A set of proteins reflecting the prognosis of patients have clinical significance since they could be utilized as predictive biomarkers and/or potential therapeutic targets. With the aim of finding novel diagnostic and prognostic markers for glioblastoma (GBM), a tissue microarray (TMA) library consisting of 62 GBMs and 28 GBM-associated normal spots was constructed. Immunohistochemistry against 78 GBM-associated proteins was performed. Expression levels of each protein for each patient were analyzed using an image analysis program and converted to H-score [summation of the intensity grade of staining (0-3) multiplied by the percentage of positive cells corresponding to each grade]. Based on H-score and hierarchical clustering methods, we divided the GBMs into two groups (n=19 and 37) that had significantly different survival lengths (p<0.05). In the two groups, expression of nine proteins (survivin, cyclin E, DCC, TGF-β, CDC25B, histone H1, p-EGFR, p-VEGFR2/3, p16) was significantly changed (q<0.05). Prognosis-predicting potential of these proteins were validated with another independent library of 82 GBM TMAs and a public GBM DNA microarray dataset. In addition, we determined 32 aberrant or mislocalized subcellular protein expression patterns in GBMs compared with relatively normal brain tissues, which could be useful for diagnostic biomarkers of GBM. We therefore suggest that these proteins can be used as predictive biomarkers and/or potential therapeutic targets for GBM.

We hypothesized that a deep-learning algorithm using HE images might be capable of predicting the benefits of adjuvant chemotherapy in cancer patients. HE slides were retrospectively collected from 1343 de-identified breast cancer patients at the Samsung Medical Center and used to develop the Lunit SCOPE algorithm. Lunit SCOPE was trained to predict the recurrence using the 21-gene assay (Oncotype DX) and histological parameters. The risk prediction model predicted the Oncotype DX score > 25 and the recurrence survival of the prognosis validation cohort and TCGA cohorts. The most important predictive variable was the mitotic cells in the cancer epithelium. Of the 363 patients who did not receive adjuvant therapy, 104 predicted high risk had a significantly lower survival rate. The top-300 genes highly correlated with the predicted risk were enriched for cell cycle, nuclear division, and cell division. From the Oncotype DX genes, the predicted risk was positively correlated with proliferation-associated genes and negatively correlated with prognostic genes from the estrogen category. An integrative analysis using Lunit SCOPE predicted the risk of cancer recurrence and the early-stage hormone receptor-positive breast cancer patients who would benefit from adjuvant chemotherapy.

Previously, we demonstrated that the homeoprotein Msx1 interaction with p53 inhibited tumor growth by inducing apoptosis. However, Msx1 can exert its tumor suppressive effect through the inhibition of angiogenesis since growth of the tumor relies on sufficient blood supply from the existing vessels to provide oxygen and nutrients for tumor growth. We hypothesized that the inhibition of tumor growth by Msx1 might be due to the inhibition of angiogenesis. Here, we explored the role of Msx1 in angiogenesis. Overexpression of Msx1 in HUVECs inhibited angiogenesis, and silencing of Msx1 by siRNA abrogated its anti-angiogenic effects. Furthermore, forced expression of Msx1 in mouse muscle tissue inhibited vessel sprouting, and application of an Ad-Msx1-transfected conditioned medium onto the chicken chorioallantoic membrane (CAM) led to a significant inhibition of new vessel formation. To explore the underlying mechanism of Msx1-mediated angiogenesis, yeast two-hybrid screening was performed, and we identified PIASy (protein inhibitor of activated STAT Y) as a novel Msx1-interacting protein. We mapped the homeodomain of Msx1 and the C-terminal domain of PIASy as respective interacting domains. Consistent with its anti-angiogenic function, overexpression of Msx1 suppressed the reporter activity of VEGF. Interestingly, PIASy stabilized Msx1 protein, whereas deletion of the Msx1-interacting domain in PIASy abrogated the inhibition of tube formation and the stabilization of Msx1 protein. Our findings suggest the functional importance of PIASy-Msx1 interaction in Msx1-mediated angiogenesis inhibition.

Angiogenesis is involved in both normal physiological and pathological conditions. Vascular endothelial growth factor (VEGF) is a major factor for promoting angiogenesis. The current anti-VEGF therapies have limited efficacy and significant adverse effects. To find novel targets of VEGFA for angiogenesis inhibition, we performed yeast two-hybrid screening and identified calpain-6 as a novel VEGFA-interaction partner and confirmed the endogenous VEGFA–calpain-6 interaction in mammalian placenta. A domain mapping study revealed that the Gly321–Asp500 domain in calpain-6 is required for the interaction with the C-terminus of the VEGFA protein. The functional significance of the VEGFA–calpain-6 interaction was explored by assessing its effect on angiogenesis in vitro. Whereas forced overexpression of calpain-6 increased the secretion of the VEGF protein and tube formation, knockdown of calpain-6 expression abrogated the calpain-6-mediated VEGF secretion and tube formation in HUVECs. Consistent with the domain mapping result, overexpressing calpain-6 without the VEGFA-interacting domain III (Gly321–Asp500) failed to increase the secretion of VEGF protein. Our results identify calpain-6, an unconventional non-proteolytic calpain, as a novel VEGFA-interacting protein and demonstrate that their interaction is necessary to enhance VEGF secretion. Thus, calpain-6 might be a potential molecular target for angiogenesis inhibition in many diseases.