Sample collection and patient information
We collected buccal mucosa samples from AP (n = 6, age 79–93 years old) and non-AP (Control; n = 6, age 66–93 years old) patients. Three samples (AP#3, AP#5, and AP#6) were collected from the patients with Parkinson’s disease, three (AP#2, AP#4, and Control#1) from those with Alzheimer's disease, and two (Control#2 and Control#6) with Lacunar infarction. The duration from the onset to the collection time varied from 1 to 18 days (Table S1). Body mass index was low in the AP group (Table 1). No significant differences were observed in the blood for CRP concentration, several white blood cells, or serum albumin concentration. While there was no difference in the number of residual teeth, the OHAT score tended to be higher in the AP group, indicating an unhealthy oral state of the AP patients.
Comparative quantitative analysis of oral proteins
After TCA precipitation and acetone wash, protein solutions were obtained with concentrations ranging from 0.65 to 2.49 mg/mL. The LC-MS/MS analysis detected a total of 3,528 proteins including 3,253, 157, and 118 proteins at high-, middle-, and low-confidence levels, respectively, based on the false discovery rate (FDR; Table S2). No significant difference was observed in the distribution of abundance (Figure S1). Among the proteins, the abundance of 179 and 181 proteins was high in the AP group and the control group, respectively (Fig. 1). Principal component analysis (PCoA) revealed that the AP #4 sample was distinct from the other 11 samples, and there was no significant difference in the PCoA profiles between the AP and control groups (Figure S2).
Using the significantly higher abundant proteins from each group, we carried out a gene ontology enrichment analysis [20, 21]. No significant results (P < 0.05 by FDR) were observed for biological process or molecular function for the 179 proteins that were higher in the AP group. In contrast, the molecular functions, nucleosome-binding and small molecule binding, were detected using the 181 proteins that were expressed higher in the control group. Although histones are involved in bacterial killing as a major component of neutrophil extracellular traps [22], the involvement of these proteins in the non-occurrence of AP is unknown.
Serum and Oral C-reactive protein (CRP)
Ouellet-Morin et al. reported a moderate-to-strong association between CRP measured in saliva and serum (r = 0.72) [23]. In the present study, there was no significant difference in CRP concentration either in the blood (P = 0.16) or oral cavity (P = 0.71) between the AP and control groups. Nonetheless, Spearman's correlation coefficient was 0.86 (P = 0.001) between blood and oral CRP levels, indicating that serum CRP may be predicted non-invasively using oral CRP values, which is consistent with the previous report [23] (Fig. 2). It is unclear why the abundance of oral CRP was relatively high in Control #3.
Detected proteins
Glycoproteins
A variety of glycoproteins play important roles in innate and adaptive immunity [24]. In the AP group, we found a low abundance of α-2-HS-glycoprotein (Fetuin; Abundance Ratio AP/Control = 0.38, P = 0.02) and β-2-glycoprotein 1 (Ratio = 0.39, P = 0.04), both of which are primarily present in blood and exhibit various physiological functions [25, 26]. Alpha-2-HS-glycoprotein, a negative acute-phase reactant [27], is decreased during the early period of sepsis but begins to increase from 72 days post-inflammation [28]. The β-2-glycoprotein 1 possesses the dual capability of up- and down-regulating complement and the coagulation systems, depending upon the external stimulus as well as CRP and thrombomodulin [26]. It is unclear whether its lower abundance is associated with the occurrence of AP.
S100 protein family
The S100 proteins, a family of calcium-binding cytosolic proteins, are known as damage-associated molecular pattern molecules and they exhibit a variety of intracellular and extracellular functions [29]. Protein S100A-7A (Ratio = 6.51, P < 0.01) and S100A-7 (Ratio = 4.33, P = 0.02) were higher in the AP group; however, there was no difference in the levels of other S100 protein family members.
Cytokines and chemokines
The LC-MS/MS detected some interleukins (ILs), however, there was no difference in IL-1α, IL-8, or IL-18 levels between the groups. Isoform 2 of IL-19 and isoform 2 of IL-36β, but not IL-36α, were higher in the AP group (Ratio > 100, P < 0.01), whereas IL-36γ also tended to be high (Ratio = 3.26, P = 0.14). For the chemokines, C-C motif chemokine 28 was low in the AP group (Ratio = 0.39, P < 0.01). The IL-36 cytokines, which include IL-36α, IL-36β, IL-36γ, and IL-36Ra, belong to the IL-1 family and exert pro-inflammatory effects on various target cells, such as keratinocytes, synoviocytes, dendritic cells, and T cells [30]. Ramadas et al. showed that IL-36γ is upregulated in airway epithelial cells and involved in chemokine (neutrophil chemoattractants CXCL1 and CXCL2) production and neutrophil influx in mice challenged with a house dust mite extract [31]. In contrast, the abundance of the IL-36 receptor antagonist protein also tended to be higher in AP samples compared with the control samples (P = 0.089). Our data suggest that various IL-36-related signaling pathways are involved in the onset of AP.
Non-salivary proteins
In the Human Body Fluid Proteome database, 2,871 proteins have been registered as saliva proteins as of May 2022 [32], and 1,973 of the 3,528 proteins detected in the present study were registered as salivary proteins in the database. Of the 1,555 non-salivary proteins, 130 were in high abundance in the AP group, whereas only six were detected with > 20% coverage. Mago Nashi Homolog 2 (Magoh2) was detected in five of the six AP samples, but not in any of the Control samples. Although Magoh proteins contribute to exon junction complexes [33], it is unclear whether the Magoh2 protein is involved in the onset of AP.
Multiplex cytokine assay
In our shotgun proteomics analysis, we did not detect peaks for IL-1β, IL-6, TNF-α, or MCP-1 (Table S2). In the multiplex cytokine/chemokine assay, the values for IL-6 and TNF-α were under the limit of detection (IL-6, < 6.80 pg/mL; TNF-α, < 0.73 pg/mL) in most of the oral samples. Using the supernatant without TCA precipitation, we also conducted a multiplex cytokine/chemokine assay. The protein concentrations of the supernatant ranged from 0.012 to 0.27 mg/mL, which were likely dependent on the strength of rubbing. Among the 13 cytokine and chemokine proteins, IL-1β, MCP-1, IL-8, and IL-18 were detected in all 12 samples. The concentrations were normalized to the total protein concentration. There were no significant differences between the AP and control groups, although MCP-1 levels tended to be lower (P = 0.065 by Mann–Whitney U test; Fig. 3).
Biomarker candidates
Besides the proteins described above, the LC-MS/MS analysis detected various proteins, some of which the function are unknown. An essential requirement for AP biomarkers is that the amount must be sufficiently high to be detected by common and convenient assays, such as ELISA, western blot analysis, or immunochromatography. Among the 360 proteins with a P-value < 0.05, we selected up to 60 proteins with a coverage of > 20%. We then selected 36 proteins that were detected in all six AP patients (Table 2). Among the 36 proteins selected, eight proteins (chloride intracellular channel protein 3, β-defensin 4A, protein S100-A7A, Purkinje cell protein 4-like protein 1, glutamine synthetase, protein S100-A7, cytosolic 5'-nucleotidase 3A, pyridoxal kinase) were higher in the AP group (Fig. 4). The remaining 28 proteins were lower in the AP group and included properdin, high mobility group nucleosome-binding domain-containing protein 5, beta-2-glycoprotein 1, C-C motif chemokine 28, and alpha-2-HS-glycoprotein. Using the 32 proteins detected in all of the samples, we prepared a cluster heat map (Fig. 5), which separated the AP and control samples into separate clusters.
Table 2
Proteins with higher abundance between the AP and control groups
Accession (Uniprot ID) | Protein name | Coverage [%] | Abundance Ratio (AP) / (Control) | P-Value |
O95833 | Chloride intracellular channel protein 3 | 83 | 6.56 | 0.001 |
Q86SG5 | Protein S100-A7A | 69 | 6.51 | 0.000 |
A6NKN8 | Purkinje cell protein 4-like protein 1 | 32 | 6.28 | 0.002 |
P15104 | Glutamine synthetase | 60 | 5.55 | 0.003 |
O15263 | Beta-defensin 4A | 36 | 5.10 | 0.011 |
P31151 | Protein S100-A7 | 87 | 4.33 | 0.024 |
Q9H0P0 | Cytosolic 5'-nucleotidase 3A | 42 | 4.16 | 0.041 |
O00764 | Pyridoxal kinase | 37 | 4.13 | 0.019 |
Q03591 | Complement factor H-related protein 1 | 60 | 0.46 | 0.037 |
O60911 | Cathepsin L2 | 62 | 0.44 | 0.023 |
P27918 | Properdin | 27 | 0.43 | 0.042 |
P82970 | High mobility group nucleosome-binding domain-containing protein 5 | 27 | 0.42 | 0.048 |
O60234 | glia maturation factor gamma | 51 | 0.41 | 0.048 |
Q92522 | Histone H1x | 28 | 0.41 | 0.015 |
P80723 | Brain acid soluble protein 1 | 80 | 0.40 | 0.024 |
P02751 | fibronectin | 24 | 0.39 | 0.015 |
P02749 | Beta-2-glycoprotein 1 | 58 | 0.39 | 0.041 |
Q9NRJ3 | C-C motif chemokine 28 | 21 | 0.39 | 0.008 |
P02765 | Alpha-2-HS-glycoprotein | 47 | 0.38 | 0.018 |
P46527 | Cyclin-dependent kinase inhibitor 1B | 62 | 0.37 | 0.006 |
P29966 | Myristoylated alanine-rich C-kinase substrate | 51 | 0.37 | 0.029 |
O76070 | Gamma-synuclein | 75 | 0.37 | 0.028 |
P58062 | Serine protease inhibitor Kazal-type 7 | 62 | 0.36 | 0.007 |
A8MWD9 | Putative small nuclear ribonucleoprotein G-like protein 15 | 25 | 0.36 | 0.030 |
P31025 | Lipocalin-1 | 61 | 0.36 | 0.025 |
Q9H1E3 | Nuclear ubiquitous casein and cyclin-dependent kinase substrate 1 | 54 | 0.35 | 0.021 |
P20930 | Filaggrin | 34 | 0.33 | 0.012 |
P02766 | Transthyretin | 73 | 0.32 | 0.009 |
P07305 | Histone H1.0 | 21 | 0.30 | 0.006 |
P56134 | ATP synthase subunit f, mitochondrial | 26 | 0.28 | 0.005 |
P07476 | Involucrin | 73 | 0.28 | 0.003 |
O75396 | Vesicle-trafficking protein SEC22b | 30 | 0.28 | 0.037 |
Q6P5S2 | Protein LEG1 homolog | 41 | 0.24 | 0.001 |
O95274 | Ly6/PLAUR domain-containing protein 3 | 45 | 0.23 | 0.001 |
O75556 | Mammaglobin-B | 49 | 0.22 | 0.004 |
P41439 | Folate receptor gamma | 53 | 0.17 | 0.026 |
The proteins were detected in all the six AP samples with > 20% coverage. P values were calculated by an ANOVA test and adjusted by the Benjamini-Hochberg method. |