Proteomic Profiling of Pre- and Post-Surgery Saliva of Glioblastoma Patients: A Pilot Investigation

Abstract

Glioblastoma multiforme (GBM) is an extremely aggressive brain tumor characterized by a high infiltration capability and recurrence rate. Early diagnosis is crucial to improve the prognosis and to personalize the therapeutic approach. This research explored, by LC-MS proteomic analysis after proteolytic digestion, the molecular profile of pre- and post-operative saliva pools from newly diagnosed (ND) GBM patients by comparing different times of collection and tumor recurrence (R). CYCS, PRDX2, RAB1C, PSMB1, KLK6, TMOD3, PAI2, PLBD1, CAST, and AHNAK, all involved in processes of tumor invasiveness and chemo- and radio-resistance, were found to depict the pre-surgery saliva of both ND and R GBM. PADI4 and CRYAB proteins, identified among the most abundant proteins exclusive of ND GBM pre-surgery saliva and classified as proteins elevated in glioma, could have a potential role as disease biomarkers. Selected panels of S100 proteins were found to potentially differentiate ND from R GBM patient saliva. TPD52 and IGKV3, exclusively identified in R GBM saliva, could be additionally distinctive of tumor relapse. Among the proteins identified in all pools, label-free relative quantitation showed statistically significant different levels of TXN, SERPINB5, FABP5, and S100A11 proteins between the pools. All of these proteins showed higher levels in both ND_ and R_T0 pre-surgery saliva with respect to CTRL and different modulation after surgery or chemo-radiotherapy combined treatment, suggesting a role as a potential panel of GBM predictive and prognostic biomarkers. These results highlight and confirm that saliva, a biofluid featured for an easily accessible and low invasiveness collection, is a promising source of GBM biomarkers, showing new potential opportunities for the development of targeted therapies and diagnostic tools.

Keywords: biomarkers; bottom-up proteomics; brain tumor; filter-aided sample preparation; glioblastoma multiforme; proteomics; saliva.

Figure 1

Venn diagram resulting from grouping analysis of the proteins identified in ND_T0, T1, and T3 GBM saliva versus R_T0 GBM and CTRL saliva.

Figure 2

Bar chart (log scale) of the label-free relative quantitation of protein area values (mean value of replicate LC-MS analysis ± SD) in ND (blue) and R (orange) T0 saliva of the 10 shared proteins identified (Table 1). On the x-axis, the proteins classified as Cancer Related are highlighted by a black rectangle. Significant differences have been determined by t-test (***, p value < 0.01; **, p value < 0.05; *, p value < 0.1).

Figure 3

(A) STRING tool protein functional network analysis of HSPE1, H2AC12, PARK7, CRYAB, YWHAB, LDHB, GLUL, HSPD1, HSP90AA1, and HSP90AB1, exclusive of ND_T0 saliva and also identified in GBM CUSA fluid from all tumor zones in our previous paper [19]. Node colors specify proteins in extracellular exosomes (purple nodes), involvement in Chaperone-Mediated Autophagy (green nodes), and HSF1-dependent transactivation (red nodes). The edges represent protein-protein associations, with each color indicating a specific type: known interactions (light blue: curated databases, purple: experimentally determined); predicted interactions (green: gene neighborhood, red: gene fusions, blue: gene co-occurrence); others (ochre: textmining, black: co-expression, light blue: protein homology). (B) Venn diagram resulting from grouping analysis of these ten proteins versus the list of brain and salivary gland classified proteins in the Human Protein Atlas database.

Figure 4

Venn diagram resulting from grouping analysis of the 108 exclusive proteins identified in the ND_T0 saliva pool versus different gene categories, namely, “gene elevated expression in glioma”, “gene only detected in glioma”, “unfavorable prognostic genes in GBM”, and “favorable prognostic genes in GBM”, downloaded from the Human Protein Atlas database.

Figure 5

Biological process gene ontology enrichment analysis of the proteins characterized in ND_T0 saliva, including the 108 exclusive proteins and the 10 also identified in R-T0 saliva (top 15 results list).

Figure 6

Reactome pathways enrichment analysis of the proteins characterized in ND_T0 saliva, including the 108 exclusive proteins and the 10 also identified in R-T0 saliva (top 15 results list).

Figure 7

Venn diagram resulting from grouping analysis of the exclusive proteins identified in the R_T0 saliva pool versus different gene categories, namely, “gene elevated expression in glioma”, “gene only detected in glioma”, “unfavorable prognostic genes in GBM”, and “favorable prognostic genes in GBM”, downloaded from the Human Protein Atlas database.

Figure 8

Label-free relative quantitation of TXN, SERPINB5, FABP5, and S100A11 protein area values (in log scale) in CTRL, ND_T0, T1, T3, and R_T0 saliva. Statistical significance was obtained by one-way ANOVA and Tukey’s multiple comparisons test (* p-value < 0.05, ** p-value < 0.01, **** p-value < 0.0001).

Figure 9

STRING tool protein–protein interaction (PPI) network (highest confidence level 0.900, disconnected nodes hidden) of the 140 salivary proteins identified in ND_ and R_T0 GBM saliva, including their exclusive as well as their shared proteins. The edges represent protein-protein associations, with each color indicating a specific type: Known interactions (light blue: curated databases, purple: experimentally determined); Predicted interactions (green: gene neighborhood, red: gene fusions, blue: gene co-occurrence); Others (ochre: text mining, black: co-expression, light blue: protein homology). In the network, the group of proteins exclusive of ND_ and R-T0 saliva and identified in both, as well as the Cancer Related proteins, are highlighted as described in the legend.