Immunoregulatory protein B7‐H3 promotes growth and decreases sensitivity to therapy in metastatic melanoma cells

B7‐H3 (CD276) belongs to the B7 family of immunoregulatory proteins and has been implicated in cancer progression and metastasis. In this study, we found that metastatic melanoma cells with knockdown expression of B7‐H3 showed modest decrease in proliferation and glycolytic capacity and were more sensitive to dacarbazine (DTIC) chemotherapy and small‐molecule inhibitors targeting MAP kinase (MAPK) and AKT/mTOR pathways: vemurafenib (PLX4032; BRAF inhibitor), binimetinib (MEK‐162; MEK inhibitor), everolimus (RAD001; mTOR inhibitor), and triciribidine (API‐2; AKT inhibitor). Similar effects were observed in melanoma cells in the presence of an inhibitory B7‐H3 monoclonal antibody, while the opposite was seen in B7‐H3‐overexpressing cells. Further, combining B7‐H3 inhibition with small‐molecule inhibitors resulted in significantly increased antiproliferative effect in melanoma cells, as well as in BRAFV600E mutated cell lines derived from patient biopsies. Our findings indicate that targeting B7‐H3 may be a novel alternative to improve current therapy of metastatic melanoma.

B7-H3 affects sensitivity to various drugs and targeted therapies in several cancer types (Jiang, Liu, Liu, Zhang, & Hua, 2016;Liu et al., 2011;Nunes-Xavier et al., 2016;Zhang et al., 2015a,b;Zhao et al., 2013), but has not been addressed in malignant melanoma. In this study, we have assessed the role of B7-H3 in the sensitivity of melanoma cells to the chemotherapeutic agent DTIC, and to clinically relevant MAPK and AKT/mTOR inhibitors: vemurafenib, binimetinib, everolimus, and triciribidine. We found that low expression or inhibition of B7-H3 renders the cells more sensitive to these drugs in addition to decreasing their glycolytic capacity. Our results suggest that targeting B7-H3 may be a novel supplement to improve current anticancer therapies in metastatic melanoma.

| B7-H3 promotes growth and glycolysis of melanoma cells
To study the role of B7-H3 in melanoma cell growth, we used FEMX-1 and SKMEL-28 cells with knockdown or overexpressed protein levels

Significance
The treatment of metastatic melanoma has experienced a shift in the past years with vemurafenib, targeting mutated BRAF, and the rise of immunotherapy. However, only a portion of patients responds to the treatment and the rate of relapse is high.
Thus, new targeted therapy is urgently needed for metastatic melanoma patients. We found that reducing or inhibiting the expression of B7-H3 in metastatic melanoma cells reduced cell growth and glycolytic capacity, and increased sensitivity to chemotherapy and various targeted therapies. Our findings indicate that targeting B7-H3 may be a novel alternative to improve current therapy of metastatic melanoma.
B7-H3 was found mainly in the cell membrane, with a weaker protein staining in the cytoplasm ( Figure S1, left panels). We did not detect any nuclear staining. In FEMX-1 cells, B7-H3 was co-localized with LAMP2 and CD63 ( Figure S1, middle panels), indicating that B7-H3 is also expressed in intracellular membrane-bound vesicles, that is, lysosomes and late endosomes. Co-localization was not found with melan-A in the pigmented SKMEL-28 cells ( Figure S1, right panels), indicating that B7-H3 is not present in melanosomes. B7-H3 was also detected in exosome extracellular vesicles ( Figure 1b). Notably, exosomes from the B7-H3-overexpressing cell lines had a higher expression of B7-H3 compared to empty vector cells, indicating that higher B7-H3 expression leads to more sorting of the protein to extracellular vesicles.
A modest decrease in cell proliferation and cell confluence was ob- This suggests that the cell glycolytic reserve is lower in B7-H3 knockdown cells. Similar ECAR profile to that of B7-H3 knockdown was found in the presence of an inhibitory anti-B7-H3 monoclonal antibody (α-B7-H3, BRCA84D) ( Figure S2). This shows that inhibiting B7-H3 expression in melanoma cells leads to decreased Warburg effect.

| B7-H3 knockdown and inhibition increase the sensitivity of FEMX-1 melanoma cells to DTIC chemotherapy and MEK and AKT/mTOR inhibitors
The role of B7-H3 in FEMX-1 (HRAS G12V mutated) cells on drug sensitivity was studied by measuring cell confluence and proliferation of FEMX-1 shSCR and shB7-H3 cells to DTIC chemotherapy and to targeted therapy with the MEK inhibitor binimetinib. FEMX-1 shB7-H3 cells were more sensitive than shSCR cells to DTIC (2.01 ± 0.29fold) and more sensitive to binimetinib (1.34 ± 0.08-fold) (

| B7-H3 knockdown and inhibition affect sensitivity of SKMEL-28 melanoma cells to vemurafenib, binimetinib, and everolimus
The role of B7-H3 in SKMEL-28 (BRAF V600E mutated) cell sensitivity to the BRAF inhibitors vemurafenib, binimetinib, and everolimus was studied using cell confluence and proliferation assays. shB7-H3 cells were more sensitive to all three inhibitors as compared to control

| Targeting B7-H3 in patient-derived BRAF V600E mutated metastatic melanoma cell lines
In cell line is more resistant to all targeted therapies tested and reinforces the potential benefit of B7-H3 inhibition in therapy-resistant tumors.
Simultaneous inhibition of MAPK and AKT/mTOR pathways did not give an additive antiproliferative effect in the HRAS G12V mutated FEMX-1 cells, which might be explained by cross talk between the two pathways. However, in BRAF V600E mutated SKMEL-28 cells, a weak but significant enhanced growth inhibition upon treatment with combinations of vemurafenib and binimetinib with everolimus was observed ( Figure 5). Interestingly, RAS mutations in melanomas have been found to activate both MAPK and PI3K/AKT/mTOR pathways, as opposed to BRAF that seems to only activate the MAPK pathway (Downward, 2003;Lasithiotakis et al., 2008). The cross talk between MAPK and PI3K/AKT/mTOR pathways in FEMX-1 cells would then depend on both MEK and AKT activation as inhibition of both causes a similar antiproliferative effect. However, treatment with B7-H3 antibody or knockdown of B7-H3 increased the growth inhibition to each of the monotreatments, but not in combination treatment, indicating that B7-H3 exerts a similar effect on both pathways.
We did not observe significant differences in cell cycle distribution upon B7-H3 knockdown (data not shown). Of note, the reduced proliferation rate in B7-H3-inhibited and knockdown cells may be associated with their loss of glycolytic capacity. Glycolytic capacity has also been proposed to be a predictor of drug sensitivity in tumor models (Mookerjee, Nicholls, & Brand, 2016). It has been shown that B7-H3 suppresses Nrf2 activity, eventually leading to promotion of aerobic glycolysis (Lim et al., 2016). Thus, loss of B7-H3 may reduce cell proliferation and increase drug sensitivity through the inability to generate enough energy of growth.
The subcellular localization of B7-H3 could be important for its functional role in tumorigenesis, but its intracellular localization has not previously been addressed in detail in melanoma cells. Here, B7-H3 was found mainly in the cell membrane, but also in the cytoplasm of melanoma cells. Importantly, we present evidence that the cytoplasmic localization of B7-H3 is within intracellular vesicles, that is, and 2 mm L-glutamine. BRAF V600E mutant patient-derived cell lines (patients 1, 3, and 4) were established from biopsies before vemurafenib treatment as previously described (Lai, Jiang, Farrelly, Zhang, & Hersey, 2012) and were a kind gift from Prof. P. Hersey (Kolling Medical Research Institute, Royal North Shore Hospital, University of Sydney, Australia). These were grown in DMEM (Invitrogen) supplemented with 2 mm L-glutamine and 5% FBS (for patients 1 and 3) and 10% FBS (for Patient 4). Generation of shSCR and shB7-H3 cell variants was previously validated for specific knockdown as previously explained (Tekle et al., 2012). The mammalian expression plasmid to generate stable overexpressing cell lines was previously explained (Nunes-Xavier et al., 2016). Whole-cell protein extracts were prepared by total cell lysis, and immunoblot were performed as described previously (Nunes-Xavier et al., 2010;Nygren et al., 2014). Antibodies  Table S1.  Table S2.

| Extracellular acidification rate
XF96 glycolysis stress test was performed using Seahorse Extracellular Flux Analyzer XF96e to measure the extracellular acidification rate (ECAR) according to the manufacturer's instructions. Cells were seeded in Seahorse plate 48 hr after splitting and cultured overnight to 80% confluence. Before measurement, the culture medium was replaced with cellular assay medium (Seahorse Bioscience) supplemented with 2 mm glutamine and incubated for 1 hr in a CO 2 -free incubator. Assays were performed according to Seahorse protocols with the final concentrations of 10 mm glucose, 1 μm of oligomycin, and 100 mm of 2-deoxy-D-glucose (2-DG) and were performed in at least triplicate wells in three independent experiments for each cell line and condition at separate days ± S.D.

| Statistical analysis
Error bars in results represent data average ± standard deviation