Current address: Neurobiology of Disease Laboratory, MassGeneral Institute of Neurodegenerative Disease, Massachusetts General Hospital, Harvard Medical School, 114 16th Street, Charlestown, MA 02129, USA.
From the Neuroscience Center, University of Helsinki, Helsinki, Finland
Prof. Heikki Rauvala, PO Box 56 (Viikinkaari 4), University of Helsinki, FIN-00014 Helsinki, Finland (fax: 358-9-19157620; e-mail: email@example.com)
Amphoterin is a ubiquitous and highly conserved protein previously considered solely as a chromatin-associated, nuclear molecule. Amphoterin is released into the extracellular space by various cell types, and plays an important role in the regulation of cell migration, differentiation, tumorigenesis and inflammation. This paper reviews recent research on the mechanistic background underlying the biology of secreted amphoterin, with an emphasis on the role of amphoterin as an autocrine/paracrine regulator of cell migration.
Amphoterin was originally isolated from the perinatal rat brain as a heparin-binding protein that promotes neurite outgrowth in brain neurones in vitro . Surprisingly, amphoterin sequence  turned out to be identical to the sequence cloned for high mobility group 1 protein (HMG-1) . The designation ‘HMG’ refers to nonhistone components of chromatin. In addition to HMG-1, amphoterin has also been called as p30, sulphoglucuronyl carbohydrate binding protein-1 (SBP-1) and differentiation enhancing factor (DEF). The nomenclature for HMG protein family has recently been revised and amphoterin is now called HMGB1 . However, in our laboratory we have used the term ‘amphoterin’ referring to secreted, extracellularly active protein and ‘HMGB1’ referring to nuclear protein serving as an architectural component of chromatin. For reasons of clarity only the term amphoterin will be used in this review.
Amphoterin (HMGB1) binds cruciform DNA, a nondouble helix form of DNA, that is generated e.g. as an intermediate in genetic recombination . Numerous studies have suggested various functions for amphoterin in the nucleus which will be discussed elsewhere in detail ( and other reviews in this issue). Interestingly, mice lacking amphoterin die only a few hours after birth, due to a defect in transcriptional enhancement of glucocorticoid receptor . However, whether amphoterin serves as a transactivator, a quasi-transcription factor or a simple structural component of chromatin is still not well understood.
Although amphoterin lacks a classical secretion signal it has been found as an extracellular protein in several independent studies aimed at identification of proteins involved in cell and tissue development [1, 2, 7–13]. Amphoterin localizes to the advancing plasma membrane of the filopodia and at the leading edge in motile cells [2, 7, 8]. Both endogenous and exogenous amphoterin (provided as a substrate-attached material) promote neurite outgrowth and antibodies against amphoterin inhibit neurite initiation [1, 2, 7, 8] suggesting an autocrine or paracrine function for amphoterin in neurite outgrowth. In the developing rat brain, amphoterin is highly expressed in migrating neurones of the cortical plate and in the ventricular zone of the cerebral cortex [14–17] suggesting involvement in neuronal migration.
Localization of amphoterin to the processes of migrating cells suggests a general role in cell motility [2, 8]. Interestingly, amphoterin mRNA is strongly expressed in migrating cells but is downregulated during cell-to-cell contact formation . Transfection of amphoterin antisense oligonucleotides or addition of anti-amphoterin antibodies to the medium prominently inhibits haptotactic transfilter migration towards laminin . Furthermore, amphoterin is expressed by a wide variety of transformed cells [8, 10, 19–23] suggesting an invasive migration of tumour cells.
A series of studies have suggested an important role for extracellular amphoterin in erythroleukaemia cell differentiation [10–12, 24]. A 6-kDa N-terminal fragment  and a proteolytically processed short peptide (residues 130–139; ) of amphoterin are responsible for amphoterin-induced differentiation of erythroleukaemia cells. In neural tissues amphoterin expression seems to correlate with an undifferentiated cell stage and early maturation [14–17]. Amphoterin induces neuronal differentiation of N18 neuroblastoma cells characterized by chromogranin expression . Chromogranins are structural components of mature secretory vesicles that are found, for e.g. in the precursor vesicles in immature axonal growth cones  suggesting that they are important factors in the generation of secretory neuronal phenotypes. Furthermore, our preliminary experiments with embryonic stem cells have shown that amphoterin is capable of inducing differentiation towards neuronal direction as manifested by expression of neurofilament light chain protein and β-III-tubulin, two commonly used markers of neuronal differentiation . Stimulation of LAN-5 neuroblastoma cells with amphoterin induces acetylcholinesterase activity to a similar extent as stimulation with retinoic acid, a well-known inducer of neuronal differentiation . In the peripheral nervous system, amphoterin mediates neurone–Schwann cell interactions important for myelination . These observations indicate that the effects of amphoterin on neuronal cells are not limited to regulation of growth processes but can be extended to regulation of cell differentiation in a wider sense.
The unexpected finding that amphoterin is a late mediator of endotoxin lethality in mice  has directed amphoterin research into completely new areas. Amphoterin was found to be secreted from monocytes and macrophages 8–32 h after stimulation with lipopolysaccharide or proinflammatory cytokines tumour necrosis factor-α (TNF-α) and interleukin-1β (IL-1β). Furthermore, delayed administration of amphoterin antibodies attenuates endotoxin lethality, and administration of amphoterin itself is lethal . Cytokine-type functions of amphoterin (reviewed elsewhere in this issue) have been further supported by evidence suggesting a role for amphoterin in haemorrhagic shock , acute inflammatory lung injury  and rheumatoid arthritis [33–36]. Amphoterin secretion is induced by proinflammatory cytokines, such as TNF-α and IL-1β, and amphoterin can serve as a macrophage-activating cytokine itself. TNF-α and IL-1β induce amphoterin secretion from pituicytes  suggesting that amphoterin might participate in the regulation of neuroendocrine and immune responses to infection.
Structure of amphoterin
Amphoterin has a highly dipolar structure (designated therefore as amphoterin) consisting of a 185-amino acid basic region followed by a cluster of 30 acidic residues in the carboxy-terminus (; Fig. 1). The 185-amino acid basic part (∼28% lysine or arginine residues) of amphoterin is subdivided in two homologous HMG boxes (Fig. 1) each ∼75 amino acids in length. Both HMG boxes have a similar α-helical structure important in DNA-binding [38, 39]. The three-dimensional structure of the whole amphoterin molecule has not been resolved to date. However, it seems likely that the dipolar nature is the apparent reason for the formation of amphoterin dimers and oligomers under physiological conditions .
The N-terminus of amphoterin (residues 6–12; see Fig. 1) contains a consensus sequence found in a variety of heparin-binding proteins  likely contributing to the heparin/heparin sulphate binding capacity of amphoterin. The region between the HMG box B and the acidic tail of amphoterin is the most distinctive feature of amphoterin when compared with the other members of the HMG protein family (residues 151–183; Fig. 1). We have recently discovered that this part of amphoterin contains a motif capable of binding to RAGE, one of the amphoterin receptors on the cell surface . Extracellular amphoterin is proteolytically processed to a short 10 amino acid peptide (residues 130–139; Fig. 1) active in erythroleukaemia cell differentiation . The highly acidic C-terminus of amphoterin plays an important role in nuclear functions [42, 43]. In addition, amphoterin contains a sequence motif (residues 12–27; Fig. 1) homologous to amyloid-β peptide, and this peptide is capable of forming amyloid fibrils and binding to amyloid-β peptide  stabilizing amyloid-β peptide oligomerization . The proinflammatory activity of amphoterin is localized to the HMG box B, especially to the residues 89–108 .
Amphoterin is an extremely well conserved protein across species. For example, human, rodent and bovine amphoterin sequences are >95% identical at the amino acid level. Human and rodent amphoterin sequences are identical at 214 of 216 amino acids (>99%). As the differences between human and rodent sequences are glutamate/aspartate changes in the acidic C-terminal tail of amphoterin, the human, mouse and rat proteins are virtually identical.
The human amphoterin gene has been mapped to chromosome 13q12 (chromosome 5 in mouse) and contains five exons and four introns [47–50]. At least seven retrotransposed pseudogenes have been found in the mouse genome . Amphoterin gene has a very strong TATA-less promoter with a maximum activity 18-fold higher than the SV40 promoter . The promoter contains several binding sites for general transcription factors such as AP1 and Sp1. However, the promoter region of amphoterin gene contains a silencer element that holds amphoterin expression at basal levels under normal circumstances .
Expression and secretion of amphoterin
Amphoterin expression can be detected at some level in almost all cell and tissue types and due to a strong promoter it can be expressed at very high levels . In addition, amphoterin expression is under strict developmental regulation. In rat brain the expression of amphoterin is high during the embryonic period and decreases after birth [1, 2, 15, 52]. The expression pattern of amphoterin in the developing rat nervous system appears to be rather widespread. Immunohistochemical studies have shown that amphoterin is expressed in the late embryonic period (E14-E21) in dividing and migrating neurones of the cortical plate and subplate, and in the ventricular zone of the cerebral cortex. During the perinatal period (P1-P15) amphoterin is expressed in neurones of external and internal granule cell layers of the cerebellum [14–17]. Expression of amphoterin in the hippocampus has also been reported . In E13 spinal cord, amphoterin is expressed in the dorsal root entry zone, the mantle layer of the ventral horn, the neuroepithelium surrounding the central canal, and in the dorsal root ganglia . In the adult retina amphoterin immunoreactivity has been detected in the inner nuclear layer and in the outer plexiform layer located between the inner and outer nuclear layers . In general, amphoterin levels in the adult central nervous system are low.
In addition to neurones, amphoterin is also abundant in oligodendrocytes and Schwann cells prior to myelinogenesis during late embryonic development [29, 55]. In the peripheral nervous system, amphoterin expression in neurones persists into adulthood whereas the expression in Schwann cells is downregulated upon neuronal contact . In the developing Xenopus nervous system abundant expression of HMG-X, an amphoterin homologue, has also been reported .
In general, the abundance of amphoterin seems to correlate with an undifferentiated cell stage and early maturation. However, in the vascular system amphoterin is expressed in several mature cell types. Platelets contain amphoterin that is exported to the cell surface during platelet activation . Amphoterin is highly expressed in mononuclear phagocytes and macrophages, and is secreted in response to proinflammatory cytokines [30, 58]. Amphoterin is also detectable in normal human serum (0.2 ng mL−1) but not in plasma . Increased amounts of amphoterin have been found in serum of patients suffering from systemic inflammation such as sepsis . Interestingly, amphoterin has been suggested to function in the interface of neuroendocrine and immune systems as it is secreted by pituicytes upon stimulation with proinflammatory cytokines TNF-α and IL-1β .
Several transformed cell lines display abundant expression of amphoterin [8, 22]. Furthermore, a large variety of tumour cells have been reported to express amphoterin at elevated levels [19, 23, 59–63].
Subcellular localization of amphoterin depends on the cell type and the state of the cell. As amphoterin is a nonhistone component of chromatin, it is regularly found in the nucleus. However, large amounts of amphoterin can be found diffusely distributed in the cytoplasm . In motile cells amphoterin becomes strongly enriched at the leading edge and extending processes of the cell [2, 8; Fig. 2]. Regulation of the subcellular localization of amphoterin is not well understood. However, studies in our laboratory have suggested that localization of amphoterin mRNA to the cell periphery might be a key step in bypassing nuclear transport of amphoterin. Localization of amphoterin mRNA shares similar features with β-actin mRNA localization, and these two mRNA particles are often found in cell processes together with ribosomes suggesting local mode of translation [18, 65]. Thus, mechanisms controlling localization of amphoterin mRNA might also be important regulators of secretion of amphoterin.
Although amphoterin lacks a classical secretion signal peptide, it is secreted independently of cell damage [1, 7, 12, 13, 30]. The classical secretory pathway involving endoplasmic reticulum (ER) and the Golgi apparatus is not involved in the secretion of amphoterin but intracellular increase of Ca2+ and protein kinase C may be required . Amphoterin is secreted in response to various stimuli. Secretion of amphoterin accompanies process extension and at least laminin directly induces secretion of amphoterin . In addition, various cytokines such as TNF-α, IL-1β and interferon-γ (IFN-γ) induce secretion of amphoterin in monocytes, macrophages and pituicytes [30, 37, 66]. Interestingly, in monocytes, amphoterin secretion occurs through a nonclassical, vesicle-mediated secretory pathway . However, as secretory endolysosomes are mainly found in haematopoietic cells, it is likely that other yet unknown mechanisms are responsible for amphoterin secretion in other cell types.
In addition to regulated secretion, amphoterin can also be released from necrotic or damaged cells. A recent report by Scaffidi et al. suggested a novel cytokine-like function for amphoterin as a local promoter of inflammation released by necrotic cells . Interestingly, it seems that in apoptotic cells nuclear amphoterin is bound firmly to the chromatin due to underacetylation of histone. Because of the retention of nuclear components within apoptotic cells until clearance by macrophages, amphoterin is never released from apoptotic cells. Thus, under physiological conditions the release of amphoterin might be the decisive signal needed to distinguish whether the cell has died due to exposure to an exogenous danger or whether it has been expelled by an endogenous mechanism.
Extracellular amphoterin-binding molecules
Amphoterin is a rather sticky protein likely due to its highly charged nature. A wide range of molecules interact with amphoterin. These include certain conformations of DNA, various carbohydrate epitopes and proteins. Amphoterin-binding molecules can be divided roughly into two categories: the molecules relevant to the nuclear functions of amphoterin and the molecules relevant to the extracellular functions of amphoterin. As the focus of this review is on the extracellular role of amphoterin, only extracellular amphoterin-binding molecules will be reviewed.
Although direct evidence is still lacking, it is likely that amphoterin clusters molecules on the cell surface in special signalling complexes (Fig. 3). Interaction of extracellular amphoterin with plasminogen activators and plasminogen in these complexes provides transient proteolytic interaction sites for the extending cytoplasmic processes that degrade the extracellular matrix surrounding the cell and thus facilitate penetration through the tissue (summarized in Fig. 3).
Amphoterin was originally isolated as a heparin-binding protein from the perinatal rat brain . Heparan sulphates, heparin-type glycans found in tissues, are found in various proteoglycans where glycosaminoglycan chains are attached to core proteins through serine residues. Proteoglycans are found on cell surfaces and as major constituents of the extracellular matrix. Various proteoglycans have been found to serve e.g. as modulators of neurite outgrowth (reviewed in [69, 70]).
Amphoterin has been reported to bind to syndecan-1 . Syndecans are transmembrane proteoglycans that are involved in the regulation of cell behaviour ranging from lipase activity and anticoagulation to growth factor signalling and cell adhesion (reviewed in [72, 73]). As amphoterin binds to syndecan-1 in a specific manner requiring heparan sulphate side chains of syndecan-1 , it is likely that amphoterin also binds to heparan sulphate structures of other syndecans (syndecan-2, -3 and -4). Syndecans can activate cell signalling pathways either as independent cell surface receptors or as coreceptors for integrins and growth factor receptors (reviewed in ). In addition to syndecans, amphoterin can bind to chondroitin sulphate side chains of phosphacan, a splice variant containing the ectodomain of the receptor-type protein tyrosine phosphatase β/ζ (RPTPβ/ζ), with a rather high affinity (Kd ≈ 0.3–0.8 nmol L−1) .
Sulphoglucuronyl carbohydrate (SGC) reacting with the monoclonal antibody HNK-1 is temporally and spatially regulated in the developing nervous system. SGCs are expressed on several neural recognition molecules of the immunoglobulin superfamily, such as neural cell adhesion molecule NCAM, L1, Tag-1 (Axonin-1) and contactin (F3/F11) (reviewed in ). SGC is also expressed on certain chondroitin sulphate proteoglycans and on some sulphoglucuronyl glycolipids. Amphoterin binds to glycolipids carrying SGC [17, 75, 76]. SGC has been implicated in adhesion of neurones and astrocytes on laminin and in the outgrowth of neuronal and astrocytic processes (reviewed in ).
Amphoterin binds to tissue-type plasminogen activator (t-PA), urokinase-type plasminogen activator and plasminogen, an inactive zymogen that produces lysine-specific serine protease plasmin upon activation [8, 9]. Interaction of amphoterin with t-PA and plasminogen enhances generation of active plasmin with similar capacity as soluble fibrin . As amphoterin is a lysine-rich protein (∼20% of amino acids) it is a sensitive target for plasmin itself. Thus, amphoterin enhances its own breakdown upon contact with t-PA/plasminogen system [8, 9]. Amphoterin and t-PA colocalize to distal tips of extending processes in motile cells creating transient proteolytic activity on the cell surface . Moreover, it seems that amphoterin-induced generation of active plasmin on the cell surface activates matrix metalloprotease (MMP) cascades further increasing local proteolytic and invasive potential of extending processes .
RAGE was originally isolated as a receptor for advanced glycation end products (AGE)  but was later found to recognize families of diverse ligands (reviewed in ). RAGE belongs to the immunoglobulin superfamily of cell surface molecules  and is known to interact with multiple signalling pathways (reviewed in [78, 80, 81]). As RAGE interacts with ligands involved in various diseases and the expression of RAGE is upregulated at sites of various pathologies, such as diabetes, atherosclerosis and Alzheimer's disease (reviewed in [81, 82]), RAGE provides a potential target for pharmacological intervention. However, RAGE expression has also been detected in developing nervous system where some of its ligands are also expressed suggesting nonpathophysiological roles for RAGE in homeostasis.
RAGE is an ∼50 kDa protein (404 amino acids in human sequence) composed of an extracellular region containing one V-type immunoglobulin (Ig) domain followed by two C-type Ig domains, a hydrophobic transmembrane-spanning domain and a highly charged, short cytoplasmic domain . The distal V-type domain that contains two potential N-glycosylation sites is thought to be responsible for ligand-binding properties of RAGE (Figs 4 and 5). The V-domain of RAGE is responsible for binding AGE  and other ligands of RAGE compete with each other for receptor binding [84–86] suggesting that all RAGE ligands bind to the same or overlapping binding site(s) in the V-domain of RAGE. Three-dimensional structural data has not yet been published for RAGE. Our molecular modelling studies using the crystal structure of vascular cell adhesion molecule-1 (VCAM-1) as a structural template suggest that RAGE V-domain has a typical Ig domain structure composed of six antiparallel β-sheets (Huttunen and Rauvala, unpublished data). One potential N-glycosylation site (Asn81) is located in close vicinity of the putative ligand binding surface of RAGE V-domain. N-glycans on the RAGE ectodomain influence amphoterin binding to RAGE .
The human RAGE gene has been mapped to human chromosome 6p21.3 (chromosome 17 in mouse) near the junction of the major histocompatibility complex class II and class III [50, 89, 90]. Several polymorphisms identified in the RAGE gene are associated with diabetic complications such as nephropathy, periodontitis and microvascular disease [91–96] and also with inflammatory diseases such as rheumatoid arthritis  and psoriasis . Particularly interesting is the G82S polymorphism which changes a glycine residue to a serine residue in the putative N-glycosylation consensus sequence (Asn81-Gly82-Ser83)[93, 94, 99]. As according to our homology modelling Asn81 is located in close vicinity of the putative ligand-binding site of RAGE and the G82S polymorphisms is expected to affect attachment of N-linked carbohydrates to Asn81, it is possible that the G82S polymorphism affects ligand binding properties of RAGE. This was also suggested by a recent report showing that cells bearing the 82S allele of RAGE display increased ligand binding followed by enhanced cytokine and MMP generation upregulating the inflammatory response when compared with the 82G allele .
The relatively short (<2 kb) promoter of RAGE gene contains binding sites for transcription factors NF-κB and Sp1 that play important roles in the regulation of RAGE expression [50, 90, 100, 101]. A typical feature of RAGE is that it is expressed at relatively low levels in homeostasis but in situations characterized by enhanced cellular activation or stress, such as diabetes , inflammation [103, 104], Alzheimer's disease  and normal development , the expression of RAGE is strikingly enhanced. The presence of NF-κB binding sites in the RAGE promoter creates a positive feedback loop in the regulation of RAGE expression as RAGE signalling itself activates NF-κB . This has been acknowledged as a crucial feature in RAGE biology as it enhances RAGE expression in environments rich in ligands creating an ascending spiral of RAGE-induced cellular perturbation in various pathological settings (reviewed in ).
Although the spatiotemporal expression pattern of RAGE within the central nervous system has not been characterized in detail, RAGE has been reported to be expressed in particular populations of cortical neurones, in the hippocampus, in spinal motor neurones of the anterior horn (especially Nissl bodies), in the ependyma and in the microvasculature of the brain. In addition, some peripheral nerves seem to express RAGE [53, 102].
RAGE interacts with a range of diverse ligands (amphoterin, AGE, amyloids and S100 proteins) suggesting that conformational determinants in the ligands are responsible for recognition of polypeptides with varied primary sequences. Amyloids form β-sheet fibrils, amphoterin and S100 family proteins are folded in α-helical conformations whereas certain common RAGE-binding epitopes have been characterized in AGE. Interestingly, each group of RAGE ligands seems to be involved in a specialized pathophysiological setting.
Amphoterin binds to RAGE in a dose-dependent and saturable manner with a Kd ≈ 5–10 nmol L−1. Amphoterin binding to RAGE can be blocked by AGEs suggesting that an overlapping binding site on RAGE is used by both ligands. sRAGE (soluble ectodomain of RAGE) and antibodies against RAGE as well as expression of cytoplasmic domain deletion mutant of RAGE efficiently inhibit neurite outgrowth on amphoterin-coated surfaces [53, Fig. 4]. RAGE colocalizes with amphoterin in the developing rat brain . Spatially colocalization was detected in the cerebral cortex, hippocampus and cerebellum, and temporally from the late embryonic period to the early neonatal period. Both amphoterin and RAGE are expressed in various transformed and tumour cells [8, 19, 22, 23]. Taguchi et al. reported that inhibition of amphoterin-RAGE interaction suppresses tumour growth and metastasis .
Ligation of RAGE on the cell surface activates multiple intracellular signalling pathways (Fig. 5). The cytoplasmic domain of RAGE is crucial for the activation of various signalling pathways induced by RAGE ligation ([22, 27, 86, 106, 107; Figs 3–5]). The 40 amino acid-long cytoplasmic domain of RAGE is highly acidic containing >30% glutamate residues sharing considerable homology with the 140-kDa splice variant of neural cell adhesion molecule (NCAM-140)  and CD20, a B lymphocyte specific cell surface molecule . Curiously, molecule(s) binding directly to the cytoplasmic domain of RAGE have remained unidentified until Ishihara et al showed that the extracellular signal-regulated kinases 1 and 2 (ERK1/2) bind to the membrane proximal region in the RAGE cytoplasmic domain .
Previous studies have identified several downstream signalling molecules responsive to RAGE ligation (summarized in Fig. 5). These include factors involved in the regulation of gene expression as well as reorganization of the actin cytoskeleton. Binding of AGEs to cell surface RAGE results in generation of reactive oxygen species and activation of the transcription factor NF-κB . RAGE-mediated activation of NF-κB depends upon upstream activation of a classical mitogen-activated protein kinase (MAPK) pathway involving the small GTPase Ras and ERK1/2 (; Fig. 5). Furthermore, two other MAP kinases, p38 MAP kinase and stress-activated protein kinase/c-Jun-NH2-terminal kinase (SAPK/JNK), are activated by RAGE . ERK1/2, JNK and p38 pathways modulate the activity of a wide array of transcription factors and thus are the major MAP kinases linking extracellular signals to the regulation gene expression (reviewed in ). The fact that RAGE is capable of activating all three major members of the MAP kinase network suggests that RAGE activation can regulate cell behaviour on a wide spectrum.
RAGE-dependent neurite outgrowth requires functional Rho-family small GTPases Cdc42 and Rac1 . Rho-family GTPases are molecular switches that control the organization and dynamics of the actin cytoskeleton (reviewed in [113, 114]) and are important regulators of neuronal growth cone guidance (reviewed in ).
In addition to RAGE, certain interleukin-1/Toll-like receptors bind amphoterin, and might be partly responsible for the cytokine like effects of amphoterin in haematopoietic cells (see other reviews in this issue).
Amphoterin-induced gene expression
As RAGE has been intensively studied in the context of diabetic and inflammatory complications, many of the genes that are regulated by RAGE are either proinflammatory cytokines or vascular adhesion molecules. Cytokines found to have increased expression in cells of monocyte/macrophage lineage in response to RAGE activation include IL-1β, IL-2 , IL-6 [116, 117] and TNF-α [86, 104, 117, 118]. Certain cytokines such as TNF-α, IL-1β and IFN-γ induce secretion of amphoterin in monocytes, macrophages and pituicytes [30, 37, 66] creating a positive regulatory feedback loop for amphoterin function in haematopoietic cells. Neurones display increased expression of macrophage colony-stimulating factor in response to amyloid-β peptide-induced RAGE activation . In addition, RAGE-induced expression of transforming growth factor-β (TGF-β) mediates epithelial-myofibroblast transdifferentiation in kidney epithelial cells . As NF-κB, a major transcription factor specifically involved in the regulation of genes involved in cellular defence mechanisms, pathogen defences, immunological responses and expression of cytokines and cell adhesion molecules (reviewed in ), is a target activated by RAGE signalling, it is probably responsible for increased expression of most cytokines upon RAGE activation. This is also the case with RAGE-mediated upregulation of VCAM-1 [119, 122], intercellular adhesion molecule-1 (ICAM-1)  and tissue factor [123, 124] – all important in the interaction of circulating blood cells with the vessel wall.
Less is known about genes regulated by RAGE signalling during neuronal development. Although some cytokines are neurotrophic (reviewed in ), it is unlikely that the trophic effects of RAGE on neuronal cells would be mediated solely through proinflammatory molecules. Ligation of RAGE by amphoterin or S100 proteins promotes survival of neuroblastoma cells through increased expression of Bcl-2, an antiapoptotic protein . Furthermore, we have recently reported identification of chromogranins and AMIGO (Amphoterin-Induced Gene and ORF) as genes upregulated by amphoterin-RAGE interaction [27, 126]. Chromogranins are structural components of mature secretory vesicles that are found in e.g. precursor vesicles in immature axonal growth cones  suggesting that they are important factors in the generation of secretory neuronal phenotypes. AMIGO defines a family of novel leucine-rich repeat proteins implicated in cell adhesion and axon tract development . Our recent results also identified CREB, a transcription factor and an immediate early gene, as a downstream target in RAGE signalling . CREB is an important regulator of neurotrophic factors, such as brain-derived neurotrophic factor, in neuronal development and plasticity (reviewed in ). As NF-κB is considered to have a critical role in the regulation of neuronal development, survival and plasticity (reviewed in ), it seems likely that these two transcription factors cooperate mediating RAGE-induced changes in neuronal gene expression.
Relationship between amphoterin and S100 proteins
S100 proteins compose a multigenic family of nonubiquitous Ca2+-binding proteins of the EF-hand type, differentially expressed in a wide variety of cell types (reviewed in [129, 130]). Various S100 proteins have been implicated in the regulation of a variety of intracellular activities such as protein phosphorylation, enzymatic activities, cell proliferation, differentiation, neoplastic transformation, dynamics of cytoskeleton constituents, the structural organization of membranes, intracellular Ca2+ homeostasis, inflammation, and in protection from oxidative cell damage. Similar to amphoterin, S100 proteins also lack a classical secretion signal but are released or secreted into the extracellular space to exert trophic or toxic effects depending on their concentration, act as chemoattractants for leucocytes, regulate macrophage activation or modulate cell proliferation. To date 19 members of the S100 protein family have been identified, some of which have both intracellular and extracellular functions. In general, S100 proteins exist as homodimers or heterodimers but they also form higher order oligomers [130, 131].
S100B is abundant in the nervous system where it is mainly expressed in astrocytes, oligodendrocytes and Schwann cells (reviewed in ). S100B is released by astrocytes into the extracellular space affecting astrocytes in an autocrine manner and neurones in a paracrine manner. The effects of extracellular S100B on target cells depend on its concentration. At nanomolar concentrations S100B has neurotrophic effects promoting neurite outgrowth [132, 133], survival of neurones during development  and neuronal regeneration after injury [135, 136]. In micromolar concentrations S100B can induce neuronal cell death either directly  or through the release of nitric oxide from astrocytes .
RAGE has been identified as a receptor for S100A12 (a.k.a. EN-RAGE; extracellular newly identified RAGE-binding protein) and S100B . In addition, S100A1 seems to be capable of binding and activating RAGE . As S100 proteins are structurally very homologous, it is likely that most, if not all, extracellular S100 proteins bind to RAGE. EN-RAGE is expressed in stimulated inflammatory cells and secreted EN-RAGE binds to RAGE on immune cells, promotes proliferation, activation of the transcription factor NF-κB, and expression of IL-1β, IL-2 and TNF-α in a RAGE- and cell-type dependent manner . Inhibition of EN-RAGE/RAGE-interaction suppresses delayed-type hypersensitivity and inflammatory colitis in murine models. Furthermore, involvement of EN-RAGE in the RAGE-mediated cellular perturbation in diabetes has been suggested .
Our results suggest that RAGE could be the receptor mediating both trophic and toxic effects of S100B . EN-RAGE is also a potent inducer of neurite outgrowth in hippocampal neurones . Although direct involvement of RAGE in the neuritogenic actions of EN-RAGE has not been shown, similar signal pathways are required for EN-RAGE-induced neurite outgrowth in a way similar to that activated by RAGE. In addition, S100A4 (a.k.a. metastasis-associated protein Mts1) has neuritogenic potential .
We have recently identified a C-terminal motif in amphoterin (residues 150–183) that binds to RAGE and is capable of promoting neurite outgrowth itself as a substrate-attached material and of inhibiting amphoterin-induced cell motility when applied in solution . Interestingly, this motif shares sequence homology with the N-termini of S100 proteins (Fig. 6). Furthermore, molecular modelling of the RAGE-binding C-terminal motif in amphoterin with the known structures of S100A12 and S100B [141, 142] suggests that amphoterin and S100 proteins may contain a structurally similar RAGE-binding motif (Fig. 6 and ). The fact that both HMG boxes of amphoterin can independently promote cell migration  suggests that in addition to RAGE other amphoterin-binding molecules might also be involved in the regulation of cell migration.
Amphoterin in tumour biology
Amphoterin is a rather general regulator of cell migration . Because amphoterin-induced extension of neuritis and other types of cellular processes was found to be coupled to proteolytic activation, amphoterin was suggested to mediate invasive migration in cells . However, RAGE is capable of regulating cell migration [104, 144]. Both amphoterin and RAGE are expressed in various transformed and tumour cells [8, 19–23]. These findings suggest that amphoterin and RAGE could be involved in the regulation of invasive migration of tumour cells. In primary tumour models, blocking RAGE function by sRAGE, anti-RAGE antibodies or expression of cytoplasmic domain deletion mutant of RAGE, suppresses tumour growth and formation of metastasis. Furthermore, in an endogenous tumour model, formation of spontaneously arising papillomas in mice expressing v-Ha-ras was markedly suppressed when sRAGE was administered to the mice . In these mouse models, the major effect of RAGE appeared to be on tumour cell migration and invasiveness, rather than apoptosis or proliferation. Interestingly, in tumours expressing wild-type RAGE, but not in those where RAGE function had been blocked, increased expression of MMP-2 and MMP-9 was detected . As the most versatile and important regulators of pericellular proteolysis, MMPs are intimately associated with tumour cell invasion (reviewed in [145, 146]).
Recently, Kuniyasu et al. reported a study of 96 cases of Japanese gastric cancer patients where they found that RAGE was expressed in 65% and amphoterin in 85% of these gastric tumours . Strikingly, the expression of RAGE had a strong correlation with the depth of tumour invasion and presence of lymph node metastasis. Histological analysis showed that RAGE expression was concentrated in the invasive front of the tumours. Although there is still limited clinical evidence available, these results strongly suggest that amphoterin–RAGE interaction in tumours is both mechanistically and clinically a relevant target that serves as a molecular checkpoint in the regulation of tumour cell invasiveness.
Conclusions and future prospects
Several studies using different cell types in culture and mouse models in vivo support the idea that amphoterin is a key regulator of invasive cell migration in the context of development, tumour cell invasion and inflammation. Secreted amphoterin is ideally positioned to regulate cell migration in an autocrine/paracrine manner as it promotes both local proteolytic activation on the cell surface and intracellular signalling pathways required for reorganization of actin cytoskeleton in motile cells. RAGE appears to be the major transmembrane receptor mediating amphoterin-dependent migration. In addition, sulphated glycans on the cell surface are likely to be involved in amphoterin-cell interactions and warrant further studies in the regulation of migration. Finally, better understanding of the molecular mechanisms underlying amphoterin/RAGE functions could yield novel therapeutic approaches to anti-tumour and anti-inflammatory strategies.