3.2 TCZ affects the concentrations of pro-angiogenic factors
Comparing the levels of pro-angiogenic factors, we noted a drop in EMMPRIN after 4 months of TCZ (Fig. 1A), an increase in NGAL (Fig. 1C), and no significant change in MMP-9 or VEGF levels (Fig. 1B, 1D). Likewise, no change occurred in the serum levels of MMP-3 and MMP-7 (data not shown). The serum levels of the anti-angiogenetic factors Tsp-1 (Fig. 1E) and endostatin (data not shown) were also unchanged. Because the angiogenic switch is turned on when the concentrations of pro-angiogenic factors exceed those of anti-angiogenic factors, we calculated the ratio between the pro-angiogenic factor EMMPRIN and the anti-angiogenic factor Tsp-1 for each patient before and 4 months after initiating TCZ, and found a significant decrease following 4 months of treatment (Fig. 1F).
To show that TCZ affected the balance between serum levels of pro- and anti-angiogenic factors, we assessed their angiogenic potential directly on endothelial cells using the tube formation assay. Serum samples before and 4 months after initiation of TCZ treatment were incubated with the endothelial cell line EaHy926, and the number of closed lumens generated, reflecting the angiogenic potential, was quantified. We show that in accordance with the EMMPRIN levels and the EMMPRIN/Tsp-1 ratio, the endothelial cells generated a reduced number of closed lumens after 4 months of TCZ treatment with thicker layers of cells between the lumens, demonstrating reduced angiogenesis (Fig. 1G, 1H).
3.3 Patients responding to TCZ treatment demonstrate reduced EMMPRIN/Tsp-1 ratio. To investigate the correlation between the effects of TCZ on angiogenic factors and the clinical response which was observed in treated patients, we stratified the patients into responders and non-responders, resulting in 15 RA patients who did not respond to TCZ treatment and 25 RA patients who responded well to TCZ treatment according to EULAR criteria [22]. Only NGAL levels were increased in the responders relative to the non-responders (Fig. 2C). Levels of VEGF, MMP-9 and Tsp-1, and surprisingly even EMMPRIN levels, were not different between responders and non-responders (Fig. 2A, 2B, 2D, 2E). However, although each one of these factors separately did not reveal a difference between responders and non-responders, the ratio between EMMPRIN and Tsp-1 was reduced in the responding patients (Fig. 2F), indicating the usefulness of this ratio in evaluating the state of angiogenesis in treated patients.
3.4 TCZ affects the serum expression of miR-146a-5p and miR-150-5p. We next asked whether miRNAs are involved in the regulation of angiogenesis in RA and whether TCZ affects their expression. We selected 9 miRNAs whose expression has been linked to angiogenesis in previous studies and which were also shown to have dysregulated expression in RA [24], and followed their expression in RA patients before and after initiation of TCZ treatment. We chose to examine the level of circulating miRNAs, as those are known to be stable and protected from RNase activity within exosomes or when complexed with serum proteins [18]. We show that no change occurred in the levels of the miRNAs tested except for the levels of miR-146a-5p and miR-150-5p which were significantly increased after 4 months of TCZ treatment relative to treatment initiation (Fig. 3). However, no difference was found in the levels of all the serum miRNAs, including miR-146a-5p and miR-150-5p, between patients considered responders and non-responders to TCZ (data not shown).
3.5 The pro-angiogenic factors are increased by the co-culture, and the anti-angiogenic factor Tsp-1 is reduced. To explore the mechanisms responsible for the changes observed in the patient serum samples after 4 months of TZC treatment, we turned to an in vitro co-culture system of HT1080 fibroblasts and U937 monocytes. Levels of secreted EMMPRIN, VEGF and MMP-9 in the U937 single cultures were minimal (Fig. 4A-C). In co-cultures, EMMPRIN and VEGF levels were synergistically elevated after 48 hours of incubation without TNFα relative to the levels in the single culture of HT1080 (both by about 1.7 fold, p < 0.05), whereas MMP-9 levels showed no significant change. The addition of TNFα increased MMP-9 levels in the co-culture (by 2.2 fold, p < 0.001), but not those of EMMPRIN or VEGF (Fig. 4A-C), consistent with the known inducing activity of TNFα on MMP-9 [25]. Levels of the anti-angiogenic factor Tsp-1 were reduced in the co-culture relative to the single culture of HT1080 (by 2 fold, p < 0.001), and the presence of TNFα further reduced them (by 1.9 fold, p < 0.05, Fig. 4D). The ratio between EMMPRIN and Tsp-1 was increased by the co-culture relative to the single culture of HT1080 cells (by 2 fold, < 0.05), and the addition of TNFα further increased it (by 2.3 fold, p < 0.05, Fig. 4E)
3.6 EMMPRIN expression promotes VGEF and MMP-9 expression in vitro, and neutralization of EMMPRIN activity reduces angiogenesis. We next asked whether EMMRPIN is directly involved in the pro-angiogenic effects of the co-culture. To this end, we incubated each cell type alone with increasing amounts of human recombinant EMMPRIN protein. In both cell lines, the addition of TNFα was necessary to elevate MMP-9 levels, and a significant increase in MMP-9 level (by about 2 fold, p < 0.05, Fig. 5A, 5C) was observed upon adding a concentration of 500 ng/ml of recombinant EMMPRIN relative to the addition of TNFα alone to each cell line. On the other hand, TNFα had no influence on the expression of VEGF, and the addition of 500 ng/ml of recombinant EMMRPIN increased VEGF in HT1080 cells (about 2 fold, p < 0.01), but not in U937 cells (Fig. 5B, 5D).
The neutralizing anti-EMMPRIN antibody was added to the two cell types co-cultured in the presence of TNFα, and after 48 hours of incubation, the accumulation of VEGF and MMP-9 in the supernatants was significantly reduced (by 1.8 and 1.4 respectively, p < 0.05, Fig. 5E, 5F).
Next, the overall contribution of EMMPRIN to the angiogenic potential of the supernatants was examined in functional in vitro assays. Conditioned media (CM) were collected from the TNFα-induced fibroblast-monocyte co-cultures, and EMMPRIN's activity was neutralized by the addition of the anti-EMMPRIN antibody. These treated CM were then incubated with EaHy926 endothelial cells, and the effect was compared to the non-neutralized CM. Neutralization of EMMPRIN activity reduced the ability of endothelial cells to form tube-like structures (by 1.6 fold, p = 0.071, Fig. 6A, 6C), or to migrate and close the gaps formed by a scratch (by 1.4, p = 0.008, Fig. 6B, 6C).
3.7 TCZ regulates miR-146a-5p which in turn regulates EMMPRIN expression in HT1080 cells. Both miR-146a-5p and miR-150-5p were elevated in the serum of RA patients following 4 months of TCZ treatment. Since we have previously demonstrated that miR-146a-5p regulates EMMPRIN expression in tumor cells, we chose to focus on this miRNA and to explore whether it regulates EMMPRIN expression in fibroblasts and monocytes, and whether TCZ treatment affects EMMPRIN and angiogenesis through this regulatory pathway. We evaluated the expression of miR-146a-5p in each cell type in single cultures and in co-cultures in the presence of TNFα, compared to the single cultures without TNFα, which served as calibrators (indicated by the dashed line). For HT1080 cells, the co-culture reduced the levels of miR-146a-5p expression (by 3.5 fold, p = 0.0024), while U937 showed no significant change (Fig. 7A).
Next, we asked whether overexpression of miR-146a-5p would affect the expression levels of EMMPRIN, VEGF and MMP-9. HT1080 cells were transfected either with the miR-146a-5p mimic or with a scrambled sequence (NC mimic), and after 24 hours the transfected cells were incubated in co-culture with TNFα. Transfection of the miR-146-5p mimic reduced EMMPRIN, VEGF and MMP-9 (by 1.3, 2.3 and 2.2 fold, respectively, p < 0.05, Fig. 7B-D), whereas the negative control did not differ from the co-culture with the non-transfected cells.
To assess the effects of TCZ on angiogenesis, we added increasing amounts of the drug to the co-cultured cells with TNFα, and observed that EMMPRIN, VEGF and MMP-9 were all increased (by 1.5 fold, p < 0.05) at a concentration of 500 ng/ml relative to co-cultured cell without the drug (Fig. 7E). Lastly, we show that TCZ at 500 ng/ml reduced miR-146a-5p expression levels in the HT1080 cells (by 1.6 fold, p < 0.001), but increased it in U937 cells (by 1.4 fold, p < 0.001). This inverse expression pattern was also qualitatively reflected by the increase in EMMPRIN expression in HT1080 cells and decreased expression in U937 cells as demonstrated by the western blot analysis (Fig. 7G).