HMGB1: Sounding the Alarm

Alarmins are a newly described and still emerging group of structurally diverse, but functionally related, molecules that include defensins, cathelicidins, eosinophil-derived neurotoxin, and HMGB1 (high mobility group box protein 1).1 All are released in response to infection and tissue damage, and mediate innate immunity and tissue repair. They promote maturation and chemotaxis of antigen presenting cells and act as endogenous adjuvants. Additional candidates for classification as alarmins include the chemokines CCL5/RANTES, CCL19/ELC and CCL21/SLC, as well as GM-CSF, IL-1 alpha, galectin-1, S100 proteins, granulysin, and hepatoma-derived growth factor (HDGF).1,2

HMGB1 is both a founding member of the alarmins and a unique member.1 It was first described as a ubiquitous and highly expressed nuclear non-histone protein. In the nucleus, HMGB1 stabilizes nucleosome formation and facilitates transcription factor binding by bending DNA. Outside the cell it may function as a potent cytokine with the ability to trigger inflammatory mediators (Figure 1). Although HMGB1 lacks a signal sequence, monocytic cells receiving inflammatory signals can acetylate HMGB1, interfering with nuclear localization signals and allowing secretion.3 Alternatively, cell necrosis, but not apoptosis, allows passive leak of HMGB1 from damaged or dying cells. HMGB1 is a chemoattractant for immature dendritic cells (DCs) and promotes their maturation.4 DCs can also secrete HMGB1, and such secretion promotes proliferation and Th1 polarization of interacting T cells.5 HMGB1 is angiogenic and promotes cardiac stem cell growth and differentiation, highlighting its potential involvement in repairing damaged tissues.6,7 Like defensins and cathelicidins, HMGB1 even has a direct and potent bactericidal activity.8

Figure 1
Figure 1. Cellular injury resulting in necrosis leads to passive HMGB1 release. Microbes, microbial particles, or pro-inflammatory cytokines may stimulate later active release from antigen-presenting cells. Pro-inflammatory HMGB1 may act as an adjuvant or assist in tissue repair. When unregulated it may contribute to pathological hyperinflammatory responses.

When unregulated, HMGB1 may contribute to immune-related pathology. An overwhelming infection causing sepsis, or tissue damage caused by trauma, ischemia, or hemorrhage, may result in life-threatening out-of-control HMGB1 responses.9-11 In spite of intensive treatment for severe sepsis, at least 30% of patients do not survive due to organ failure and cardiac arrest. In mouse or rat models of sepsis or hemorrhage, inhibiting HMGB1 has been effective in increasing survival.12 Surprisingly, one such inhibitor is a portion of the HMGB1 molecule itself. Of two DNA interaction sites, box B can substitute for the cytokine-like properties of the whole molecule, while box A can inhibit inflammation.12 Recently, the vascular thrombin binding protein thrombomodulin (TM) has been shown to bind and sequester HMGB1, offering protection from its ill effects and at least partially explaining the anti-inflammatory effects of TM.13 Therapeutic strategies based on one or more of these inhibitors are attractive, especially considering that the HMGB1 levels peak later than 24 hours after the initiation of sepsis, potentially allowing time for treatment to occur.

The receptor utilized by HMGB1 is not entirely clear and may vary by context. For instance, the receptor for advanced glycation end products (RAGE) has a high affinity for HMGB1, and RAGE is implicated in many HMGB1 effects.1,2,5 Depending on the setting, toll-like receptors (TLR), TLR-4 and TLR-2, are likely cofactors or primary receptors of HMGB1, and additional receptors have also been proposed.1,2,14 Recently, it has been suggested that bacterial DNA or other contaminants in bacterially expressed HMGB1 may activate TLR and complicate interpretation results in some studies.2

The alarmins, including HMBG1, are endogenous promoters of the immune response to injury or infection. An increasing appreciation of their functions and mechanisms of activity will further our understanding of both normal and pathogenic inflammatory responses.


  1. Oppenheim, J. J. & D. Yang (2005) Curr. Opin. Immunol. 17:359.
  2. Harris, H. E. & A. Raucci (2006) EMBO Reports 7:774.
  3. Bonaldi, T. et al. (2003) EMBO J. 22:5551.
  4. Yang, D. et al. (2006) J. Leuk. Biol., Sep 11 [Epub ahead of print].
  5. Dumitriu, I. E. et al. (2005) J. Immunol. 174:7506.
  6. Mitola, S. et al. (2005) J. Immunol. 176:12.
  7. Germani, A. et al. (2006) J. Leuk. Biol., Aug 28 [Epub ahead of print].
  8. Zetterstrom, C. K. et al. (2003) Pediatr. Res. 52:148.
  9. Qin, S. et al. (2006) J. Exp. Med. 203:1637.
  10. Ueno, H. et al. (2004) Am. J. Resp. Crit. Care Med. 170:1310.
  11. Goldstein, R. S. et al. (2006) Shock 25:571.
  12. Yang, H. et al. (2004) Proc. Natl. Acad. Sci. USA 101:296.
  13. Abeyama, K. et al. (2005) J. Clin. Invest. 115:1267.
  14. Yu, M. et al. (2006) Shock 26:174.