatively supported the role of vitamin metabolism, nutrients, and bacteria in recurrent aphthous stomatitis. We demonstrated the usefulness of salivary biopsy for obtaining systemic and local information. From a clinical perspective, we specifically suggest the participation of proteins involved in the metabolism of vitamins B9 and B12, the nutrients nitrogen and selenium, and the activity of Neisseria meningitidis. Our evidence deepens the existing knowledge on the etiopathogenesis of recurrent aphthous stomatitis, using an approach that has never been utilized in this context. The conceptual framework that we have established here should be confirmed with subsequent experiments that seek, first, to define a state of systemic susceptibility, which would allow the establishment of lesions in the presence of an etiological candidate agent. Subsequent research may enable the development of preventative approaches for this disease, which are currently unable to be achieved.

Data availability
All supplementary files are included in this article and are publicly available at: https:// doi. org/ 10. 5281/ zenodo. 44828 98. The mass spectrometry proteomic data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository 87 with the dataset identifier PXD026401 (https:// www. ebi. ac. uk/ pride/ archi ve/ proje cts/ PXD02 6401). All other data supporting the findings of this study are available from the corresponding author on reasonable request.