High Mobility Group Box-1 and Cancer Jessica E

Review

Masquerader: High Mobility Group Box-1 and Cancer Jessica E. Ellerman,1Charles K. Brown,1Michael de Vera,1Herbert J. Zeh,1Timothy Billiar,1 Anna Rubartelli,2 and MichaelT. Lotze1

Abstract Since its identification a third of a century ago, the high-mobility group box-1 (HMGB1) protein has been linked to varied diverse cellular processes, including release from necrotic cells and secretion by activated macrophages engulfing apoptotic cells. Initially described as solely chromatin-associated, HMGB1 was additionally discovered in the cytoplasm of several types of cultured mammalian cells 6 years later. In addition to its intracellular role, HMGB1has been identified extracellularly as a putative leaderless cytokine and differentiation factor. In the years since its discovery, HMGB1has also been implicated in disease states, including Alzheimer’s, sepsis, ischemia-reperfusion, arthritis, and cancer. In cancer, overexpression of HMGB1, particularly in con- junction with its receptor for advanced glycation end products, has been associated with the proliferation and metastasis of many tumor types, including breast, colon, melanoma, and others. This review focuses on current knowledge and speculation on the role of HMGB1in the development of cancer, metastasis, and potential targets for therapy.

The closing years of life are like the end of a masquerade party, when sepsis (13), ischemia-reperfusion injury (14), arthritis (15), the masks are dropped. and cancer (16). Here we will focus on current knowledge Cesare Pavese concerning the role ofHMGB-1 in the development ofcancer, metastasis, and as a potential target for therapy. The high-mobility group box-1 (HMGB1) protein is present in almost all eukaryotic cells. It was first identified by HMGB1Gene/Protein and Regulation ofExpression Goodwin, Sanders, and Johns in 1973 as one ofa ‘‘group of The HMGB1 gene consists ofsix exons and is located on the chromatin-associated protein with high acidic and basic amino human chromosome 13q12 (17). Southern blot analysis ofthe acid contents’’ (1). As an ancient protein that probably human genomic DNA reveals several pseudogenes and one active originated before the split between the animal and plant gene within the genome (18). Expression ofthe human HMGB1 kingdoms more than 500 million years ago (2), it is gene is under the control ofa very strong TATA-less promoter, ubiquitously present in the nucleus ofalmost all mammalian which has one major start site at 57 nucleotides upstream from cells and is highly conserved between species (3). In 1979, the first exon-intron boundary. This TATA-less promoter is one of Michael Bustin showed that HMGB1 could also be found in the the highest expressing mammalian promoters and exhibits an 18- cytoplasm (4, 5). Because then, we have come to recognize fold greater transcriptional activity than that of the SV40 many additional functions possessed by this versatile protein. promoter. Immediately, upstream ofthe transcriptional start site Within the nucleus, HMGB1 stabilizes nucleosomes and is a silencer element that represses the transcriptional activity of regulates transcription ofmany genes (6, 7). As a cytokine- the HMGB1 promoter by >80%. The transcriptional repression like factor, HMGB1 is secreted by macrophages (8, 9), mature of the silencer element is offset by an enhancer element found dendritic cells (ref. 10), and natural killer cells (11) in response in the first intron of the human HMGB1 gene and can increase to injury, infection, or other inflammatory stimuli. The the HMGB1 activity by 2-fold to 3-fold (19). biological importance ofHMGB1 is underscored by its multi- Members ofthe HMG superfamilyare typically 25 to 30 kDa functionality, as well as pathologic conditions in man with homologous basic domains that mediate DNA binding, implicated by its dysregulation: Alzheimer’s Disease (12), the HMG boxes (20). The human HMGB1 gene encodes a 215–amino acid polypeptide. HMGB1 is highly conserved among species, with over 98% identity between rodent, bovine, Authors’ Affiliations: 1Department of Surgery, Molecular Genetics, and and human proteins (21, 22). Structurally, HMGB1 is Biochemistry, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania composed ofthree domains: two positively charged domains and 2Cell Biology Unit, Instituto Nazionale per la Ricerca sul Cancro y Genoa, Italy (A and B boxes) and a negatively charged carboxyl terminus Received 8/6/06; revised 12/30/06; accepted 1/29/07. (acidic tail; Fig. 1; refs. 23, 24). Functionally, A and B boxes are Grant support: 1 PO1CA 101944-01 (M.T. Lotze) Integrating NK and DC into Cancer Therapy, 1 R21 CA115059-01 (D.L. Bartlett) Isolated Hepatic Perfusion DNA-binding domains. The B box contains the cytokine-like with Oxaliplatin, and Associazione Nazionale per la Ricerca sul Cancro activity, inducing macrophage secretion ofadditional proin- (A. Rubartelli). flammatory cytokines (25). This cytokine activity is antago- Requests for reprints: Michael T. Lotze, University of Pittsburgh, G.27a Hillman nized by recombinant A box. The first 21 amino acid residues Cancer Center, 5117 Centre Avenue Pittsburgh, PA15213. Phone: 412-623-6790; Fax: 412-692-2520; E-mail: [email protected]. (26–46) ofthe recombinant B box represent the minimal F 2007 American Association for Cancer Research. peptide sequence that retains the cytokine-like activity. The doi:10.1158/1078-0432.CCR-06-1953 protein structure involved in the binding ofHMGB1 with

Clin Cancer Res 2007;13(10) May 15, 2007 2836 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. HMGB1 and Cancer

Fig. 1. Structure of HMGB1protein. HMGB1is a 215^ amino acid (AA) protein of f30 kDa. Structurally, HMGB1is composed of three domains: two positively charged domains (A box and B box) and a negatively charged carboxyl terminus (Acidic tail). Functionally, A and B boxes are DNA-binding domains. Recombinant analysis shows that B box contains ‘‘cytokine’’activity by inducing macrophage secretion of proinflammatory cytokines. This cytokine activity is antagonized by recombinant A box.The first 21 ^ amino acid residues (AA89-109) of the recombinant B box represent the minimal peptide sequence that retains cytokine activity.The protein structure involved in the binding of HMGB1with RAGE is located between amino acid residues 150 and 183.These various functional domains of HMGB1are retained conservatively in many species, suggesting multiple critical roles in biological function intracellularly and extracellularly, precluding changes in the molecule. receptor for advanced glycation end products (RAGE) is located HMGB1 as a Nuclear Protein between amino acid residues 150 and 183. HMGB1 expression and subcellular localization varies in HMGB1 binds the minor groove ofDNA facilitatingthe individual cell types and during cell activation (47). Control of assembly ofsite-specificDNA-binding proteins, including HMGB1 gene transcription has been linked to several steroid nuclear hormone/nuclear hormone receptor complexes and hormones involved in development, including glucocorticoids, p53 or p73 transcriptional complexes (57, 58). As it binds estrogens, and progestins. HMGB1 transcription is up-regulated DNA, nuclear HMGB1 plays an important role in the by IFN-g, tumor necrosis factor (TNF)-a, and transforming facilitation of protein-protein interaction and recognition of growth factor-h in matured THP-1 macrophages and in human DNAdamageintheprocessofmismatchrepair(59). peripheral blood monocytes. Posttranscriptional control of HMGB1acts to recruit proteins required for the early steps in mRNA stability or translation is likely important as well (47). mismatch repair enabling physical interactions with exonu- HMGB1 undergoes posttranslational modifications, such as cleases, helicases, polymerases, and ligases at or before the phosphorylation and ADP ribosylation; whereas acetylation excision ofmispaired nucleotides (59). HMGB1 enhances (48, 49) had potential modifications. ADP ribosylation of HMG heteroduplex bending by exonucleases, exposing the DNA proteins regulates gene transcription and may be necessary substrate to attack by helicases and/or nucleases (59). for cell recovery from DNA damage (50, 51). Acetylation allows Nuclear HMGB1 also facilitates the DNA-binding and for the generation of tetramer formation of histone-histone transcriptional activity ofsteroid receptor members ofthe complexes and the stimulation ofcomplex formationwith DNA nuclear receptor family of transcription factors, in particular, polymerase a in the process ofDNA replication (52). the estrogen receptors ERa and ERh. Both are responsive Acetylation by histone acetyltransferases at the K2 position to HMGB1 via enhancement ofreceptor DNA-binding affinity results in enhanced affinity to distorted DNA (49, 53). DNA and transcriptional activity dependent on the C-terminal bending ability is attributed to modification of K81 located in extension region ofthe estrogen receptor DNA-binding domain close proximity to B box (49, 54, 55). Acetylation modulates (60). In its role as a ‘‘superadaptor,’’ HMGB1 has been HMGB1 interactions with other proteins, regulates entry into proposed to act through ‘‘hit-and-run’’ mechanism with these the nucleus, determines several biological activities, and steroid hormone receptors. It has been suggested that the regulates its inflammatory potential (56). HMGB protein acts as a transient and unstable component of

www.aacrjournals.org 2837 Clin Cancer Res 2007;13(10) May 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Review the estrogen receptor/estrogen response element complex, allows HMGB1 to ‘‘change its mask’’ and acquire a role as a enhancing binding interactions at two interfaces. Through its cytokine-like agent. From the cytosol, HMGB1 is released into binding in the vicinity ofthe estrogen response element to open the extracellular environment as a leaderless cytokine. It is the minor groove, HMGB1 allows for rearrangement of the C- synthesized without a leader sequence, which is used by terminal extension residues. Overall this facilitates access to and conventional cytokines to allow delivery into the rough interactions in the minor groove, which is energetically less endoplasmic reticulum, and Golgi apparatus for release through accessible in the absence ofHMGB1. Due to the factthat the the cell membrane. Instead, HMGB1 release is possibly mediated estrogen receptor and HMGB1 proteins are undergoing a by specialized organelles that belong to the endolysosomal dynamic association and dissociation with DNA, the estrogen compartment (64). Active secretion initiates execution ofcellular receptor maintains this binding affinity only as long as HMGB1 response programs in activated macrophages and monocytes remains available. Through this mechanism, these data suggest (56). Via nuclear localization signals 1 and 2, HMGB1 is that HMGB1 proteins may play a critical role in the translocated from the cytosol of cells into the nucleus to bind enhancement ofthe binding ofsteroid hormone receptors to DNA. During the immunologic challenge, macrophages are their cognate response elements and the regulation ofmany activated, resulting in the acetylation ofnuclear localization estrogen-responsive genes and a vital role in the biology of signal sites within HMGB1-enabling cytosolic retention, fol- estrogen-sensitive cancers (61). lowed by packaging into secretory lysosomes and liberation into In the context ofcancer, nuclear HMGB1 is responsible in part the extracellular environment (Fig. 2; refs. 8, 64). for the efficacy of platinum-based therapies through its affinity Passive release ofHMGB1 into the extracellular space occurs for DNA adducts, preventing interaction with the mutation during unscheduled cell death, where HMGB1 diffuses out of the repair machinery. In particular, HMG domain proteins inhibit leaky membranes ofthe necrotic cells (65). In this role, HMGB1 repair ofthe 1,2 intrastrand d(GpG) crosslink by the human serves as a ‘‘necrotic marker’’ or damage-associated molecular excision nuclease, suggesting that levels ofHMG proteins in pattern for recognition by cells of the innate immune system and tumors may influence susceptibility to platinum therapy and is a surrogate for the extent of injury initiating tissue repair (56). serve as a potential screen for therapeutic efficacy (62). However, Passive release ofHMGB1 is not observed during type I apoptotic the effect of HMGB1on platinated DNA remains unclear, as cell death, in which HMGB1 is tightly sequestered within cisplatin-resistant human cancer cells mediate resistance apoptotic bodies, thereby preventing its release. Apoptotic through enhanced expression ofthe HMGB1 gene product (63). cellular chromatin is not only digested at nucleosomal intervals but also is hypoacetylated with phosphorylation ofhistone H2B HMGB1 as a Masquerader: Taking on an (47), enhancing tight HMGBI binding. Extracellular Role Outside the cell, HMGB1 masquerades as a cytokine to activate endothelial cells (Fig. 3), promoting angiogenesis and HMGB1 release into the extracellular environment is mediated extravascular emigration ofinflammatorycells and stem cells, by active and passive mechanisms. It has been suggested that thereby initiating inflammation (66, 67). In this extracellular HMGB1 is hyperacetylated on many ofits 43 lysine residues to role, HMGB1 induces concentration-dependent activity as an allow migration into the cytosol (16), although this has yet to be early mediator after acute injury and late mediator of other confirmed by others. This enzymatic modification presumably varied inflammatory processes (68). Like other cytokines, it has

Fig. 2. HMGB1as a damage-associated molecular pattern molecule (DAMP). In response to tissue damage, stress, and degradation of the tissue matrix, HMGB1 acts at the RAGE receptor to initiate many cellular processes. HMGB1-RAGE signaling results in nuclear translocation of NF-nB (15), angiogenesis, tumorigenesis, and metastasis. HMGB1-RAGE also results in endothelial cell activation, smooth muscle cell migration, and mesangioblast recruitment, migration and proliferation. Other molecules suggested as damage-associated molecular pattern molecules include the heat shock proteins, the S100 family of proteins, purine metabolites including ATP and uric acid, and degraded matrix components such as hyaluronan and heparin sulfate.

Clin Cancer Res 2007;13(10) May 15, 2007 2838 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. HMGB1 and Cancer

Fig. 3. HMGB1interactions with endothelial cells.When released from necrotic cells or secreted from activated monocytes/macrophages, HMGB1acts at endothelial cells to activate the ERK1/ERK2 and stress-activated mitogen-activated protein kinase pathways (c-Jun-NH2-kinase and p38), further activating NF-nB and Sp1. Endothelial cell activation results in the secretion of cytokines, such asTNF-a, which has the ability to enhance endothelial cells to further increase cytokine secretion. ERK1/ERK2 and mitogen-activated protein kinase pathways act to up-regulate receptors such as RAGE and cellular adhesion receptors which can also positively feedback to further increase receptor activation. Fibrinolysis regulators activated by endothelial cell/HMGB1interactions allow for the conversion of plasminogen to plasmin, activating proteolytic degradation of HMGB1 (72).This research was originally published in Blood (72) F the American Society of Hematology. both beneficial and pathologic effects, confining infection and based on its positively charged amino terminal domains and a damage and allowing healing and regeneration ofinjured tissues negatively charged carboxy terminus (74). Membrane-bound (69). Monocytes, macrophages, natural killer, and myeloid/ amphoterin activates RAGE-expressing cells, leading to neurite plasmacytoid dendritic cells secrete HMGB1 in response to outgrowth and survival in part through increased expression of pathogen or damage-associated molecules with resultant extra- B-cell lymphoma 2. In addition to the role ofHMGB1 in cellular HMGB1-recruiting myeloid dendritic cells, plasmacytoid neurite outgrowth, additional studies have suggested a role in dendritic cells, macrophages, neutrophils, and CD4+ T cells. In cell motility and invasive behavior (75). This notion is its role as a late acting mediator ofinflammation,HMGB1 supported by observations that HMGB1 is up-regulated in induces the release of proinflammatory cytokines from mono- immature cells, migrating growth cones, and edges ofmigrating cytes (9, 65), such as interleukin (IL)-12, IL-6, IL-1a, and IL-8, malignant cells (75, 76). HMGB1 expression is high in and matures myeloid dendritic cells, up-regulating CD40, CD54, migrating cells but is down-regulated as a consequence of CD80, CD83, and MHC class II molecules (70, 71). In cell-to-cell contact as a sensor for contact-dependent inhibition endothelia, dose-dependent HMGB1 stimulation results in in migrating cells (75, 77). HMGB1 also binds tissue-type increased expression ofintercellular adhesion molecule 1, plasminogen activator and plasminogen through their lysine- vascular cell adhesion molecule, RAGE, tissue-type plasminogen binding kringle domains, resulting in plasmin production and activator, IL-8, TNF (52), MCP-1, and plasminogen activator enhancing penetration and invasion into tissues (75, 76, 78). inhibitor 1 (72). In primitive mesangioblasts and bone marrow–derived stem cells, HMGB1 serves as a migration target HMGBI as a Superadaptor Promoting Interaction from damaged tissues, thereby promoting wound repair (71). withReceptorsandSignaling Excessive extracellular amounts ofHMGB1 can also cause uncontrolled inflammatory responses leading to tissue injury/ HMGB1 signals alone or together with other molecules as a organ failure, disruption of endothelial barrier functions, linker or superadaptor. HMGB1 binds heparin (75), syndecan- vascular leakage, and tissue hypoperfusion (69). Pretreatment 1 (79), and phosphacan (80), acting to link phosphacan to cell with HMGB1 in ischemia-reperfusion models decreases subse- surface adhesion molecules (80). HMGB1 is a sticky protein quent inflammatory responses, demonstrating the ability of and is postulated to exist in a ‘‘molten globule-like’’ state that HMGB1 to limit innate immune mechanisms (73). facilitates its binding to other proteins. Several receptors have been identified for HMGBI including HMGB1 as a Regulator ofCell Motility and RAGE, toll-like receptors (TLR) 2 and 4, syndecan (79), a Metastasis specific receptor, tyrosine phosphatase, (80), and thrombomodulin (Tables 1 and 2; ref. 81). In 1991, a membrane-bound molecule expressed on growing HMGB1 binds RAGE expressed on neurons, endothelium, neurites was identified by Rauvala et al. and named amphoterin smooth muscle, monocytes, macrophages, T cells, and immature

www.aacrjournals.org 2839 Clin Cancer Res 2007;13(10) May 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Review dendritic cells (56). As a member ofthe immunoglobulin phenotype in endothelial cells after activation of ERK1/ERK2 superfamily, the RAGE gene, like TNF and lymphotoxin, is intracellular signaling via binding ofextracellular RAGE and located within the MHC class III region near the junction with other cell surface receptors (88). When triggered by extracellular the class II MHC complex in humans and mice (82, 83). It is HMGB1, RAGE mediates endothelial cell activation (89), composed ofthree extracellular immunoglobulin like domains, smooth muscle cell migration (90), and mesangioblast a single pass transmembrane domain, and a short highly migration and proliferation (91). Interestingly, RAGE knockout charged cytoplasmic domain essential for signal transmission models result in a phenotype ofdecreased innate immunity but (84). RAGE expression mirrors developmental processes with intact adaptive immunity as illustrated by the finding that high expression during embryonic development and down- RAGE deletion provides protection from the lethal effects of regulation with age (85–87). RAGE increases are observed at septic shock but does not protect from inflammatory reactions the extremes ofage, possibly due to the accumulation ofits such as those in delayed-type hypersensitivity (92). HMGB1 ligands or as a compensatory protective mechanism (86, 87). knockout mice display a phenotype ofnormal organ develop- After RAGE interaction, HMGBI induces phosphorylation of ment and cell growth but die within 24 h ofbirth from extracellular signal-regulated kinase (ERK; ref. 88) and nuclear hypoglycemia (93). Cell lines lacking HMGB1 grow normally, translocation ofnuclear factor- nB (NF-nB; ref. 56) induces but the activation ofgene expression by the glucocorticoid neurite outgrowth (85) and activates GTPases ofthe Rho receptor (encoded by the gene Grl1) is impaired. This suggests family, Cdc42 and Rac, implicated in the cytoskeletal remodel- that HMGB1 is critical for proper transcriptional control by ing in cell movement (84). HMGB1 induces a proangiogenic specific transcription factors (93).

Table 1. Cancer prognosis and correlated HMGB1-RAGE signaling

Cancer Selected references Comments Breast (36, 58) Increased HMGB1 protein expression occurs in breast carcinomas compared with normal tissue. Varying responses of estrogen positive breast cancer may depend on HMGB1 as a modulator of estrogen receptor/estrogen response element binding. Colon (30 – 35) HMGB1 has been defined as an antiapoptotic oncoprotein in human colon cancer due to its over expression and inhibition of apoptosis. Levels are extremely elevated in tumor samples compared with normal tissue, along with the antiapoptotic NF-nB gene product c-IAP2. Coexpression of RAGE and HMGB1 enhanced migration and invasion by colon cancer cell lines. Colon cancer lines expressing HMGB1 have also been a model illustrating that macrophage infiltration into cancers is suppressed, providing a mechanism for tumor destruction of the innate immune system. RAGE expression is associated with atypia, adenoma size, metastasis to lymph nodes/distal organs, and poor prognosis at any stage. Melanoma (20, 106-108) HMGB1 is linked to the initial development of melanoma and is strongly up-regulated in melanoma compared with primary melanocytes. HMGB1 binds to the melanoma inhibitory activity promoter to enhance promoter activity in melanoma cells, with HMGB1 up-regulation closely paralleling malignant transformation. Prostate (27, 63) Expression of HMGB1 and RAGE post luteinizing hormone – releasing hormone therapy was increased in metastatic cases, however not in nonmetastatic cases. The majority of metastatic cases showed coexpression of RAGE and HMGB1 in tumor cells and stromal cells. Pancreatic (27, 63) Human pancreatic cancer lines with high metastasis ability displayed strong expression of RAGE and MMP-9 compared with normal cell lines. MMP-9 and RAGE are now considered indicators of metastasis ability. HMGB1 has been officially grouped as a metastasis specific gene in pancreatic cancer. Lung (37 – 40, 103) Provides an exception to the norm of HMGB1-RAGE over expression in human tumors. RAGE-HMGB1 is down-regulated compared with normal control tissues. sRAGE administration did not suppress tumor growth as typically found in other tumors. RAGE is strongly reduced at both the mRNA and protein levels with down tumors. RAGE is strongly reduced at both them RNA and protein levels with down-regulation paralleling higher tumor stages. Down-regulation was also observed in benign neoplasms.

NOTE: The role of HMGB1 in cancer varies depending on the tumor type. HMGB1-RAGE interaction is typically up-regulated in tumors and is linked to metastasis and poorer prognosis. Lung tumors, however, show down-regulation of RAGE and HMGB1, and administration of lung tumors with RAGE and HMGB1 causes decreased proliferation and metastasis.

Clin Cancer Res 2007;13(10) May 15, 2007 2840 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. HMGB1 and Cancer

Table 2. HMGB-1 Receptors and signaling

Receptor Comment Reference RAGE Interactions with HMGB1 result in MAP kinase (84, 102) activation, enhanced tumor growth, metastases, and release of MMPs. RAGE-HMGB1 also interact with the Rho family of GTPases, Cdc42 and Rac to regulate cell motility. Activates Ras and NF-n (members of the MAP kinase serine/threonine phosphorylation pathway) in inflammatory responses. TLR2/TLR4 Interaction with HMGB1 results in activation of macrophages (73, 94) and maximal stimulation of NF-nB activity in macrophages, along with induction of neovascularization in periods of innate immune system activation. Syndecan HMGB1 links syndecan to the ECM, participates in cell adhesion (79) and migration of simple epithelial cells, particularly in early cell spreading. Phosphacan/protein-Tyr This receptor participates in proteoglycan-mediated regulation (80) phosphatase g/h of cell adhesion, neurite growth and cell migration during central nervous system development. It links phosphacan in the extracellular matrix or the transmembrane phosphatase on adjacent cells, to cell surface glycoproteins such as contactin to which phosphacan alone binds only minimally. Plasminogen This activates plasmin via formation of ternary complexes, also (78) binds tissue-type plasminogen activator.

NOTE: Many studies have illustrated the ability of HMGB1 to signal alone or in concert with other molecules as a linker. Several receptors have been identified for HMGB1 including RAGE, TLR2, TLR4, syndecan (58), a specific receptor-tyrosine phosphatase (59), and thrombomodulin (60). The signaling mechanisms and end events of these receptor interactions are still not totally clear.

Transfection experiments suggest that TLR2 and TLR4 syndrome (97, 98). Recently, extremely elevated levels of receptors are also involved in HMGB1-induced NF-nB activa- thrombomodulin have been identified in the setting of tion (14, 56). The ability ofnecrotic cells to maximally myocardial infarction and stroke (99). stimulate NF-nB activity in macrophages is at least in part dependent on a TLR-mediated signaling pathway; and both Role ofHMGB1 in Cancer TLR2 and TLR4 signaling have been implicated in the activation ofmacrophages by HMGB1 (14). Signaling through The notion that necrotic tumor cells could provide signals to TLRs has also been suggested as a mechanism ofHMGBI in- enhance the growth ofremaining viable ones has been duction ofneovascularization during periods ofinnate immune considered for over 50 years (100). The initial discovery of system activation (81, 94). HMGB1, through TLR activation, HMGB1-RAGE interactions in the modulation ofneurite increases maturation and cytokine secretion from myeloid outgrowth suggested that HMGB1-RAGE signaling might also and plasmacytoid dendritic cells, initiating secretion of be involved in cancer metastasis. Indeed, HMGB1-RAGE IL-18. Natural killer cell secretion ofHMGB1 drives den- interactions mediated invasion, migration, and the growth dritic cell maturation, in turn protecting dendritic cell from and spread ofimplanted C6 gliomas (101). Blockade ofthis natural killer–mediated lysis. HMGB1 is secreted by acti- interaction by soluble or mutated RAGE resulted in suppression vated macrophages and monocytes after exposure to lipopoly- ofmigration ofthe glioma, forcingthe tumor to undergo saccharide, TNF-a,IL-1h,orIFN-g, but not migration prolonged dormancy with decreased proliferation, invasion, inhibitory factor, IL-6, or macrophage inflammatory protein and matrix metalloproteinase (MMP) activity (102). Initially it (MIP)-1h (56). was hypothesized that increased RAGE-HMGB1 activity was Thrombomodulin-HMGB1 interactions have also been linked elevated in all cancers; however, recent studies have revealed to neovascularization (Fig. 4). Thrombomodulin is an endo- that this increase is not always the case, such as in lung thelial cell membrane glycoprotein (95, 96) containing five cancer, in which reduced RAGE expression is found (Table 3; domains: D1 (an N-terminal lectin-like region with antiin- ref. 103). flammatory properties, mediating cell adhesion and binding With confirmation of increased HMGB1 levels in most tumor HMGB1; ref. 97), D2 (a region consisting of six epidermal cells, studies were done to determine whether HMGB1 varied in growth factor–like structures, critical for anticoagulant activity), tumor samples when compared with that found in normal D3 (a glycosylated domain), D4 (a transmembrane region), and cells. HMGB1 from Guerin ascites tumor cells were five times D5 (the cytoplasmic tail; ref. 96). The N-terminal domain of more poly(ADP) ribosylated when compared with control rat thrombomodulin sequesters HMGB1, inhibiting its interactions liver cell HMGB1 (104). Tumor HMGB1 was also highly with RAGE. Increased serum thrombomodulin and impedance acetylated, whereas normal tissue HMGB1 was not. Although ofHMGB1 proangiogenic effects through RAGE has been acetylation can induce conformational changes and HMGB1 suggested as a cause ofpoorer outcomes after acute coronary oligomerization, this change was later identified as an attribute

www.aacrjournals.org 2841 Clin Cancer Res 2007;13(10) May 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Review ofproliferatingtissues in general and not restricted to tumor Prostate cancer. Expression ofHMGB1 and RAGE was cells (104). Whether HMGB1 can be linked to general cellular examined in prostatectomy specimens from 40 patients with proliferation or solely metastasis is still debated. At minimum, pT3 prostate cancer (metastatic and nonmetastatic) preopera- HMG proteins are involved in transcriptional regulation ofa tively treated with luteinizing hormone–releasing hormone number ofgenes with potential roles in malignancy, such as agonist (28). HMGB1 expression was detected in cells from TNF (105). 27% ofmetastatic cases and in none ofthe nonmetastatic cases. Melanoma. Changes in transcriptional regulation represent HMGB1 was detected in prostatic stromal cells of63% derived an early event in tumorigenesis arising within the tumor from metastatic cases compared with 11% nonmetastatic cases. microenvironment after epigenetic and genetic changes in the RAGE expression was detected in cells from 73% of metastatic tumor cell (20). Melanoma development and progression is cases and 33% ofnonmetastatic cases ( P = 0.0244). Most linked to several transcription factor modifications, including metastatic cases coexpressed both HMGB1 and RAGE in tumor cAMP-responsive element binding protein, C-terminal binding cells or in tumor and stromal cells (73%). Androgen protein, microphthalmia transcription factor, NF-nB, as well as deprivation provided by the luteinizing hormone–releasing HMGB1 (20). Melanoma cell lines overexpress HMGB1 at the hormone therapy resulted in paracrine interaction between mRNA and protein levels when compared with control cancer and stromal cells likely through RAGE-HMGB1 interac- melanocytes,andthisislinkedtothedevelopmentof tion in patients with advanced prostate cancer (28). HMGB1- melanoma (106). HMGB1 binds to a specific region within RAGE interactions have been examined in several prostate the melanoma inhibitory activity promoter to enhance cancer cell lines, in hormone-refractory prostate cancer tissues, promoter activity in melanoma cells, with repression of and in normal prostatatic tissues (29). HMGB1 mRNA was HMGB1 expression leading to suppression ofcell invasion expressed in all three cell lines, with DU145 cells, from a (20). In melanoma cells, HMGB1 up-regulates expression ofthe hormone-independent prostate cancer cell line expressing the melanoma inhibitory activity protein, which closely parallels highest level ofRAGE mRNA. Untreated prostate and hormone- malignant transformation (106). Melanoma inhibitory activity refractory prostate carcinomas in turn expressed higher RAGE expression in vivo correlates with the progressive malignancy of and HMGB1 mRNA levels than normal prostatatic tissue (29). melanocytic tumors (107, 108), further confirmed by finding Colon cancer. Overexpression ofHMGB1 in human colon increased melanoma inhibitory activity protein levels in carcinoma with resultant inhibition ofapoptosis suggests the patients with metastatic melanomas (26, 109). role ofHMGB1 as an antiapoptotic oncoprotein (30). HMGB1 Pancreatic cancer. Several human pancreatic cancer cell protein levels are significantly elevated in human colon carci- lines with varying metastatic potential were examined for nomas when compared with control matched normal tissues, expression ofRAGE and MMP-9, both associated with tumor leading to increased NF-nB activity and increased expression of progression and metastasis (27). Lines with high metastatic the antiapoptotic NF-nB target gene product c-IAP2 (inhibitor of ability displayed strong expression ofRAGE and MMP-9 apoptosis), which acts to suppress TNF (30). IAP proteins bind compared with minimal expression ofboth proteins in and inhibit caspases resulting in the downstream prevention of nonmetastatic lines. For this reason, MMP-9 and RAGE are cytochrome C release in mitochondria and complete inhibition indicators ofmetastatic ability in pancreatic cancer cells of ofthe apoptosis pathway (30). Suppression ofcaspase-9 and possible therapeutic importance (27). caspase-3 activity is linked to HMGB1 overexpression, providing

Fig. 4. Mode of cell death is important for immune cell recruitment and activation. When it is time to die, human cells have three predominant fates: apoptosis (I), autophagy (II ), and necrosis (III ). Upon necrotic death, HMGB1is released into the extracellular space resulting in inflammatory cell recruitment and activation. Upon apoptotic death, HMGB1contents are initially sequestered in apoptotic bodies in the nucleus. Later, these bodies are ingested by phagocytes and the HMGB1is released. The exact role of HMGB1in autophagy is unknown; however, it is hypothesized that the mode of release of HMGB1is similar to that in apoptosis.

Clin Cancer Res 2007;13(10) May 15, 2007 2842 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. HMGB1 and Cancer

Table 3. Human cell line models of cancer and HMGB1 interactions

Cell line Comments Reference Melanoma (primary cutaneous): Western blotting showed up-regulation of HMGBI (106) Mel Im, Mel Ei, Mel Wei, in all nine cell lines versus NHEM control melanocytes. Mel Ho, Mel Juso, SK-Mel-3 Keratinocyte staining showed increased HMGB1 staining in malignant melanoma versus NHEM controls. Overall, overexpression of HMGBI protein and mRNA was measured in the malignant melanoma lines compared with NHEM. Metastases of malignant melanoma: HTZ19d, Mel Ju, Sk-Mel-28 Pancreatic: melanoma inhibitory RAGE was strongly expressed in melanoma inhibitory (27) activity, PaCa-2 and PANC-1 activity PaCa-2 and PANC-1, both with high metastatic with high metastatic ability. ability. On the contrary, RAGE was expressed little BxPC-3 has low metastatic ability in BxPC-3 that has low ability. Colon: WiDr, RKO WiDr secreted HMGB1, showed RAGE activity. Endogenous (32, 58) HMGB1 was found in RKO colon cancer lines, with additional HMGB1 transfection providing additional protection against apoptosis. Lung: NCI-H358, NCi-A549, All cell lines were characterized by a low level of RAGE (103) and NCI-H322 mRNA and lack protein expression. Overexpression of RAGE resulted in diminished proliferation of NCI-H358. Breast: MCF-7 (estrogen +) HMGB1 is expressed at a slightly elevated level of 2-fold to (19, 124) 3-fold in estrogen (+) breast cancer cells. Transcription of the human HMGB1 gene in the breast cancer MCF-7 cells starts at one major site 57 nucleotides upstream from the first exon-intron boundary. The 2-fold to 3-fold increase in HMGB1 expression in breast cancer cells stimulated by estrogen may result from the action of the enhancer elements in intron 1. Several novel estrogen-tethered platinum (IV) complexes were synthesized, evaluated for their ability to up-regulate HMGB1, and screened for cytotoxicity against breast cancer cell lines. All complexes induced the overexpression of HMGB1 in estrogen receptor (+) MCF-7 cells.

a possible explanation for the interference of apoptotic Breast cancer. Increased HMGB1 protein expression is machinery through inhibition ofthe Bak pathway (30). found in primary human breast carcinomas compared with Coexpression ofRAGE and HMGB1 enhanced migration and normal tissues, with HMGB1 mRNA present during mouse invasion by colon carcinoma cells Colo320, DLD1, WiDr, and mammary development with low amounts measured during TCO with RAGE and HMGB1 protein and mRNA measured at periods ofapoptotic activity (58). As noted previously, nuclear similar levels (31). Colon cancer tumors expressing HMGB1 HMGB1 facilitates DNA binding and transcriptional activity of have less extensive macrophage infiltration (32), and HMGB1 the ERa and ERh through enhancement ofbinding estrogen derived from the WiDr cell line induces growth inhibition and responsive elements (36, 60, 61). Breast cancer response to apoptosis in macrophages, suggesting a role for HMGB1 in the endocrine therapy does not always correlate with the quantity inhibition ofhost immune responses within the tumor ofexpressed steroid hormone receptor (36). This lack of microenvironment (31). Phosphorylation levels ofRac and correlation may be due to mediation ofestrogen receptor– c-Jun-NH2-kinase/stress-activated protein kinase SAPK were binding to estrogen-responsive elements oftarget genes by increased in macrophages similarly treated with HMGB1- HMGB1 and other adaptor proteins. Northern blot analyses of containing media derived from colon cancer cells (32). HMGB1 transcripts in breast cancer samples revealed a strong Three nuclear matrix proteins were found to be absent in intertumoral variation, which in turn could explain varied normal colonic tissue and altered in colorectal cancer: HMGB1, responses among estrogen receptor–positive tumors (36). creatine kinase B, and heterogeneous nuclear ribonucleoprotein Lung cancer. Lung carcinomas provide an exception to the F (33). Enhanced expression ofHMGB1 correlated with poor ‘‘norm’’ ofHMGB1-RAGE overexpression foundwithin most prognosis. RAGE expression has also been strongly associated other human tumors. Nonsmall cell carcinoma studies estab- with atypia and increased size ofcolorectal adenomas, lished a role for RAGE in the regulation of differentiation in colorectal metastases to lymph nodes and distal organs, and lung tissue characterized by down-regulation ofRAGE (87, poor prognosis at any colorectal cancer stage (34). RAGE was 103). Loss ofHMGB1-RAGE–mediated regulation oftumor found in 19%, 81%, and 100% of Dukes’ B, C, and D cases, cell migration and invasive processes is associated with respectively, correlating with invasiveness and poor outcome increased aggressiveness oftumor behavior in the lung (37, when coexpressed with the ligand (35). 87, 103). The lung is one offewtissues (38, 87) in which RAGE

www.aacrjournals.org 2843 Clin Cancer Res 2007;13(10) May 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Review is constitutively expressed at high levels. Soluble RAGE (sRAGE) tumor growth but also metastasis ofC6 gliomas in immuno- administration does not suppress lung cancer growth and competent mice supports this hypothesis (102). This finding has metastasis as has been found in other tumor types (87, 102). a wide range ofimplications forpotential therapies based on the This variation may be due to the ability ofsRAGE to intercept blockade ofHMGB1-RAGE signaling. Several therapeutic strat- the interaction ofRAGE ligands with other receptors (103). egies have been suggested based on the properties ofHMGB1, RAGE is strongly reduced at both the mRNA and protein level RAGE, and their interactions: (a) the administration ofsRAGE, in nonsmall cell lung carcinomas when compared with (b) antibodies to RAGE or TLR2 receptors or to the ligand normal lung tissues, with down-regulation ofRAGE correlat- (HMGB1/amphoterin), (c) administration ofthe inhibitory A ing with higher tumor-node-metastasis stages. This finding did box ofHMGB1 which antagonizes the B box functionalactivity, not depend on the histologic subtypes; squamous cell lung (d) or small molecule targeting ofsignaling molecules (Fig. 5; carcinoma and adenocarcinoma both showed comparable refs. 35, 41). Manipulation of ERK, signal transducers and down-regulation (103). Down-regulation was also observed in activators oftranscription, myeloid differentiationprotein 88, benign neoplasms ofthe lung including hamartomas. IL-1 receptor–associated kinase, or other pathways involved in Examination ofthe human lung cancer line NCI-H358 promoting inflammation after HMGB1 signaling are also in vitro and in vivo also illustrated that overexpression of possibilities (Table 4; ref. 35). Further investigation is needed full-length human RAGE is associated with diminished tumor to dissect the extent ofextracellular activities ofHMGB1 in cancer growth with RAGE-overexpressing cells evidenced by smaller and the proteins it interacts with to more fully understand these tumors in spheroid cultures in vitro and in athymic mice therapies and their implications in cancer therapeutics. in vivo (103). RAGE down-regulation could be a necessary HMGB1-neutralizing antibodies. Administration ofanti- process in the formation and reorganization of tissue in lung HMGB1 antibody before and shortly after endotoxin exposure tumors. However, the exact mechanism by which RAGE increases the survival ofexposed mice (9). This response is dose contributes to decreased tumor growth remains unclear. dependent, with a higher survival rate correlating with Several explanations have been suggested, including posttran- increased frequencyofadministration ofthe anti-HMGB1 scriptional regulation, the existence oftruncated secretory antibody (9). Neutralizing anti-HMGB1 antibodies inhibit isoforms, varied extracellular levels of RAGE and its ligands release ofTNF- a and IL-6 by blocking extracellular HMGB1 based on tumor location, and the activation by multiple but does not prevent HMGB1 secretion. Ischemic insult-causing ligands with synergistic effects between the proinflammatory hepatic injury is preventable in mouse models by treatment and prosurvival actions (39, 40, 66, 103). with anti-HMGB1 antibodies, most likely through TLR4- mediated inflammatory suppression (15, 68). The role for Potential Therapies Targeting HMGB1 NF-nB in ischemia-reperfusion induced apoptosis in the liver and its interaction with HMGB1 in this circumstance is not Nonprogrammed cell death, rather than hyperproliferation, is currently established. However, inhibition ofNF- nBafter an underexplored mechanism mediating cancer development hepatectomy results in massive apoptosis, impaired liver (35). RAGE-HMGB1 signaling blockade suppresses not only function, and decreased survival (15, 42). Because of response

Fig. 5. Possible therapeutic strategies targeting HMGB1. Several therapies have been suggested based on the properties of HMGB1, RAGE, and their interactions. Extracellular HMGB1can be blocked by administration of sRAGE, antibodies to the RAGE orTLR2 receptors or the ligand (HMGB1/amphoterin), or administration of the inhibitory A box of HMGB1which antagonizes the B box functional activity (35, 41). Intracellular HMGB1can be blocked at the nuclear level by platinum therapy. At the cytosolic level, HMGB1can be blocked from entering the extracellular space by ethyl pyruvate, acetylcholine/nicotine, or stearoyl LPC. By blocking HMGB1at any one of these locations, further release of inflammatory mediators or tissue repair cytokines can potentially be inhibited. Modified from Cytokine Growth Factor Review (15), with permission from Elsevier.

Clin Cancer Res 2007;13(10) May 15, 2007 2844 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. HMGB1 and Cancer

Table 4. Therapeutic strategies

Therapy Prototypes and advantages Disadvantages Selected references HMGB1 sequestration Platinum therapy: sequesters Myeloid suppression, ototoxicity (122, 123) HMGB1 into the nucleus in nephrotoxicity apoptotic death Targets of Extracellular HMGB1 A box HMGB-1: antagonistic, is more A boxHMGB-1: short (9, 110) efficacious than polyclonal HMGB-1 half life antibodies in sepsis and arthritis Anti-HMGB-1 Ab: inhibits TNFa, IL-6, Anti HMGB-1 Ab: does not increases shock survival rate prevent HMGB 1 secretion Targeting receptors sRAGE: inhibits tumor proliferation, sRAGE: has shown promise (43, 69) invasion and expression of MMPs, in animal models only blocks endogenous tumor growth thus far. Stearoyl LPC; mechanism of action unknown Stearoyl LPC: inhibits HMGB1 secretion Stearoyl LPC: exact mechanism from macrophages and protects mice of action remains unknown from lethal sepsis Acetylcholine/nicotine: like stearoyl LPC Acetylcholine/nicotine: potentially toxic Inhibition of HMGB1 release Ethyl pyruvate: inhibits the release of Unknown efficacy outside of septic (46) TNF and HMGB-1, attenuates shock models activation of p38 and NF-nB (95)

NOTE: Several therapies have been suggested for inflammatory diseases and cancer that rely on properties of HMGB1 and RAGE. These therapies are grouped according to their ability to sequester HMGB1, target extracellular HMGB1, target receptors such as RAGE, or inhibit HMGB1 release. These therapies vary in their efficacy and availability for human administration. variability from NF-nB blockade, more consistent outcome may Truncated forms of HMGB1. Structural characteristics of be obtained by blocking initial HMGB1 secretion rather than HMGB1 revealed an ‘‘A box’’ with possession ofantagonistic targeting its presence in the extracellular space. properties to the full length or the agonistic B box of HMGB1 sRAGE and RAGE-HMGB1 blocking strategies. Blockade of (Fig. A; refs. 25, 110). To capitalize on this function, truncated RAGE signaling pathways could also result in attenuation of forms of recombinant HMGB1 containing the A box were tumor development and growth. Several strategies that block tested and found to have antagonistic properties (111). A-box HMGB1-RAGE signaling have been reported, such as the treatment has been evaluated in a sepsis and arthritis model administration ofextracellular ligand binding domain of and found to be equally efficacious as treatment with sRAGE, administration ofblocking Fab fragmentsderived from polyclonal anti-HMGB1 antibodies; however, the short half- anti-RAGE and/or anti-HMGB1 IgG, and generation ofstably life of A box has compelled the need for structural modifica- transfected C6 glioma expressing sRAGE (15). sRAGE in tions (25, 111, 112). A-box therapy has not been tested in particular acts as a decoy to prevent RAGE signaling and has murine tumor models. been used successfully in animal models (43). Human sRAGE Stearoyl (18:0) lysophosphatidylcholine. Stearoyl (18:0) lyso- generated by alternative splicing ofthe RAGE transcript is able phosphatidylcholine has potential as a therapeutic agent for use to bind heparin, enabling distribution in the extracellular in systemic inflammatory states due to its ability to inhibit matrix and at the cell surface (43). Human sRAGE can also be HMGB1 secretion from activated macrophages and protect influenced by proteinases similar to sRAGE found in mice (43). mice from lethal sepsis (69, 113, 114). Lysophosphatidylcho- HMGB1-RAGE signaling inhibition suppresses activation of line treatment ofactivated macrophages abolished HMGB1 p44/p42, p38, SAP/MAP kinases, and NF-nB (39, 43–45, 84), cytoplasmic accumulation and subsequent extracellular secre- along with molecular effectors linked to tumor proliferation, tion (115). Lysophosphatidylcholine binds to the G protein– invasion, and expression ofMMPs (15). sRAGE administration coupled G2A receptor to increase intracellular calcium ions; blocks endogenous growth ofmany tumors, including murine however, the exact mechanism ofaction remains unknown (16, papillomas (102). 69, 116). Ethyl pyruvate. Ethyl pyruvate inhibits the release ofTNF a7-Nicotinic acetylcholine receptor agonists. The nervous and HMGB1 from endotoxin-stimulated RAW 264.7 murine system actively regulates innate immune responses and macro- macrophages, as well as attenuates activation ofboth the p38 phage production ofinflammatorycytokines (69). Vagus nerve mitogen-activated protein kinase and NF-nB signaling path- activation inhibits macrophage release ofHMGB1 (69, 117). ways. Ethyl pyruvate treatment ofseptic mice decreased Acetylcholine could itself be responsible for the antiinflamma- circulating levels ofHMGB1 (46). Pretreatment with ethyl tory mechanism. Indeed, acetylcholine was the first inhibitor of pyruvate also prevented endotoxin lethality and inhibited the HMGB1 secretion from macrophages to be described (69, 115, release ofTNF and HMGB1 (46). Despite these findings,the 118). This finding led to the use of nicotine to activate the exact mechanism ofaction ofethyl pyruvate as a HMGB-1 inhibitor is unknown and its use is currently being explored in 3 murine inflammatory bowel disease models. 3 S. Dave¤ , M.T. Lotze, and S.E. Plevy, manuscript in preparation.

www.aacrjournals.org 2845 Clin Cancer Res 2007;13(10) May 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Review a7-nicotinic acetylcholine receptor, the receptor responsible for PDCs and macrophages that have been identified within the modulation of antiinflammatory effects (69, 119). Macro- tumors wear many masks with some identified roles and phage treatment with nicotine showed inhibition ofHMGB1 others that are still a mystery. HMGB1 is itselfthe great secretion and prevented activation ofthe NF- nB signaling masquerader—it is puzzling how one protein could be pathway (69, 115, 118). Mice exposed to septic conditions responsible for so many roles, both within and outside cells. possessed reduced HMGB1 levels in serum and improved HMGB1 clearly plays a role in cancer development and survival after nicotine therapy (69, 115). Currently, nicotine metastasis, with RAGE-HMGB1 signaling promoting spread of therapy is not feasible, considering its potential toxicity and most tumor types. However, the full extent of that role remains nonspecific receptor activation. However, an agonist of the cryptic along with the possibility ofits interaction with other a7-nicotinic acetylcholine receptor with increased specificity signaling systems. The development ofnew therapies will aid in could present as an alternative therapy (69). the process ofuncovering the extent ofHMGB1 involvement in Platinum therapy. Platinum therapy characteristically creates the numerous cellular processes in which it is hypothesized to two adducts: a 1,2 intrastrand d(GpG) crosslink and a minor 1,3 be involved, including cancer. intrastrand d(GpTpG) adduct (62). Both ofthese adducts are reparable by an excision repair system. HMG domain motifs Note Added in Proof bind specifically to the major platinum DNA dGpG adducts, creating a shield against the human excision nucleases accom- At the Death, Danger and Immunity Meeting held at the plished through the acidic domain (120, 121). In this role, Institut Pasteur, March 8 to 9, 2007, Laurence Zitvogel cisplatin serves as a sequestration mechanism for HMGB1 and addressed the question ofwhether TLR signaling plays a role further validates its potential as an anti-HMGB1 agent. However, in immunogenic cell death. Dr. Zitvogel presented data that the this anti-HMGB1 property cannot be extended to all chemo- response to an irradiated ovalbumin expressing EL4 lymphoma therapeutics. Specifically, the alkylating agent N-methyl-N¶- was TLR4 and MyD88 dependent, but without an apparent nitro-N-nitrosoguanidine induces necrotic cell death with the role for TLR2/TLR7/TLR9/TRIF. Interestingly, large amounts of release ofinflammatorymediators, including HMGB1 (121). HMGB1 could be identified in the supernatants of various One well-known side effect of platinum therapy is ototoxicity, tumor cell lines treated with doxorubicin or radiation. Dr. and recently, HMGB1 has been shown to play a role in cisplatin- Zitvogel showed that HMGB1 binds to TLR4 using pull-down induced ototoxicity in rats (122). The increase in HMGB1 was immunoblots from dying cells. In addition, Dr. Zitvogel was 299 prevented by the administration ofa known protective agent able to show that in breast cancer patients with the AspGly against cisplatin toxicity, the L-methionine (122, 123), suggest- TLR4 polymorphisms, the time to progression was faster in ing that L-methionine might also block HMGB-1 activities. those without the mutated allele. Thus, normal TLR4 is protective, presumably because ofits ability to respond more Conclusions rapidly to the DAMP, HMGB1.

Cancer evolution enables poseurs ofcell damage to promote Acknowledgments reparative angiogenesis, stromagenesis, and epithelial proliferation in a grand masquerade party, with many ofthe involved We thank Drs. Ramin Lotfi, Sebnem Unlu, and Norimasa Ito for helpful discus- players not yet unmasked. Even inflammatory cells including sions.

References 1. Goodwin G. A new group of chromatin-associated 9. Wang H, Bloom O, Zhang M, et al. HMG-1as a late 17. Ferrari S, Finelli P, Rocchi M, Bianchi ME.The active proteins with a high content of acidic and basic amino mediator of endotoxin lethality in mice. Science 1999; gene that encodes human high mobility group 1 pro- acids. Eur J Biochem 1973;38:14^ 9. 285:248 ^ 51. tein (HMG1) contains introns and maps to chromo- 2. Laudet V, Stehelin D, Clevers H. Ancestry and diver- 10. Dumitriu IE, Baruah P,Valentinis B, et al. Release of some 13. Genomics 1996;35:367 ^ 71. sity of the HMG box superfamily. Nucleic Acids Res high mobility group box 1 by dendritic cells controls T 18. WenL,HuangJK,JohnsonBH,ReeckGR.Ahuman 1993;21:2493^ 501. cell activation via the receptor for advanced glycation placental cDNA clone that encodes nonhistone chro- 3.VaccariT, Beltrame M, Ferrari S, Bianchi ME. Hmg4, a end products. J Immunol 2005;174:7506 ^ 15. mosomal protein HMG-1. Nucleic Acids Res 1989;17: new member of the Hmg1/2 gene family. Genomics 11. Semino C, Angelini G, Poggi A, Rubartelli A. NK/ 119 7 ^ 2 14 . 1998;49:247 ^ 52. iDC interaction results in IL-18 secretion by DCs at 19. Lum HK, Lee KL. The human HMGB1promoter is the synaptic cleft followed by NK cell activation and 4. Bustin M, Neihart NK. Antibodies against chromo- modulated by a silencer and an enhancer-containing release of the DC maturation factor HMGB1. Blood somal HMG proteins stain the cytoplasm of mammali- intron. Biochim Biophys Acta 2001;1520:79 ^ 84. 2005;106:609^16. an cells. Cell 1979;16:181 ^ 9. 20. Poser I, Bosserhoff AK. Transcription factors 12. Takata K, Kitamura Y, Kakimura J, et al. Role of high involved in development and progression of malignant 5. Isackson PJ, Bidney DL, Reeck GR, Neihart NK,Bustin h mobility group protein-1 (HMG1) in amyloid- homeo- melanoma. Histol Histopathol 2004;19:173 ^88. M. High mobility group chromosomal proteins isolated stasis. Biochem Biophys Res Commun 2003;301: from muclei and cytosol of cultured hepatoma cells are 699 ^ 703. 21. Paonessa G, Frank R, Cortese R. Nucleotide sequence of rat liver HMG1cDNA. Nucleic Acids Res similar. Biochemistry 1980;19:4466 ^ 71. 13. Qin S,Wang H,Yuan R, et al. Role of HMGB1in apo- 1987;15:9077. 6. Park JS, Arcaroli J,Yum HK, et al. Activation of gene ptosis-mediated sepsis lethality. J Exp Med 2006; expression inhuman neutrophils by high mobility group 203:1637 ^ 42. 22. Yotov WV, St-Arnaud R. Nucleotide sequence of a box 1 protein. Am J Physiol Cell Physiol 2003;284: 14. Tsung A, Hoffman RA, Izuishi K, et al. Hepatic ische- mouse cDNA encoding the nonhistone chromosomal C870 ^ 9. mia/reperfusion injury involves functional TLR4 sig- high mobility group protein-1 (HMG1). Nucleic Acids 7. Stros M, Ozaki T,Bacikova A, Kageyama H, Nakagawara naling in nonparenchymal cells. J Immunol 2005;175: Res 1992;20:3516. A. HMGB1and HMGB2 cell-specifically down-regulate 7661 ^ 8. 23. Bustin M, Lehn DA, Landsman D. Structural fea- the p53- and p73-dependent sequence-specific trans- 15. Ulloa L, Messmer D. High-mobility group box 1 tures of the HMG chromosomal proteins and their activation from the human Bax gene promoter. J Biol (HMGB1) protein: Friend and foe. Cytokine Growth genes. Biochim Biophys Acta 1990;1049:231 ^ 43. Chem 2002;277:7157 ^ 64. Factor Rev 2006;17:189 ^ 201. 24. Weir HM, Kraulis PJ, Hill CS, Raine AR, Laue ED, 8. BonaldiT,TalamoF,ScaffidiP,etal.Monocyticcells 16. Lotze MT,Tracey KJ. High-mobility group box1 pro- Thomas JO. Structure of the HMG box motif in the hyperacetylate chromatin protein HMGB1to redirect it tein (HMGB1): nuclear weapon in the immune arsenal. B-domain of HMG1. EMBO J1993;12:1311^9. towards secretion. EMBO J 2003;22:5551 ^60. Nat Rev Immunol 2005;5:331 ^ 42. 25. Li J, Kokkola R,Tabibzadeh S, et al. Structural basis

Clin Cancer Res 2007;13(10) May 15, 2007 2846 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. HMGB1 and Cancer

for the proinflammatory cytokine activity of high cription factor nuclear factor-nB. Diabetes 2001;50: cancer and the immune response to unscheduled cell mobility group box 1. Mol Med 2003;9:37 ^ 45. 2792 ^ 808. death. J Immunother 2005;28:1 ^ 9. 26. van Groningen JJ, Bloemers HP, Swart GW. Identi- 45. Hofmann MA, Drury S, Fu C, et al. RAGE mediates 66. Scaffidi P, MisteliT, Bianchi ME. Release of chroma- fication of melanoma inhibitory activity and other dif- a novel proinflammatory axis: a central cell surface tin protein HMGB1by necrotic cells triggers inflamma- ferentially expressed messenger RNAs in human receptor for S100/calgranulin polypeptides. Cell tion. Nature 2002;418:191 ^ 5. melanoma cell lines with different metastatic capacity 1999;97:889 ^ 901. 67. Rovere-Querini P, Capobianco A, Scaffidi P, et al. by messenger RNA differential display. Cancer Res 46. UlloaL,OchaniM,YangH,etal.Ethylpyruvatepre- HMGB1 is an endogenous immune adjuvant released 1995;55:6237^ 43. vents lethality in mice with established lethal sepsis by necrotic cells. EMBO Rep 2004;5:825 ^ 30. 27. Takada M, Hirata K, Ajiki T, Suzuki Y, Kuroda Y. and systemic inflammation. Proc Natl Acad Sci U S A 68.Tsung A, Sahai R,Tanaka H, et al. The nuclear factor Expression of receptor for advanced glycation end 2002;99:12351 ^ 6. HMGB1 mediates hepatic injury after murine liver products (RAGE) and MMP-9 in human pancreatic 47. Muller S, Ronfani L, Bianchi ME. Regulated expres- ischemia-reperfusion. JExp Med 2005;201:1135 ^ 43. cancer cells. Hepatogastroenterology 2004;51: sion and subcellular localization of HMGB1, a chroma- 69. Mantell LL, Parrish WR, Ulloa L. Hmgb-1as a thera- 928 ^ 30. tin protein with a cytokine function. J Intern Med peutic target for infectious and inflammatory disor- 28. KuniyasuH,ChiharaY,KondoH,OhmoriH,UkaiR. 2004;255:332^43. ders. Shock 2006;25:4 ^ 11. Amphoterin induction in prostatic stromal cells by 48. Sterner R,Vidali G, AllfreyVG. Studies of acetylation 70. Rouhiainen A, Kuja-Panula J, Wilkman E, et al. Reg- androgen deprivation is associated with metastatic and deacetylation in high mobility group proteins. ulation of monocyte migration by amphoterin prostate cancer. Oncol Rep 2003;10:1863 ^ 8. Identification of the sites of acetylation in HMG-1. (HMGB1). Blood 2004;104:1174^ 82. 29. Ishiguro H, Nakaigawa N, Miyoshi Y, Fujinami K, J Biol Chem 1979;254:11577 ^ 83. 71. Messmer D, Yang H, Telusma G, et al. High mobility KubotaY,Uemura H. Receptor for advanced glycation 49. Pasheva E, Sarov M, Bidjekov K, et al. In vitro acet- group box protein 1: an endogenous signal for dendrit- end products (RAGE) and its ligand, amphoterin are ylation of HMGB-1and -2 proteins by CBP: the role of ic cell maturation and Th1 polarization. J Immunol overexpressed and associated with prostate cancer the acidic tail. Biochemistry 2004;43:2935 ^ 40. 2004;173:307 ^ 13. development. Prostate 2005;64:92 ^ 100. 50. Tanuma S, Yagi T, Johnson GS. Endogenous ADP 72. Fiuza C, Bustin M, Talwar S, et al. Inflammation- 30.Volp K, BrezniceanuML, Bosser S, et al. Increased ribosylation of high mobility group proteins 1 and 2 promoting activity of HMGB1on human microvascular expression of high mobility group box 1 (HMGB1) is and histone H1 following DNA damage in intact cells. endothelial cells. Blood 2003;101:2652^ 60. associated with an elevated level of the antiapoptotic Arch Biochem Biophys 1985;237:38 ^ 42. 73. Izuishi K,Tsung A, Jeyabalan G, et al. Cutting Edge: c-IAP2 protein in human colon carcinomas. Gut 2006; 51. Tanuma S, Johnson GS. ADP-ribosylation of nonhis- 55:234 ^ 42. High-Mobility Group Box 1 Preconditioning Protects tone high mobility group proteins in intact cells. J Biol against Liver Ischemia-Reperfusion Injury. J Immunol 31. Kuniyasu H, Chihara Y, Kondo H. Differential effects Chem 1983;258:4067^70. 2006;176:7154 ^8. between amphoterin and advanced glycation end 52. Dimov SI, Alexandrova EA, Beltchev BG. Differen- 74. Merenmies J, Pihlaskari R, Laitinen J, Wartiovaara J, products on colon cancer cells. Int J Cancer 2003; ces between some properties of acetylated and non- 104 :722 ^ 7. Rauvala H. 30-kDa heparin-binding protein of brain acetylated forms of HMG1protein. Biochem Biophys (amphoterin) involved in neurite outgrowth. Amino 32. Kuniyasu H, Yano S, Sasaki T, Sasahira T, Sone S, Res Commun 1990;166:819^ 26. acid sequence and localization in the filopodia of the Ohmori H. Colon cancer cell-derived high mobility 53. Ugrinova I, Pasheva EA, Armengaud J, Pashev IG. advancing plasma membrane. J Biol Chem 1991;266: group 1/amphoterin induces growth inhibition and In vivo acetylation of HMG1protein enhances its bind- 16722 ^ 9. apoptosis in macrophages. Am J Pathol 2005;166: ing affinity to distorted DNA structures. Biochemistry 751 ^ 60. 75. Rauvala H, Huttunen HJ, Fages C, et al. Heparin- 2001;40:14655 ^60. binding proteins HB-GAM (pleiotrophin) and ampho- 33. Balasubramani M, Day BW, Schoen RE,Getzenberg 54. Teo SH, Grasser KD,Thomas JO. Differences in the terin in the regulation of cell motility. Matrix Biol 2000; RH. Altered expression and localization of creatine ki- DNA-binding properties of the HMG-box domains of 19 :377 ^ 87. nase B, heterogeneous nuclear ribonucleoprotein F, HMG1andthesex-determiningfactorSRY.EurJBio- high mobility group box 1protein in the nuclear matrix 76. Parkkinen J, Raulo E, Merenmies J, et al. Ampho- chem 1995;230:943^50. associated with colon cancer. Cancer Res 2006;66: terin, the 30-kDa protein in a family of HMG1-type 55. Paull TT, Haykinson MJ, Johnson RC. The nonspe- 763 ^ 9. polypeptides. Enhanced expression in transformed cific DNA-binding and -bending proteins HMG1and cells, leading edge localization, and interactions with 34. SasahiraT, AkamaY,Fujii K, Kuniyasu H. Expression HMG2 promote the assembly of complex nucleopro- plasminogen activation. J Biol Chem 1993;268: of receptor for advanced glycation end products and tein structures. Genes Dev 1993;7:1521 ^34. 19726 ^ 3 8. HMGB1/amphoterin in colorectal adenomas.Virchows 56. Ulloa L, Messmer D. High-mobility group box 1 Arch 2005;446:411^ 5. 77. Fages C, Nolo R, Huttunen HJ, Eskelinen E, Rauvala (HMGB1) protein: Friend and foe. Cytokine Growth H. Regulation of cell migration by amphoterin. J Cell 35. Lotze MT, DeMarco RA. Dealing with death: Factor Rev 2006;17:189 ^ 201. Sci 2000;113:611^ 20. HMGB1as a novel target for cancer therapy. Curr Opin 57. Thomas JO, Travers AA. HMG1and 2, and related Investig Drugs 2003;4:1405 ^ 9. 78. Parkkinen J, Rauvala H. Interactions of plasminogen ‘architectural’ DNA-binding proteins. Trends Biochem and tissue plasminogen activator (t-PA) with ampho- 36. Flohr AM, Rogalla P, Meiboom M, et al.Variation of Sci 2001;26:167 ^ 74. terin. Enhancement of t-PA-catalyzed plasminogen HMGB1expression in breast cancer. Anticancer Res 58. BrezniceanuML, Volp K, Bosser S, et al. HMGB1 activation by amphoterin. J Biol Chem 1991;266: 2001;21:3881 ^ 5. inhibits cell death in yeast and mammalian cells and is 16730 ^ 5. 37. Huttunen HJ, Fages C, Kuja-Panula J, Ridley AJ, abundantly expressed in human breast carcinoma. 79. Salmivirta M, Rauvala H, Elenius K, Jalkanen M. Rauvala H. Receptor for advanced glycation end FASEB J 2003;17:1295 ^ 7. Neurite growth-promoting protein (amphoterin, p30) products-binding COOH-terminal motif of amphoterin 59. Yuan F, Gu L, Guo S, Wang C, Li GM. Evidence for binds syndecan. Exp Cell Res 1992;200:444 ^ 51. inhibits invasive migration and metastasis. Cancer Res involvement of HMGB1 protein in human DNA mis- 2002;62:4805 ^ 11. 80. Milev P, Chiba A, Haring M, et al. High affinity bind- match repair. J Biol Chem 2004;279:20935 ^ 40. ing and overlapping localization of neurocan and 38. Brett J, Schmidt AM, Yan SD, et al. Survey of the 60. Melvin VS, Harrell C, Adelman JS, Kraus WL, phosphacan/protein-tyrosine phosphatase-~/h with distribution of a newly characterized receptor for Churchill M, Edwards DP. The role of the C-terminal tenascin-R, amphoterin, and the heparin-binding advanced glycation end products in tissues. Am J extension (CTE) of the estrogen receptor a and h growth-associated molecule. J Biol Chem 1998;273: Pathol 1993;143:1699^712. DNA binding domain in DNA binding and interaction 6998^7005. 39. ArumugamT, Simeone DM, Schmidt AM, Logsdon with HMGB. J Biol Chem 2004;279:14763 ^ 71. 81. MitolaS,BelleriM,UrbinatiC,etal.Cuttingedge: CD. S100P stimulates cell proliferation and survival via 61. Das D, Peterson RC, Scovell WM. High mobility extracellular high mobility group box-1 protein is a receptor for activated glycation end products (RAGE). group B proteins facilitate strong estrogen receptor proangiogenic cytokine. J Immunol 2006;176:12^ 5. J Biol Chem 2004;279:5059 ^ 65. binding to classical and half-site estrogen response 82. Sugaya K, Fukagawa T, Matsumoto K, et al. Three 40. Donato R. Intracellular and extracellular roles of elements and relax binding selectivity. Mol Endocrinol genes in the human MHC class III region near the junc- S100 proteins. Microsc ResTech 2003;60:540 ^ 51. 2004;18:2616^32. tion with the class II: gene for receptor of advanced 41. AgrestiA,LupoR,BianchiME,MullerS.HMGB1inter- 62. Huang JC, Zamble DB, Reardon JT, Lippard SJ, glycosylation end products, PBX2 homeobox gene actsdifferentially withmembersof the Relfamilyoftran- Sancar A. HMG-domain proteins specifically inhibit and a notch homolog, human counterpart of mouse scriptionfactors. Biochem Biophys Res Commun 2003; the repair of the major DNA adduct of the anticancer mammary tumor gene int-3. Genomics 1994;23: 302:421^6. drug cisplatin by human excision nuclease. Proc Natl 408^19. 42. Iimuro Y, Nishiura T, Hellerbrand C, et al. NFnB Acad Sci U S A1994;91:10394 ^ 8. 83. Schmidt AM, Stern DM. Receptor for age (RAGE) prevents apoptosis and liver dysfunction during liver 63. Nagatani G, Nomoto M,Takano H,etal.Transcriptional is a gene within the major histocompatibility class III regeneration. J Clin Invest 1998;101:802^ 11. activationof thehuman HMG1geneincisplatin-resistant region: implications for host response mechanisms in 43. Hanford LE, Enghild JJ,Valnickova Z, et al. Purifica- human cancer cells. Cancer Res 2001;61:1592^7. homeostasis and chronic disease. Front Biosci 2001; tion and characterization of mouse soluble receptor 64. Gardella S, Andrei C, Ferrera D, et al. The nuclear 6:D1151^60. for advanced glycation end products (sRAGE). J Biol protein HMGB1is secreted by monocytes via a non- 84. Huttunen HJ, Fages C, Rauvala H. Receptor for ad- Chem 2004;279:50019^ 24. classical, vesicle-mediated secretory pathway. EMBO vanced glycation end products (RAGE)-mediated 44. Bierhaus A, Schiekofer S, Schwaninger M, et al. Rep 2002;3:995 ^ 1001. neurite outgrowth and activation of NF-nB require Diabetes-associated sustained activation of the trans- 65. Zeh HJ III, Lotze MT. Addicted to death: invasive the cytoplasmic domain of the receptor but different

downstream signaling pathways. J Biol Chem 1999; cytokine’’. J Immunol 2006;176:4512; author reply sepsis with antagonists of endogenous high-mobility 274:19919^ 24. 4513. group box 1. Proc Natl Acad Sci U S A 2004;101: 85. Hori O, Brett J, Slattery T, et al. The receptor for 99. Goldstein RS, Gallowitsch-Puerta M, Yang L, et al. 296 ^301. advanced glycation end products (RAGE) is a cellular Elevated High-Mobility Group Box 1Levels in Patients 112. Kokkola R, LiJ, Sundberg E, et al. Successful treat- binding site for amphoterin. Mediation of neurite out- with Cerebral and Myocardial Ischemia. Shock 2006; ment of collagen-induced arthritis in mice and rats by growth and co-expression of rage and amphoterin in 25:571 ^ 4. targeting extracellular high mobility group box chro- the developing nervous system. J Biol Chem 1995; 100. Revesz L. Effect of tumour cells killed by X-rays mosomal protein 1 activity. Arthritis Rheum 2003;48: 270:25752 ^ 61. upon the growth of admixed viable cells. Nature 2052^ 8. 86. Kokkola R, Andersson A, Mullins G, et al. RAGE is 19 5 6;178 :13 91 ^ 2. 113 . Chen G, Li J, Qiang X, et al. Suppression of the major receptor for the proinflammatory activity of 101. YuW, KJ, Ossowski L. Reduction in surface uroki- HMGB1 release by stearoyl lysophosphatidylcho- HMGB1 in rodent macrophages. Scand J Immunol nase receptor forces malignant cells into a protracted line:an additional mechanism for its therapeutic 2005;61:1 ^ 9. state of dormancy. J Cell Biol 1997;3:767 ^ 77. effects in experimental sepsis. J Lipid Res 2005;46: 87. Bierhaus A, Humpert PM, Morcos M, et al. Under- 102. Taguchi A, Blood DC, delToro G, et al. Blockade of 623 ^ 7. standing RAGE, the receptor for advanced glycation RAGE-amphoterin signalling suppresses tumour 114. YanJJ, JungJS, LeeJE, et al.Therapeutic effects of end products. J Mol Med 2005;83:876 ^ 86. growth and metastases. Nature 2000;405:354 ^ 60. lysophosphatidylcholine in experimental sepsis. Nat 88. Yang H,Wang H, Czura CJ,Tracey KJ.The cytokine 103. Bartling B, Hofmann HS, Weigle B, Silber RE, Med 2004;10:161 ^ 7. activity of HMGB1. J Leukocyte Biol 2005;78:1 ^ 8. Simm A. Down-regulation of the receptor for ad- 115. Wang H, Liao H, Ochani M, et al. Cholinergic ago- 89. Treutiger CJ, Mullins GE, Johansson AS, et al. High vanced glycation end-products (RAGE) supports nists inhibit HMGB1release and improve survival in ex- mobility group 1B-box mediates activation of human non-small cell lung carcinoma. Carcinogenesis 2005; perimental sepsis. Nat Med 2004;10:1216^ 21. endothelium. J Intern Med 2003;254:375 ^ 85. 26:293^ 301. 116. XuY. Sphingosylphosphorylcholine and lysophos- 90. Degryse B, Bonaldi T, Scaffidi P, et al. The high mo- 104. Alexandrova EA, Beltchev BG. Differences phatidylcholine: G protein-coupled receptors and re- bility group (HMG) boxes of the nuclear protein between HMG1proteins isolated from normal and tu- ceptor-mediated signal transduction. Biochim HMG1induce chemotaxis and cytoskeleton reorgani- mourcells. Biochim BiophysActa1987;915:399 ^ 405. Biophys Acta 2002;1582:81 ^ 8. zation in rat smooth muscle cells. JCell Biol 2001;152: 105. Fashena SJ, Reeves R, Ruddle NH. A poly(dA-dT) 117. Tracey KJ. The inflammatory reflex. Nature 2002; 119 7 ^ 2 0 6 . upstream activating sequence binds high-mobility 420:853 ^9. 91. Palumbo R, Sampaolesi M, De Marchis F, et al. Ex- group I protein and contributes to lymphotoxin (tumor 118. Ulloa L. The vagus nerve and the nicotinic anti- tracellular HMGB1, a signal of tissue damage, induces necrosis factor-h) gene regulation. Mol Cell Biol 1992; inflammatory pathway. Nat Rev Drug Discov 2005;4: mesoangioblast migration and proliferation. J Cell Biol 12:894^903. 673 ^ 84. 2004;164:441 ^ 9. 106. Poser I, Golob M, Buettner R, Bosserhoff AK. 119. Critical Therapeutics IPR, Critical Therapeutics ini- 92. Harris EH, Raucci A. Alarming news about danger: Upregulation of HMG1leads to melanoma inhibitory tiates Phase II clinical study of CTI-01. 2005. Workshop on innate danger signals and HMGB1. activity expression in malignant melanoma cells and 12 0. Mitkova E, Ugrinova I, Pashev IG, Pasheva EA.The EMBO Reports 2006;7:774^ 8. contributes to their malignancy phenotype. Mol Cell inhibitory effect of HMGB-1protein on the repair of 93. Calogero S, Grassi F, Aguzzi A, et al. The lack of Biol 2003;23:2991 ^ 8. cisplatin-damaged DNA is accomplished through the chromosomal protein Hmg1 does not disrupt cell 107. Golob M, Buettner R, Bosserhoff AK. Character- acidic domain. Biochemistry 2005;44:5893 ^ 8. growth but causes lethal hypoglycaemia in newborn ization of a transcription factor binding site, specifical- 121. ZongWX,DitsworthD,BauerDE,WangZQ, mice. Nat Genet1999;22:276^80. ly activating MIA transcription in melanoma. J Invest Thompson CB. Alkylating DNA damage stimulates a 94. Frantz S,Vincent KA, Feron O, Kelly RA. Innate im- Dermatol 2000;115:42 ^ 7. regulated form of necrotic cell death. Genes Dev munity and angiogenesis. Circ Res 2005;96:15^26. 108. JacobK,WachF,HolzapfelU,etal.In vitro modu- 2004;18:1272^ 82. 95. Esmon CT, Owen WG. The discovery of thrombo- lation of human melanoma cell invasion and prolifera- 12 2. Li G, LiuW, Frenz D. Cisplatin ototoxicity to the rat modulin. J Thromb Haemost 2004;2:209 ^ 13. tion by all-trans-retinoic acid. Melanoma Res 1998;8: inner ear: a role for HMG1and iNOS. Neurotoxicology 96. Presta M. Response to Comment on ‘‘Cutting 211^ 9. 2006;27:22^ 30. edge: extracellular high mobility group box-1 pro- 109. Bosserhoff AK, Hauschild A, Hein R, et al. Elevated 12 3. Li G, Frenz DA, Brahmblatt S, et al. Round window tein is a proangiogenic cytokine’’. J Immunol 2006; MIA serum levels are of relevance for management of membrane delivery of L-methionine provides protec- 8:4513. metastasized malignant melanomas: results of a Ger- tion from cisplatin ototoxicity without compromising 97. Abeyama K, Stern DM, Ito Y, et al. The N-terminal man multicenter study. J Invest Dermatol 2000;114: chemotherapeutic efficacy. Neurotoxicology 2001; domain of thrombomodulin sequesters high-mobility 395^6. 22:163 ^76. group-B1 protein, a novel antiinflammatory mecha- 110. Erlandsson Harris H, Andersson U. Mini-review: 12 4. Barnes KR, KutikovA, Lippard SJ. Synthesis, char- nism. J Clin Invest 2005;115:1267 ^ 74. The nuclear protein HMGB1as a proinflammatory me- acterization, and cytotoxity of a series of estrogen 98. Alleva L. Comment on ‘‘Cutting edge: extracellular diator. Eur J Immunol 2004;34:1503 ^ 12. tethered platinum (IV) complexes. Chem Biol 2004; high mobility group box-1protein is a proangiogenic 111. Yang H, Ochani M, Li J, et al. Reversing established 11: 5 5 7 ^ 6 4 .

Clin Cancer Res 2007;13(10) May 15, 2007 2848 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2007 American Association for Cancer Research. Masquerader: High Mobility Group Box-1 and Cancer

Jessica E. Ellerman, Charles K. Brown, Michael de Vera, et al.

Clin Cancer Res 2007;13:2836-2848.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/13/10/2836

Cited articles This article cites 121 articles, 46 of which you can access for free at: http://clincancerres.aacrjournals.org/content/13/10/2836.full#ref-list-1

Citing articles This article has been cited by 24 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/13/10/2836.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/13/10/2836. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.