Role of High Mobility Group Box 1 (HMGB1) redox state on cardiac fibroblasts functions and tissue remodeling after myocardial infarction.

High mobility group box 1 (HMGB1) is a nuclear factor that when secreted, is able to signal tissue damage. HMGB1 is also implicated in cardiac regeneration and remodeling after myocardial infarction (MI). The study of the redox state of its cysteines led to the identification of different redox isoforms of HMGB1, with different functions: the totally reduced form, named fully reduced HMGB1 has chemotactic effect; the partially oxidized form, disulfide HMGB1, induces cytokines expression; the sulfonyl HMGB1 is terminally oxidized and apparently inactive. The present study aims to examine the role of HMGB1 redox isoforms and the non-oxidizable mutant 3S-HMGB1 on human cardiac fibroblasts (hcFbs) functions in vitro and evaluate their effects on cardiac remodeling using a mouse model of MI. First we studied the effect of HMGB1 on hcFbs migration. We found that fully reduced HMGB1 and 3S-HMGB1, but not disulfide HMGB1, were able to induce hcFbs migration. Moreover, only the 3S-HMGB1 was not susceptible to oxidation and, indeed, its chemotactic effect was maintained also in presence of H2O2. Among the main receptors of HMGB1, hcFbs showed the membrane expression of CXCR4 but not TLR4 and RAGE. In order to investigate the contribution of CXCR4 to hcFbs migration, cells were treated with AMD3100, an inhibitor of CXCL12/CXCR4 binding, and we found that migration induced by fully reduced HMGB1 was abolished but not in response to 3S-HMGB1, demonstrating that HMGB1-induced migration of hcFbs is CXCR4-dependent. In addition, cxcr4-/- mouse embryonic fibroblasts did not migrate in response to fully reduced HMGB1 or 3S-HMGB1, confirming the previous result. HcFbs released basal levels of CXCL12 and the treatment with a blocking antibody for CXCL12 abolished migration induced by fully reduced HMGB1 but not by 3S-HMGB1 showing that 3S-HMGB1 does not require CXCL12 to bind CXCR4. Finally, biacore analysis demonstrated the higher affinity of 3S-HMGB1 for CXCR4 compared to fully reduced HMGB1 or its natural ligand CXCL12. Moreover, both fully reduced HMGB1 and 3S-HMGB1 affected hcFbs-matrix adhesion in a dose dependent way, while neither disulfide HMGB1 or 3S-HMGB1 modulated pro-inflammatory cytokines levels, probably because of lack of TLR4 expression, essential for this function. In vivo experiments showed that infarcted mice receiving fully reduced HMGB1 exhibited a significant recovery of cardiac function while treatment with 3S-HMGB1 determined an increase in LV dilation and infarct size and a worsening in LV function, compared to the vehicle. Finally, only fully reduced HMGB1-treated mice presented a thicker wall of the infarcted area, no sign of cardiomyocytes hypertrophy and increased arteriole density. Our results demonstrated that HMGB1-induced migration of hcFbs is CXCR4 dependent and 3S-HMGB1 does not require CXCL12 to bind the receptor, unlike fully reduced HMGB1. Moreover, the 3S-HMGB1 is active also in oxidizing conditions that may occur soon after MI. Whether these are the reasons for the 3S-HMGB1-dependent worsening of the infarcted heart function observed in vivo has to be determined.