Insulin-Like Growth Factor Factor Binding Protein-2 is a Novel Mediator of p53 Inhibition of Insulin-Like Growth Factor Signaling

Abstract

The p53 tumor suppressor induces cellular growth arrest and apoptosis in response to DNA damage by transcriptionally activating or repressing target genes and also through protein-protein interactions and direct mitochondrial activities. In 1995, insulin-like growth factor binding protein (IGFBP)-3 was identified as one of the genes transcriptionally activated by p53. IGFBP-3 is one of six closely related IGFBP’s, with additional IGFBP-related proteins belonging to the IGFBP superfamily. Here we show that IGFBP-2 is also a p53 target. Like IGFBP-3, IGFBP-2 secretion is reduced when p53^+/+ lung cancer cells are transfected with human papillomavirus E6, which targets p53 for degradation. IGFBP-2 mRNA is induced by irradiation in vivo in a p53-dependent manner. p53 protein binds IGFBP-2 intronic sequences in an electrophoretic mobility shift assay, and activates transcription in a luciferase assay. Loss of IGFBP-2 inhibits the ability of p53 to inhibit the activation of extracellular signal-regulated kinase (ERK)1 by IGF-I. Thus, p53 effects on the IGF axis are more complex than previously appreciated, and overall transform the axis from IGF-mediated mitogenesis to growth inhibition and apoptosis. This has significant implications for how growth hormone and IGF-I can induce growth without also inducing cancer.

INTRODUCTION

p53’s importance as a tumor suppressor has been established through three major lines of evidence: p53 is the most frequently mutated gene in human cancers (and those cancers that do not harbor mutations in p53 itself frequently involve mutations in regulators of p53 function),¹ germline p53 mutations lead to the heritable Li-Fraumeni cancer syndrome,² and various mouse models have demonstrated how perturbations in p53 function can affect carcinogenesis and tumor behavior.³ Normally p53 is a constitutively repressed protein that is stabilized and activated by DNA damage, oncogenic stress and hypoxia to induce cell cycle arrest and apoptosis.^4-6 Mechanisms of p53 action include protein-protein interactions as well as transactivation and repression of target genes.⁷ Activated p53 tetramerizes and binds sequence-specific sites within multiple genes through its central DNA-binding domain, where the majority of p53 mutations occur in human cancers.¹

Among the known p53 transcriptional targets is the insulin-like growth factor binding protein (IGFBP)-3. Two p53-binding sites, named Box A and Box B, were identified in IGFBP-3’s first and second introns, respectively.⁸ Another p53-responsive element was found later within the promoter region 70 bp upstream of the TATA box,⁹ and in vitro IGFBP-3 promoter hypermethylation, as observed in human hepatocellular carcinomas, suppressed p53 binding and p53-induced IGFBP-3 expression in a human hepatoblastoma cell line.¹⁰ As its name implies, IGFBP-3 was discovered as the principal binding protein of IGF-I in the circulation, and its functions were initially understood as related to IGF; IGFBP-3 prolongs the half-life of circulating IGF-I, inhibits IGF-I transfer from the circulation to tissue sites of action and locally modulates the amount of free IGF available to interact with the type 1 IGF receptor (IGF1R).¹¹ Thus, IGFBP-3’s role in growth arrest and apoptosis was attributed to its mostly inhibitory sequestration of IGF from the IGF1R. Newer evidence supports the concept that IGFBP-3 can also mediate growth arrest and apoptosis via IGF-independent mechanisms.^11,12

Combining IGFBP-3’s role as the p53 target within the IGF axis together with its pluralistic means of affecting cell growth suggests that IGFBP-3 should be important to apoptosis. As shown by various p53 point mutants that selectively lost the ability to transactivate some but not all of the p53 targets, ability of p53 to transactivate IGFBP-3was needed for full induction of apoptosis.¹² Furthermore, addition of IGF-I or specific IGFBP-3 inhibitors blocked p53-induced apoptosis during serum starvation.¹³ However, unlike p53^-/- mice, IGFBP-3^-/- mice do not show increased spontaneous cancer formation. This raises the question of redundancy of IGFBP-3 with other p53 targets. Indeed, the null phenotype of the IGFBP-3^-/- mice, together with their compensatory increases in circulating levels of some of the other IGFBP’s, raises the question of redundancy within the IGFBP family.¹⁴ We therefore sought to examine whether p53 regulates another IGFBP.

In the present studies we have identified IGFBP-2 as a direct target for p53-mediated transcriptional activation. IGFBP-2 was found to be induced following DNA damage in vivo in a tissue-specific manner. To determine the impact of IGFBP-2 regulation by p53, we found that silencing of IGFBP-2 blocked the p53-mediated suppression of phospho-ERK expression in cells exposed to IGF-I. The present studies reveal IGFBP-2 as a novel target of p53 and suggest a role for IGFBP-2 in mediating p53’s inhibition of IGF-I signaling.

MATERIALS AND METHODS

Cell lines

As previously reported,^13,15 two sister cell lines created from the p53^+/+ human lung carcinoma cell line (H460) differ in their p53 content; H460 cells stably transfected with a plasmid containing the gene for E6 (H460-E6) contain far less p53 than H460 cells stably transfected with the empty plasmid as control (H460-neo). H460 cells were maintained at all times in RPMI 1640 medium (Life Technologies, Inc.-Gibco BRL, Grand Island, NY) with 500 μg/ml G418 (Mediatech, Inc., Herndon, VA) for continued transfectant selection pressure. PC-3 cells (American Type Culture Collection, Manassas, VA) were cultured in F12K medium (Mediatech) as recommended by the ATCC.

IGFBP Western ligand blot and immunoblot

H460 cells were plated at equal densities, serum-starved and the conditioned medium collected at times shown. For detection of any and all IGFBPs in the samples, Western Ligand Blot probing with ¹²⁵I-IGF-I and ¹²⁵I-IGF-II was performed as described.¹⁶ For confirmation of the 27 kDa band as IGFBP-2, IGFBP-2 immunoblotting with high affinity purified anti-human IGFBP-2 antibody (Upstate Biotechnology; Lake Placid, NY) was performed as described.¹⁶ Recombinant human IGFBP-3 (Upstate Biotechnology; Lake Placid, NY) was analyzed in the last lane as control for antibody specificity.

Mice and treatments

Healthy six to seven week-old female mice, two p53^-/- and two p53^+/+ (Jackson Laboratories, Bay Harbor, ME), each received 5 Gy total body irradiation from a ¹³⁷Cesium γ-source at a dose rate of 1.532 Gy/min, or a single intraperitoneal injection of sterile 1X PBS. The eight mice were euthanized at 6 and 24 h using an approved protocol (University of Pennsylvania Institutional Animal Care and Use Committee), which followed recommendations of the Panel on Euthanasia of the American Veterinary Medical Association. Tissues were harvested, snap frozen in liquid nitrogen, and kept at -80°C until used for the Taqman RT-PCR experiments.

Taqman real time quantitative RT-PCR

Total cellular RNA was isolated by standard procedures from radiation-sensitive tissues, and 1 μg was reverse transcribed and amplified using TaqMan Reverse Transcription Reagents according to the manufacturer’s protocol (Applied Biosystems, Foster, CA). Parallel samples were similarly treated, but without adding reverse transcriptase (-RT), to provide a control for basal detection of genomic DNA by the probe in the RT-PCR reaction. Using Primer Express Version 1.0 (Applied Biosystems), the following primer and probe sequences were designed for IGFBP-2:IGFBP-2+ 5′-AAAAGAGACGCGTGGGCA-3′, IGFBP-2(-) 5′-CCCTCAGAGTGGTCGTCATCA-3′, and probe 6FAM-CACCCCACAGCAGGTTGCAGACA. GAPDH primers and VIC-labeled probe were obtained from Applied Biosystems.

RT-PCR reactions were carried out in 96 well plates using the ABI PRISM 7700 Sequence Detection System (Applied Biosystems), as previously described for other p53 target genes.^17,18 Each experimental condition was run in quadruplicate replications, with one of each condition’s -RT as control. Three reactions of Taqman master mix without any template served as negative control for the RT-PCR reaction. The standard curve method was used to quantify the amount of IGFBP-2 in each reaction according to manufacturer’s protocol (Applied Biosystems). Serial dilutions of mouse prostate total RNA, obtained from Clontech Laboratories, Inc. (Palo Alto, CA), underwent RT-PCR, and triplicate replications were used to generate the standard curve. All IGFBP-2 measurements were normalized to GAPDH mRNA concentrations to control for any loading differences.

Electrophoretic mobility shift assay (EMSA)

Double-stranded oligomers corresponding to the putative hIGFBP-2 sites in Table 1 were synthesized (Integrated DNA Technologies, Inc., Coralville, IA) and labeled with ³²P (purchased through the Environmental Health and Radiation Safety Program of the University of Pennsylvania). The labeled oligomers were purified via 12% non-denaturing TAE PAGE, and their radioactivities were normalized. p53 protein was obtained, as previously reported, by preparing nuclear extracts of p53^-/- Saos-2 cells infected with Ad-p53; nuclear extracts from sister cells infected with Ad-lacZ served as negative controls.¹⁹ The mouse anti-human p53 monoclonal antibody pAb421 (Ab1, Oncogene Science, Cambridge, MA) was added to all conditions to activate sequence-specific DNA binding by p53, and EMSA was performed as previously reported.²⁰

Table 1.

Potential p53 binding sites in hIGFBP-2

start nt #	first 1/2-site sequence	% match	second 1/2-site sequence	% match	# nts between 1/2 sites	site #
PROMOTER
-1734	GGGCATGTTT	100	GTACATGTAA	70	2	5
-413	AAACTAGGAG	70	AAACAAGGCT	90	0	6
INTRON 1
4079	GGGCATGGTG	80	GAGCATGCCT	100	0	2
10292	AGGCATGCAC	90	CACCATGCCT	80	0	3
13445	AGGCATGTGC	90	CACCATGCCC	80	0	4
21067	GAGCTTGCTG	90	GGGCTTGTCT	100	0	1
INTRON 2
998	GAGCTTGCGT	90	GCGCGTGCTT	80	3	7
1011	GCGCGTGCTT	80	GGGCCTGCTG	80	3	8

Sequences within the IGFBP-2 gene were evaluated by their homology to the p53 consensus sequence²⁵ (% match) and ranked in descending order of likelihood as a possible p53 binding site. nt, nucleotide.

Luciferase assay for p53-inducibility

Double stranded oligomers corresponding to each of the putative p53 binding sites identified by the EMSA experiments were individually ligated into the luciferase reporter, pGL3-promoter vector (Promega Corporation, Madison, WI). The reporter was contransfected with different concentrations of pCEP4-p53 into p53^-/- PC-3 cells; empty pCEP4 was added to equalize the total DNA transfected to 2 μg.^20,21 Following 24 h incubation, cells were lysed, luciferin (Promega Corporation) added, and luciferase activity measured by a luminometer (3010 Moonlight Luminometer, Pharmingen, San Diego, CA). Empty pGL3 was run as negative control, and PG13-Luc, containing 13 copies of the p53 consensus sequence, served as positive control.^20,21

IGFBP-2 knockdown

To knockdown the expression of IGFBP-2 in PC-3 cells, a lentivirus-based shRNA system (pll3.7 and helper plasmids; kindly provided by Dr. Luk van Parijis, Massachusetts Institute of Technology, Boston, MA) was used to establish stable cell sublines: PC-3/BP-2i, where the gene for green fluorescent protein (GFP) was introduced together with the IGFBP-2 inhibiting sequence, and PC-3/vec, where only GFP was introduced. The specific targeting sequence for IGFBP-2, GGAGGCCTGGTGGAGAACC, was designed using the Whitehead Institute siRNA selection program (http://jura.wi.mit.edu/bioc/siRNAext/). Infected cells were selected by their fluroescence, and the degree of IGFBP-2 expression was determined by Western immunoblotting as described above. Immunoblotting for tubulin (mouse anti-human-α-tubulin-antibody, Santa Cruz, Santa Cruz, CA) was performed as loading control.

Effects of p53 on IGF signaling

p53^-/- PC-3 cells were seeded, 2.5 × 10⁴ per well, in 96 well plates and allowed to reach 80% confluence. Culture medium was changed to serum-free, and adenoviral preparations with p53 and GFP expression components (Ad-p53) or GFP only (Ad-GFP) were added at an MOI of 1300; these adenovirus constructs had been created previously in our laboratory as reported.²² Following 24 h incubation, infecting medium was replaced with fresh serum-free medium and hIGF-I (Sigma, St Louis, MO) was added at 500 ng/ml for 15 min. Cells were fixed in 4% formaldehyde (Fisher Scientific, Boston, MA), and the amounts of phosphorylated and total ERK were quantified by ELISA per manufacturer’s instructions (Cellular Activation of Signaling ELISA (CASE)™ Kits for ERK1/2 T202/Y204, Superarray Bioscience Corporation, Frederick, MD). To control for cell number, replicates were further stained with the kit’s cell staining solution, as directed in the manual.

RESULTS

Screening for p53 effects on the IGFBPs

A useful model for exploring p53 effects takes advantage of E6, a protein produced by tumor-associated human papillomaviruses (HPV types 16 and 18) that binds p53 and targets it for ubiquitin-dependent degradation.²³ E6 also represses p53-dependent gene activation, independently of inducing p53 degradation, by inhibiting the acetylation of p53.²⁴ As previously reported,^13,15 two sister cell lines created from the human lung carcinoma cell line (H460) that contains fully functional wild-type p53 differ in their p53 content:H460 cells stably transfected with a plasmid containing the gene for E6 (H460-E6) contain far less p53 than H460 cells stably transfected with the empty plasmid as control (H460-neo). H460-E6 cells were compared to H460-neo cells by Western ligand blotting with radiolabeled IGF-I and IGF-II, the traditional method for detecting IGFBPs.¹⁶ As observed in Figure1A, transfection with E6 significantly decreased secretion of two IGFBPs. The 44-kDa band corresponds to IGFBP-3, as expected.¹³ The molecular weight of the 31-kDa band suggests that it is IGFBP-2. For confirmation, the same membranes were washed and reprobed by immunoblotting with anti-IGFBP-2-antibody (Fig.1B). The lowest band in Figure 1A corresponds to IGFBP-1, whose levels were not affected by the p53 status of the cell.

Figure 1.

IGFBP-2 is a novel p53 transcriptional target. A and B, Transfection with E6 reduces IGFBP-2, as well as IGFBP-3, secretion by H460 cells. H460 cells were stably transfected with E6 (E6) to cause p53 degradation, or with empty plasmid (neo) as control. Conditioned media were harvested at indicated times of serum starvation. Western ligand blot (A) shows the levels of secreted IGFBPs. Left-most lane shows the molecular weight marker, and right-most lane is 1 μg recombinant human IGFBP-3 as positive control. IGFBP-1 levels are unaffected by E6, and serve as loading controls. Immunoblotting with anti-IGFBP-2-antibody (B) confirms that the second IGFBP decreased by E6 in A is IGFBP-2. The recombinant human IGFBP-3 now serves as negative control for antibody specificity. (C) In vivo IGFBP-2 induction by irradiation (thymus) is p53-dependent. p53^+/+ and p53^-/- mice were treated, two each, with 5 Gy irradiation or PBS for control (ctl), and euthanized 6 or 24 h later. IGFBP-2 mRNA was measured by Taqman quantitative RT-PCR, normalized to GAPDH, and shown as fold induction relative to untreated wild type mice. Each column represents the mean ± standard deviation of quadruplicate measurements. (D) DNA binding activity of p53 protein to four hIGFBP-2 sequences. Electrophoretic mobility shift assay (EMSA) was performed with ³²P-labeled oligomers (numbers 1 through 4) corresponding to hIGFBP-2 sites with the highest degree of homology to the p53 consensus sequence (numbered 1 through 4 in Table 1, and indicated in the hIGFBP-2 gene map above). hIGFBP-2 consists of 4 exons (indicated by black squares), totaling 1.4 kb, and 3 introns, whose lengths are indicated beneath each bracket.⁵⁰ Each oligomer was incubated with no protein (probe), or cell lysate from p53^-/- Saos-2 cells infected with Ad-lacZ (negative control) or Ad-p53 (p53). Samples were resolved on a 4% polyacrylamide gel. The three arrows indicate, from top to bottom, the p53-retarded band, a nonspecific band and the free probe. (E)Transcriptional activation by p53 of site 1 ligated in a luciferase reporter construct (pGL3). The reporter, containing site 1 (pGL3-site 1), no insert (pGL3; negative control) or 13 copies of the p53 consensus sequence (PG13-Luc; positive control), was cotransfected with increasing concentrations of a p53-expressing plasmid into p53^-/- cells. Luciferase activity was measured following 24 h incubation, and shown as fold induction relative to the no pCEP4-p53 condition.

In vivo assay for p53-dependent induction of IGFBP-2

E6 may reduce IGFBP-2 secretion through pluripotent effects that do not involve p53. To test the hypothesis that it is p53-mediated, p53^-/- versus p53^+/+ mice were compared. Mice were treated with 5Gy irradiation or with intraperitoneal injection of sterile PBS as control, and total mRNA was isolated from organs known to be sensitive to radiation-induced apoptosis. As shown for the thymus (Fig. 1C), IGFBP-2 mRNA levels increased 14-fold at 6 h and 38-fold at 24 h following irradiation in the wild type mice; irradiation failed to induce IGFBP-2 in the p53^-/- mice. For reference, similar experimental conditions at 24 h led to about a twenty-fold induction of known p53 targets such as IGFBP-3, Bax and KILLER/DR5.^17,18 This effect was tissue-specific. IGFBP-2 mRNA concentration doubled in the small intestine following irradiation in the p53^+/+ but not p53^-/- mice (data not shown). Although statistically significant, the magnitude of this effect was far smaller than in the thymus. Irradiation did not induce IGFBP-2 mRNA in the spleen.

p53-binding sites within the IGFBP-2 gene

p53 transcriptionally activates its target genes by sequence-specific binding to intronic and promoter sites.²⁵ Human IGFBP-2 cDNA, but not the entire gene, has been cloned and fully sequenced (Genebank Accession number NM000597).²⁶ Human IGFBP-2 cDNA sequences were used to BLAST search the human genome database to reveal the gene’s intronic and promoter sequences. Incidentally, the BLAST search also produced an 84% identity hit with human IGFBP-5, which was further along the same contig. This attests to the high degree of homology among the conserved first exons of the IGFBPs. Examination of the human IGFBP-2 intronic and promoter sequences revealed eight sites with high homology to the p53 consensus sequence²⁵ (Table 1). Electrophoretic mobility shift assay was performed to determine if p53 binds each of the putative sites identified in hIGFBP-2. Retardation of probe migration with p53, but not lacZ adenovirus infected cell lysates, occurred for sites 1 through 4 (Fig. 1D); similar shifts were not observed for sites 5 though 8.

Transcriptional activation by p53

More than binding, the ultimate question is whether p53 can activate transcription from the putative sites. Double stranded oligomers corresponding to sites 1 through 4 were individually ligated into the luciferase reporter, pGL3-promoter vector, and cotransfected with different concentrations of pCEP4-p53 into p53^-/- PC-3 cells. Empty pGL3 vector served as negative control, and PG13-Luc (pGL3 containing 13 copies of the p53 consensus sequence)^20,21 as positive control. Increasing concentrations of p53 increased luciferase activity 72 to 85-fold for site 1, which was comparable to the 81-fold induction seen for PG13-Luc (Fig. 1E). These results were replicated with two different pGL3/site 1 constructs. Similar transactivation of luciferase activity was not observed for sites 2 though 4.

Functional significance of IGFBP-2 induction by p53

With so many transcriptional targets of p53, we sought to determine the functional significance of IGFBP-2 as a mediator of p53’s effects on cell signaling. Successful shRNA knock-down of IGFBP-2 expression in PC-3 cells was confirmed by IGFBP-2 immunoblot (Fig. 2A). Addition of hIGF-I to PC-3 cells induced a significant increase in phosphorylated, i.e., activated, ERK (Fig. 2B). Similar increase in activated Akt could not be demonstrated (data not shown), likely due to the already high level of Akt from the homozygous deletion of PTEN in PC-3 cells.²⁷ Prior infection with p53-expressing adenovirus prevented the IGF-I-induced ERK activation in PC-3 cells, but not when IGFBP-2 expression was knocked down (Fig. 2B).

Figure 2.

IGFBP-2 is a mediator of p53 inhibition of ERK activation by IGF-I. (A) Western immunoblot shows successful silencing of IGFBP-2 expression in PC-3 cells infected with a lentivirus-based shRNA construct (PC-3/BP-2i) relative to uninfected PC-3 cells and cells infected with empty vector (PC-3/vec). α-tubulin immunoblot is shown below as a loading control. (B) Ratio of phosphorylated to total ERK, normalized to cell number, was measured by ELISA after 15 min ± IGF-I stimulation. The cells had been pretreated with 24 h infection with adenoviral p53 (Ad-P53) or Ad-GFP as control. PC-3/vec cells express IGFBP-2 while PC-3/BP-2i cells do not, as shown in (A). Each column represents the mean ± SEM of six measurements, and differences compared by unpaired t-test.

DISCUSSION

Our results are the first to identify IGFBP-2 as a direct transcriptional target of p53, and to demonstrate the importance of IGFBP-2 in mediating p53’s inhibition of IGF signaling, at least in cell culture. Like IGFBP-3, IGFBP-2 secretion is decreased by E6, IGFBP-2 mRNA is induced by irradiation in vivo in a p53-dependent and tissue-specific manner, and p53 binds and transactivates transcription from at least one site within the first intron of IGFBP-2. IGFBP-2 and IGFBP-3 both belong to the IGFBP superfamily, which consists of six high-affinity IGFBPs and at least ten lower-affinity IGF binders, termed IGFBP related proteins (IGFBP-rPs).²⁸ All IGFBPs and IGFBP-rPs share a highly conserved cysteine-rich amino-terminal domain that binds IGF; the IGFBPs also share a conserved cysteine-rich carboxy-terminal that contributes to their unique high-affinity IGF binding. In the genes for all these proteins, the highly conserved cysteine-rich amino-terminal domain is encoded by a single exon. The modular nature of these genes has led to the suggestion that they have all evolved by exon shuffling from a common ancestral gene.²⁸

The modular gene structure and functional studies have led to our understanding of the IGFBP superfamily as a whole, though much remains to be elucidated regarding the particularities of the individual proteins. Because the binding affinity of IGF for the IGFBPs exceeds that for IGF1R, the IGFBPs competitively inhibit IGF signaling and are therefore generally growth inhibitory. However, disruption of the IGFBP tertiary structure by a reduction of disulfide bonds (as in the IGFBP-rPs) or by proteolysis significantly lowers their affinity for IGF. Thus, the IGFBPs are the primary regulators of IGF action by modulating the amount of bioavailable IGF for binding to and activating IGF1R. Less is understood about their IGF-independent actions, which are presumably conferred by their nonhomologous domains. Additional individuality derives from the different ontogenetic expression profiles and varied regulators of the different IGFBPs. For example, IGFBP-3 and IGFBP-5 are the only IGFBPs that contain nuclear localization signals,²⁹ and while IGFBP-3 is the principal circulating IGFBP, IGFBP-2 concentrations are highest in seminal plasma and cerebrospinal fluid.³⁰ As further examples, circulating levels of IGFBP-3 rise during childhood to peak during puberty and steadily decline thereafter, whereas IGFBP-2 levels are highest in infancy and old age. Hepatic expression of IGFBP-3 is mainly regulated by growth hormone while insulin is the principal regulator of hepatic IGFBP-1 expression.

Within this general model, p53 exerts multiple direct effects that altogether lead to growth inhibition. p53 activates transcription of IGFBP-3 and, as shown by our results, IGFBP-2. p53 was shown to transcriptionally repress, by inhibiting DNA binding of the TATA-box-binding protein (TBP) subunit of the TFIID transcription factor, to the initiator region of the IGF1R gene³¹ and the third promoter (P3) of the IGF2 gene.³² Thus, p53 effects converge on decreased IGF/IGF1R signaling by reducing both IGF bioavailability and IGF1R density. For example, IGF-I-induced tyrosine phosphorylation of IGF1R and IRS-1, two early steps in IGF/IGF1R signaling, were reduced by tetracycline-inducible p53 expression in an osteosarcoma cell line,³³ and over expression of wild type but not mutant p53 in murine hematopoetic cells decreased the number of IGF1Rs and increased cellular sensitivity to apoptosis caused by interleukin-3 withdrawal.³⁴ In contrast, a p53 mutant (p53mt249) was shown to upregulate both ligand (IGF-II) and receptor (IGF1R) in Hep3B human hepatocellular carcinoma cells in vitro, leading to enhanced IGF1R and IRS-1 phosphorylation, thymidine incorporation and cell growth.³⁵

By decreasing IGF/IGF1R signaling, the p53 effects on the IGF axis serve its function as a tumor suppressor. Although current evidence does not support a causal role of IGF in cancer, epidemiologic studies of cancer risk factors, in vivo tumor models and in vitro experiments on altered cellular signaling all suggest that IGF signaling can contribute to cancer progression and aggressiveness.^36,37 IGF/IGF1R signaling proceeds through the receptor tyrosine kinase pathway and the phosphoinositide 3 kinase (PI3-K)/Akt pathway, both of which are commonly altered in the neoplastic process.³⁸ Cross-talk and coordination between the p53 and IGF-I-Akt-TOR pathways have been proposed as means of integrating growth factor signaling, changes in nutrient levels and stress signals into regulation of cell growth, mitogenesis and apoptosis.³⁹ Our paper adds to the current understanding by showing that IGFBP-2 is a novel transcriptional target of p53 and loss of IGFBP-2 prevents p53’s inhibition of IGF-I signaling through the receptor tyrosine kinase pathway (Fig. 3).

Figure 3.

Model of novel p53 effects on the IGF axis. p53 stimulates transcription of IGFBP-2 (lower left corner). IGFBP-2 is secreted and binds IGF-I, thereby preventing IGF1R stimulation through competitive inhibition. IGF1R is an α₂β₂ tyrosine kinase membrane receptor. Ligand binding of IGF1R activates the PI3 kinase/Akt (left) and MAP kinase (right) pathways, leading to cell survival and mitogenesis. IGFBP-2 silencing in vitro prevented p53 inhibition of ERK activation (phosphorylation) by IGF-I signaling.

By reducing IGF while inducing IGFBP-2 and IGFBP-3, p53 may not only dampen IGF/IGF1R signaling, but also lead to enhanced IGF-independent activities of the IGFBPs. Accumulating evidence shows that IGFBP-3 can induce apoptosis in an IGF-independent fashion and suggests IGFBP-3 can serve a protective role against cancer.^12,37 Whether the IGF-independent activities of IGFBP-2 are also cancer protective remain unclear, as very little is known about them. IGF-independent activities of IGFBP-2 have been suggested by its interactions with α5β1-integrin,⁴⁰ its cytosolic uptake,⁴¹ and its nuclear translocation.^42,43 Although IGFBP-2 has been shown to inhibit proliferation in multiple cell lines in IGF-dependent cell culture systems,⁴¹ experimentally created IGFBP-2 overexpression in vitro increased invasiveness of ovarian⁴⁴ and bladder⁴⁵ cancer cell lines, and increased proliferation of Y-1 murine adrenocortical tumor cells⁴⁶ and DU145 prostate cancer cells.⁴⁷ IGFBP-2 expression has been shown to be elevated in multiple tumors (including prostate, colon, adrenocortical, mammary, ovarian, brain, thyroid, and hepatic cancers), to often correlate positively with tumor grade and/or stage and to be increased in the serum of these patients.⁴¹ However, as these are correlations, it is unclear if the IGFBP-2 elevations serve as a marker of disease or actually play a pathogenic role. In vivo, IGFBP-2 overexpression led to reductions in body length and weight gain and brain weight, especially on a transgenic high growth hormone/IGF-I background,⁴⁸ while IGFBP-2 knockout mice had selective increase in liver and decrease in spleen weights.⁴⁹

In summary, we have shown that IGFBP-2 is a novel direct transcriptional target of p53 and that IGFBP-2 is induced by irradiation in vivo in a p53-dependent and tissue specific manner. p53 is known to inhibit IGF signaling by decreasing both receptor (IGF1R) expression and ligand bioavailability (repressing transcription of IGF-II and stimulating transcription of IGFBP-3). Here we provide evidence that p53’s inhibition of IGF-I-induced ERK phosphorylation in vitro is lost when IGFBP-2 expression is silenced. Thus, our work provides yet another example of potentially important links between the p53 and IGF pathways. The p53 network is the primary tumor suppressing mechanism in humans, while the growth hormone/IGF axis is the principal mediator of somatic growth. The intricate balance between these opposing pathways has been implicated in human growth, cancer, aging and longevity.

ABBREVIATIONS

IGF

insulin-like growth factor

IGFBP

insulin like growth factor binding protein

IGF1R

the type 1 IGF receptor

ERK

extracellular signal-regulated kinase

GFP

green fluorescent protein

IGFBP-rPs

IGFBP-related proteins

PI3-K

phosphoinositide 3 kinase