Editorial: The Blind Men and the Elephant—A Parable for the Study of Insulin-Like Growth Factor Binding Proteins

We have all heard the parable of the blind men and the elephant: three blind men are led to an elephant and asked to determine, by touch alone, what the animal must be like. The first blind man grasps the elephant’s trunk and announces that the creature must resemble a giant snake; the second man feels the tough hide of the elephant and describes the animal as a giant tortoise; the third blind man feels the elephant’s tusks and determines that it must resemble an enormous walrus. The insulin-like growth factors (IGFs) may well be endocrinology’s elephant (and, by inference, we are its blind men). First identified in 1957 by Salmon and Daughaday (1) and named sulfation factor because of their ability to stimulate the incorporation of radiolabeled sulfate into proteoglycans, these proteins were believed to mediate much, if not all, of GH’s anabolic actions. Other laboratories, however, identified seemingly related peptides that simulated insulin’s metabolic actions but could not be inhibited by antiinsulin antibodies, hence the name nonsuppressible insulin-like activity (NSILA) (2). A third set of investigators, studying the mitogenic activity of serum-free conditioned medium from BRL-3A rat hepatoma cells, identified multiplication-stimulating activity (3). In time, all three sets of investigators recognized the existence of one or more common peptides, which was renamed somatomedin in 1972 (4), and, with the elucidation of the amino acid structures of two peptides, was given the nomenclature IGF-I and -II (5, 6). A similar set of events characterized the identification of the IGF binding proteins (IGFBPs) (7). Early studies had indicated that when plasma was fractionated under neutral conditions, both the NSILA and sulfation activity comigrated in relatively high molecular weight fractions (8). Eventually, it became apparent that there existed a large, GH-dependent binding protein and a smaller, GH-independent binding protein (9). The characterization of the latter protein, eventually identified as IGFBP-1, resulted from studies of at least four seemingly distinct proteins: 1) an IGFBP found in amniotic fluid (10); 2) an IGFBP identified in the conditioned medium of a human hepatoma cell line (HEP-G2) (11); 3) a placental protein originally isolated from soluble extracts of term human placenta and adjacent membranes (PP12) (12); and 4) a major secretory protein of the stromal cells of the decidua (pregnancy-associated endometrialα 1-globulin; endometrial protein 14) (13). It was not until the late 1980s that it became apparent that all of these proteins corresponded to what was eventually termed IGFBP-1 (14). Over the next few years, five additional IGFBPs were identified, bringing the seemingly complete complement of IGFBPs up to a total of six (15–17). In 1993, Murphy et al. (18), employing subtraction hybridization to characterize changes in gene expression between normal leptomeningeal cells and meningiomas, identified three genes whose expression was altered in meningioma cell lines and tumors. The complementary DNA (cDNA) clone designated mac25 (meningioma-associated cDNA-25) was found to hybridize to a single 1.1-kb messenger RNA isolated from cultured normal leptomeningeal cells; decreased levels of this 1.1-kb transcript were found in several meningioma cell lines and in meningioma solid tumors. It was speculated that the predicted 281 amino acid preprotein had important growth-regulatory function. A search of the GenBank database indicated that the identified nucleotide sequence was seemingly novel but was homologous to several members of the IGFBP family, especially in the amino-terminus, where 11 of the 12 cysteines found in IGFBPs 1–6 are conserved. These observations were extended in 1995 by Swisshelm et al. (19), who used differential display to identify high expression of mac25 in senescent mammary epithelial cells and virtual absence in estrogen receptor (ER)-positive mammary carcinoma cell lines. Up-regulation of mac25 expression by retinoic acid in normal mammary epithelial cells further supported a putative role for mac25 as a tumor-suppressor or sensenscence factor. Sequencing of the cDNA seemingly supported the GenBank sequence identified previously by Murphy et al. (18). The potential growth-regulatory actions of mac25, its up-regulation by retinoic acid in mammary epithelial cells, and its potential homology to IGFBPs 1–6, stimulated Oh and colleagues (20) to clone the cDNA for mac25 from Hs578T mammary carcinoma cells (ER-) and synthesize the mature protein in a baculovirus expression system. Molecular manipulation using PCR resulted in the addition of the eight amino FLAG epitope tag at the 3′ end of the mac25 cDNA, to facilitate protein purification. The resulting protein was found to begin with the residues SSSDT; as predicted, 11 of the 12 amino-terminal cysteines were conserved. Affinity cross-linking studies and Western ligand blotting performed under nondenaturing conditions demonstrated that the protein was capable of binding both IGF-I and IGF-II, although with lower affinity than is the case for IGFBPs 1–6, and based on its structural and functional properties, the name IGFBP-7 was proposed. Subsequent immunoblotting studies with a polyclonal antibody generated against the mac25/IGFBP-7 baculovirus protein demonstrated that IGFBP-7 is produced by Hs578T cells and can be found in a variety of biological fluids (21). In parallel with the investigations of Murphy et al. (18), Swisshelm et al. (19), and Oh et al. (20), two groups of researchers in Japan were studying seemingly different proteins. In 1994, Akaogi and colleagues (22) identified a 30-kDa secreted protein from the human bladder carcinoma cell line EJ-1. This“ tumor-derived adhesion factor” (TAF) was found to promote the attachment and spreading of cells from the rat liver line, BRL, and the human umbilical vein endothelial cell line ECV-304. At the same time, Yamauchi et al. (23) reported the purification and cloning of “prostacyclin-stimulating factor” (PSF) from serum-free conditioned medium of human diploid fibroblast cells. These studies were based on prior observations that fibroblast conditioned medium was capable of stimulating prostacyclin (PGI2) and that this stimulatory activity was similar to that of plasma-derived serum, namely, heat-stable, acid-labile, trypsin-sensitive, and possessing an affinity for heparin. Interestingly (and ironically), initial purification steps resulted in copurification with the high-affinity IGFBPs, and an IGF-I-affinity column was used to separate PSF from the traditional IGFBPs. A cDNA coding for PSF was cloned and sequenced, revealing an apparently novel preprotein of 282 amino acids. A structural similarity with IGFBP-4, which is also found in fibroblast conditioned medium, was noted, and the success of purification was attributed to the use of the IGF-I-affinity column. The question arises, naturally, as to the similarity, if not identity, of one or more proteins given four different names over a period of less than 5 yr: mac25, TAF, PSF, and IGFBP-7. Reevaluation of the nucloetide sequence for mac25 entered into GenBank indicates a striking similarity with that of PSF (23). Four nucleotides differ in the signal peptide region, resulting in three amino acid substitutions; one nucleotide differs in the amino-portion of the molecule, resulting in an R (mac25) for K (PSF) substitution. Finally, in the carboxy-terminus, a nucleotide was seemingly erroneously inserted into the sequence of mac25, resulting in a frame shift: the 5 carboxy-terminal residues of mac25 (ASEKR) are replaced by 10 amino acids in PSF (PVKKGEGAEL). Recent genomic cloning of mac25 confirms that the nucleotide sequence is, indeed, identical to that of PSF (Hwa, and R. G. Rosenfeld, unpublished data). One should ask, what is the biological significance of this seemingly serendipitous identification of the same gene/protein by four groups, each assigning it a totally distinctive name. For the IGFBPs, these observations are of considerable potential importance, for several reasons: 1) they suggest that the IGFBPs are not limited to six in number and may constitute a superfamily of structurally and physiologically related genes and proteins (24); 2) they indicate that this IGFBP superfamily may include both high-affinity members (i.e. IGFBPs 1–6) and low-affinity members (IGFBP-7 and, perhaps, others); 3) the fact that mac25/IGFBP-7/TAF/PSF clearly has a range of biological activities in both normal and transformed cells that may be independent of the ability of this protein to bind IGF provides circumstantial evidence for the hypothesis that the conventional IGFBPs may also have important IGF-independent actions, in addition to their IGF-dependent effects. Evidence for IGF-independent actions is strongest for IGFBP-3, which has been found to potently inhibit the replication of certain human breast cancer cell lines and to promote apoptosis in prostate cancer cells (25–27). The observation that IGFBP-3 binds to some cancer cells with high specificity and affinity (28), and the recent finding that IGFBP-3 is a functional ligand for the the type V transforming growth factor-β receptor (29), provide further evidence for this hypothesis. IGFBP-1, additionally, has been shown to stimulate cell migration in a monolayer wounding assay, presumably by binding to the α5β1-integrin receptors on cell surfaces (30). The data currently available, buttressed by the observations that mac25/TAF/PSF is, potentially, a seventh IGFBP, strongly suggest that all of the IGFBPs are derived from a common ancestor gene/protein and, over the course of evolution, have acquired the ability to regulate cell growth by both IGF-dependent and IGF-independent actions. We are, indeed, blind men trying to identify the elephant. Perhaps, this kind of scientific convergence is a necessary and, ultimately, healthy part of research. It also, however, suggests that blindness is an appropriate metaphor. We tend to wear blinders that, while they may help focus our vision for research, also prevent us from appreciating the complexity, intricacy, and broader implications of the biological systems that we investigate. Addendum: At the recent 4th International IGF Symposium, held in Tokyo, it was proposed that mac25/IGFBP-7 and other low-affinity IGF binders be categorized as IGFBP-related proteins (IGFBP-RPs), until further characterization of their physiological role in the IGF system. Accordingly, mac25/IGFBP-7 will be termed provisionally“ IGFBP-RP-1.”