br protein fragments can passively di use through the endoth
protein fragments can passively diﬀuse through the endothelial barrier in the blood . However, the facile clearance of the low MW tumor-secreted proteins by kidney AZD8931 limits their detection. We have previously shown that liposomes intravenously injected in healthy mice tend to interact with low MW plasma proteins . To examine if our previous finding applies also in the case of protein coronas extracted from tumor-bearing mice, the Relative Protein Abundance (RPA%) of each identified protein was calculated and proteins were then classified according to their molecular weight. As illustrated in Fig. 2C, proteins with MW < 60 kDa accounted for approximately 70% of the protein coronas formed, in both healthy and tumor-inoculated mice. Re-markably, analysis of the in vivo protein coronas increased the identi-fication of proteins with MW < 40 kDa, in comparison with plasma control analysis (Fig. 2C).
To explore the potential exploitation of the in vivo protein corona to sample disease-related proteins we compared protein coronas formed in healthy and melanoma-bearing mice, aiming to identify diﬀerentially abundant proteins. As revealed by the Venn diagram of Fig. 2A, the composition of protein corona was diﬀerent in the absence and pre-sence of melanoma tumors, with 157 and 122 proteins uniquely found, respectively. Moreover, common proteins between the liposomal cor-onas formed in healthy and melanoma-bearing mice (n = 775) i (Fig. 2A), were not equally abundant, as depicted by the RPA (%) values of Figs. 2d and S4 and Fig. S1. These results strengthened our hy-pothesis that proteomic analysis of the liposomal coronas reveals dif-ferences in healthy and diseased states and prompted us to further in-vestigate if the above variations in the abundance of plasma proteins can be exploited for cancer diagnostics.
To assess the reproducibility of protein corona formation we com-pared the total number of identified proteins between the three biolo-gical replicates. As shown in Fig. S2, ∼60% and 55% of corona proteins were common between the three biological replicates in healthy and melanoma-bearing mice respectively. To identify those proteins that were reproducibly diﬀerentially abundant between healthy and mela-noma-bearing mice in all the three biological replicates, further statis-tical analysis of the raw LC-MS/MS data was carried out using Pro-genesis software. The Relative Protein Expression (fold change) and the reliability of measured diﬀerences (ANOVA, p value) were calculated and are presented in Fig. 3A and Table S5. Progenesis statistical analysis revealed 384 individual proteins diﬀerentially expressed (n = 136 up-regulated; n = 248 downregulated) in tumor compared to healthy corona samples, with a p value < 0.05. Astonishingly, the same com-parative analysis performed for plasma samples, revealed only 6 po-tential biomarker proteins (Table S6). Overall, the above data suggest that proteomic analysis of the liposomal protein coronas eliminated the issue of albumin masking, increased the total number of plasma pro-teins identified and enabled the discovery of a significantly higher number of diﬀerentially abundant proteins, in comparison to plasma control analysis.
To investigate whether the above 384 corona proteins have been previously associated with melanoma cancer, Ingenuity Pathway Analysis (IPA) was performed. Disease and function IPA search revealed the association of 170 diﬀerent proteins with melanoma cancer path-ways (with a p value of 1.24E-6) (Fig. 3B and Table S7). According to IPA (Fig. 3B), only 2 of them have been described in the literature as
potential biomarkers for skin cancer (epidermal growth factor receptor and fibronectin 1). The ability of intravenously injected liposomes to surface-capture low abundance, disease specific molecules was further reinforced by the classification of the 170 melanoma-associated corona proteins according to their cellular localisation. The detection of in-tracellular proteins, released from the tumor microenvironment into the blood, is a major challenge in proteomic analysis, due to their ex-tremely low concentration in plasma . As shown in Fig. 3B, 73 melanoma-associated proteins were found to be intracellular (n = 71 cytoplasm; n = 2 nucleus). Despite their involvement with previously described melanoma-associated cancer pathways, none of the in-tracellular corona proteins identified in this study has been reported in the literature as potential biomarker for melanoma diagnosis. Further evaluation will be necessary in order to define the sensitivity, specificity and predictive capabilities of these candidate biomarker proteins and to demonstrate their clinical utility.