Role of insulin-like growth factor binding proteins in controlling IGF actions

https://doi.org/10.1016/S0303-7207(98)00024-0Get rights and content

Abstract

The insulin-like growth factors (IGF) stimulate growth in multiple connective tissue cell types. The capacity of IGF-I and -II to access cell surface receptors is controlled by insulin-like growth factor binding proteins (IGFBPs). Connective tissue cells synthesize four of the IGFBPs (IGFBP-2 through -5). Synthesis is controlled by growth hormone and several other growth factors. In addition to regulating synthesis, other variables regulate the abundance of the IGFBPs including specific serine proteases that are produced for each form of IGFBP. Following cleavage, the IGFBPs have reduced affinity for IGF-I and -II, thus allowing release to receptors. Variables that regulate the amount of proteolysis have been shown to regulate IGF action. In addition to being proteolytically cleaved, three forms of IGFBPs (IGFBP-2, -3 and -5) can associate with extracellular matrix (ECM). In the case of IGFBP-5 binding to ECM, its affinity is lowered substantially allowing IGF to better equilibrate with the receptors. This event results in a potentiation of IGF-I action on fibroblasts and smooth muscle cells (SMC). In summary, IGFBPs are important molecules for regulating the bioavailability of IGF-I and -II to receptors. Understanding the variables that regulate their abundance may lead to a better understanding of the factors that regulate IGF action in skeletal tissues.

Introduction

Insulin-like growth factor-I (IGF-I) and -II circulate in relatively high concentrations in plasma, but are also synthesized by multiple cell types and released into interstitial fluids. Since both locally secreted IGF-I and -II can promote cell growth presumably by autocrine or paracrine mechanisms and concentrations of these peptides that are present in blood can function as classic endocrine hormones, e.g. being transported out of the circulation to distant sites of action, understanding the variables that regulate the tissue distribution of IGFs and their accessibility to receptors has been an important topic of study. It is generally assumed that systemic IGFs are required for normal systemic growth and development, but that locally synthesized IGFs may be important both in embryogenesis and for regulation of growth and anabolism in disorders such as osteoporosis, osteoarthritis, atherosclerosis, diabetic nephropathy and retinopathy (Jones and Clemmons, 1995, LeRoith et al., 1995). In these disorders, there appears to be localized growth dysregulation rather than systemic overgrowth since the pathology is confined to a small area. The IGFs have been implicated in such disorders because many of them involve the response to injury (Jennische et al., 1987a). Following injury, there is generally an increase in IGF synthesis in connective tissue cell types (Jennische and Hannson, 1987). Thus, following nerve transection, muscle crush injury, cartilage freezing or other types of local trauma, such as balloon catheter denudation, there is a large wave of increased synthesis of IGF-I in response to the injury stimulus that correlates with maximum proliferative response (Hansson et al., 1986, Hansson et al., 1987, Jennische et al., 1987b, Cercek et al., 1990). However, other than synthesis of IGF-I following injury, the variables that regulate how much IGF-I is available to bind to receptors and thus, stimulate IGF-mediated signal transduction pathways and growth are not well defined.

A major variable determining the amount of IGF-I that is available to receptors is the IGF binding proteins. Six forms of high affinity IGF binding proteins (IGFBPs) have been isolated and their complete sequences determined (Rechler, 1993). In general, human connective tissue cells that are involved in the response to injury, synthesize IGFBP-3, -4, and -5 (Camacho-Hubner et al., 1992). In some cases, they may synthesize IGFBP-2 or -6, but generally IGFBP-3, -4, and -5 are the most prevalent. All three of these forms of IGFBPs bind to IGF-I and -II with higher affinity than the Type I receptor. Therefore, under equilibrium conditions, if there is unsaturated binding capacity, the IGF in the pericellular environment will preferentially bind to the unsaturated binding protein. Binding protein-associated IGF-I is in equilibrium with the receptors, but the extent to which IGF-I can access the receptors is influenced by the amount of high affinity carrier and the amount of IGF-I since under normal conditions, receptor number varies minimally. Our laboratory has focused on the variables that regulate the synthesis and secretion of IGFBPs in two connective tissue cell types—human dermal fibroblasts and porcine aortic smooth muscle cells. However, much of the data that we have published regarding fibroblasts is also applicable to chondrocytes and osteoblasts based on the work of other investigators (Mohan et al., 1989, Ernst and Rodan, 1990, Conover and Kiefer, 1993a).

Studies using either Northern blotting, or metabolic labeling and immunoprecipitation have shown that smooth muscle cells and fibroblasts express mRNA for IGFBP-3, -4, and -5 and synthesize these proteins (Camacho-Hubner et al., 1992, Duan et al., 1996). In addition, smooth muscle cells synthesize IGFBP-2, but they do not synthesize IGFBP-3 (Cohick et al., 1995). In contrast, IGFBP-3 is the most abundant form of IGFBP that is synthesized by normal human skin fibroblasts. In general, peptide growth factors are poor stimulants of IGFBP synthesis by this cell type. TGF-β causes some modest increase in IGFBP-3 synthesis, but other growth factors are relatively weak (Duan et al., 1996). Likewise, IGF-I causes only relatively modest (60%) increase in IGFBP-3 mRNA abundance. The major variables that appear to control the abundance of IGFBP-3 are culture density with decreasing concentrations per cell being synthesized as density increases and cellular aging. Senescent cells appear to secrete large quantities of IGFBP-3 relative to non-senescent fibroblasts (Goldstein et al., 1991). In epithelial tissue retinoic acid appears to be a potent inducer of IGFBP-3 (Adamo et al., 1992), however, its effects in connective tissue cells are less impressive. Following its synthesis and secretion, IGFBP-3 is stable in fibroblast culture medium, that is, although IGFBP-3 proteases have been found to exist in multiple extracellular fluids, cultured fibroblasts do not appear to release proteolytic activity. Therefore, synthesis appears to be the predominant variable regulating the amount of IGFBP-3 in the microenvironment. Chondrocytes (DiBattista et al., 1996) and osteoblasts (Lalou et al., 1994) also synthesize IGFBP-3 and in general, the regulatory variables that have been identified for fibroblasts are applicable to these two cell types.

IGFBP-5 synthesis is also regulated in connective tissue cell types. However, these cells show some differences in regulation. Specifically, in fibroblasts, IGFBP-5 synthesis is not stimulated by IGF-I, but IGF-I appears to have some capacity to inhibit proteolytic cleavage (Camacho-Hubner et al., 1992). In contrast, in smooth muscle cells (Duan et al., 1996), chondrocytes, and osteoblasts (Conover et al., 1993b) IGF-I appears to be a potent stimulant of IGFBP-5 synthesis (Fig. 1). This increase in synthesis appears to be mediated at a transcriptional level, since IGF-I causes no change in mRNA stability, but directly increases gene transcription. Other growth factors such as PDGF, TGF-β, heparin binding epidermal growth factor (EGF) and FGF have minimal effects on IGFBP-5 synthesis by these cell types. As for IGFBP-3, retinoic acid appears to be a potent inducer of IGFBP-5 synthesis, at least by these cell types in culture.

IGFBP-4 synthesis is also regulated in these cell types. This has been most extensively studied in osteoblasts where it has been shown that prostaglandin E2 increases IGFBP-4 synthesis as does parathyroid hormone. In smooth muscle cells, insulin has been shown to be a potent stimulant of IGFBP-4 but EGF and FGF cause minimal changes (Cohick et al., 1995). EGF stimulated a large increase in IGFBP-4 synthesis if cells were serum starved for 72 h prior to initiation of the experiment. TGF-β caused a 100% increase in IGFBP-4 mRNA abundance and increased peptide concentrations by ≈130% after 24 h of serum starvation. Therefore, serum deprivation greatly altered the response to growth factors such as PDGF and TGF-β.

Clearly, a major variable that regulates the abundance of IGFBP-2, -3, -4, and -5 in the connective tissue cell pericellular environment is the presence of serine proteases. Although these proteins are all degraded by matrix metalloproteases (Fowlkes et al., 1994), this appears to be a relatively non-specific effect in that MMPs-2 and -9 have been shown to degrade each form of binding protein to some extent. Further evidence against these being predominant proteases for IGFBP-4 or -5 derives from the fact that when directly compared with serine proteases that are present in smooth muscle cell and fibroblast conditioned medium they account for much less of the total proteolytic activity. Finally, purification of the IGFBP-5 proteolytic activity from conditioned medium resulted in proteases that had inhibitor profiles that were typical for members of the chymotrypsin family of serine proteases and not typical of MMPs (Nam et al., 1996). These data suggest that fibroblasts, smooth muscle cells, osteoblasts and chondrocytes appear to secrete serine proteases that are calcium dependent that specifically cleave IGFBP-4 and -5. In contrast, IGFBP-3 appears to be minimally cleaved by conditioned medium from chondrocytes, fibroblasts or osteoblasts.

Specific proteases have been identified in smooth muscle cell and fibroblast media for IGFBP-2, -4 and -5. These proteases are so active, that in the case of smooth muscle cells, only IGFBP-2 is present in significant amounts in an intact form and IGFBP-4 and -5 appear to be dominantly fragments. For fibroblasts, the IGFBP-4 and -5 proteases also appear to be very active and there is intact IGFBP-4 or -5 present in the medium. Exposure of fibroblasts to IGF-I results in an increase in the amount of intact IGFBP-5 in the medium. IGF-I does not directly inhibit the activity of the protease, but it may induce a protease inhibitor. It has no effect on IGFBP-5 synthesis (Camacho-Hubner et al., 1992). In smooth muscle cells, IGF-I increases IGFBP-5 synthesis 5–8-fold and this is adequate to exceed the capacity of the protease and intact protein is detectable (Duan et al., 1996) (Fig. 2). These serine proteases that have been isolated from both fibroblast and smooth muscle cell conditioned medium and appear to be identical. The IGFBP-5 protease is a 95 kDa calcium dependent serine protease (Fig. 3). Its activity is inhibited by α1 antichymotrypsin and EDTA. It is activated by calcium, but not by zinc (Nam et al., 1994).

The protease that degrades IGFBP-4 is distinct. It is 52 kDa as determined by zymography and its activity is markedly accentuated by binding of IGF-I or -II to IGFBP-4 (Conover et al., 1993c, Parker et al., 1995). That is, in the absence of cells and in the presence of the protease in vitro, addition of IGF-I or -II to the incubation medium results in rapid increase in cleavage of the IGFBP-4 substrate (Fig. 4). In contrast, if des-IGF (an IGF isoform that does not bind to binding protein) is included in the incubation medium, there is no acceleration of proteolysis, thus suggesting that IGF-I binding to a binding protein is required for this effect to be detected. That IGF-I itself is not binding to protease was excluded by passing partially purified proteolytic activity over an IGF affinity column and showing that none bound. The actual determinants of protease synthesis have not been identified because pure antibodies or cDNA probes for these proteases do not exist.

The physiologic consequences of proteolysis have been probed to some extent. Initially, studies were descriptive. These studies compared the effects of one form of binding protein on IGF actions using cells that did or did not secrete a given form of protease. By this method of deduction, certain inferences were made indirectly regarding the function of the protease. In general, it has been assumed that cleavage of IGFBPs results in liberation of IGF-I and enhancement of cellular responsiveness. This was clearly shown in pig smooth muscle cells as compared with fibroblasts that secreted a protease for IGFBP-4. The muscle cells were minimally inhibited by concentrations of IGFBP-4 that clearly inhibited IGF-I action on the fibroblasts. An analysis of the conditioned medium showed the appearance of fragments of the muscle medium, but not the fibroblast medium at these concentrations (Cohick et al., 1993). Similar findings were determined for SV-40 transformed fibroblasts compared with normal fibroblasts for IGFBP-4 (Conover et al., 1993c). More recently, direct studies have confirmed these findings. Specifically, the protease cleavage site in IGFBP-4 was identified by two separate groups (Chernausek et al., 1995, Conover et al., 1995). Each group used mutagenesis to create protease-resistant forms of IGFBP-4 and then determined that these were potent inhibitors of IGF-I action on the respective cell types, whereas forms that were readily cleaved were much less potent. More recently, this finding has been extended to IGFBP-5. The cleavage site was identified as occurring between Lys138,139 within the protein. When these residues were mutated to asparagines, the protein was shown to be resistant to proteolysis. Addition of concentrations of ≈4:1 molar excess of IGFBP-5 over IGF-I resulted in complete attenuation of IGF-I actions in this cell type (Fig. 5). Protein synthesis, DNA synthesis and induction of IGFBP-5 by IGF-I were all inhibited (Imai et al. in press). Furthermore, transfection of the cDNA construct containing this mutagenized form of IGFBP-5 in the pig smooth muscle cells resulted in expression of a stable form of IGFBP-5 in the medium and cells so expressing this stable mutant were much less responsive to IGF-I in terms of its capacity to increase protein synthesis. Therefore, inhibition of proteolytic cleavage of IGFBP-5 results in attenuation of IGF-I actions.

In addition to proteolytic cleavage, another property of mesenchymal cells regarding IGFBP physiology is focal localization on either cell membrane or in extracellular matrix (ECM). IGFBP-3, to a limited extent, has been shown to localize in cell ECM. When its affinity is compared with IGFBP-5, IGFBP-5 clearly binds with much higher affinity. The physiologic significance of ECM localization is that when IGFBP-5 is localized in the ECM, its affinity for IGF-I and -II is reduced by ≈15-fold. This reduction of affinity allows better equilibration of IGF-I with all surface receptors. Furthermore, the IGFBP-5 that is bound to ECM is resistant to proteolysis, probably because its cleavage site is part of the binding domain; thus, making it inaccessible to the protease. The physiologic significance of this finding is that following enrichment of the amount of IGFBP-5 within the ECM, there is a potentiation rather than inhibition of the cellular growth response to IGF-I. When fibroblast matrix was enriched in IGFBP-5, the cellular growth response to IGF-I could be potentiated by 2.1-fold suggesting that amplification of the response was due to the amount of IGFBP-5 that could be focally concentrated.

Further studies have determined the specific amino acids that are involved in localization of IGFBP-5 in ECM. Most of these are localized in the region between amino acids 201 and 218, the so-called heparin binding domain because this region mediates binding to glycosaminoglycans. The basic residues Arg207 and Arg214 appear to be the two most important residues responsible for ECM association. Arg217 and Arg218 have a minor contribution (Parker et al., 1995). Mutants in which these basic amino acids have been converted to neutral ones are much less effective in binding to ECM and in enhancing IGF-I action. The moieties in ECM that bind to IGFBP-5 are incompletely characterized. Clearly, this heparin binding motif binds avidly to glycosaminoglycans; specifically to glycosaminoglycans that are similar in structure to heparan sulfate (Arai et al., 1994a). The structural property that appears to be the most important is the presence of O-linked sulfate groups on the 2 or 3 position of iduronic acid ring. The presence of these groups causes glycosaminoglycans to bind avidly to IGFBP-5 and this binding contributes to ECM localization by heparan sulfate containing proteoglycans (Imai et al. in press). Furthermore, it inhibits proteolytic cleavage suggesting this may be part of the mechanism by which ECM-associated IGFBP-5 is protected (Arai et al., 1994b).

Other proteins that are not proteoglycans but are major components of ECM also bind to IGFBP-5. Plasminogen activator inhibitor-1 (PAI-1) appears to also bind to IGFBP-5 with high affinity. Therefore, non-heparan sulfate containing non-proteoglycan proteins may bind to IGFBP-5 with high affinity and focally localize it in ECM (Nam et al., 1997). Furthermore, since plasmin is an important protease for IGFBP-5, it is possible that PAI-1 binding helps to protect IGFBP-5 from cleavage by plasminogen that has been converted to plasma.

Connective tissue cells constitutively synthesize small quantities of IGFBP-4 and therefore the variables that regulate its synthesis probably have minimal inhibitory effects on IGF-I actions. The factors that regulate the secretion and distribution of IGFBP-4 protease may have an important role in limiting IGFBP-4's ability to inhibit cellular responses to IGF-I. In contrast, positive modulation of IGF-I action on fibroblasts or smooth muscle cells may be mediated by IGFBP-3, and IGFBP-5 as well as IGFBP-2. IGFBP-3, when associated with fibroblast surfaces, has been associated with result in slight potentiation of IGF action and IGFBP-2 has similar properties. IGFBP-5 has been studied in more detail. Clearly, a 4:1 molar excess of intact IGFBP-5 to IGF-I is a potent growth inhibitor. The IGFBP-5 protease is so active that it rarely allows this concentration of intact IGFBP-5 to be present in medium. However, rates of IGFBP-5 synthesis by osteoblasts, chondrocytes and fibroblasts are relatively rapid and this allows some of the synthetic product to be directly deposited in the ECM where it is protected from proteolysis. Therefore, factors such as IGF-I stimulation in increased IGFBP-5 synthesis may preferentially result in further potentiation of the cellular response to IGF-I by increasing release of protein into both media and ECM. Since the media protein would be rapidly degraded, but the matrix protein preserved, this would act to further facilitate IGF-I action. Whether or not this hypothesis is definitively operative in vivo is a critical question for future studies.

In summary, the IGF binding proteins that are synthesized by connective tissue cells can regulate IGF pericellular distribution and the amount of IGF that is available to bind to cell surface receptors. By providing such a regulatory mechanism, these cells may modulate the effect of IGF-I in mediating connective tissue cell growth during normal growth and development in childhood and in pathophysiologic processes such as osteopenia and atherosclerosis.

Section snippets

Acknowledgements

The author thanks Ms Dottie McQueen for her help in preparing the manuscript. These studies were supported by grants from the National Institutes of Health (HL-56250 and AG-02331).

References (32)

  • B. Cercek et al.

    Induction of insulin-like growth factor I messenger RNA in rat aorta after balloon denudation

    Circ. Res.

    (1990)
  • W.S. Cohick et al.

    Vascular smooth muscle cells synthesize two forms of insulin-like growth factor binding proteins which are regulated differently by the insulin like growth factors

    J. Cell. Physiol.

    (1993)
  • W.S. Cohick et al.

    Regulation of insulin like growth factor (IGF) binding protein-2 synthesis and degradation by platelet derived growth factor and IGFs is enhanced by serum deprivation in vascular growth muscle cells

    J. Cell. Physiol.

    (1995)
  • C.A. Conover et al.

    Regulation of biological effect of endogenous insulin like growth factor binding protein 5 in human osteoblast cells

    J. Clin. Endocrinol. Metab.

    (1993)
  • C.A. Conover et al.

    Regulation of insulin-like growth factor binding protein-5 messenger ribonucleic acid expression and protein availability in rat osteoblasts

    Endocrinology

    (1993)
  • C.A. Conover et al.

    Post translational regulation of insulin like growth factor binding protein 4 in normal and transformed human fibroblasts

    J. Clin. Invest.

    (1993)
  • Cited by (0)

    View full text