TNFα expressed on the surface of microparticles modulates endothelial cell fate in rheumatoid arthritis

Background Rheumatoid arthritis (RA) is associated with a high prevalence of atherosclerosis. Recently increased levels of microparticles (MPs) have been reported in patients with RA. MPs could represent a link between autoimmunity and endothelial dysfunction by expressing tumor necrosis factor alpha (TNFα), a key cytokine involved in the pathogenesis of RA, altering endothelial apoptosis and autophagy. The aim of this study was to investigate TNFα expression on MPs and its relationship with endothelial cell fate. Methods MPs were purified from peripheral blood from 20 healthy controls (HC) and from 20 patients with RA, before (time (T)0) and after (T4) 4-month treatment with etanercept (ETA). Surface expression of TNFα was performed by flow cytometry analysis. EA.hy926 cells, an immortalized endothelial cell line, were treated with RA-MPs purified at T0 and at T4 and also, with RA-MPs in vitro treated with ETA. Apoptosis and autophagy were then evaluated. Results RA-MPs purified at T0 expressed TNFα on their surface and this expression significantly decreased at T4. Moreover, at T0 RA-MPs, significantly increased both apoptosis and autophagy levels on endothelial cells, in a dose-dependent manner. RA-MPs did not significantly change these parameters after 4 months of in vivo treatment with ETA. Conclusions Our data demonstrate that MPs isolated from patients with RA exert a pathological effect on endothelial cells by TNFα expressed on their surface. In vivo and in vitro treatment with ETA modulates this effect, suggesting anti-TNF therapy protects against endothelial damage in patients with RA. Electronic supplementary material The online version of this article (10.1186/s13075-018-1768-8) contains supplementary material, which is available to authorized users.


Correlation between TNF and clinical parameters
The outcome of our analyses was to evaluate if sTNF percentage (independent variable) was predictor of indexes of disease activity status: DAS28 or CDI or HAQ or SW or TJ, respectively. Analysis was performed by considering each variable at a time in a linear regression analysis model by using SPSS package.

Expression of sTNF on MPs subsets
After characterization of TNF on the surface of MPs, we decided to perform the same analysis on MPs subset. So, our results showed that EMPs expressed a percentage of sTNF that was significantly greater than those of PMPs (p=0.0115). LMPs expressed a high percentage of TNF but not significantly different respect to the others subset ( Fig. 1).
So, the major source of sTNF-MPs seems to be, as expected, the endothelial portion.

Determination of serum TNF and correlation with sTNF-MPs
Results of ELISA test showed that at T0 and T4 RA patients showed serum TNF levels which didn't significantly correlate with the percentage of TNF expressed on MPS (Fig. 2a,b). This result could be due to the fact that we use different methods and different units of measurement.
At T4, only 5 of the 20 patients tested showed a small decrease of serum TNF respect to T0, while in the remaining cases it significantly increased ( p=0.006) (Fig. 2c).
Instead, the TNF bound on the surface of the MPs significantly decreased after therapy with ETA (p<0.0001) (Fig. 2d). In light of these results we could hypothesize that the superficial expression of TNF on microparticles, unlike soluble TNF, could be used as a therapy response marker. Obviously to confirm this we need a far greater cohort of patients.

TNF as predictor of indexes of disease activity status: DAS28 or CDI or HAQ or SW or TJ.
Although the cohort of RA patients was small, to confirm the results about correlation between sTNF -MPs and clinical parameters described in the text, we re-analyzed our data reinforcing previous results, as shown in the supplementary Table 1.In addition, any single association was tested by bootstrap correction (P adj -value) based on 1000 bootstrap samples, with the aim to adjust raw p-values and obtain more robust estimates of standard errors and confidence intervals of the TNF values included in each model. Supplementary Table 1 R 2 -value was used to evaluate the percentage of variation of DAS28 or CDI, or HAQ, or SW, or TJ explained by TNF .
β-coefficients values were used to evaluate the degree of change in DAS28 or CDI, or HAQ, or SW, or TJ for every 1-unit of change of TNF .
F-test and P-value F-test were used to evaluate the significant amount of variance of R 2 -variation of DAS28 or CDI, or HAQ, or SW, or TJ explained by TNF variation.
Padj-value was based on 1000 bootstrap samples.

FIG.1. Expression of sTNF on MPs subsets.
Relative distribution of the percentage of the expression of TNF on the surface of RA-MPs subsets. Comparisons between groups were performed with Student's t-test. *p < 0.05. Graph at the left bottom shows the variation of serum TNF concentration in RA patients at T0 and T4(c). Graph at right bottom shows the change of percentage in TNF expression on the surface of the MPs in RA patients at T0 and at T4 (d).
Comparisons between groups were performed with Student's t-test. *p < 0.05.