HMGB1: Chemoattraction and Tissue Repair
The role of HMGB1 as a chemoattractant is not restricted to the cells of the immune system. Other cell types, including the mesoangioblasts, heart myocytes, smooth muscle cells, fibroblats, monocytes and endothelial cells have been shown to migrate in response to HMGB1 [35-37] (S.
Barnay, unpublished results). These observations suggested that HMGB1 might contribute to tissue repair and/or remodeling processes, concomitantly or following its participation in damage signaling and immune activation.In agreement with this model, an increased regeneration has been observed in vivo in various pathological tissues treated by HMGB1, including infarcted heart [36] and the skin of diabetic mice [38]. However, HMGB1 chemoattractant properties may have some drawbacks, since HMGB1/RAGE interactions have also been implicated in cell growth and tumor metastasis [39-42].
To be or not to be... A Cytokine?
When K. Tracey and colleagues discovered HMGB1 contribution to severe sepsis, it seemed quite obvious that it was a new and important pro-inflammatory cytokine. Experiments conducted in vitro and in animal models of sepsis [10,43] lead to the exciting proposal that HMGB1 was secreted by activated macrophages and, in turn, could act as a secondary activator of immunocompetent cells. Therefore, HMGB1 activation was considered as a central part of a classical pro-inflammatory response involving TNF, IL-1 and IL-6. Moreover, the delayed pattern of HMGB1 secretion (several hours after the initiation of sepsis) compared to the rapid accumulation of TNF and IL-1 opened very promising therapeutic possibilities, using either antibodies against HMGB1, recombinant inhibitory HMGB1 box A or chemical inhibitors of HMGB1 secretion [10,44].
This model was also strengthened by the accumulation of HMGB1 in the sera of patients with sepsis, the highest concentrations of HMGB1 being correlated with the severity of sepsis. Recent data using highly purified forms of the protein however challenged the hypothesis of HMGB1 being a pro-inflammatory cytokine “by its own”.
Investigating the secretion of pro-inflammatory mediators by smooth muscle cells exposed to HMGB1, our laboratory provided evidences that highly purified forms of HMGB1 had barely no pro-inflammatory activity in this model, whereas it could cooperate very efficiently with IL-1β to promote the secretion of soluble phospholipase A2 (sPLA2) [45]. In the same respect, Rouhiainen and coll. confirmed that the pro- inflammatory activity (i.e. its ability to induce TNF-α secretion and nitric oxide release from mononuclear cells) of recombinant HMGB1 was mainly due to its natural ability to bind to phosphatidylserine [46]. In fact, there are now accumulating evidences that HMGB1 exerts its pro-inflammatory activity mainly by forming non-covalent complexes with pro-inflammatory molecules such as LPS or CpG-DNA [47-49].
In the case of CpG DNA, it has been suggested that HMGB1 ability to form complexes and to interact with RAGE may promote the subsequent interaction of CpG DNA to its final receptor, TLR-9.
Altogether, these results may arise questions about some conclusions drawn from experiments conducted with HMGB1 purified from E.coli under native conditions, since this protein can contain minute amounts of LPS, phosphatidylserine, but also significant amounts bacterial nucleic acids as shown by our group (S. Thierry and V. Marechal, unpublished, Fig. 5). Importantly, these contaminants do not appear to induce artefacts nor cooperate with HMGB1 in cell migration assays.