Dimethylfumarate (DMF) has been shown to reduce melanoma growth and metastasis in animal models. We addressed the question of whether DMF is as effective in its antitumor activity as the US Food and Drug Administration-approved alkylating agent dacarbazine (DTIC). We also tested the possibility of an improved antitumoral effect when both therapeutics were used together. Using our severe combined immunodeficiency (SCID) mouse model, in which xenografted human melanoma cells metastasize from primary skin sites to sentinel nodes, we show that these treatments, alone or in combination, reduce tumor growth at primary sites. Our main finding was that metastasis to sentinel nodes is significantly delayed only in mice treated with a combination of DTIC and DMF. Subsequent experiments were able to show that a combination of DTIC/DMF significantly reduced lymph vessel density in primary tumors as examined by real-time PCR and immunohistochemistry. In addition, DTIC/DMF treatment significantly impaired melanoma cell migration in vitro. In vivo, DTIC/DMF therapy significantly reduced mRNA expression and protein concentration of the promigratory chemokines CXCL2 and CXCL11. In addition, our data suggest that this xenotransplantation model is suitable for preclinical testing of various combinations of antimelanoma agents.

The response of blood vessels to physiological and pathological stimuli partly depends on the cross talk between endothelial cells (EC) lining the luminal side and smooth muscle cells (SMC) building the inner part of the vascular wall. Thus, the in vitro analysis of the pathophysiology of blood vessels requires coculture systems of EC and SMC. We have developed and validated a modified three-dimensional sandwich coculture (3D SW-CC) of EC and SMC using open μ-Slides with a thin glass bottom allowing direct imaging. The culture dish comprises an intermediate plate to minimize the meniscus resulting in homogenous cell distribution. Human umbilical artery SMC were sandwiched between coatings of rat tail collagen I. Following SMC quiescence, human umbilical vein EC were seeded on top of SMC and cultivated until confluence. By day 7, EC had formed a confluent monolayer and continuous vascular endothelial (VE)-cadherin-positive cell/cell contacts. Below, spindle-shaped SMC had formed parallel bundles and showed increased calponin expression compared to day 1. EC and SMC were interspaced by a matrix consisting of laminin, collagen IV, and perlecan. Basal messenger RNA (mRNA) expression levels of E-selectin, angiopoietin-1, calponin, and intercellular adhesion molecule 1 (ICAM-1) of the 3D SW-CC was comparable to that of a freshly isolated mouse inferior vena cava. Addition of tumor necrosis factor alpha (TNF α) to the 3D SW-CC induced E-selectin and ICAM-1 mRNA and protein induction, comparable to the EC and SMC monolayers. In contrast, the addition of activated platelets induced a significantly delayed but more pronounced activation in the 3D SW-CC compared to EC and SMC monolayers. Thus, this 3D SW-CC permits analyzing the cross talk between EC and SMC that mediate cellular quiescence as well as the response to complex activation signals.

Melanoma cells can switch between distinct gene expression profiles, resulting in proliferative or invasive phenotypes. Signaling pathways involved in this switch were analyzed by gene expression profiling of a cohort of 22 patient-derived melanoma cell lines. CDH1 negativity was identified as a surrogate marker for the invasive phenotype. CDH1 expression could be turned on and off by modulating activity of p38 or its downstream target MK2, suggesting that this pathway controls melanoma progression. Mechanistically, MK2 inhibition prevented melanoma-induced vascular barrier disruption, reduced the expression of PODXL and DEL-1, and prevented vascular dissemination in vivo. PODXL and DEL-1 expression in patients with melanoma were associated with poor survival and thus can be used as prognostic markers. Downstream targets of MK2 may thus serve as candidate therapeutics.

Cell-cell junctions of blood endothelial cells are critical barriers in inflammatory diseases. Endothelial tight junctions (TJs) control barrier function, and the cytoplasmic adaptor protein cingulin connects TJs to signalling pathways. However, local events at TJs during inflammation are largely unknown. In this study, we investigate the local response of TJ adaptor protein cingulin and its interaction with Rho guanine nucleotide exchange factor H1 (GEF-H1) upon vascular barrier disruption to find a new approach to counteract vascular leak. Based on transendothelial-electrical-resistance (TEER) measurements, cingulin strengthened barrier integrity upon stimulation with histamine, thrombin, and VEGF. Cingulin also attenuated myosin light chain 2 (MLC2) phosphorylation by localising GEF-H1 to cell junctions. Using cingulin phosphomutants, we verified that the phosphorylation of the cingulin head domain is required for its protective effect. Increased colocalisation of GEF-H1 and cingulin was observed in the vessels of vasculitis patients compared to those in healthy skin. Our findings demonstrate that cingulin can counteract vascular leak at TJs, suggesting the existence of a novel mechanism in blood endothelial cells that protects barrier function in diseases.

DEL-1 (developmental endothelial locus-1) induces integrin signaling and recognizes phosphatidylserine exposed on apoptotic cells. We show that DEL-1, which is thought to be a secreted molecule, is not found in melanoma cell culture supernatants but is exported by an endosomal pathway and released via small extracellular vesicles (sEV). Proteomics of DEL-1 positive sEV, but not of DEL-1 negative sEV contain proteins associated with poor survival in cancer. To determine whether DEL-1 is suitable to predict treatment responses, we isolated sEV from plasma of melanoma patients before and 90 days of treatment with checkpoint inhibitors. Although we could not detect DEL-1 in plasma sEV even in patients with progressive disease (most likely due to the very low protein yield from sEV isolated from 1ml plasma), the principal component analysis allowed a clear differentiation between controls and patients as well as between patients before and after treatment. Interestingly, in one patient with complete regression, in the post treatment sample, the protein expression profile remained in the pre-treatment cluster. The low protein yield of patient sEV and the low patient number are clear limitations of this study, but results demonstrate that this method could have the potential to predict treatment responses.