First printed in R&D Systems' 1996 Catalog.
Alternative messenger RNA splicing is a well known cellular mechanism for generating multiple, functional forms of a basic protein motif. Associated with split genes (those possessing introns and exons), alternative splice events are used by the cell in activities as diverse as enzyme production, cell surface antigen expression and growth factor secretion.1-3 Both normal and abnormal activities have been associated with the production of variant protein forms (or isoforms) from a common gene and these range from changes in cellular function to the induction of overt pathology.4,5 Recently, a genetic locus was identified that has the potential to be the most extensively-spliced growth factor gene known to date. This gene, localized to chromosome 8p11-p22 in the human, is now known to be a member of the epidermal growth factor family of molecules.6-8 Although the molecules that comprise the epidermal growth factor (EGF) family are classically described as the prototypical polypeptide mitogens, the known activities ascribed to this family clearly transcend those of simple mitosis.9 Among the members of the EGF family are EGF itself, plus betacellulin, amphiregulin, TGF alpha and heparin-binding EGF.7,8 Three characteristics common to all these family members include proteolytic processing of their transmembrane proforms7,8,10-12, an ability to bind to the 170 kDa EGF receptor (EGFR)7,8,12, and the presence of a conserved six-cysteine, two-glycine, 36-37 amino acid long motif (cysteines 1 - 6) that is collectively referred to as an EGF-like domain.7 This domain is composed of three loops, formed by disulfide bonds between Cys 1 & 3, 2 & 4, plus 5 & 6; a configuration that is suggested to account for each molecule's ability to bind to the EGFR.7,10,13,14 Subsequent to the isolation of the EGFR, studies on rat neuroblastomas revealed the presence of a previously unknown 185 kDa transforming glycoprotein that showed striking homology to the already known EGFR. When this molecule was later isolated in the human, and again found to be related to the human epidermal growth factor receptor, it was named HER2 (HER/HER1 is the original EGFR).7,15 In the rat, the same proto-oncogene (receptor) was alternatively named neu for its neuroblastoma origin.16 Following receptor identification, and based on an apparent ability to phosphorylate HER2, 44 kDa glycoprotein ligands from both rat and human were reported.17,18 In the rat model, neu activation resulted in breast cancer cell differentiation and maturation rather than proliferation, and the ligand was termed NDF for neu differentiation factor. In the human, the HER2 ligand stimulated proliferation of a different breast cancer cell line, and the factor was named heregulin (regulator of HER2). Notably, HER2-ligands did not activate HER118, and HER1 (EGFR) ligands did not activate HER2.7,8,19 The ability of heregulin (HRG) to phosphorylate HER2 turned out to be an anomaly; the initially observed HER2 phosphorylation pattern was the result of HER2 receptor heterodimerization with the now known HER3 and HER4 receptor molecules.20-22 Thus, HRG is no longer believed to be the ligand for HER2 but for the recently discovered 160 kDa HER3 and 180 kDa HER4 receptors (see below). The ligand for HER2 remains in question.21-24 Finally, two other independent lines of investigation also identified HRG, albeit under different names. In the chicken, a 42 kDa factor was isolated that induced the synthesis of acetylcholine receptors in embryonic skeletal muscle, and this was initially called ARIA for acetylcholine receptor inducing activity.19,25 In addition, and as a consequence of studies investigating Schwann cell mitosis using bovine pituitary mitogens, a 59 kDa human Schwann cell (or glial) mitogen was isolated which was named glial growth factor II (or GGF-II).26 Given the neural associations of the HRG molecule (acetylcholine receptors, glial growth, neuroblastoma-associated receptor, and its identification in neurons), neuregulin (NRG) has been proposed as an alternative name for the heregulin family members.26
There is considerable species cross-reactivity within the HRG family of molecules.28,29 There are at least 12 different HRG isoforms currently known among bovine, human, rat, and chicken, with the potential for additional members.27,30 Based on the pattern of alternative splice events, the HRG gene is believed to consist of at least 13 exons that give rise to a series of highly variable domains.30 Although the "standard" HRG molecule is described as consisting of six domains, it is perhaps useful to consider heregulin a seven-domain, proform transmembrane glycoprotein.27 In order, from the N-terminus, HRG contains 1) a 50 amino acid (aa) residue N-terminal region, 2) an approximately 60-100 aa residue immunoglobulin-like (Ig-like) domain, 3) an approximately 35-40 aa residue "spacer" segment (for carbohydrate attachment), 4) a 60 aa residue (six cysteine) EGF-like domain, 5) a highly variable 1 to 35 aa residue juxtamembrane segment (sometimes grouped with the EGF-domain), 6) a 23 aa residue transmembrane segment, and 7) a variable-length cytoplasmic tail.19,27,31 Given the presence of both an EGF-like domain and an Ig-like segment, HRG has been classified as both an EGF family member and an Ig superfamily member29 Relative to HRG variability, isoforms are identified by differences in the EGF-like domain, the juxtamembrane segment, the cytoplasmic tail, and the N-terminal configuration. Within the EGF-like region, the third loop between cysteines #5 and #6, plus the following 9 to 12 aa residues define the major subtypes for HRG. When this entire stretch contains 21 aa residues (excluding cysteine #5), the HRG molecule is considered an alpha form (HRG alpha); when there are only 18 aa residues in this region, the molecule is designated a beta form (HRG beta).18,19 The above difference between the alpha and beta forms is reinforced by their low degree of amino acid identity over this short stretch (33%), the association of alpha forms with connective tissue cells and beta forms with nervous tissue cells, and a decided difference in their mitogenic activities. HRG beta is much more effective in inducing mitosis in HER3- and HER4-transfected cells.19,32 As with members of the EGF-family, the overall EGF-like domain is believed to confer receptor-binding ability, and the three additional aa residues between cysteines 3 and 4 in HRG relative to EGF account for the inability of HRG to bind to the EGFR.19 A number of variations also exist within the juxtamembrane segment. Within this region, and not including eight invariant C-terminal aa residues, the number of aa residues varies from 1 to 27. Although there are as many as 27 residues, only five variations are currently known, and each has a numeral to designate its length. A "1" indicates nine amino acids in the variable juxtamembrane segment, a "2" represents only one aa residue, a "3" indicates eleven aa residues, a "4" denotes 27 aa residues, and a "5" represents fourteen amino acids.27 Notably, all type "3" forms terminate after the eleventh aa residue, with the remaining eight constant juxtamembrane aa residues plus the TM and cytoplasmic region being absent. This variant is either secreted (as in GGF-II) or retained internally.18,19,27,30 The third major variable region is the cytoplasmic domain. All transmembrane HRGs contain at least 157 aa residues in their cytoplasmic tails. When this is the extent of their intracellular domain, the molecules are considered "c" forms. If there are an 37 additional aa residues (for a total of 194), then the molecule is a "b" form. If there are an additional 217 aa residues (for a total of 157 + 217 = 374), the molecule is considered an "a" form HRG. Although all three "a", "b" and "c" forms occur with alpha type EGF-like domains, only form "a" variants exist with beta type EGF-like domains.19 Given the above, heregulins are now often described in terms of variations in their EGF-like, juxtamembrane, and cytoplasmic domains. For example, HRG alpha 2a indicates an alpha EGF-like domain, one amino acid in the variable juxtamembrane region, and a 217 aa residue extension to the standard 157 aa residue cytoplasmic tail.
The final variable region lies in the N-terminus. Three possibilities exist; 1) a combined signal sequence/kringle-type domain found in GGF-II, 2) a signal sequence-less stretch of approximately 50 aa residues found in almost all HRGs, and 3) a unique, approximately 30 aa residue stretch found only in ARIA.25,27 Following transmembrane expression, most HRGs undergo proteolytic cleavage at both the N-terminus and C-terminus (juxtamembrane segment), resulting in the release of mature HRG.18,31,33 Given the diversity of names, some common equivalencies include HRG beta 1/NRG beta 1/NDF beta 1a, HRG beta 2/NRG beta 2/NDF beta 2a, HRG alpha /NRG alpha 2/NDF alpha 2a, HRG beta 3/NRG beta 3/NDF beta 3, and NDF/NRG alpha 2/NDF alpha 2c.12,19,27 GGF-II is sometimes referred to as HRG beta K, with K standing for the kringle structure in the N-terminus.19 Cells known to express HRG include neurons and astrocytes34, fibroblasts35, and keratinocytes.36
Although the HER2 receptor was originally used to identify heregulin family members, it is now known that only the HER3 and HER4 receptors truly bind heregulin.37 To date, the ligand(s) for HER2 is unknown. It is possible, however, that the two newly discovered EGF-family members (epiregulin and/or SMDF [sensory and motor neuron-derived factor]) may turn out to be HER2 activators.38,39 This is particularly true for SMDF which is an unusual beta 3 heregulin variant.39
The HER3 receptor is a 160 kDa, 1323 aa residue transmembrane glycoprotein that has been identified in keratinocytes, melanocytes, skeletal muscle cells (on the motor end plate), embryonic myoblasts and Schwann cells.40-42. As with other EGF family receptors, HER3 contains an extracellular region (624 aa residues) that is organized into four domains; an N-terminal 161 aa residue domain (I), followed by a 174 aa residue cysteine-rich region (II), a 108 aa residue ligand-binding domain (III), and a second 166 aa residue cysteine-rich domain.40,43. Relative to the other HER receptors, HER3 is exceptional in that it contains the longest cytoplasmic tail (678 aa residues vs. 633 aa residues in HER4 and 580 aa residues in HER2).24,44 Additionally, it lacks the "standard" intrinsic tyrosine kinase domain.40,45 This has led to speculation that HER3 requires an additional co-receptor for its functioning. It is known that HER3 alone binds HRG beta with a Kd declines to 0.1 - 0.02 nM, depending upon the ratio of HER3 to HER2.45 Thus, it appears that HER2 (and perhaps other HERs) can provide some sort of cooperative influence on heregulin binding. This suggestion is further supported by the recognition that HRG does not activate HER3 unless another HER is present.37,45,47 The best-studied model involves HER3/HER2, where HER3 phosphorylation is accompanied by HER2 phosphorylation, suggesting that HER2 is activated by HER3-HRG binding followed by HER2 cross-phosphorylation of the bound HER3 molecule.21,37,48 In general, and consistent with the above, multiple heterodimeric combinations of HER do exist, and it is believed that diverse signal transduction pathways may be activated, depending upon the various heterodimer combinations.37,47 Finally, and as with the other EGF receptors, HER(3) has an alternative designation termed erbB(3) which derives from the B gene product of the avian erythroblastosis virus. Since erbB corresponds to the EGFR, erbB3 thus corresponds to HER3, etc.
The HER4 receptor is a 180 kDa, 1284 aa residue transmembrane glycoprotein which is expressed by neurons49, Schwann cell precursors50 and Schwann cells42, and skeletal muscle24. As with HER3, HER4 has a characteristic EGFR extracellular organization with four domains, the second through fourth (II-IV) of which share 55-65% amino acid sequence identity with HER324. Unlike HER3, HER4 has a classic intrinsic tyrosine kinase motif, and undergoes phosphorylation following HRG binding22,24 The Kd for direct HRG binding to HER4 is reported to be 5 nM.51 As is the case with HE3, HER4 is also proposed to form multiple dimers with other HERs, and HER4 phosphorylation is suggested to induce HER2 phosphorylation.37,46,52
The number of functions attributed to HRG are many, and are reflected in its diversity of names. For instance, her/neuregulin (ARIA) has now been demonstrated to modulate the number of acetylcholine receptors at the neuromuscular junction. Presumably the source of HRG is motor neuronal cells, with axonal transport being responsible for its concentration in the region of the motor nerve terminal.53,54 Notably, muscle cells which respond to HRG express both HER2 and HER3, consistent with the receptor crossactivation model proposed above.53 In addition, the heregulin GGF-II has been shown to be a potent mitogen for cultured glia (Schwann) cells, a cell type which, again, expresses HER2 and HER3 (plus HER4).26,50 Effects on Schwann cells are not limited to proliferation, however. Schwann cell precursors will also undergo maturation when exposed to NDF.50 This effect is also seen in astrocytes where HRG has limited mitogenic activity, but pronounced maturation-stimulating activity.34 Outside the nervous system, HRG has documented activity on keratinocytes, with HRG beta forms demonstrating to have strong mitogenic effects, and HRG alpha forms acting as maintenance factors.36 In vivo studies with keratinocytes also reveal significant HRG effects. During wound repair, HRG alpha 2, but not alpha 1, beta 1 or beta 2, materially increases keratinocyte migration and epidermal thickening. This thickening is not attributable to proliferation, however, and suggests alpha forms influence keratinocytes through their recruitment and differentiation.35 Perhaps most importantly, only alpha forms seem to be synthesized by fibroblasts at the site of wounding, reinforcing the suggestion that alpha forms are associated with mesenchymal tissue (connective tissue), while beta forms are generated in the nervous system.
The EGF receptor family of tyrosine kinase-linked receptors, including EGFR, erbB2/neu, erbB3, and erbB4, are frequently overexpressed in breast cancer and may mediate tumor cell proliferation.7,14 This overexpression has been suggested to be correlated with poor patient prognosis. HRGs as ligands for some of these receptors apparently also play important roles in the regulation of cancer cell proliferation and in neuronal function.7,14