IGF-1 was originally discovered because of its ability to stimulate sulfation of cartilage proteoglycans.¹ The administration of GH to hypophysectomized animals resulted in induction of a substance in serum that was a potent stimulant of cartilage sulfation. In contrast, when GH was added to cartilage in vitro, it had minimal bioactivity. This suggested that a separate growth factor was induced in the serum of these animals. Purification of this substance showed that its amino acid sequence was similar to insulin and led to studies which showed that it could stimulate growth in vivo.²

Molecular technology made it possible to determine the structure of the IGF receptors, and further studies have identified the signal transduction pathways that are linked to each receptor. Both the insulin and IGF-1 receptors can activate similar signaling molecules, although in many cases, each receptor induces a subset of signaling molecules that are distinct. A more complete description of the relative roles of IGF-1, insulin, and GH in growth and metabolic regulation ultimately awaits the elucidation of all the proteins that are induced by activation of each receptor, which will lead to a better understanding of their respective target cell roles, as well as their relative hierarchical importance in growth regulation.

IGF-1 Gene and Protein Structures

The insulin-like growth factor-1 gene is a complex, multi-component gene with 6 exons. The gene structure is shown in Fig. 17-1. The first and second exons encode the 5′ untranslated and pre-propeptide regions of IGF-1. Exon 3 encodes the distal propeptide sequence and the regions of the mature peptide that are homologous to the B chain of insulin, the region homologous to the C peptide and to the A chain region. Exon 4 encodes a D extension peptide. The fifth and sixth exons are shuffled and can encode one of two sequences termed IGF-1A and IGF-1B. This alternative splicing occurs in multiple tissues, and both IGF-1A and IGF-1B have been found to be secreted by specific cell types in culture.

FIGURE 17-1 Structure of the human insulin-like growth factor 1 (IGF1) gene and the precursor proteins it encodes. The black boxes that are shown represent exons. The portions of each exon that encode parts of the precursor protein are shown by lines. The IGF-1A and IGF-1B precursor forms are represented by boxes. The B, C, A, and D domains of the mature peptides are noted.

Several forms of IGF-1 mRNA are transcribed, and at least four specific transcripts have been detected in tissues.³ The most abundant IGF-1 transcript (6 kb) contains multiple polyadenylation sites and a long 3′ untranslated sequence. The abundance of this transcript is regulated by GH. GH increases transcription of IGF-1 by inducing STAT5b which binds to an intronic region between exons 2 and 3 and initiates transcription. Several different fetal and tissue-specific promoters of IGF-1 have been identified, and they account for distinct transcript patterns in various tissues and the appearance of various forms at specific periods in development.⁴ Other abundant transcripts include a 3.2 kb transcript, a 2.7 kb transcript, and a 0.9 kb transcript. Stimuli other than GH have been shown to influence the abundance of these transcripts in various tissues.⁵ The small, 0.9-kb transcript is one source of the mature 70 amino acid IGF-1 peptide. This transcript is present in the liver and is an important source of the peptide that is present in the systemic circulation. Alternative processing of IGF-1 mRNA following its transcription has been shown to occur in multiple tissues and may be physiologically relevant in specific situations, such as muscle repair after injury.⁶ Variable polyadenylation sites and regulation of processing of the 3′ untranslated RNA extensions have been demonstrated and can result in different-length transcripts.⁷

The polypeptide structures of three members of the IGF gene family are shown in Fig. 17-2. Mature IGF-1 and IGF-2 contain 70 and 67 amino acids, respectively. Proinsulin has a longer C peptide region compared to IGF-1 or IGF-2. The sequence in this region is not conserved. The A chain and B chain peptide regions are of similar length. The sequences in this region are 41% and 43% homologous with proinsulin. IGF-1 and 2 contain D-domain extensions of 8 and 6 amino acids, respectively. Unlike proinsulin, IGF-1 and 2 are not cleaved into two-chain polypeptides during intracellular processing, but rather they are secreted as intact single-chain proteins. Forms of IGF-1 have been isolated from serum and from cell culture supernatants that contain the E peptide extensions (e.g., both A and B), but the relative abundance of these forms in most tissues is unknown. The frequency of processing of the E peptide domains is unclear, since longer forms of IGF-1A or IGF-1B have been shown to be secreted by cells in culture. However, some cells do not secrete IGF-1 with the E peptide extension.

FIGURE 17-2 The sequences of proinsulin, insulin-like growth factor 1 (IGF-1), and IGF-2. The sequences are divided into the B, C, A, and D domains.

Specific amino acids within the IGF-1 molecule have been shown by site-directed mutagenesis to account for receptor and/or binding protein association (Table 17-1). Specifically, tyrosine 24, tyrosine 60, and to some extent tyrosine 31 are required for IGF-1 receptor recognition.⁸ The tyrosines at positions 24 and 60 are conserved within IGF-2, but tyrosine 31 is not present. The residue within the proinsulin sequence that is homologous to Tyr24 (e.g., Phe25) is important for insulin binding to its receptor. Tyrosine 60 appears to be necessary for IGF-1 to maintain a stable conformation. Studies using mutant forms of IGF-1 with large deletions indicate that the region between residues 24 and 37 contains the primary receptor binding site.⁹ Mutations in this region have very little effect on binding protein affinity. More recent crystallographic¹⁰ and NMR studies¹¹ have confirmed the importance of these residues for receptor binding. These studies highlighted the importance of Phe16 and Leu54 for ligand-induced activation and suggest they are required by full activation of the receptor kinase. These studies have also shown the importance of specific residues in the C domain, particularly Arg35 and Arg36.¹² Alanine scanning mutagenesis has confirmed the importance of Phe23 and 25, Tyr31, Arg36, Arg 37, Val44, and Tyr60, the residues that compose the binding site. These studies also identified a secondary site composed of Glu9, Asp12, Phe16, Leu54, and Glu58. Substitution for these residues resulted in a 33- to 100-fold reduction in receptor affinity.¹³

Table 17-1

Specific Amino Acids in Insulin-Like Growth Factor 1 (IGF-1) That Mediate Binding Protein and Receptor Association

Region of IGF-1	Ligand Interaction
B chain Glu3, Thr4, Gln15, Phe16	Required for binding to IGF-binding proteins (IGFBPs) 1-6
A chain Phe 49, Arg50, Ser51	Required for optimal binding to IGFBP-1, 2, 4, 5
Tyr24, Tyr31, Tyr60,	IGF-1 receptor
Tyr24-Arg37	Contains the primary receptor binding site
Tyr60	Necessary for a stable conformation

IGF-1 and IGF-2 contain 4 amino acids that are the primary determinants of their high affinity for IGF-1 binding proteins. These include the amino acids at positions 3, 4, 15, and 16 of the B chain region of IGF-1 and the homologous residues 6, 7, 18, and 19 in IGF-2.¹⁴ These residues are critical for recognition by all six forms of IGF-binding proteins (IGFBPs). Mutant forms of IGF-1 that contain substitutions of proinsulin residues in these four positions have a nearly total loss of binding protein activity. In addition, residues at amino acids 49, 50, and 51 in the A chain are important for recognition by four of the six high-affinity binding proteins. The major exception is IGFBP-3, wherein only the B chain residues appear to be important. Recent x-ray crystallographic studies of the IGF-1/IGFBP-4 complex have confirmed the importance of these residues. They confirmed that they are the primary sites within IGF-1 that interact with IGFBP-4 and explained how the interaction between IGF-1 and IGFBP-4 interferes with receptor binding.¹⁵ Studies of the tertiary structures of IGF-1 and IGF-2 have shown that these residues are surface exposed. A recent NMR study showed that two of three helices that are present in IGF-1 form a surface-exposed hydrophobic patch that contains these A- and B-chain residues. A peptide that bound to this patch inhibited IGF-1/IGFBP binding. The C-peptide regions in each of the three proteins are divergent, and this accounts for most of the heterogeneity of sequence between IGF-1 and IGF-2. The three disulfide linkages are conserved in all three peptides.

The structure of IGF-1 is highly conserved across species. Bovine IGF-1 is identical to human, and rat differs by only three amino acids. IGF-1-like molecules have been detected in all vertebrates that have been analyzed, and even species as low on the phylogenetic tree as Caenorhabditis elegans contain IGF-1-like molecules. Computer modeling studies have indicated that the three-dimensional structure of IGF-1 is probably similar to that of insulin, which (except for the C-peptide) has been analyzed by x-ray crystallography.¹⁰ Except for residues contained in the C-peptide, many of the IGF-1 residues that bind IGF-1 receptor are also present in insulin. The high affinity of IGF-1 for the IGF-1 receptor as compared to insulin is explained by the presence of the C-peptide.^12,13,14 In contrast, the higher affinity of the insulin receptor for insulin is accounted for by residues 4, 15, 49, and 51. If these insulin residues are substituted for those of IGF-1, this IGF-1 mutant has an affinity for the insulin receptor that is equal to insulin.¹⁶ Different forms of IGF-1 have been found to be present in human serum and tissues. The most extensively studied form is des-1-3 IGF-1, which occurs in brain and in serum. This IGF variant has much lower affinity for IGFBPs and therefore is more biologically active.

IGF-1 Receptor

The IGF-1 receptor is ubiquitously present and has been shown to be present in cell types derived from all three embryonic lineages. When animal tissues are analyzed, the receptor is detected uniformly, thus accounting for IGF-1’s ability to stimulate growth of all tissues. The receptor number per cell is tightly controlled and maintained in a narrow range of 20,000 to 35,000. This may be an important regulatory function, since cellular transformation in response to IGF-1 usually requires >1,000,000 receptors per cell,¹⁷ whereas cells that have <100,000 receptors per cell rarely induce tumors in experimental animal models. Thus the variables that regulate IGF-1 receptor number may be important in terms of the genesis of neoplasia.

Hormones such as GH, FSH, LH, progesterone, estradiol, and thyroxine have been shown to increase IGF-1 receptor expression.¹⁸ Similarly, PDGF, EGF, FGF, and angiotensin II up-regulate receptor expression in specific cell types.¹⁹ Following hormone binding, there is down-regulation of receptor number with internalization of receptors. However, possibly due to IGF binding proteins, the rate of internalization of IGF-1 receptors is substantially slower than that of other growth factors such as epidermal growth factor or insulin.

The biochemical structure of the receptor is similar to other polypeptide growth factor receptors (Fig. 17-3). The receptor is a heterotetrameric glycoprotein composed of two ligand binding subunits, termed alpha subunits, that contain 706 amino acids and two beta subunits that contain 627 amino acids. Only the beta subunits have a transmembrane domain (see Fig. 17-3). In man, the protein is translated from a single mRNA transcript derived from a gene that contains 21 exons located on chromosome 15, Q25-Q26.²⁰ The prepropeptide is 1367 amino acids, and the signal peptide is removed cotranslationally. The precursor is cleaved between Lys708/Arg709 to generate the alpha and beta subunits. These are linked together by disulfide bonds to form the heterotetrameric receptor. Amino acid sequence comparison with the insulin receptor reveals 46% amino acid identity.²⁰

FIGURE 17-3 Structural characteristics of the insulin, insulin-like growth factor 1 (IGF-1), and hybrid receptors.

The alpha subunit contains three domains that are essential for ligand binding. These have been termed leucine rich (LR), cysteine rich (CR), and carboxy terminal (CT) domains. The receptor binds IGF-1 with a mean equilibrium dissociation constant (KD) of 10⁻⁹ M. IGF-2 binds with sixfold lower affinity, and insulin with a 200- to 300-fold lower affinity.²¹ The composition of crystal structure of the first three domains of the alpha subunit of both the IGF-1 and insulin receptors shows that there are two important differences, one in LR1 and one in the CR1, that account for these differences in ligand binding.²² Mutagenesis studies have shown that the residues Asp8, Tyr28, His30, Leu33, Phe58, Tyr79, and Phe90 within the LR1 domain are important for binding.^23,24 The CR region contains four residues that are essential to maintaining high affinity. A short C-terminal region (692-702) is also very important, since changes in 7 of these 10 residues reduced binding affinity 10- to 30-fold.^24,25 Following site one (LR-CR-CT) contact, the ligand becomes immobilized then cross-links through its second binding domain to a distinct site on the second monomer, thus resulting in high-affinity binding.

The beta subunit of the receptor is composed of an insert domain followed by two fibronectin repeat domains, then a transmembrane domain between positions 906 and 929 that is followed by an intracytoplasmic domain. This region contains intrinsic tyrosine kinase activity and critical sites of tyrosine and serine phosphorylation. The tyrosine kinase (TK) domain is 84% homologous with the insulin-receptor TK domain. The catalytic domain contains an ATP binding motif and a catalytic lysine at position 1003. Substitution for this lysine abolishes IGF-1-stimulated biological actions. Ligand binding to the alpha subunit triggers a conformational change that leads to autoactivation. This in turn leads to trans subunit autophosphorylation wherein a specific tyrosine 1135 on one beta subunit is transphosphorylated by the TK activity located on the paired beta subunit. This nonphosphorylated tyrosine is autoinhibitory, and its phosphorylation leads to kinase activation and transphosphorylation of the paired tyrosine 1135 on the corresponding subunit, followed by sustained TK activation.²⁶

There are at least six important tyrosines contained within the cytoplasmic domain that are phosphorylated by the intrinsic tyrosine kinase. The most important is a triple tyrosine motif at positions 1131, 1135, and 1136. Substitutions for these tyrosines abolish IGF-1 signaling.24,25 Crystal structure analysis has shown that phosphorylation of all three tyrosines is required to obtain the optimal conformation.²⁷ Following activation of the intrinsic tyrosine kinase activity, the enzyme autophosphorylates tyrosine 950 in the beta subunit, which forms a binding site for two important intracellular substrates, insulin receptor substrate 1 (IRS-1) and insulin receptor substrate 2 (IRS-2).²⁸ Substitution for this residue attenuates IRS-1 phosphorylation. Following IRS-1 binding to Tyr950, the IGF-1R kinase phosphorylates specific sites on IRS-1 that provide binding sites for adaptor proteins, such as Grb-2, which in turn leads to Ras activation. Other kinases, such as phosphotidylinositol-3 (PI-3) kinase are activated by binding to phosphorylated IRS-1. Mutation of tyrosine 1316 in the beta subunit abrogates the ability of IGF-1R to activate PI-3 kinase. The receptor can also directly phosphorylate other substrates, including Shc, Crk, and Grb-10.²³ Phosphorylation of beta subunit residues 1280 and 1283 is necessary for binding to 14-3-3, an additional signaling intermediate, and for mediating IGF-1’s anti-apoptotic activity. The NPXY motif located near the transmembrane domain is required for internalization.

Chimeric receptors that contain heterodimers of the IGF-1 and insulin receptor have been described.²⁹ These dimers are disulfide linked. Receptor hybrids have been detected in several tissues and cell types. It is possible that they exist in all cells in which both IGF-1 and insulin receptors are expressed. The ligand specificity and affinity properties of hybrid receptors are much closer to those of the IGF-1 receptor as compared to the insulin receptor. Hybrid receptor activation has been shown to lead to stimulation of signal transduction in vitro³⁰; however, the biological significance of hybrid receptor activation in tissues in whole animals has not been determined. Following IGF-1 activation of the receptor, it undergoes endocytosis. This is regulated in part by the adaptor protein 2 complex.³¹ Following its recruitment to endosomes, the receptor is cleaved by a cysteine protease, and ligand is released.³² Ubiquitination also regulates this process, and two E3 ligases, Nedd4 and MDM2, have been shown to play a role. Nedd4 binds to IGF-1R through Grb-10 and MDM2 through beta arrestin; thus these molecules also play a role in IGF-1R degradation.

The IGF-1 receptor has been overexpressed in several types of cells in culture. Receptor overexpression enhances growth in soft agar and tumor formation in nude mice.¹⁷ Studies using antisense oligonucleotides to lower IGF-1 receptor number have confirmed its importance for growth and transforming activity of human tumor cells.³³ Importantly, deletion of specific tyrosines, such as tyrosines 1280 and 1281, results in a marked diminution in the transforming property of the IGF-1 receptor, although mitogenesis in vitro is still preserved.³⁴ Additionally, the receptor is important for IGF-1’s ability to modulate the effects of other growth factors. Mouse fibroblasts containing deficient numbers of IGF-1 receptors do not undergo DNA synthesis in response to the addition of epidermal growth factor. Similarly, overexpression of the EGF and PDGF receptors does not lead to proliferation of fibroblasts in soft agar in the absence of IGF-1 receptors,³⁵ and reexpression of the IGF-1 receptor allows proliferation to occur. Large T-antigen induction by the cellular transforming virus SV-40 requires expression of the IGF-1 receptor, and wild-type Ras activation has less of an effect on cellular transformation if the IGF-1 receptor is absent.³⁶ Likewise, Src oncogene expression results in transforming activity only in the presence of an IGF-1 receptor.

The IGF-1 receptor has an important role in normal development and normal fetal growth. Animals that have had the IGF-1 receptor deleted by homologous recombination are born 40% of normal size.³⁷ These animals are not viable at birth, due to hypoplasia of diaphragmatic muscle. Defects in the development of the nervous system, skin, and bones have been noted. These developmental abnormalities apparently occur relatively late in gestation. Fibroblasts obtained from these embryos have a markedly attenuated growth response compared to fibroblasts from normal embryos.

The receptor is also important for prevention of apoptosis. IGF-1 and its receptor support the viability of nonproliferating cells in culture, such as neurons. The extent of apoptosis that can be induced in neurons by osmotic hyperglycemia, ischemia, or potassium shock is dependent upon normal IGF-1 receptor expression, suggesting that it is neuroprotective.³⁸ Hematopoietic cells that undergo apoptosis if IL-3 is withdrawn are protected by IGF-1 exposure if IGF-1 receptors are present. Plating tumor cells on a surface that does not allow ligand binding to integrins results in susceptibility to apoptosis, and this susceptibility can be reversed by incubation with IGF-1.³⁹ In contrast to the IGF-1 receptor, overexpression of the insulin receptor is nontransforming. Likewise, overexpression of a chimeric receptor bearing the beta subunit of the insulin receptor is nontransforming, but if the IGF-1 receptor beta subunit is expressed with the insulin receptor alpha subunit, then mitogenic activity of insulin is detected at much lower ligand concentrations, and this receptor construct allows transformation to occur.⁴⁰

Receptor-Mediated Signal Transduction

Following activation of the intrinsic tyrosine kinase activity and phosphorylation of tyrosine 950, the docking protein IRS-1 binds directly to the receptor (Fig. 17-4). The functionally similar protein IRS-2 has been shown to bind by a similar mechanism.⁴¹ Following binding, IRS-1 is tyrosine phosphorylated by the receptor at multiple sites, creating docking motifs that are critical for binding of intracellular proteins that contain Src homology-2 (SH-2) domains. These domains contain approximately 100 amino acids that share sequence similarity to cellular oncogene Src. Six of the tyrosines in IRS-1 occur within YXXM sequences, a recognition motif for some SH-2 domains. IRS1 gene deletion in mice results in a major decrease in body weight, with proportionate reduction in liver, heart, and spleen.⁴² Activation of signaling pathways that lead to enhanced IRS-1 degradation result in attenuation of IGF-1 signaling.

FIGURE 17-4 The two major signaling pathways that are used by the insulin-like growth factor 1 (IGF-1) receptor. These include the MAP kinase (shaded) and PI-3 kinase (open) pathways. P-110 and P-85 represent the major subunits of phosphatidylinositol-3′-kinase.

Signaling proteins that bind directly to the phosphorylated tyrosines on IRS-1 include the adaptor proteins Grb-2 and p85. Grb-2 forms a complex with the Ras-activating protein Son of Sevenless (SOS), and this complex leads to subsequent p21 Ras activation, which activates Raf and downstream components of the MAP kinase pathway.⁴³ Activation of this pathway is important for the mitogenic function of IGF-1.

IRS-1 phosphorylation also results in binding of the p85 regulatory subunit of PI-3 kinase, and this leads to binding of the catalytic subunit p110 and its activation. This results in generation of inositol triphosphate and activation of protein tyrosine kinase B.⁴⁴ This kinase activates mTOR and P70 S6 kinase, which leads to activation of protein translation. This pathway is also important for IGF-1-induced increases in cell motility and for inhibition of apoptosis. AKT also phosphorylates GSK-3 beta, leading to its inactivation, which is important for several responses that include stimulation of glucose transport.

The IGF-1 receptor can directly phosphorylate Shc, and this leads to association of Grb-2, which activates Ras and MAPK independently of IRS-1. Although Shc can be directly phosphorylated by the IGF-1 receptor, in certain situations such as following glucose-induced oxidative stress, Shc activation proceeds by a different mechanism. In vascular endothelial or smooth muscle cells, hyperglycemic stress leads to increased secretion of ligands for the αVβ3 integrin. Oxidative stress also leads to activation of c-Src. αVβ3 activation results in translocation of activated c-Src to a plasma membrane–associated scaffolding protein, SHPS-1. The IGF-1 receptor phosphorylates SHPS-1, which results in recruitment of activated Src, which recruits Shc and phosphorylates it.⁴⁵ Since IRS-1 signaling is markedly down-regulated by hyperglycemia, this mechanism allows full MAP kinase activation in response to IGF-1 even in the absence of IRS-1 activation. An additional signaling molecule that is activated by the receptor is Crk, a Grb-2-like protein, with SH-2 and SH-3 domains. Crk then activates Grb-2 and SOS after it is phosphorylated by the IGF-1 receptor.²¹ Other signaling pathways that have been shown to be activated by IGF-1 include protein kinase-C, phospholipase-C, and direct stimulation of calcium-permeable ion channels. Activation of these proteins leads to activation of downstream signaling cascades, including G-protein activation. Additional signaling molecules that have been shown to interact with the IGF-1 receptor include RACK-1⁴⁶ and Grb-10.⁴⁷

Since there is specificity between insulin and IGF-1 in terms of their metabolic and growth-promoting actions, it was presumed that major differences would be detected in the signal transduction pathways that each hormone utilized. However, IGF-1 and insulin-receptor kinase domains are 84% identical, and similar residues are autophosphorylated. Presumably, during normal growth or stimulation of glucose transport, either distinct domains are activated in the IRS-1 and IRS-2, or separate combinations of signaling pathways are activated. However, in pathophysiologic states such as hyperglycemia, IGF-1 receptor activation of MAP and PI-3 kinase is enhanced in some cells, whereas insulin-receptor signaling is inhibited. Other differences in signaling have also been reported. Activation of Crk-2 is specific for the IGF-1 receptor.²¹ Since Crk-2 has transforming activity, its activation may partially account for the ability of overexpression of IGF-1 receptors to be transforming. Insulin and IGF-1 receptors have been shown to utilize different G-protein signaling components.⁴⁸ Activation of Src kinase results in phosphorylation of the IGF-1 receptor but not the insulin receptor. In summary, multiple signaling events are activated in response to IGF-1 receptor stimulation. The best characterized are those that lead to MAP or PI-3 kinase activation, but other pathways may be important in specific physiologic or pathophysiologic situations.

Blocking specific functions of intracellular signaling pathways has been shown to attenuate specific IGF-1 actions. The PI 3–kinase pathway appears important for glucose transport and for cell migration, and specific inhibitors of PI-3 kinase have been shown to inhibit these IGF-1-stimulated effects.44,49 Similarly, the MAP kinase pathway appears to be the predominant pathway for mitogenesis and rescue from apoptosis.³⁹ Protein kinase C also appears to be essential for IGF-1-stimulated cell migration and stimulation of the transcription of specific genes. However, the requirement of stimulation of a specific pathway for a specific function is not absolute, since the results generated using specific inhibitors of each pathway support the conclusion that there are overlapping functions. In addition to interactions between the IGF-1 and insulin receptor–linked signaling pathways, several other signaling pathways have been shown to influence IGF-1-stimulated signaling events. Several hormones and growth factors such as EGF, angiotensin II, aldosterone, and estrogen have been shown to modulate IGF-1 receptor–linked signaling events.^50–52 Conversely, cellular activation by IGF-1 has been shown to result in transactivation of the androgen receptor EGFR, VEGFR, and the chemokine receptor CXR4.^53,54 In addition, postreceptor signaling pathway cross-talk has been demonstrated for the GH receptor, estrogen receptor, progesterone receptor, glucocorticoid receptor, and multiple cytokine pathways such as TNFα,⁵⁵ which induces tissue refractoriness to IGF-1 in states of cachexia.

IGF-2 Mannose-6 Phosphate Receptor

The IGF-2/cation-independent mannose-6 phosphate receptor is a single-chain, membrane-spanning glycoprotein that contains 2451 amino acids. It binds mannose-6 phosphate residues on lysosomal enzymes as well as IGF-2. There is a large extracellular domain, a 23 amino acid transmembrane domain, and a 164 residue carboxy terminal intracytoplasmic domain. The extracellular domain is composed of 15 repeating motifs. Motifs 7 to 9 bind mannose-6 phosphate, and motif 11 contains the IGF-2 binding region.⁵⁶ Analysis of this region shows that Tyr1542, Glu1544, Phe1567, Thr1520, and Ile1572 come in close contact with IGF-2, and mutagenesis studies have confirmed its importance for binding.⁵⁷ Intracellularly, this receptor functions to translocate newly synthesized lysosomal enzymes into endosomes. On the cell surface, it binds to mannose-6 phosphate–containing extracellular glycoproteins, which are endocytosed into endosomes. The receptors are then recycled back to the cell surface. Proteins other than lysosomal enzymes shown to bind to this receptor include proliferin, thyroglobulin, and latent transforming growth factor beta (TGF-β). Binding of latent TGF-β has been shown to result in cleavage of the inactive form into active TGF-β. In adipocytes, it has been shown that insulin is a potent stimulant of redistribution of mannose-6 phosphate receptors from intracellular locations to the plasma membrane. The receptor binds IGF-2 with an affinity in the range of KD 1 to 3 nM. The affinity for IGF-1 is 80-fold lower, and the receptor does not bind insulin. Mannose-6 phosphate–containing proteins bind to a site that is distinct from IGF-1 or IGF-2, and the receptor can bind both types of ligands simultaneously. Once IGF-2 is bound, it is internalized and degraded. The extracellular portion of the receptor can be proteolytically cleaved in certain cell types, and the cleavage product is released. This soluble form has been detected in plasma; however, the physiologic significance of its release into plasma has not been determined.

The role of this receptor in IGF physiology is incompletely understood. Deletion of the receptor or mutations that result in loss of IGF-2 binding result in death of fetal mice.⁵⁸ The receptor is subject to parental imprinting, such that only the maternal allele of the IGF-2 receptor and the paternal allele of IGF-2 are expressed. Therefore, mice that inherit a receptor allele containing a mutation from the mother have functionally altered IGF-2 receptors. These mice develop severe edema in utero prior to death.⁵⁸ They are also larger than fetuses of comparable developmental age. If IGF-2 is deleted concomitantly, 50% of the fetuses survive birth; however, postnatal survival is poor. The hypothesis has been that these mice lack the putative scavenging function of the IGF-2 receptor and accumulate toxic levels of IGF-2. Although the scavenging function of the receptor is well accepted, it is clear that this receptor does not mediate the actions mediated by the IGF-1 receptor, such as growth stimulation. In most systems, inhibition of the IGF-1 receptor is sufficient to completely block the mitogenic response to IGF-1 or IGF-2 stimulation. Increases in calcium flux have been shown to occur following stimulation of 3T3 cells by IGF-2 binding to this receptor. Additionally, the receptor has been shown to activate GTP binding proteins, but the exact functional significance of these effects is undetermined. The cytoplasmic portion of the receptor encodes regions that are necessary for specific subcellular localization and endocytosis, as well as binding to GTP binding proteins.⁵⁹ Partitioning of the receptor following internalization can be hormonally regulated. Treatment with insulin was found to cause a rise in the fraction of surface receptors without a change in total number. Mannose-6 phosphate stimulates a similar increase, and this can be blocked by pretreatment with pertussis toxin, implying both stimulatory and inhibitory GTP binding protein regulation.

IGF-Binding Proteins

A characteristic of IGF-1 and IGF-2 that distinguishes them from proinsulin is the ability to bind to high-affinity IGF binding proteins (IGFBPs). The IGFBPs are a family of six proteins that each have high affinity for IGF-1 and IGF-2.⁶⁰ In each case, this affinity is greater than the affinity of the type 1 IGF receptor for IGF-1. One or more members of this family is present in all extracellular fluids. Therefore, they control the ability of IGF-1 and IGF-2 to bind to receptors. In addition to this property, the major functions of the IGFBPs include: (1) transporting the IGFs in the vasculature, (2) controlling their access to the extravascular space, (3) controlling tissue localization and distribution, and (4) controlling access to receptors and thereby modulating the biological responses of cells to IGF-1.

The gene structure of the IGFBPs shows that each of the six forms contains four exons.⁶¹ The mRNA species range in size from 1.4 kb (IGFBP-2) to 6 kb (IGFBP-5). Their protein structures show great similarity. Of the 18 cysteines, all are conserved in 5 of the 6 binding proteins. IGFBP-4 has 2 additional cysteines, and IGFBP-6 has only 16 cysteines. If the cysteine structure is disrupted, IGF-1 binding is markedly attenuated. All are secreted proteins and contain a hydrophobic leader sequence. The affinity of each protein for IGF-1 and IGF-2 is shown in Table 17-2. The greatest difference is in IGFBP-6, which has a 40-fold higher affinity for IGF-2.

Table 17-2

Affinities of Insulin-Like Binding Proteins (IGFBPs) for Insulin-Like Growth Factor 1 (IGF-1) and IGF-2

There is a high degree of sequence homology in both the N-terminal and C-terminal domains of each protein.⁶¹ Similarly, the sequences in these regions are highly conserved across species. In contrast, the middle third sequence diverges completely. This is important functionally because this is the major site of proteolytic cleavage for IGFBPs. Two of the proteins are N-glycosylated, and glycosylation sites occur in the middle third of the sequence, thereby providing specificity for this property among the different proteins. Recent structural studies have yielded a great deal of information regarding the IGF binding sites and the specific residues that are responsible for IGF binding. Two-dimensional NMR studies of IGFBPs showed that a hydrophobic pocket in the amino terminals (residues 49 to 74) contained six amino acids that form the binding pocket R49, V50, K68, L70, L72, L74.¹¹ Mutagenesis studies confirmed the significance of this region for IGFBP-5 binding and showed that similar residues in IGFBP-3 had a similar function.⁶² Similarly, mutagenesis of the residues in IGFBP-2 that are comparable to Leu70, 73, 74 in IGFBP-5 results in a major decrease in IGF binding. A specific domain in the C-terminus of these proteins also contributes to IGF binding. The C-terminal binding site contribution to net affinity of the entire protein is greater for IGFBP-1 and 2.⁶³ Recent studies have suggested there is strong cooperativity between these domains which contributes to high-affinity binding of the full-length proteins, and that covalent linkage between the N and C terminus is necessary for maximal affinity.⁶⁴ The residues in IGF-2 that bind to the C-terminal domain binding site in IGFBP-6 are similar to those that bind the IGF receptor, and this probably accounts for the ability of the IGFBPs to inhibit IGF-1 binding to its receptor.

Specific Properties of Each Form of IGFBP

IGFBP-1 contains 235 amino acids and is not glycosylated. It contains an Arg-Gly-Asp near its carboxy terminus which mediates binding to the α5β1 integrin.⁶⁵ IGFBP-1 has been detected in multiple types of extracellular fluids. The affinities of IGFBP-1 for IGF-1 and IGF-2 are nearly equal.

IGFBP-2 contains 289 amino acids and is not glycosylated. Its sequence is highly conserved across species, especially in the C-terminus.⁶¹ It has an Arg-Gly-Asp sequence near its carboxy terminus, and it has been shown to bind to cell surfaces. Following cleavage, its affinity for IGF-1 and 2 is greatly reduced.

IGFBP-3 contains 266 amino acids and is variably N-glycosylated.⁸¹ This accounts for its varying molecular weight estimates between 43 and 56 kD. There are three potential N-linked glycosylation sites. Digestion within N-glycanase reduces the estimated molecular mass to 34 kD. Glycosylation does not alter the affinity of this protein. IGFBP-3 contains a highly basic region between residues 216 and 244 (in which 10 of 18 amino acids are basic). This region accounts for its heparin-binding activity and its ability to adhere to glycosaminoglycans.⁶⁶

IGFBP-4 contains 237 amino acids. It is N-glycosylated and therefore has a mass estimate of 28 kD in the glycosylated form and 24 kD in the nonglycosylated form. Glycosylation does not affect the affinity for IGF-1 or 2. IGFBP-4 is cleaved in most physiologic fluids to 16- and 14-kD fragments that have a reduced affinity for IGF-1 and 2.⁶⁷

IGFBP-5 has 252 amino acids and is the most highly conserved form of IGF binding protein, with 97% homology in sequence between the mouse and human forms.⁶⁸ It is most closely related in sequence to IGFBP-3 (e.g., 50% homology in the amino and carboxy terminal ends). IGFBP-5 contains the same heparin-binding domain as IGFBP-3 between amino acids 201 and 218.⁶⁹ This sequence mediates its binding to extracellular matrix, and some specific ECM proteins that bind IGFBP-5 have been defined.⁶⁹ IGFBP-5 is O-glycosylated and has size estimates between 31 and 34 kD. This protein has a high affinity for IGF-1 and IGF-2. It is proteolytically cleaved into a 22 kD fragment in physiologic fluids that has a much lower affinity for these ligands.

IGFBP-6 has 216 amino acids, and the human form has 16 cysteines. The protein is O-glycosylated. It has a high affinity for IGF-2 compared to IGF-1, but the physiologic significance of this difference has not been ascertained.⁷⁰ IGFBP-6 is proteolytically cleaved in physiologic fluids.

Control of IGF-1 Concentrations in Serum

Prior to the availability of IGF-1 to administer to humans, the primary means of assessing IGF-1’s effects on anabolism was to make inferences from changes in plasma or tissue IGF-1 concentrations. Correlations between IGF-1 levels and parameters of anabolism, such as growth rate, rates of total body protein synthesis, and nitrogen balance were undertaken, and inferences were drawn based on changes in IGF-1 serum levels in response to variables such as GH administration. These studies formed the basis of several principles of IGF-1 physiology that have been confirmed directly by manipulation of IGF-1 expression in transgenic animals or by infusion of IGF-1 into animals and humans and indirectly by measurements of changes in IGF-1 concentrations in states of GH deficiency or excess.

Age is an important determinant of the normal serum IGF-1 concentrations. Plasma concentrations rise from very low levels (20 to 60 ng/mL) at birth to peak values between 212 and 638 ng/mL at puberty.⁷¹ The concentrations then fall rapidly in the second decade, reaching a mean value of 284 ng/mL by age 20 and then decline more slowly over each decade (Fig. 17-5). They are reduced to <50% of the 20-year-old value by age 60 years. A portion of this change is due to age-dependent changes in GH secretion. Although the change in GH may account for much of the decline that occurs during adulthood, it does not account for all of the major increase that occurs during childhood.

FIGURE 17-5 Serum concentrations of insulin-like growth factor 1 (IGF-1) in healthy subjects, aged birth to 75 years. The means are shown as solid lines, and the 95% confidence intervals are shown as rectangles.

There are important genetic determinants of plasma IGF-1 concentrations. Studies in twins have shown that approximately 40% of each individual’s IGF-1 variability can be accounted for on the basis of undefined genetic factors which are linked to height.⁷² There is a very close correlation between IGF-1 concentrations and statural height in many different types of populations that have been studied, and these appear to be due, at least in part, to this genetic factor. This genetic determinant is independent of intrinsic GH secretion. Recently a polymorphism in the IGF1 gene that occurs in 12% of Caucasians has been shown to be associated with a lower mean serum IGF-1 concentration (∼30% reduction) and a decreased final adult height (e.g., ∼2 cm). The presence of this polymorphism in individuals >60 years was associated with a twofold increase in the prevalence of type 2 diabetes and an increased incidence of heart attacks and strokes.⁷³

The major hormonal determinant of plasma IGF-1 concentrations is growth hormone. Children with definitive evidence of growth hormone deficiency (GHD) usually have IGF-1 values that are below the 95% confidence interval.74,75 Because values vary so much throughout childhood, however, age-adjusted normative data are required to interpret low plasma IGF-1 values (see Fig. 17-5). Consideration of developmental stage (skeletal age) is also important for interpreting low values.⁷⁶ In children, normal IGF-1 value is strong evidence that GH deficiency is not present. Conversely, a low IGF-1 is very suggestive of GHD, but it does not definitively prove that GHD is present.⁷⁷ Other causes of growth retardation can be associated with a low IGF-1, although causes such as constitutional growth delay are usually associated with normal levels. Administration of GH to patients with GHD results in a substantial rise in IGF-1, and this occurs during the 4 to 6 hours following an injection. The values peak at 24 hours and then begin to attenuate. Because GH also increases the plasma concentrations of IGFBP-3 and a third protein termed acid labile subunit (ALS), which binds both IGFBP-3 and the IGFs, the formation of this ternary complex accounts for the extended duration of the change in serum IGF-1. The IGF-1 response of a short child to GH administration has not proven to be a useful diagnostic test of GH deficiency.^78,79 In spite of these problems in interpreting low values, basal IGF-1 measurements have proven very useful for as a screening test for selecting individuals who should undergo stimulation testing to assess their GH secretory response.^74,75

In states of GH excess, IGF-1 values are invariably increased. The mean IGF-1 for patients with acromegaly is seven times the normal age-adjusted control value.⁸⁰ The sensitivity and specificity of a single IGF-1 measurement for accurately diagnosing acromegaly in patients older than 20 years is >97%.⁸¹ The severity of the IGF-1 abnormality appears to correlate with disease activity, and values correlate with measurement of soft tissue growth, such as heel pad thickness.⁸⁰ IGF-1 measurements are useful in monitoring the response to therapy and correlate well with residual GH secretion in these patients.⁸² Generally, if 24-hour mean GH values are less than 1.6 ng/mL, then IGF-1 will be within the age-adjusted 95% confidence interval. IGF-1 values are also elevated during the last trimester of pregnancy, presumably due to increases in placental GH secretion.

Another hormonal variable that controls IGF-1 concentrations is thyroxine. Plasma IGF-1 concentrations are low in severe thyroxine deficiency and rise with thyroid hormone replacement.⁸³ Serum IGF-1 values are not suppressed in Turner’s syndrome, and estrogen replacement results in little change. Prolactin has a weak, stimulatory effect on plasma IGF-1. In subjects who are severely growth hormone deficient, prolactin concentrations of 200 ng/mL or greater can maintain IGF-1 in the normal range.⁸³

Nutritional status is an important determinant of plasma IGF-1 concentrations. Adequate caloric and protein intake have to be maintained in order to maintain an adequate serum IGF-1, both in children and adults.⁸⁴ Fasting for 3 days results in substantial reduction in total serum IGF-1 and a blunted response to the administration of GH.⁸⁵ Ten days of fasting results in a 70% decrease in plasma IGF-1. Following a 5-day fast, values decline by 53%, and subjects must be refed for at least 8 days for values to return to normal. During fasting and refeeding, the changes in IGF-1 correlate with changes in nitrogen balance.⁸⁵ These changes are due to both energy and protein deficiency. An energy intake of 20 Kcal/kg is required to maintain a normal IGF-1, whereas an intake of 0.6 gm/kg of protein is required. The energy must be supplied as at least 100 gm of carbohydrate. Similarly, the quality of the protein intake (e.g., the amount of essential amino acids) is an important determinant of IGF-1 if the protein intake is below 0.5 g/kg/day. Children with severe protein-calorie malnutrition have low IGF-1 values that respond to treatment.⁸⁵ Other catabolic conditions, such as hepatic failure, inflammatory bowel diseases, or renal failure, are associated with low serum IGF-1 concentrations.^86,87 Insulin is an important determinant of IGF-1 concentrations. Although it is difficult to differentiate between nutritional regulation and insulin action, insulin perfusion of the liver in diabetic animals results in a substantial increase in plasma IGF-1. Patients with poorly controlled type 1 diabetes mellitus have low normal IGF-1s that rise into the normal range with adequate insulin treatment.⁸⁸ Furthermore, in poorly controlled type 1 diabetes, there is a correlation between hemoglobin A₁C values and IGF-1. Similarly, patients with severe insulin resistance have low IGF-1 values.⁸⁹

Control of IGFBP Concentrations in Blood and Extracellular Fluids

IGFBP-3 is the most abundant form of IGFBP in plasma. It has the highest affinity for IGF-1 and IGF-2. It also binds to ALS, and the ternary complex that is formed has a long half-life. These characteristics explain why IGFBP-3 accounts for most of the binding protein activity in blood.⁹⁰ The IGFBPs in plasma perform three functions. The first is to act as transport proteins for the IGFs. The second is to regulate their half-lives, and the third is to provide a specific means for transcapillary transport into extravascular fluid compartments.

The plasma concentrations of IGFBP-3 are regulated by GH. IGFBP-3 concentrations are low in patients with GHD and increase as a function of GH secretion.⁹¹ This increase is partially due to a direct effect of GH on IGFBP-3 synthesis; however, it is also because the half-life of IGFBP-3 is prolonged by binding to the two other proteins to form a ternary complex (consisting of IGF-1 or IGF-2, IGFBP-3, and ALS). ALS is an 88-kD glycoprotein containing several leucine-rich domains that are known to facilitate protein/protein interactions, and it is this domain structure that accounts for its binding to IGFBP-3.⁹² Since IGF-1 and ALS synthesis are also increased by GH, all three components are increased, and this acts to prolong the half-life of each component. The binding of IGF-1 to this complex in plasma functions to prolong its half-life from 6 minutes in the free form, which is similar to that of insulin, to 16 hours. The prolongation of the half-life of ALS-associated IGF-1/IGFBP-3 complexes is also due to the fact that this macromolecular complex (150 kD) cannot freely cross capillary barriers, and therefore it is not excreted by the kidney. If sufficient IGF-1 and IGFBP-3 are infused to exceed the binding capacity of ALS, then their half-lives are shortened substantially, indicating that it is the ternary complex that maintains the stability and prolongs their half-lives. The molar concentration of IGFBP-3 in serum is generally equal to the sum of IGF-1 and IGF-2, and therefore it is usually saturated. The affinity of IGFBP-3 for IGF-1 and IGF-2 is not lowered by binding to ALS, and its high affinity and its long half-life account for the fact that 75% of the IGF-1 and 2 in plasma is carried in this complex. The exact function of this large storage pool of IGF-1 and 2 in serum is unknown. However, it is clear that changes in the IGF-1 concentrations within this large complex correlate with the anabolic response to GH administration. Plasma IGFBP-3 levels are elevated in patients with acromegaly and low in patients with GH deficiency, as are ALS levels.^91,93,94 Age is an important determinant of IGFBP-3 concentrations, and serum IGFBP-3 varies with age in a manner similar to IGF-1.⁷⁶

Hormones other than GH can influence the synthesis of IGFBP-3 and therefore its plasma concentration. IGFBP-3 is low in prepubertal males and increases following testosterone administration. It decreases 40% following menopause and can be increased in postmenopausal females with physiologic estrogen replacement.⁹⁵ IGFBP-3 concentrations are low in patients with hypothyroidism and increase 55% following administration of thyroxine.⁹⁶

Insulin enhances the IGFBP-3 synthesis response to GH, but it does not appear to have a direct effect. Insulin also stimulates ALS secretion, and severe diabetes results in reduced ALS levels and reduced ternary complex formation. Although GH directly stimulates IGFBP-3 and ALS synthesis, infusion of IGF-1, while increasing serum IGFBP-3 transiently, acts to suppress its concentrations over time by suppressing GH release from the pituitary and thereby lowering ALS synthesis.⁹⁷

IGFBP-3 abundance in serum is also regulated by protease activity.⁹⁸ Several proteases that degrade IGFBP-3 have been described, including PSA and plasmin, but the exact identity of the serum protease has not been determined. Protease concentrations are abundant in human pregnancy serum⁹⁹ and are also present in GH-resistant states such as diabetes.¹⁰⁰ Proteolytic cleavage reduces the affinity of IGFBP-3 greatly, and the IGF-1 that is released binds to unsaturated IGFBP-1, 2, and 4, wherein it can equilibrate much more readily with the interstitial fluids. Therefore, an important function of proteases that cleave IGFBP-3 may be to liberate IGF-1 and IGF-2 from the IGFBP-3/ALS complex and allow them to bind to lower-affinity forms of IGFBPs that can cross capillary barriers, thus facilitating a more favorable equilibrium with the extravascular space.

The next most abundant IGFBP in plasma is IGFBP-2. The affinity of IGFBP-2 for IGF-1 is less than IGFBP-3, and its plasma concentrations are substantially lower. IGFBP-2 concentrations are inversely regulated by GH; that is, they are high in GHD, suppressed with administration of GH, and reduced in acromegaly.¹⁰¹ Unlike IGFBP-3, IGFBP-2 does not bind to ALS, and there is no ternary complex in plasma; therefore, its half-life when bound to IGF-1 is approximately 90 minutes. It is not saturated, and excess binding capacity exists. Intact IGFBP-2 crosses the capillary barriers. Hepatocytes appear to be the major source of serum IGFBP-2, and the abundance of its mRNA in liver is regulated in parallel with its plasma concentrations.¹⁰² Hypophysectomy in experimental animals results in a major increase in hepatic IGFBP-2 mRNA expression. GH administration to normal or GH-deficient humans results in substantial lowering of plasma IGFBP-2.⁶⁰ IGF-1 is a major regulator of IGFBP-2 concentrations in serum. Following IGF-1 administration to GH-deficient humans or patients with diabetes, there is a three- to fourfold increase in IGFBP-2.⁶⁰ Plasma IGFBP-2 concentrations are also increased by IGF-2, and they are elevated in patients with retroperitoneal tumors that produce IGF-2.¹⁰³ Hepatic IGFBP-2 mRNA expression is significantly increased in diabetic rats and suppressed with insulin administration. Severely limiting nutrient intake in humans results in increases in plasma IGFBP-2, as does poorly controlled type 1 diabetes.¹⁰⁴ The response to nutrient restriction is dependent upon protein intake, since it can be mimicked with low-protein diets that contain a normal caloric content, and IGFBP-2 expression in animals is increased during protein restriction.^105,106 Since the half-life of the IGF-1 bound to IGFBP-2 is considerably less than IGF-1 bound to IGFBP-3, it has been assumed that IGF-1 that is bound to IGFBP-2 is in more rapid equilibrium with IGF-1 in the extravascular space.

The third most abundant protein in serum is IGFBP-1. IGFBP-1 also circulates in binary complexes with IGF-1 and IGF-2. Its affinity for the two growth factors is coequal (see Table 17-2). IGFBP-1 is acutely regulated by insulin.¹⁰⁷ Insulin-deficient states such as fasting or type 1 diabetes are associated with very high concentrations of IGFBP-1, whereas administration of insulin or ingestion of a meal results in marked suppression.¹⁰⁸ Major sites of synthesis of IGFBP-1 are highly restricted, and the liver is the principal site of synthesis, although kidney, maternal placenta, and uterus are other sources of this peptide. Plasma concentrations are controlled primarily by hepatic synthesis and release. Hepatic synthesis is primarily under the control of insulin.¹⁰⁷ IGFBP-1 in blood is unsaturated, and therefore IGFBP-1 is proposed to be a major modulator of free IGF-1 levels, particularly in response to food intake. Postprandially, changes in serum insulin result in a four- to fivefold decrease in IGFBP-1. This is due to direct suppression of hepatic synthesis. Insulin directly affects IGFBP1 gene transcription, and there is an insulin-response element in the 5′ flanking region of the IGFBP1 gene.¹⁰⁹ IGFBP-1 crosses intact capillary beds, and the amount that crosses in a fixed time period is dependent upon ambient insulin concentrations.¹¹⁰

Because IGFBP-1 can bind free IGF-1, it has been proposed to have a gluco-regulatory function, that is, since IGF-1 enhances insulin sensitivity, factors that lead to excessive IGFBP-1 could lead to reduced insulin sensitivity. In states of significant insulin resistance, there is enhanced phosphorylation of IGFBP-1, which increases its affinity for IGF-1 and therefore results in further attenuation of IGF-1’s ability to enhance insulin sensitivity.¹¹¹ Both fasting and diabetes have been shown to cause disproportionate increases in serum IGFBP-1 concentrations.^104,108 In addition, administration of glucocorticoid increases IGFBP-1 through a direct effect on IGFBP1 gene transcription.¹⁰⁹ Administration of large concentrations of IGFBP-1 to hypophysectomized rats results in slight increases in glucose concentrations, suggesting that IGFBP-1 may have some role in regulating the insulin-like actions of IGF-1.¹¹²

The exact roles of IGFBP-1 and IGFBP-2 in controlling the distribution of the IGFs has not been determined. In catabolic states, such as nutritional deprivation, GH deficiency, or renal failure, IGFBP-1 and IGFBP-2 levels are increased. Similarly, in these conditions, the amount of IGF-1 that is bound to IGFBP-3 is decreased.¹¹³ Therefore, they can become the major serum binding component.

IGFBP-4 concentrations in serum have been shown to correlate with changes in bone physiology. Specifically, in states of low bone turnover and in states associated with low parathyroid hormone concentrations, serum IGFBP-4 concentrations are increased. There is a correlation between sunlight exposure and IGFBP-4, suggesting that vitamin D or one of its active metabolites regulates IGFBP-4.¹¹⁴

IGFBP-5 exists in serum mostly as proteolytic fragments, and intact IGFBP-5 is present at very low concentrations. The fragments that are present have very low affinity for IGF-1 and IGF-2, and therefore their plasma concentrations are unlikely to be major regulators of IGF-1 action. IGFBP-5 in plasma binds to ALS, and its concentrations are regulated by growth hormone and IGF-1. Both intact IGFBP-5 and its major fragment increase substantially when GH is administered to GH-deficient patients.¹¹⁵

Circulating IGFBP-6 levels are lower in females than males, but estrogen does not change its concentration.¹¹⁶ IGFBP-6 increases with physical stress, and serum concentrations are elevated in patients with critical illnesses.¹¹⁷ Similarly, they are increased in renal failure.¹¹⁸

Control of IGF-1 Synthesis in Tissues

While it is beyond the scope of this chapter to discuss the expression of IGF-1 in all tissues that have been studied, some general principles are important for a fundamental understanding of the autocrine/paracrine mediated actions of this growth factor. Connective tissue cells within a given tissue or organ are often the origin of IGF-1 transcripts. In-situ hybridization studies have shown that fibroblasts and other cells of mesenchymal origin are the primary extrahepatic source of IGF-1 in vivo.¹¹⁹ Importantly, the abundance of this transcript in connective tissue cells is increased in response to GH, and its synthesis is also regulated by factors that are released in response to injury, such as PDGF.¹²⁰

Cartilage and Bone

In cartilage, both GH and fibroblast growth factor have been shown to be potent stimuli of IGF-1 synthesis by prechondrocytes.¹²¹ Its synthesis is most abundant in those cells that are actively differentiating; when chondrocytes reach the hypertrophic state, IGF-1 synthesis is diminished. Fetal chondrocytes, during development, have been shown to be an abundant source of IGF-1 mRNA.

Similar to cartilage, bone osteoblasts are a source of IGF-1 peptide, and it is synthesized in fetal calvarial tissue.¹²² GH also increases IGF-1 synthesis by osteoblasts. IGF-1 synthesis rates correlate with changes in osteoblast DNA synthesis, type I collagen synthesis, and synthesis of other components of bone extracellular matrix.¹²³ Several bone trophic factors, such as bone morphogenic proteins, stimulate the synthesis of IGF-1.¹²⁴ In bone, IGF-1 mRNA expression is down-regulated by glucocorticoids. In contrast, estrogen stimulates the expression of IGF-1 in osteoblasts.¹²⁵ PTH also stimulates IGF1 gene transcription, and its effect is mediated through cAMP induction, which enhances IGF1 gene transcription.¹²⁶ In contrast, the bone growth factors, FGF, PDGF, and TGF-β, down-regulate IGF-1 expression. IGF-1 appears to be an important factor for erythropoiesis. Red cell mass is decreased in IGF-1-deficient humans and is restored to normal with IGF-1 administration.¹²⁷ Erythroid precursor cells synthesize IGF-1, and its synthesis can be stimulated in these cells both by GH and erythropoietin. Similarly, granulocyte precursor cells synthesize IGF-1 mRNA, and this is stimulated by granulocyte/macrophage colony-stimulating factor.

Reproductive Tract

IGF-1 expression is decreased in the ovary of the hypophysectomized rat, and ovarian expression increases in response to GH. Estrogen can increase ovarian IGF-1 expression, and this has been localized primarily to the granuloma cells of the early follicle.¹²⁸ IGF-1 receptors are also present in these follicular cells, indicating the possibility for an autocrine loop. Follicular fluid contains IGF-1 and IGF-2 peptides, and their concentrations are increased following FSH administration. Several studies suggest that the effects of IGF-2 predominate over IGF-1 in the ovary, and much more IGF-2 is produced in that organ. In the oviduct, IGF-1 and IGF-2 have been shown to be present in oviductal fluid. Oviductal cells express mRNAs encoding both IGF-1 and IGF-2, as well as IGF-1 receptors. Endometrium normally expresses IGF-1 mRNA, and in rats, a 20-fold increase can be induced with estradiol administration.¹²⁹ Estrogen induces IGF-1 expression primarily in the epithelium, whereas progesterone induces it in the endometrial stroma. In the late proliferative phase, IGF-1 mRNA is present almost exclusively in the stroma. Similarly, IGF-1 receptor mRNA is up-regulated during the secretory phase of the menstrual cycle. The testes express IGF-1 mRNA, and the source of origin is the Leydig cell. IGF-1 expression by Leydig cells is down-regulated by interleukin 1 and stimulated by LH.

Neural Tissue

Circulating plasma IGF-1 crosses the blood-brain barrier. However, much of the IGF-1 that is present in CSF is believed to arise from IGF-1 synthesis within the CNS. The major sites of IGF-1 mRNA are the Purkinje cells of the cerebellum, the olfactory bulb, and the hippocampus.¹³⁰ The retina is also a site of postnatal expression. Astroglial cells in the cerebellum are also an important site of IGF-1 synthesis. Immunohistochemical staining has shown that IGF-1 is transported along axons and dendrites and that IGF-1 peptide is present in the choroid plexus. Factors that regulate IGF-1 synthesis in peripheral tissues such as nutrition, thyroid hormone, and estrogen also regulate CNS IGF-1 expression.¹³¹ TNFα down-regulates IGF-1 expression.

Skeletal Muscle

IGF-1 mRNA is expressed in the satellite cells and myoblasts of skeletal muscle.¹³² Following an ischemic or toxic injury, there is a major increase in IGF-1 mRNA expression.¹³³ The wave of increase of expression after skeletal muscle injury coincides with the appearance of regenerating tissue and rapid cell division. Work-induced hypertrophy in muscle can also lead to an increase in expression of IGF-1 and IGF-2, indicating that this change is GH independent.¹³⁴ The IGF-1B transcript is selectively increased. Cardiac muscle is also a site of IGF-1 synthesis, and it is increased in models of cardiac hypertrophy that have been induced either by pressure or volume overload.¹³⁵ Blood vessels are also an important site of IGF-1 synthesis. Both endothelial and smooth muscle cells contain IGF-1 mRNA. Pressure overload, oxidative stress, and angiotensin II increase IGF-1 expression.¹³⁶ Following mechanical injury to blood vessels, there is an increase in IGF-1 expression by smooth muscle cells.¹³⁷