Insights into immune responses in oral cancer through proteomic analysis of saliva and salivary extracellular vesicles

Abstract

The development and progression of oral cavity squamous cell carcinoma (OSCC) involves complex cellular mechanisms that contribute to the low five-year survival rate of approximately 20% among diagnosed patients. However, the biological processes essential to tumor progression are not completely understood. Therefore, detecting alterations in the salivary proteome may assist in elucidating the cellular mechanisms modulated in OSCC and improve the clinical prognosis of the disease. The proteome of whole saliva and salivary extracellular vesicles (EVs) from patients with OSCC and healthy individuals were analyzed by LC-MS/MS and label-free protein quantification. Proteome data analysis was performed using statistical, machine learning and feature selection methods with additional functional annotation. Biological processes related to immune responses, peptidase inhibitor activity, iron coordination and protease binding were overrepresented in the group of differentially expressed proteins. Proteins related to the inflammatory system, transport of metals and cellular growth and proliferation were identified in the proteome of salivary EVs. The proteomics data were robust and could classify OSCC with 90% accuracy. The saliva proteome analysis revealed that immune processes are related to the presence of OSCC and indicate that proteomics data can contribute to determining OSCC prognosis.

Figure 1. Differentially expressed proteins in whole saliva from human oral cancer.

LFQ intensity values of the salivary proteins identified from oral cancer patients were compared to the LFQ values of salivary proteins identified from healthy individuals, and proteins with a significant differential abundance were detected using the ANOVA (p < 0.05) method. Clustering analysis of the data was performed using a Euclidian distance method for the z-score normalized LFQ values. The z-score normalized values are shown in a heat map that represents the variation in protein abundance between the analyzed samples. Groups of oral cancer and healthy individuals and the corresponding protein names are indicated.

Figure 2. Network of overrepresented GO terms in the dataset of differentially expressed proteins in the whole saliva proteome.

Proteins identified as differentially expressed in the whole saliva of patients with oral cancer and healthy individuals were subjected to a GO term enrichment analysis using the ClueGO plugin. Overrepresented GO terms (p-value < 0.05) for Biological Processes, Cellular Components and Molecular Function categories were visualized using the Cytoscape suite. The most significant GO terms within the dataset of significantly enriched terms are indicated in red font.

Figure 3. Differentially expressed proteins in response to the presence or absence of tumors in patients diagnosed with oral cancer.

LFQ intensity values (log2) of the salivary proteins identified from diagnosed oral cancer patients with tumors (with lesion) were compared to patients without tumors (no lesion). Differentially expressed proteins were detected using the ANOVA method (p < 0.05). A Euclidean distance clustering analysis of these data was performed, and the differential protein abundance is shown in the heat map of normalized z-score values.

Figure 4. Extracellular vesicles isolated from saliva.

Extracellular vesicles were isolated from the saliva of healthy and oral cancer patients using ultracentrifugation. (A) Nanoparticle tracking analysis of the extracellular vesicles isolated from healthy individuals (C4, C6, C10) and oral cancer patients (T2, T5, T16); (B) Immunochemical detection of the protein flotilin-1, a protein marker of extracellular vesicles, in the whole saliva proteome (S) and in the proteome of extracellular vesicles (EV); (C) Transmission electron microscopy of isolated extracellular vesicles using a negative staining method.

Figure 5. Proteins from extracellular vesicles with a significant differential abundance between healthy and oral cancer groups.

Extracellular vesicles were isolated from the saliva of healthy and oral cancer individuals, and the EV protein content was analyzed by shotgun proteomics. Proteins with a significant differential abundance between healthy and oral cancer groups were identified using the label-free quantification (LFQ) method (ANOVA, p < 0.05). The Euclidean distance clustering method was used to visualize the variations in the protein abundance between the samples and is shown as normalized z-score log2 LFQ values in a heat map.

Figure 6. PPIA survival function from lifespan data.

Kaplan-Meier curves demonstrate that patients with a PPIA low profile succumbed to the disease much earlier than those with a high profile. After 1.5 years, approximately 40% of the low profile patients survived, but all subjects in the high intensity group were alive at follow-up. A log-rank test shows significant differences between pairs, with p-value < 0.05. The number of individuals included in the analysis is shown in the figure legend.