|Home | About | Journals | Submit | Contact Us | Français|
Insulin-like growth factor II (IGF2) is perhaps the most intricately regulated of all growth factors characterized to date. Its gene is imprinted—only one allele is active, depending on parental origin—and this pattern of expression is maintained epigenetically in almost all tissues. IGF2 activity is further controlled through differential expression of receptors and IGF-binding proteins (IGFBPs) that determine protein availability. This complex and multifaceted regulation emphasizes the importance of accurate IGF2 expression and activity. This review will examine the regulation of the IGF2 gene and what it has revealed about the phenomenon of imprinting, which is frequently disrupted in cancer. IGF2 protein function will be discussed, along with diseases that involve IGF2 overexpression. Roles for IGF2 in sonic hedgehog (Shh) signaling and angiogenesis will also be explored.
In the early 1900s, the innovative surgeon Alexis Carrel experimented with maintaining tissues and whole organs in vitro, hoping to advance techniques in organ transplantation. Carrel observed that certain tissue extracts could induce cell proliferation, and he published his findings with this disclaimer:
“Possibly the finding of the activating power of tissue extracts will have no immediate practical application. Nevertheless, it may be indirectly useful by leading to the discovery of some of the factors determining the growth of tissues and of the unknown laws of cell dynamics… .”
Carrel was mistaken that this finding would have no practical application—rather, it pioneered the discipline of tissue culture and the widespread use of serum to support in vitro cell growth. He was right, however, that this “activating power” would eventually lead to the discovery of growth factors, many of which were isolated and characterized in the decades that followed. Two of these factors, which were structurally similar to insulin, had many effects on cell growth and differentiation. In 1987, after 30 years of confusing nomenclature, these proteins were designated as insulin-like growth factor I (IGF1) and insulin-like growth factor II (IGF2) (Table 1).
The IGFs regulate cell growth and differentiation in many species. The anabolic functions of growth hormone are largely mediated by IGF1, which designates IGF1 as a major determinant of somatic growth . Rare mutations in the human IGF1 gene lead to severe growth inhibition and mental retardation . Igf1-null mice are born at 60% of normal birth weight, and the few that survive to adulthood are less than one-third the size of normal mice [12, 13]. On the other hand, IGF2 is virtually dispensable for post-natal development in mice, since Igf2 expression is almost entirely limited to the embryo in rodents . At birth, Igf2-null mice are also growth-impaired but are otherwise normal, and subsequent growth proceeds at normal rates .
These studies support a somewhat redundant role for IGF2; furthermore, its designation as the “second” IGF seems to have relegated it to a lesser role than IGF1. However, IGF2 is the predominant IGF in adult humans (reviewed in ), and inappropriate IGF2 expression is implicated in a growing number of diseases (reviewed in ). The importance of IGF2 is highlighted by its complex and multifaceted regulation. The gene that codes for IGF2 is imprinted such that only one allele is expressed, depending on parental origin . Besides the intriguing mechanisms that surround its imprinted expression, IGF2 is further modulated by a concert of differentially expressed proteins and receptors that determine IGF availability (reviewed in ). This review will examine the complex epigenetic regulation of the IGF2 gene and provide a broad introduction to IGF2 signaling. The ability of IGF2 to stimulate cell proliferation and differentiation will be reviewed, which will lead to a discussion on its involvement in various cancers and other diseases. The angiogenic functions of IGF2 will be addressed, and conclude with a proposal that IGF2 is a key mediator facilitating the angiogenic activity of sonic hedgehog (Shh).
Igf2 is widely expressed during murine embryonic development and is particularly important in placental growth . As with many genes that regulate placental development, Igf2 is imprinted, or expressed monoallelically, and active only on the paternally inherited allele. Igf2 is highly expressed in the mouse embryo, but levels decline dramatically after birth; in adult mice, Igf2 transcripts are detectable only in the choroid plexus and leptomeninges, where expression is biallelic . IGF2 is also imprinted in humans, but is expressed biallelically in the choroid plexus, leptomeninges, and perhaps the developing retina . However, human IGF2 is also expressed in the adult, with transcripts arising from an adult-specific promoter . The corresponding region in the mouse Igf2 gene contains two pseudoexons and what appears to be a remnant of this adult-specific promoter—which may explain why Igf2 expression ceases after birth in mice but not in humans .
Almost all known imprinted genes occur in clusters with one or more reciprocally imprinted genes (reviewed in ). The mouse Igf2 gene lies on the distal region of chromosome 7 with the oppositely imprinted, non-coding gene H19. Igf2 and H19 share a set of enhancers that act on either gene, depending on parental origin. In eukaryotic DNA, promoters generally harbor regions dense with CpG dinucleotides, which are targets of methylation. These “CpG islands” are often methylated in inactive promoters. On the paternal chromosome, the H19 promoter region is methylated and inactive; this methylation and expression pattern is passed on when cells divide. Because this inheritance of gene expression patterns is achieved without altering the DNA sequence, it is called epigenetic.
The Igf2 promoter is not methylated on the maternal chromosome, so another mechanism must account for silencing Igf2. Several kilobases (kb) upstream of the H19 promoter is a differentially methylated region (DMR) that, when deleted, reactivates Igf2 on the maternal chromosome . This region, also called the imprinting control region (ICR), was found to harbor binding sites for CCCTC binding factor (CTCF), an insulator protein that demarcates active and inactive chromatin domains (reviewed in ). Methylation of the CG-rich CTCF binding sequence prevents CTCF binding. Thus, on the paternal chromosome, the DMR/ICR is methylated, CTCF is excluded, and the enhancers act on the Igf2 promoter. Conversely, on the maternal chromosome, CTCF forms a chromatin insulator that blocks the enhancers from activating Igf2 (Figure 1A).
Murine Igf2 was the first gene found to be imprinted, and has served as a model of allele-specific gene repression—the most extreme example being X chromosome inactivation, where one X is silenced in each somatic cell of XX female mammals to equalize gene dosage with XY males . Igf2 imprinting and X chromosome inactivation are the most well-studied mechanisms of epigenetic regulation, and the parallels between these mechanisms give insight into the epigenetic alterations that are abundant in cancer.
X chromosome inactivation generally occurs in a random fashion and silences either X; however, in some mammals and in certain tissues of others, the paternal X is always silenced. In either random or imprinted X chromosome inactivation, the X that is destined to be silenced expresses the non-coding Xist RNA, which covers the chromosome and mediates silencing (reviewed in ).
Xist lies in a region called the X inactivation center (XIC) along with another noncoding gene, Tsix, is transcribed antisense to Xist and expressed on the active X chromosome . Not long after CTCF was shown to regulate imprinting at the Igf2/H19 locus, a similar mechanism was found at the Xist/Tsix locus. In a region implicated in controlling both random and imprinted X chromosome inactivation, functional methylation-sensitive CTCF binding sites were identified (Figure 1B). This region was later found to contain developmentally specific enhancers  and to be differentially methylated in vivo . CTCF has since been demonstrated to control imprinting at several other gene domains, and putative binding sites have been discovered in several other imprinted loci . However, not all imprinted genes contain functional CTCF binding sites. It is proposed that another multifunctional transcription factor, yin yang 1 (YY1), functions as a methylation-sensitive insulator that mediates allele-specific gene activation or silencing at some loci. YY1 has been found to control imprinting at the human SNURF-SNRPN locus within the Prader-Willi syndrome and Angelman syndrome locus, and the PEG3, Gnas, and Nespas genes ( and references therein). Interestingly, it was shown recently that YY1 is a cofactor for CTCF in X chromosome inactivation . Because both CTCF and YY1 are ubiquitously expressed, it is possible that tissue- and developmentally-specific imprinting of Igf2 is accomplished through a combination of these factors.
The similarities between Igf2/H19 and Xist/Tsix regulation have additional implications for other regulatory mechanisms that may be aberrant in cancer. The X chromosomes initiate silencing after forming a transient interchromosomal complex (reviewed in ). This pairing phenomenon has also been observed with the Igf2/H19 region, in which CTCF mediates interchromosomal colocalization and induces trans effects on a non-homologous chromosome . Interchromosomal pairing may increase the frequency of mitotic recombination, which can account for both heritable and sporadic mutations . Because CTCF mediates interchromosomal pairing of the IGF2/H19 region, it may very well facilitate such mitotic recombination events. X chromosome inactivation has also drawn attention in the field of cancer research with the recent discovery of X-linked tumor suppressor genes; when mutated, these can lead to hemizygosity in males and skewed X inactivation in females (reviewed in ). One gene, FOXP3, codes for a forkhead family transcription factor that represses the HER-2/ErbB2 oncogene . Interestingly, the forkhead transcription factors are targets of the PI3-kinase pathway, which is activated by IGF signaling (reviewed in ). The other X-linked tumor suppressor, WTX, was frequently inactivated in Wilms’ tumor, a disease also associated with disrupted IGF2 imprinting .
X chromosome inactivation can have other implications for Igf2/H19 regulation as well. There is mounting evidence that non-coding (especially antisense) RNAs regulate allele-specific gene expression (reviewed in ). Multiple sense and antisense transcripts have been detected in the mouse Igf2 5’ region, and the major antisense transcript, Igf2AS, is paternally expressed and noncoding . An antisense message transcribed from a homologous region near human IGF2 encodes a putative 273-amino acid protein of unknown function . It remains unclear whether IGFAS regulates IGF2 or H19 imprinting; nonetheless, it may have biological importance. In Wilms’ tumor, IGF2AS is highly expressed and demonstrates sporadic loss of imprinting [42, 43]. As stated before, disrupted IGF2 imprinting is implicated in a number of diseases, and can be attributed to increased gene dosage and subsequent increases in IGF2 signaling, which will be discussed in the following section.
The IGFs signal primarily through the type I IGF receptor (IGF1R), but there is significant crosstalk between the IGF and insulin systems as certain variants of the insulin receptor (IR) have been shown to bind IGFs (Figure 2). The alternatively spliced IR-A isoform, which is expressed predominantly during embryogenesis , binds insulin and IGF2 (but not IGF1) with high affinity . IGF2 can also stimulate insulin-like metabolic responses by binding the classical IR-B isoform; furthermore, functional heterodimers can form between IGF1R and the IR isoforms (reviewed in ). Thus, tissue-specific effects of insulin and the IGFs may be accomplished through differential expression of the receptors and receptor hybrids. Though IGF1R is activated more efficiently by IGF1 , the ability to signal through IR potentially gives IGF2 a broader range of biological functions than IGF1.
IGF2 has high affinity for another receptor, IGF2R, and is its principal ligand (Figure 2). However, IGF2R does not transduce a signal; rather, it serves mainly to limit IGF2 bioavailability by targeting IGF2 for degradation (reviewed in ). Interestingly, the IGF2R gene is also imprinted—but it is maternally expressed (reviewed in ).
Whereas insulin circulates freely in the bloodstream, the IGFs are found in complexes with the IGF binding proteins (IGFBPs). Six different IGFBPs have been identified, and each binds the IGFs with significantly higher affinity than IGF1R. The expression patterns of the various IGFBPs differ both spatially and temporally, and they have distinct activities (Table 2). Thus, IGFBPs are important modulators of IGF action, availability, and tissue distribution (reviewed in ). Differential expression of IGFBPs, as well as differential expression of IGF receptors and receptor hybrids, may govern the cell- and tissue-specific actions of IGFs.
IGF1 and IGF2 are well known for their mitogenic activities. Almost all cell types express IGF1R, so the IGFs can stimulate growth and differentiation in many tissues (reviewed in ). Upon binding to IGF1R, the IGFs trigger the receptor tyrosine kinase activity, which leads to phosphorylation of itself and its major substrate, the insulin receptor substrate 1 (IRS-1). Phosphorylated IRS-1 can activate the Ras/Raf/MAPK and PI3-kinase/Akt cascades, and depending on the cell type, stimulate proliferation, differentiation, or both (reviewed in ). PI3-kinase activation can lead to anti-apoptotic signals, and components of this pathway are frequently amplified or mutated in cancers (reviewed in ).
The role of IGF2 in muscle development has been studied extensively. IGF2 is upregulated early in MyoD-induced in myocyte differentiation, and signals in an autocrine loop to activate PI3-kinase and Akt . IGF2 inhibition leads to reduced expression of MyoD target genes, which suggests that IGF2 is essential for amplifying and maintaining MyoD efficacy . IGF2 is also essential in bone development, where it promotes proliferation and differentiation of bone cells. Down-regulation of IGF2 most likely accounts for the decrease in bone mass observed with cortisol use . Thus, IGF2 has great therapeutic potential in wound and fracture healing.
Growth in the developing mouse embryo is largely governed by IGF2. When a targeted Igf2 deletion is transmitted paternally, mouse embryos inherit only the inactive maternal allele and are born runted . Conversely, IGF2 overexpression, achieved by disrupting the inhibitory Igf2r , by deleting H19 , or by transactivating Igf2 , leads to fetal overgrowth and malformations with characteristics that resemble Beckwith-Wiedemann syndrome (BWS, discussed below).
IGF2 is regulated precisely to ensure monoallelic expression in most tissues , which emphasizes the importance of gene dosage. Normal development requires accurate expression, and many disorders can be attributed to an abnormally high dose of IGF2 caused by loss of imprinting (LOI). BWS is one such disease, characterized by fetal and neonatal overgrowth, and is often accompanied by an increased risk of childhood cancers (reviewed in ). BWS patients almost always have mutations in the chromosome 11p15.5 region, a large cluster of imprinted genes that includes IGF2 and p57KIP2 (Figure 3). Most of these mutations affect imprinting; quite often, biallelic IGF2 expression and H19 methylation are observed (reviewed in ). BWS usually occurs sporadically, but in rare familial cases IGF2 LOI may be caused by deletions of the CTCF binding sites in the maternal IGF2/H19 ICR [59, 60].
Disrupted imprinting is perhaps the most common observation in cancer (reviewed in ), and IGF2 overexpression is a recurring theme. Wilms’ tumor, a childhood cancer of the kidney, is often associated with defects in the WT1 gene, which encodes a transcriptional repressor of IGF2 . Wilms’ tumor is also associated with mutations in the 11p15.5 region that affect IGF2 imprinting: altered IGF2 expression accounts for nearly 50% of all cases of Wilms’ tumor, and IGF2 LOI is found in the vast majority (90%) of pathological cases . IGF2 LOI has also been observed in many other cancers. Both benign and malignant breast lesions show biallelic IGF2 expression, and altered imprinting of IGF2 is has been identified in hepatoblastoma, lung cancer, cervical carcinoma, rhabdomyosarcoma, choriocarcinoma, and testicular cancer ( and references therein).
The epigenetic mutations associated with cancer, such as aberrant methylation or LOI, may magnify the effects of genetic mutations or even have causal roles. In either case, epigenetic changes have potential value for assessing disease risk and prognosis. In a mouse model of intestinal cancer, where the adenomatous polyposis coli (Apc) gene is mutated, supplementary Igf2 LOI increases the incidence of intestinal hyperplasia. The clinical relevance of this is corroborated by the fact that patients with IGF2 LOI also have an increased risk of developing colorectal cancer . Alterations involving CTCF may also be informative. Elevated CTCF expression levels have been reported in breast cancer, where it is postulated to have anti-apoptotic actions . Gene activation by a CTCF homolog is observed in lung cancer [67, 68], and methylation changes in CTCF binding sites have also been reported in osteosarcoma . Because epigenetic changes such as LOI and demethylation are among the earliest evens in cancer progression (reviewed in ), assays for epigenetic biomarkers may allow for early detection, prevention, and treatment of cancer.
Igf2 overexpression sometimes occurs without apparent LOI or gene duplication. Other factors, such as sonic hedgehog (Shh), can also transcriptionally activate Igf2. Shh is a developmental morphogen involved with patterning and organ specification, and its signaling pathway is mutated in several diseases (reviewed in ). The Shh cascade culminates in the activation of Gli, a transcription factor that induces several target genes (Figure 4).
Shh has been demonstrated to upregulate Igf2 both in vitro and in vivo. When mouse mesenchymal cells are treated with Shh or transfected with Gli1, Igf2 mRNA is upregulated . A Ptc-deficient mutation in mice, which results in constitutive Gli activation, increases IGF2 protein levels and also the formation of medulloblastomas and rhabdomyosarcomas . It is not entirely clear how Shh induces Igf2 expression. Though putative Gli-binding sites have been identified in the mouse Igf2 promoter , it is not known whether these sites are functional, or if they exist in the human VEGF promoter. However, functional Gli sites have been documented in the human IGFBP-6 promoter . IGFBP-6 specifically binds IGF2 and is generally thought to have anti-proliferative properties. Nonetheless, like most of the IGFBPs (Table 2), IGFBP-6 can have contrasting activities, and has also been shown to be anti-apoptotic and tumorigenic (reviewed in ).
IGF2 itself may provide an oncogenic signal in some systems, such as the mouse mammary gland, where transgenic Igf2 overexpression induces adenocarcinomas . In mouse models of rhabdomyosarcoma and medulloblastoma, Igf2 alone is insufficient to generate tumors; however, it can enhance the tumorigenic potential of Shh [73, 77]. Interestingly, tumors often overexpress the IR-A variant, which binds IGF2 with high affinity; thus, concomitant IGF2 and IR-A overexpression can potentially generate an autoproliferative loop . Taken together, these observations substantiate the hypothesis that IGF2 can supply the “second hit” necessary for oncogene-induced tumors .
Angiogenesis, or blood vessel growth, is another critical element of tumor progression that may involve IGF2. Oxygen, nutrients, and metabolic wastes can simply diffuse in and out of small tumors, but growth beyond a critical size (1 mm3) requires a vascular network (reviewed in ). Areas of hypoxia within tumors induce the expression of angiogenic factors, which prompt an influx of vessels from surrounding tissues. Neovascularization also facilitates the spread of cancer cells to other tissues; thus, there is a correlation between high metastatic potential and tumor vascularity (reviewed in ).
Vascular endothelial growth factor (VEGF) has a central function in both normal and pathological neovascularization, and its expression is upregulated in tumors (reviewed in ). Hypoxia-inducible factors (HIFs) are principle mediators of VEGF upregulation, though VEGF mRNA levels are also increased via message stabilization . Transcriptional regulation also occurs through other cis elements in the VEGF promoter, and can be instigated by various growth factors, hormones, and oncogenes (reviewed in ).
Though studies of the IGFs in vascular development are limited, IGF2 may participate in angiogenesis through its ability to upregulate VEGF. In hepatocellular carcinomas cells, hypoxia-induced VEGF expression is increased by IGF2, which is itself upregulated by HIFs . Other studies have suggested that IGF2 signaling upregulates VEGF in part by increasing HIF levels [85, 86]. Because reciprocal upregulation of IGF2 and HIF has been demonstrated , they may act in synergy to induce VEGF expression. Though the mechanisms remain unclear, the ability to induce VEGF accentuates the importance of IGF2 in tumor development.
IGF2 may also be involved in the pathological neovascularization that characterizes proliferative diabetic retinopathy (PDR) and retinopathy of prematurity (ROP). Several studies have implicated IGF1 in retinopathy (reviewed in ), but IGF2 has been largely overlooked—despite reports of 10- to 30-fold more IGF2 in the vitreous of diabetic patients than IGF1 ( and references therein). A recent study showed that IGFBP-3 suppressed retinal neovascularization irrespective of IGF1 levels , which supported the long-standing notion that IGFBPs can act independently of IGF signaling through IGF1R (reviewed in ). However, the potential contribution of IGF2 needs to be examined—specifically, its interactions with other receptors (such as IR-A variant) and whether these interactions are subject to IGFBP regulation. Clearly, the likely role of IGF2 in retinopathy calls for further exploration.
In recent years, Shh has been identified as an angiogenic factor. Studies in zebrafish reveal vascular defects in Shh-mutant embryos [91, 92], and place Shh upstream of VEGF signaling during arterial differentiation . The cascades induced by Shh also appear to regulate vessel formation in mammals. In the mouse embryo, indian hedgehog (Ihh), a Shh homolog, has been suggested to be critical for early vasculogenesis [94, 95]. In Shh-deficient mice, the developing lung is poorly vascularized ; conversely, Shh overexpression in the neural tube results in hypervascularization . Shh can also induce angiogenic factors (including VEGF) and promote neovascularization in adult mice . Thus, vessel formation may depend on the ability of Shh to induce VEGF. Though the exact mechanism remains elusive, it may very well involve IGF2, which is a downstream target of the Shh cascade  and has a demonstrated ability to synergize with Shh [73, 77]. Moreover, IGF2 has also been shown to induce VEGF [84–86]. Thus, IGF2 may mediate the angiogenic effects of Shh, and provide the critical link between Shh and VEGF.
Though interest in IGF2 has been somewhat skewed towards the study of gene regulation and imprinting, it is likely to attract attention from other fields as studies implicate IGF2 in an increasing number of diseases. The complexity of IGF2 regulation indicates that overexpression can occur at multiple levels. Since IGF2 is pivotal in many developmental and pathological processes, its multifaceted regulation presents a number of potential therapeutic targets.
Because imprinting defects are now recognized as common in the pathogenesis of cancer, the mechanisms surrounding IGF2 imprinting are likely to gain interest as well. Perhaps the most thoroughly studied of known imprinted genes, IGF2 has yielded valuable insight into other epigenetic gene regulatory mechanisms—namely X chromosome inactivation, which also gained significance with the discovery of X-linked tumor suppressors (reviewed in ). These studies highlight the multifactorial nature of cancer, in which IGF2 may have a pivotal role. More importantly, they suggest that imprinting and X inactivation are not just interesting epigenetic phenomena, but have considerable functional relevance.
Wendy Chao recently received her Ph.D. in genetics from Harvard Medical School, where her research focused on gene regulation and epigenetics. Her interests in community education have led to credits on ABC News, The Learning Channel, and in U.S. Department of Justice publications. She has been involved in science mentoring partnerships with local schools, and is on the board of advisors of the Foundation for Art and Healing in Boston, Massachusetts. Wendy is also a freelance science writer and Contributing Editor of The Scientist magazine.
Patricia D'Amore received her Ph.D. in Biology from Boston University in 1977. She was a postdoctoral fellow at Johns Hopkins Medical School before moving to the Children's Hospital in Boston where she is currently a Research Associate in Surgery. In 1998, she became Professor of Ophthalmology (Pathology) at Harvard Medical School and a Senior Scientist at the Schepens Eye Research Institute. She is the recipient of numerous awards including the Jules & Doris Stein Research to Prevent Blindness, Senior Scientific Investigator Award, Cogan Award, and the A. Clifford Barger Excellence in Mentoring Award. She is currently the Associate Director of Research and the Ankeny Scholar of Retinal Molecular Biology at Schepens. Dr. D'Amore's research focuses on understanding the mechanism of vascular growth and development. She is the author of 112 publications.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Wendy Chao, Schepens Eye Research Institute 20 Staniford Street, Boston, MA 02114, TEL: 617.912.2558, FAX: 617.912.0128, Email: wendy.chao/at/schepens.harvard.edu.
Patricia A. D’Amore, Departments of Pathology and Ophthalmology, Harvard Medical School Schepens Eye Research Institute 20 Staniford Street Boston, MA 02114, TEL: 617.912.2559, FAX: 617.912.0128. Email: patricia.damore/at/schepens.harvard.edu.