The interactions of tumor cells with their neighboring endothelial cells present in their surrounding environment has emerged as an increasingly relevant factor in tumor progression during angiogenesis, intravasation at the primary tumor site, and adhesion and extravasation at the site of metastasis [26–28]. The available information indicates that the soluble factors secreted by tumor cells can alter the phenotype of different cell types, modifying their activity and provoking tissue destruction, tumoral cell migration and dissemination . We have previously reported that HUVECs treated with soluble factors secreted by tumoral cells (TSF's), can adhere U937 cells and that this response is linked to the activation of NF-κB and the expression of cell adhesion molecules [19, 30].
Since Jaffe established the methodology for the culture in 1973, HUVECs had been the principal model to the studies of physiological and pathological process involved endothelial cells. For instance, previous studies have investigated functional differences between HUVECs (human umbilical vein endothelial cells) and HDMECs (human dermal microvascular endothelial cells) with respect to, upregulation of adhesion molecules in response to cytokines, stimulation and expression of surface antigens or mechanical properties of leukocytes rolling. However, at the moment does not exist clear functional differences. Histological studies of the expression of adhesion molecules, as such as VCAM-1 in primary vascular tumoral tissue, can serve to compare endothelial models with the behavior of cells in vivo [31–33].
Our current working hypothesis is that the tumor cells can use adhesion molecules, such as VCAM-1, to interact with and adhere to the endothelial monolayers, essentially emulating leukocytes during the inflammatory reaction. In this work, we compared the increase in adhesive capacity of HUVEC's with the increased expression of different VCAM-1 isoforms.
A cell adhesion assay (Figure 1) was used to compare the pro-adhesive phenotype of HUVECs induced by TSF's. Since different tumor cell lines present variable adhesion to unstimulated endothelial cells, we used the promyelocytic human cell line U937 as a probe of the induction of pro-adhesive phenotype in response to the different TSF's. This assay showed that the factors derived from the cell line ZR75.30 (TSFZR75.30) were as effective as TNF in activating the endothelial phenotype, using the concentration of TSF's in which the percentage of adhesion (1 μg/ml-50%), was the highest and did not have differences statistically significatives, with respect to another concentrations (0.5 μg/ml-38%, 0.25 μg/ml-24%, 0.125 μg/ml-19%, 0.0625 μg/ml-18%). TSF's by other three breast cancer cell lines were prepared and tested in the same cell adhesion assay (T47D, MDA MB 435 and MDA MB 231). Addition of 1 μg/ml of TSF's from either induced different fold increase of adhesion: 1.3 ± 0, 1.6 ± 0.2, 2.2 ± 0.1 respectively.
TNF is recognized as the most important physiological stimulus for the activation of signaling pathways that lead to the translocation of NF-κB into the nucleus, for several cell types [34–36]. In HUVECs, TNF and TSFZR75.30 both induced the translocation of NF-κB to the nucleus, although the TSF's only stimulated the system by about 50% in comparison with TNF (Figure 5). However, the amount of VCAM-1 expressed was slightly higher in TSFZR75.30 treated cells, suggesting that expression of this adhesion molecule could result from the recruitment of other transcription factors activated by the mixture of elements present in the TSF's. The analysis of the mixture of TSFZR75.30 revealed very low levels of TNF, along with an abundance of cytokines such as IL-6 and IL-8 that could be responsible for NF-κB activation.
The expression of adhesion molecules, such as VCAM-1, in response to chemokines and cytokines is essential in the acute inflammatory response and represents a clear sign of an activated endothelial phenotype [35, 37–39]. Unidimensional and bidimensional western blots analysis [40, 41] revealed that TSFZR75.30 was able to induce the expression of VCAM-1a (Mr 81 kDa/pI 5.1) [NCBI access number NP_001069] and VCAM-1b (Mr 71 kDa/pI 5.0) [NCBI access number NP_542413] in HUVECs in a similar magnitude as TNF (Figure 2A).
In addition, the westerns showed four new isoforms: isoform x in uni-dimensional gels (Figure 2B) and isoforms c, d, and e in bi-dimensional gels (Figure 4). Isoforms c, d and e were present both, in cells treated with TNF as well as in those treated with TSFZR75.30, although isoform d was bearly visible in cells treated with TNF. We conclude that TSFZR75.30 promote a stronger expression of all isoforms compared to the induction mediated by TNF. In an attempt to determine if the isoforms contained N-glycosylations, we interfered with the formation of dolicholpyrophosphate N-acetylglucosamine, the first step in the synthesis of N-linked glycoproteins, by using tunicamycin. Under this condition, the protein portion of glycoproteins will be synthesized completely devoid of N-glycosylations . VCAM-1a has six N-glycosylation sites, whereas removal of exon 5 in VCAM-1b eliminates the second of these sites. Proteins lacking N-glycosylation have been reported to have decreased stability in the endoplasmic reticulum and hence, are more easily exported and degraded in the cytoplasm by the proteasome. This is a likely explanation for the decreased cellular content of isoform "a" (40% and 70% decrease with TNF or TSFZR75.30 respectively) and the disappearance of isoform "b" in the presence of tunicamycin (Figure 2B and Figure 4).
The isoform "x" (Mr ~75-77 kDa) (Figure 2B), became visible only in the presence of tunicamycin. Considering that one N-glycosylation modification corresponds to an added weight of 3 kDa and that VCAM1a (90-95 kDa) has six N-glycosylation sites, tunicamycin treatment could lead to isoforms that are up to 18 kDa smaller. Hence, isoform "x" could correspond to the full length core protein (9 exons) lacking all N-glycosylation modifications. At this point we cannot discard that the TSFZR75.30 induce altered glycosylation compared to that indicated by TNF. The fact that tunicamycin pretreatment abolished 50% of the cell adhesion induced by TSFZR75.30 indicates that N-glycosylated proteins such as VCAM-1 play an important rol in cell adhesion, other cell adhesion molecules such as E-selectin and ICAM-1 are likely involved in this process.
We assigned VCAM-1b to the spot with a Mr ~80 kDa/pI 4.6, the spot with Mr ~83 kDa/pI 5.1-5.2 we labeled as isoform "c", which could correspond to the core protein of isoform "b", with different pI resulting from differential states of sialo-glycosylation at any of the five remaining N-glycosylation sites. Alternatively, isoform "c" could also result from the loss of exons 2 or 8, leading to a protein with a similar Mr as "b", but with an increase in the density of charge due to the preservation of all the reported glycosylations sites (Figure 2A).
The appearance of isoform "d", which was overexpressed in cells treated with TSFZR75.30, could be of potential clinical use as a biological marker for indicating the abnormal activation of endothelial cells by tumoral factors. Isoform "d" had the lowest molecular weight, which was suggestive of a smaller protein generated by alternative splicing or proteolytic processing. Interestingly when we interfered the process of N-glycosylation, isoform "d" disappeared, and a new isoform "e" (Mr ~77 kDa/pI 6.0) appeared. The fact that these two isoforms ("d" and "e") have the same apparent Mr suggests that they both correspond to the same core protein. It is likely, that the isoform "x", identified in the Figure 2-B, corresponds to isoform e, since both were visible only in the presence of tunicamycin. According to the reported exon structure of VCAM-1, we evaluated the possibilities for the expected proteins when exons 2 (92 amino acids), 3 (107 amino acids), or 8 (89 amino acids) were eliminated. The predicted proteins had the following Mr/pI values: 68928 kDa/5.13, 66997 kDa/5.14, and 69376 kDa/5.01, indicating that none of them could produce the observed "e" isoform, which had a Mr/pI value of ~77 kDa/6.0. This analysis further supports the idea that isoform "e", is encoded by all nine exons but lacks all N-glycosylation modifications. In addition, although we cannot discard other types of posttranslational modifications such as phosphorylation that could also explain the differences in pI of the different isoforms, these modifications have not been previously described for VCAM-1. Expression of VCAM-1 isoforms in tumors has not been well studied. In the past half-century, numerous studies have dealt with the effects of TSF's on endothelial cells. These studies have demonstrated that malignant cells produce a host of factors, most notably VEGF, that favor growth and vascular permeability, facilitating the spread of tumors [43–45]. In addition to cytokines and chemokines, our study also detected significant amounts of VEGF secreted by the breast cancer cell line ZR75.30. The complex mixture of soluble factors secreted by these cells reflects the multifactorial nature of signals emitted by tumor cells that can influence endothelial behavior . The specific combination of cytokines, chemokines and growth factors, observed in the TSF's could serve as a signature to distinguish between tumor cells with different metastatic or invasive potentials in breast cancer.