Tadatsugu Taniguchi, Department of Immunology, Graduate School of Medicine and Faculty of Medicine, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-0031, Japan. (fax: +81-3-5841-3375; e-mail: firstname.lastname@example.org).
Abstract. Yanai H, Ban T, Taniguchi T (The University of Tokyo; and Japan Science and Technology Agency; Tokyo, Japan). Essential role of high-mobility group box proteins in nucleic acid-mediated innate immune responses (Symposium). J Intern Med 2011; 270: 301–308.
Central to protective and pathological immunity is the activation of innate immune responses upon recognition of nucleic acids by transmembrane Toll-like receptors (TLRs) and cytosolic receptors. In mammals, the transmembrane pattern recognition receptors TLR3, TLR7 and TLR9 recognize double-stranded RNA, single-stranded RNA and hypomethylated DNA, respectively, while the retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs), RIG-I and MDA5 are known to be cytosolic RNA-sensing receptors. In addition, cytosolic DNA-sensing receptors that include DAI, RIG-I/MDA5 and AIM2 also trigger innate immune responses. High-mobility group box (HMGB)1, 2 and 3 proteins, which also bind immunogenic nucleic acids, are generally involved in the nucleic acid receptor-mediated activation of innate immune responses. There is a hierarchy in the nucleic acid-mediated activation of immune responses, wherein the selective activation of the nucleic acid-sensing receptors is contingent on the more promiscuous sensing of nucleic acids by HMGBs. The aim of this review is to summarize this novel feature of HMGB proteins, as essential frontline instigators of nucleic acid-mediated activation of innate immune responses. In addition, we will discuss the therapeutic implications of these findings.
The high-mobility group box 1 (HMGB1) protein, the first identified member of the HMGB family, is present in almost all eukaryotic cells and is highly conserved between species. It was originally described as a nonhistone chromosomal protein . As an ancient protein that probably originated before the split between the animal and plant kingdoms, HMGB1 is ubiquitously expressed in the nucleus of almost all mammalian cells where it stabilizes nucleosomes and regulates the transcription of many genes. After the original discovery, Rauvala and colleagues isolated HMGB1 protein in a membrane-bound form from the embryonic brain during development, thereby revealing the versatile function of this protein beyond its role in maintaining chromosomal integrity . Twenty years later, this protein is now considered to be central to inflammation-associated events, as a unique mediator of the innate immune responses. As a cytokine-like factor, HMGB1 is secreted by macrophages, mature dendritic cells (DCs) and natural killer cells in response to injury, infection or other inflammatory stimuli. The biological and clinical importance of HMGB1 is underscored by its multifunctionality, as well as by the dysregulation of this protein in a number of pathological conditions, including sepsis, ischaemia–reperfusion injury, arthritis and cancer [3–6]. Moreover, currently, the HMGB family is known to comprise three distinct genes for HMGB1, HMGB2 and HMGB3 .
Recently, HMGBs have attracted attention for their role in nucleic acid-mediated immune responses. During microbial infection or tissue damage, DNA and RNA potently activate the innate immune response through nucleic acid-sensing pattern recognition receptors (PRRs) [8, 9]. In mammals, the transmembrane Toll-like receptors (TLR) namely TLR3, TLR7 and TLR9 recognize double-stranded RNA, single-stranded and short double-stranded RNAs and hypomethylated DNA, respectively [8–11], whereas the retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs), RIG-I and melanoma differentiation-associated protein 5 (MDA5) are best known as RNA-sensing receptors in the cytosol [12–14] (Fig. 1). In addition, cytosolic DNA-sensing receptors, which include DNA-dependent activator of interferon (IFN) regulatory factors (IRFs) (DAI), absent in melanoma 2 (AIM2) and IFN-inducible protein 16 (IFI16)/IFN activated gene 204 (IFI204), also trigger the innate immune systems [15–22] (Fig. 1). Furthermore, it has been shown that RLRs also participate in the cytosolic DNA-sensing system [23–25]. A hallmark of the innate immune response activated by these receptors is the induction of type I IFN [8, 26], pro-inflammatory cytokines and chemokines, except for that by AIM2, which is a critical component of the inflammasome that typically promotes the secretion of interleukin (IL)-1β [15–18] (Fig. 1). Until recently, it was assumed that each of these receptor class functions more or less independently of one another, and no universal or shared mechanism of activation for these nucleic acid receptor classes had been suspected. Yanai et al.  recently reported that HMGB proteins (HMGB1, 2 and 3) function as universal sentinels for nucleic acids.
HMGB proteins and signal transduction
The structure of each HMGB protein is highly conserved (>80% amino acid identity) with a molecular mass of around 25 kDa and several domains including two DNA-binding domains (HMG boxes A and B) and an acidic tail enriched with glutamate and aspartate residues [28–30] (Fig. 2). Whereas HMGB1 is expressed ubiquitously in almost all cells [31, 32], the expression of HMGB2 and HMGB3 is limited. HMGB2 is widely expressed during embryonic development, but its expression is restricted mainly to lymphoid organs and testis in adult mice , and the expression of HMGB3 is limited to embryos and haematopoietic stem cells . Of the three HMGB proteins, HMGB1 is highly expressed in the nucleus where the protein regulates chromatin structure and gene transcription, but it is also present in the cytosol and even in extracellular fluid [35, 36]. HMGB1 has become the focus of attention in the field of immunity because of its participation in the activation of several immune receptors including TLRs [3, 27, 37–44]. For instance, recombinant HMGB1 protein binds to lipopolysaccharide (LPS) and efficiently transfers it to soluble and membrane-bound CD14 . This binding of HMGB1 to LPS is thought to enhance the production of pro-inflammatory cytokines via activation of TLR4, the signalling receptor for LPS/CD14. In addition, it has been reported that HMGB1 secreted by cells shows cytokine-like activity by acting on TLR2 and receptor for advanced glycosylation end products (RAGE) [4, 6, 36, 38, 42, 43, 45–48]. HMGB1 is also involved in TLR9 signalling, as evidenced by the fact that activation of TLR9 is substantially decreased in macrophages and DCs deficient in HMGB1 . These reports suggest that HMGB1 plays a role in PRR-mediated innate immune responses.
Binding of HMGB proteins to immunogenic nucleic acids
During biochemical screening to identify cytosolic proteins that bind to B-form DNA (poly(dA-dT)·poly(dT-dA); referred to as B-DNA ), which is a synthetic ligand for cytosolic DNA sensors of innate immune responses, HMGB1, HMGB2 and HMGB3 were found to be the best candidates . Subsequent in vitro pull-down analysis revealed the direct binding of HMGB proteins to B-DNA. These findings thus raised the possibility that HMGBs play a role in the cytosolic nucleic acid-sensing systems. In support of this, it was found that HMGB1 also binds to DNA derived from calf thymus, bacteria and viruses, all of which also activate, albeit more weakly compared to B-DNA, the cytosolic-sensing pathway [20, 49]. Corresponding to the weaker activity, the binding affinity of these types of DNA was also weaker compared to that of B-DNA . In addition, HMGB1 also binds to immunogenic, synthetic RNAs, such as double-stranded RNA poly(I:C) and single-stranded RNA poly(U), but does not bind to imiquimod (R837), a non-nucleic acid agonist for TLR7. By contrast, HMGB2 binds to immunogenic DNAs but not RNAs, whereas HMGB3 binds to both DNA and RNA. It is interesting to note that HMGB1 interacts preferentially with DNA that has a branched or bent structure [50, 51]. In addition, a recent report has shown that HMGB1 interacts with Escherichia coli 5S rRNA and the group I intron ribozyme from Azoarcus pre-tRNA, both of which have branched RNA features . Thus, it is conceivable that some structural features of immunogenic DNAs and RNAs may reflect their different affinities for binding to the three HMGB proteins. Further study is required to clarify why HMGB1 and HMGB3, but not HMGB2, bind to immunogenic RNAs.
HMGB proteins are essential for cytosolic DNA- or RNA-mediated innate immune responses
The contribution of HMGB proteins to cytosolic nucleic acid-mediated activation of innate immune responses was first examined using cells derived from gene-targeted mice for HMGB1 or HMGB2 . Analysis of mouse embryonic fibroblasts (MEFs) from HMGB1-deficient mice (Hmgb1−/− MEFs) showed a significant defect in mRNA induction for type I IFN, IL-6 and RANTES in response to cytosolically delivered B-DNA or poly(I:C), whereas the response to the non-nucleic acid TLR4 agonist LPS was normal, showing that the defect is selective to these nucleic acid-mediated responses. In addition, mRNA induction was reduced in Hmgb2−/− MEFs only when the cells were stimulated with B-DNA, which is consistent with the selective interaction of HMGB2 with DNA. Involvement of HMGB3 in these responses is also suggested by the observation that type I IFN gene induction by B-DNA was still detectable, albeit at low levels, in Hmgb1−/− MEFs expressing a small interfering RNA (siRNA) that specifically targets HMGB2. Further studies in MEFs expressing an siRNA designed to interfere with the expression of all HMGB proteins (HMGB-deficient MEFs) showed more severe impairment of mRNA induction for type I IFN and pro-inflammatory cytokines upon cytosolic stimulation by B-DNA or poly(I:C) compared to the similarly stimulated Hmgb1−/− and Hmgb2−/− MEFs. Thus, these findings indicate that all HMGB proteins are involved in cytosolic nucleic acid-activated innate immune responses. In macrophages, cytosolic DNA stimulation induces the formation of the inflammasome via activation of AIM2, triggering the secretion of pro-inflammatory cytokines such as IL-1β [15–18]. The involvement of HMGBs in the DNA–AIM2–inflammasome pathway was underscored by the observation that B-DNA-induced secretion of IL-1β is significantly impaired in Hmgb1−/− macrophages and in RAW264.7 cells expressing siRNA that suppresses the expression of all HMGBs . Therefore, HMGBs are required for the full activation of innate immune responses by these classes of cytosolic nucleic acid-sensing receptors.
HMGB proteins in cytosolic signalling pathways and anti-viral responses
Which signalling pathways activated via the cytosolic nucleic acid receptors are affected by the absence of HMGBs? It has been shown that IRF3 and NF-κB transcription factors are both activated by B-DNA or poly(I:C) stimulation [8, 20, 26, 49]. In HMGB-deficient MEFs, the formation of the IRF3 dimer, a hallmark of IRF3 activation [53, 54], was strongly suppressed upon B-DNA or poly(I:C) stimulation . Likewise, the activation of extracellular signal-regulated kinase and NF-κB was also suppressed in these cells. Thus, in the context of the nucleic acid-mediated activation of innate immune responses, HMGBs function upstream of the cytosolic receptor signalling pathways. Furthermore, as described previously, cytosolic nucleic acid-sensing systems are crucial for anti-viral immunity [8, 26]. Infection of HMGB-deficient MEFs with vesicular stomatitis virus or herpes simplex virus type 1 resulted in reduced type I IFN gene induction and higher viral replication than in control cells . Although further study is required to clarify where and how HMGBs bind to virus-derived nucleic acids, these findings support the notion that sensing of these nucleic acids by HMGBs is critical for effective responses.
HMGBs have an essential role in nucleic acid-mediated activation of TLRs
In the light of previous reports showing a requirement for HMGB1 in CpG DNA-TLR9-activated innate immune responses [40, 43] and our observation  that HMGB1 binds to poly(I:C), poly(U) and also CpG-B oligodeoxynucleotide (CpG-B ODN; [55, 56]), which activate TLR3, TLR7 and TLR9, respectively, HMGB1 evidently plays a role in the activation of all nucleic acid-sensing TLRs. Indeed, in conventional DCs from HMGB1-deficient mice, mRNA induction of pro-inflammatory cytokines upon poly(I:C) or CpG-B ODN stimulation was impaired, whereas the induction levels upon TLR4 stimulation by LPS were normal . As gene activation by poly(I:C) or CpG-B ODN was more severely impaired in HMGB-deficient RAW 264.7 cells, it is likely that HMGB2 and HMGB3 also participate in the nucleic acid-mediated TLR responses. Furthermore, in plasmacytoid DCs, a small subset of DCs that produce high quantities of type I IFN via TLR7 or TLR9 stimulation, type I IFN mRNA induction in response to stimulation with poly(U) or another TLR9 agonist CpG-A ODN was impaired in the absence of HMGB1, whereas the induction was largely unaffected by stimulation with the non-nucleic acid TLR7 agonist R837, which does not show any binding affinity for HMGBs. Thus, these findings indicate a role of HMGBs in the activation of all nucleic acid-sensing TLR signalling.
Based on our findings  and those of others [40, 43], we have proposed that HMGB proteins serve as universal sentinels for nucleic acids (promiscuous sensing) and are required for the full-blown nucleic acid-induced activation of innate immune responses mediated by the more discriminative PRRs (discriminative sensing)  (Fig. 3). What is the biological significance of the function of HMGBs? One possible explanation is that nucleic acid-bound HMGBs may function as co-ligands for the receptors; in other words, the binding of nucleic acids to HMGBs may be a precondition for the subsequent recognition by and activation of their cognate receptors. As mentioned previously, this is perhaps analogous to the role of LPS-binding protein and CD14, which are important for the recognition by TLR4 of LPS derived from gram-negative bacteria [44, 57, 58]. In support of this notion, an association between HMGB1 and TLR9 has indeed previously been shown , and we have evidence that HMGB1 was co-immunoprecipitated with RIG-I upon nucleic acid-stimulation (H. Yanai, T. Ban and T. Taniguchi, 2011, unpublished observation). Additionally, a requirement of HMGB1 for eliciting TLR9 signalling in cooperation with RAGE was recently reported . It is also an interesting possibility that, like TLR9 signalling, TLR3 and TLR7 signalling as well as cytosolic nucleic acid-sensing receptors may involve HMGB1–RAGE signalling. However, further study is required for a better understanding of the detailed mechanism of action of HMGB proteins.
Interference of innate immune responses by nucleotide analogues with high binding affinity for HMGB proteins
Excessive or uncontrolled activation of immune responses can cause damage to the host. Ample evidence suggests the involvement of nucleic acid-mediated immune responses in the exacerbation of certain autoimmune diseases such as systemic lupus erythematosus (SLE) [59–64]. Thus, inhibitors of the nucleic acid-mediated immune responses are thought to be useful for suppressing the symptoms of autoimmune disease. In addition to the TLR9 agonist CpG-B ODN, we found that a TLR9 antagonist, a base-free phosphorothioate deoxyribose homopolymer (termed PS; ), also binds to HMGB1 strongly . Given that HMGB proteins are universal sentinels for nucleic acid-sensing receptors as described previously, these compounds may have the potential to interfere with innate immune responses evoked by other immunogenic nucleic acids. Indeed, MEFs showed a hyporesponsiveness to cytosolic B-DNA or poly(I:C) stimulation when pretreated with CpG-B ODN, which alone does not evoke immune response in these cells, whereas their response to LPS remained normal . Furthermore, even in TLR9-deficient plasmacytoid and conventional DCs, a profoundly reduced response was observed upon pretreatment of the cells with CpG-B ODN or PS followed by TLR3 or TLR7 stimulation with their cognate nucleic acid ligands . By contrast, TLR7 activation by R837 remained unaffected by pretreatment with CpG-B ODN or PS. Although the possibility that CpG-B ODN- or PS-mediated interference of nucleic acid signalling may involve additional mechanisms that cannot be excluded, these findings lend further support for the universal role of HMGBs in nucleic acid-mediated innate immune responses and suggest that the development of inhibitors targeting HMGB proteins, i.e., promiscuous nucleic acid sensors, could be useful to suppress nucleic acid-mediated pathological immune responses.
Implications and future prospects
HMGB1 has been implicated in a variety of biologically important processes including transcription, DNA repair, differentiation, development and extracellular signalling. The data discussed in this review demonstrate an additional role of the HMGB family of proteins in nucleic acid-mediated innate immune responses; HMGBs function as universal sentinels of immunogenic nucleic acids in which such nucleic acids are first recognized promiscuously by HMGBs and then subsequently delivered to more discriminative PRRs, such as TLRs and cytosolic receptors. In speculating on the origin of such an intricate recognition and signalling mechanism, it is interesting to note that the chromatin architectural factors Nhp6A and Nhp6B in Saccharomyces cerevisiae belong to the HMGB superfamily and appear to be functionally related to mammalian HMGB1 . Given such a long history of evolution, the interaction between HMGBs and nucleic acids could have served as a primitive mechanism by which eukaryotic cells maintained their chromosomal integrity, on the one hand, and sensed the invading foreign nucleic acids, on the other. The latter may have constituted a primitive mechanism to protect the integrity of the host’s genome by simply binding to foreign nucleic acids or facilitating their degradation. It may be worth investigating whether HMGBs have nuclease activity or are associated with any nucleases. If this is the case, the nuclease activity may be insufficient to eliminate ‘non-self’ nucleic acids, derived from pathogens or nonpathogenic species; hence, later signalling mechanisms could have evolved to involve the induction of genes for cytokines such as type I IFNs to protect the surrounding cells from invasion. During this evolutionary process, HMGBs continued to serve as ‘sentinels’, acting as critical components to activate the nucleic acid-induced signalling pathway. It is clear that this idea requires further elaboration. In any case, promiscuous sensing of nucleic acids by HMGBs appears to be a more primitive feature of sensing ‘non-self’, but is critical for subsequent pattern recognition by PRRs. Characterization of the function of HMGBs and the detailed molecular events in the hierarchical sensing system is required to further understand the significance and advantage of the promiscuous sensing system.
Further, this line of research would provide new insights into the control of immune response with many therapeutic implications. Currently, it is hoped that antibodies against HMGB1 will become available for clinical use, translating the physiologically relevant findings of experimental models to clinical application in HMGB-associated inflammatory diseases [3–6]. In the light of the findings reviewed here, another approach may be to generate nonimmunogenic oligonucleotides with high affinity for HMGBs. Clearly, such oligonucleotides would be useful with regard to nucleic acid-mediated pathological immune responses, such as in SLE [67–70], but one may also expect that these oligonucleotides will inhibit other HMGB-mediated signalling events by introducing conformational change in proteins through tight interactions. Finally, because HMGB1-deficient mice die soon after birth, it would be interesting to generate mice with conditional null-mutations in the HMBG1 loci, to gain further insights into the function of HMGB1 and other members of this family protein.
Conflict of interest statement
None of the authors has any conflicts of interest to declare.
We thank David Savitsky and Hideo Negish for advice and constructive discussion. This work was supported in part by a Grant-In-Aid for Scientific Research on Innovative Areas, and Global Center of Excellence Program ‘Integrative Life Science Based on the Study of Biosignaling Mechanisms’ from the Ministry of Education, Culture, Sports, Science, and Technology of Japan, and by Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Agency (JST). T.B. is a research fellow of the Japan Society for the Promotion of Science.