Id2 Mediates Tumor Initiation, Proliferation, and Angiogenesis in Rb Mutant Mice

Generation of Id2-Rb double-mutant mice and animal tissue preparation.

Rb^+/− and Id2^+/− mice, both in a mixed C57B6/129Sv background, were crossed to generate Id2^+/−; Rb^+/− mice that were than crossed with each other to generate Rb^+/− mice with intact Id2 copies or Rb^+/− mice lacking both Id2 alleles. Mice were genotyped by PCR and maintained in compliance with Institutional Animal Care and Use Committee guidelines. For generation of the survival curve, mice were observed daily and sacrificed when moribund, which included the following criteria: excessive loss of weight, inability to move and seizures. The Logrank test was performed to establish statistical significance (StatView version 5).

Histopathology and immunohistochemistry.

Pituitary glands were excised from mice, fixed in 10% neutral formalin, and embedded in paraffin, and 5-μm sections in the horizontal plane were serially cut. Embryos were harvested from timely pregnant mice at E15.5, fixed overnight in 10% formalin, and processed as above. Sagittal sections were prepared for evaluation of the pituitary intermediate lobe. For analysis of early focal tumor lesions, sections of the whole pituitary from Rb^+/− and Id2^−/−; Rb^+/− 3-month-old mice were stained with hematoxylin and eosin (H&E) and the number of foci was scored in each section using a magnification of ×60 to ×100. The number of cells in each independent early tumor focus was determined by counting the number of tumor cells in each focus in the sections in which the tumor focus was present. Statistical significance was determined by using an unpaired t test. The proliferation index in tumor lesions was calculated by counting the percentage of bromodeoxyuridine (BrdU)-positive nuclei in individual tumor lesions, using nine serial sections from each 3-month-old and 4-month-old Rb^+/− and Id2^−/−; Rb^+/− mouse, respectively. The proliferation and mitotic indices in the embryonic pituitary intermediate lobe were calculated by counting the percentage of BrdU-positive nuclei out of the total number of nuclei in the pituitary intermediate lobe by using four serial sections from each pituitary. Significance was tested using an unpaired t test. Antibodies used in immunohistochemical analysis were Id2 (C-20) from Santa Cruz Biotechnology, MASH-1 from Pharmingen, proopiomelanocortin (POMC) from Cortex Biochem, p27^Kip1 from Transduction Laboratories, BrdU from Chemicon, VEGF AF-493-NA from R&D systems, thrombospondin-1 (TSP-1) from Neomarkers, PECAM from BD Pharmingen, and hypoxia-inducible factor 1 alpha subunit from Novus Biological. Immunostaining was performed on formalin-fixed, paraffin-embedded serial sections by standard procedures. Avidin-biotin peroxidase complex techniques were used for primary antibody detection (Vectastain kit; Vector Laboratories). Staining was developed using diaminobenzidine (brown precipitate). Slides were counterstained with hematoxylin. Controls included no primary antibody and/or normal rabbit or mouse serum and tissues from Id2-null mice.

Vascular analysis of pituitary tumors.

Tumor-bearing animals were anesthetized and then immediately infused with 5.0 ml of a curable orange latex injection compound, Microfil MV-122 (Microfil). The compound was allowed to cure at 4°C overnight before the tumors were harvested. The tissue was cleared in increasing concentrations of glycerol, as specified in the manufacturer's protocol. The vascular architecture was visualized with an SZX12 Olympus microscope.

Cell culture.

Neuroblastoma cells were cultured in Dulbecco modified Eagle medium with 10% fetal bovine serum. REF-52 cells were infected with an adenovirus vector expressing Id2 or Ad-green fluorescent protein vector at a multiplicity of infection (MOI) of 100. Cells were harvested at the indicated times and processed for Northern blot and Western blot analyses. To generate SH-N and IMR32 cells expressing ectopic Id2, cells were transfected with pCMV-Id2 and pCMV-Flag-Id2 or the corresponding empty vectors and selected with G418. For differentiation experiments, LAN-1 cells were exposed to 2 μM retinoic acid (Sigma).

siRNA.

LAN-1 cells were transfected with pSuper.retro vectors (OligoEngine) by using Lipofectamine 2000 (Invitrogen). The expression vectors encode a small interfering RNA siRNA expression cassette including the H1 promoter of RNA polymerase III. The targeted Id2 sequences were CATGAACGACTGCTACTCC in pSuper.Retro112 and GGTGAGCAAGATGGAAAT in pSuper.Retro171. After transfection, cells were selected in puromycin (2 μg/ml) for 48 h and subsequently subjected to Western blot analysis or plated for clonogenic assay.

Western blot analysis

Cell lysates were prepared in RIPA buffer containing protease and phosphatase inhibitors, and proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and finally visualized by enhanced chemiluminescence (Amersham-Pharmacia). The antibodies used in this study have been described previously (24, 26).

Northern blot analysis.

Total RNA was isolated using the Trizol reagent (Invitrogen). Then, 20 μg of total RNA from each sample was separated by electrophoresis and transferred to a Nitran Plus membrane (Schleicher & Schuell). Membranes were stained with methylene blue for evaluation of RNA loading and probed with the coding portion of the mouse or human VEGF gene.

Loss of Id2 delays pituitary tumorigenesis in Rb^+/− mice.

To ask whether Id2 contributes to malignant transformation induced by loss of Rb, we crossed Rb^+/− mice with Id2^+/− mice and produced double mutants Id2^+/−; Rb^+/−. These mice were used as founders to generate Rb^+/− and Id2^−/−; Rb^+/− animals. A comparison of the survival curves of Rb^+/− and Id2^−/−; Rb^+/− mice revealed a clear interaction between the two genes. Id2^−/−; Rb^+/− mice showed prolonged survival compared to Rb^+/− mice (median survival of Rb^+/− mice, 276 days versus 334 days for Id2^−/−; Rb^+/− mice; Logrank P value = 0.0001) (Fig. 1A). At the time of death, histological analysis established the presence of pituitary tumors in all animals.

In the course of this study, we observed that the clinical signs of pituitary tumors such as coat hyperpigmentation consequent to production of α-melanocyte-stimulating hormone (α-MSH) by Rb-null tumor cells, were remarkably delayed in Id2-null mice. However, the latency between early clinical symptoms and death was shorter than that in Rb^+/− mice. Regardless of the Rb status, the Id2 mutant mice are smaller than their littermates from the neonatal period and exhibit symptoms of intestinal dysfunction (48, 64). The suboptimal physical conditions of Id2-null mice may reduce their tolerance to the increased intracranial pressure associated with pituitary tumors. Thus, the observed difference in survival between Rb^+/− and Id2^−/−; Rb^+/− mice may underestimate the contribution of Id2 to the development of pituitary cancer. We reasoned that if loss of Id2 delays tumorigenesis in Rb^+/− mice, pituitary tumors should develop later and/or display slower growth in the absence of Id2. We sacrificed Rb^+/− mice with different Id2 backgrounds on postnatal day 200 (P200) and performed gross examination of the pituitary gland. As shown in Fig. 1B, 100% of Rb^+/− mice (nine of nine) but only 14% of Id2^−/−; Rb^+/− animals (one of seven) had grossly enlarged, irregular glands suggestive of established tumors (P = 0.0002, Fisher's exact P value). Three (30%) of ten Id2^+/−; Rb^+/− mice had normal pituitaries. Histologic analysis from age-matched animals showed extensive tumor growth with entirely subverted pituitary glands in the Rb^+/− genotype. Conversely, Id2^−/−; Rb^+/− pituitaries contained considerably fewer tumor cells, clustered in individual foci, and still had recognizable intermediate lobes (Fig. 1C). These findings are consistent with delayed development and/or slower progression of tumors in the compound mutant animals.

Loss of Id2 induces premature differentiation and cell cycle arrest of Rb^+/− melanotropes.

The intermediate lobe of the pituitary gland can be first identified in the mouse on embryonic day 14 (E14), a time when melanotrope precursors undergo active proliferation (38). The proliferation of melanotropes decreases significantly after birth and ceases almost completely after postnatal day 35 (P35). Since cell proliferation is required for stochastic loss of the residual Rb allele, the initiating event in melanotroph carcinogenesis that leads to Rb LOH has been traced to embryonic and early postnatal life (38). To ask whether Id2 is required for progression of the cell cycle in the intermediate lobe of the pituitary gland during this time window, we labeled E15.5 Rb^+/− and Id2^−/−; Rb^+/− embryos with BrdU and performed immunostaining for BrdU (Fig. 2A). Quantitation of the proliferative index in the developing intermediate lobe determined by BrdU staining demonstrated that 14.6% of nuclei in Id2^−/−; Rb^+/− pituitaries (n = 5) were labeled, compared with 22.2% of nuclei for Rb^+/− controls (n = 4, P = 0.005; unpaired t test), (Fig. 2B). The mitotic index confirmed this finding (Rb^+/−, 2.6 ± 0.6%; Id2^−/−; Rb^+/−, 1.3 ± 0.1%; P = 0.008). Interestingly, Id2 was highly expressed in proliferating melanotropes at E15.5 but its expression was extinguished in the postmitotic cells of the adult pituitary (Fig. 2C; also see Fig. 4A).

Differentiation of the melanotroph lineage is marked by expression of POMC, the precursor of α-MSH (51). Neurogenic bHLH proteins are essential activators of POMC promoter expression (44, 45). The bHLH factor MASH1 is expressed in mature melanotropes with a pattern fully corresponding to that of POMC expressing cells (31). As described for the neurogenic bHLH transcription factors in the nervous system, MASH1 is probably a component of the hierarchy of multiple bHLH factors involved in determination and differentiation of the melanotroph lineage in the pituitary gland (47). Because the best-characterized biochemical function of Id2 is the inhibition of bHLH-mediated transcription, we hypothesized that reduced proliferation of Id2-null melanotropes might be associated with premature differentiation. MASH1 expression was barely detectable in Rb^+/− melanotropes on E15.5 but was markedly increased in the intermediate lobe of Id2^−/−; Rb^+/− pituitaries (Fig. 2D). Expression of POMC was correspondingly elevated in Id2-null melanotropes (Fig. 2E). Differentiation induced by neurogenic bHLH transcription factors is associated with increased expression of the cdk inibitor p27^Kip1 (10). Expression of p27^Kip1 followed the same pattern as that of MASH1 and POMC, with significantly higher levels in the absence of Id2 (Fig. 2F). At the level of sensitivity of the immunohistochemical analysis, we failed to detect significant differences in the expression of MASH1, POMC, and p27^Kip1 between Rb^+/+ and Rb^+/− melanotropes (data not shown). These results indicate that loss of Id2 in the intermediate lobe of the mouse pituitary relieves bHLH proteins from an inhibitory control and induces premature activation of the bHLH transcription cascade. The outcome is premature exit from the cell cycle and differentiation of Rb^+/− melanotropes at a time when cell cycle progression is essential for LOH at the residual Rb allele and tumor initiation.

Loss of Id2 decreases number and size of early focal lesions and impairs proliferation of pituitary tumor cells in Rb^+/− mice.

Following Rb LOH, the earliest morphologically distinct pituitary tumor cells are detected as microscopic foci of atypical cells (early atypical proliferates [EAPs]) (38, 56). Each EAP is thought to originate from a single genetic event, and by P90 all Rb^+/− mice have multiple tumor foci that later expand, fuse, and replace most of the intermediate lobe. To determine the role of Id2 in the early events of pituitary tumorigenesis, we analyzed pituitary glands from Rb^+/− and Id2^−/−; Rb^+/− 90-day-old mice (Table 1; Fig. 3). To assess accurately the number and size of individual EAPs, we performed serial sectioning of the entire glands, stained sections with H&E, and analyzed each section to determine the number of foci and the total number of tumor cells within each focus. In the absence of Id2, the number of tumor foci was reduced by 50% (P = 0.003) (Table 1; Fig. 3A). Furthermore, the average number of tumor cells contained in each Id2^−/−; Rb^+/− focus was approximately half of that found in the Rb^+/− focus (P = 0.011) (Table 1; Fig. 3B). Remarkably, loss of Id2 led to a fourfold reduction of the total number of tumor cells per pituitary (P = 0.0006) (Table 1). To determine the mechanism of compromised growth in Id2-deficient tumor cells, we examined cell proliferation and apoptosis in 3- and 4-month-old Rb^+/− and Id2^−/−; Rb^+/− pituitary tumors, respectively. In the absence of Id2, BrdU labeling was significantly reduced in the pituitary tumor foci (Fig. 3C). Quantitation of the proliferation index in four pairs of Id2^−/−; Rb^+/− and Rb^+/− pituitaries demonstrated that 6.0 and 16.6% of nuclei stained positive for BrdU in Id2^−/−; Rb^+/− and Rb^+/− controls, respectively (P = 0.027) (Fig. 3D). However, the analysis of apoptosis by TUNEL assays and examination of hematoxylin-and-eosin stained sections revealed low levels of cell death with no differences between the two genotypes (data not shown). These findings suggest that impaired proliferation and premature differentiation of Id2-null melanotropes decrease the probability of the initiating Rb LOH event, thus leading to the reduced number of early tumor foci observed in Id2^−/−; Rb^+/− pituitaries. They also indicate that, at an early stage of tumor progression, Id2 contributes to an increase in the cell number and size of early tumor lesions because it provides Rb-null tumor cells with a proliferative advantage without an obvious effect on apoptosis.

Deregulated expression of Id2 and G1 cyclins in pituitary tumorigenesis.

To determine whether Id2 protein contributes to tumor progression, we analyzed the expression of Id2 in pituitary tumors from Rb mutant mice. We also examined the expression of Id1 and Id3, which are unable to interact with Rb (25). Expression of Id genes is high during embryogenesis and declines in postnatal life (21, 22, 24). Accordingly, the three Id proteins were absent in adult pituitaries from normal mice. Id1 and Id3 remained undetectable in Rb^+/− and Id2^−/−; Rb^+/− pituitary tumors, but the Id2 level was significantly elevated in the tumors from Rb^+/− mice (Fig. 4A). Because the expression of Id1 and Id3 in mouse tumors is localized predominantly to tumor endothelial cells, we analyzed the expression of Id2 at the cellular level (33, 50). In pituitary tumors from Rb mutant mice, Id2 was abundant in the tumor cells but was notably absent or barely detectable in the tumor endothelium (Fig. 4B). Interestingly, Id2 was expressed in three of three primary retinoblastomas, which are embryonal tumors carrying mutations of the Rb gene in humans (Fig. 4C). These results indicate that, among Id proteins, Id2 activity is selected for in mouse pituitary tumorigenesis initiated by loss of Rb. They also suggest that Id1 and/or Id3 fail to compensate for loss of Id2 in pituitaries of Id2^−/−; Rb^+/− mice. To gain insights into the different mechanisms of pituitary tumor growth in Rb^+/− and Id2^−/−; Rb^+/− mice, we examined the expression of cell cycle components. The G1 cyclins (cyclins D1, E, and A) were analyzed by Western blotting. Consistent with the postmitotic state of the adult mouse pituitary, cyclins were undetectable in normal glands. The levels of cyclins E and A increased in tumors from Rb^+/− mice but not in those from Id2^−/−; Rb^+/− mice (Fig. 4D). Interestingly, cyclin D1 was the only G1 cyclin that accumulated in Id2^−/−; Rb^+/− tumors, where it reached levels even higher than those found in Rb^+/− tumors (Fig. 4A). Overall, these results support the notion that pituitary tumorigenesis in Rb^+/− mice requires deregulated expression of Id2 to promote progression of the cell cycle.

Id2 mediates pituitary tumor angiogenesis.

One of the most important attributes of tumor progression is the ability to develop new blood vessels (5). There is wide acceptance of the notion that, beyond a critical size, the “angiogenic switch” is required for growth of the tumor mass. The intermediate lobe of the pituitary gland is poorly vascularized (41). However, Rb-null tumors undergoing rapid expansion became highly angiogenic (Fig. 1B). We determined the extent of tumor vascularization by perfusing Rb^+/− and Id2^−/−; Rb^+/− animals carrying advanced tumors with curable orange latex. The large tumors arising in Rb^+/− pituitaries were highly vascularized, with large vessels and extensive capillary networks. In contrast, the small tumors that arose in Id2^−/−; Rb^+/− mice contained very few vessels and occasionally exhibited gross hemorrhage and necrosis (Fig. 5A). When pituitary tumors were stained for PECAM, an endothelial cell marker, the number of positive vascular structures was visibly higher in Rb^+/− than Id2^−/−; Rb^+/− mice (Fig. 5B). Impaired angiogenesis in Id2-null pituitary tumors may be caused by higher expression of antiangiogenic molecules and/or decreased ability to upregulate proangiogenic factors. A recent study reported that the antiangiogenic protein thrombospondin-1 (TSP-1) is a target of transcriptional repression by Id1 and that its expression is increased in Id1-null mice (59). However, TSP-1 immunostaining was not elevated in Id2-null pituitary tumors (Fig. 5C), thus making TSP-1 upregulation an unlikely mechanism for the angiogenic defect of Id2^−/−; Rb^+/− tumors.

Among several positive regulators of tumor angiogenesis, increased production of VEGF is a crucial rate-limiting step, required for generation of new blood vessels (12, 19, 62). Therefore, we investigated whether induction of VEGF occurred during pituitary tumor progression. VEGF is absent in the intermediate lobe but is detectable in the posterior lobe of the pituitary gland in rodents (41) (Fig. 5E). However, VEGF immunostaining revealed that tumorigenesis of the intermediate lobe in Rb^+/− pituitaries resulted in dramatic upregulation of VEGF expression. Here, VEGF was associated mostly with tumor cells although blood vessels were also stained (Fig. 5D, left panels). Interestingly, VEGF was undetectable in Id2^−/−; Rb^+/− pituitary tumors, even when we focused our analysis on the most advanced lesions observed in these mice (Fig. 5D, right panels). Western blot analysis confirmed the defective activation of VEGF production by Id2-null tumors (data not shown). To test whether differential expression of VEGF is an intrinsic attribute of the two genotypes, we analyzed pituitary tumors of similar size and at an early stage of development (3 months for Rb^+/− mice and 4 months for Id2^−/−; Rb^+/− mice). VEGF was abundant in Rb^+/− tumor foci but absent in Id2^−/−; Rb^+/− tumor foci (Fig. 5E). Interestingly, although large necrotic areas were clearly visible in Id2^−/−; Rb^+/− tumors, expression of HIF-1α, the key factor promoting expression of VEGF in hypoxic cells, was undetectable in those lesions (Fig. 5F). The strikingly different ability of Rb^+/− and Id2^−/−; Rb^+/− tumors to initiate expression of VEGF at an early, avascular stage of tumor progression, combined with the absence of VEGF even in advanced Id2-null tumors, suggests that Id2 is essential for production of VEGF during pituitary tumorigenesis.

Deregulated Id2 is sufficient and necessary for cellular proliferation and expression of VEGF.

To address whether Id2 could induce the expression of VEGF, we first infected REF52 fibroblasts with an adenovirus vector expressing Id2 and green fluorescent protein in cis from an internal ribosomal entry site (Ad-Id2). Infection with Ad-Id2 markedly enhanced the expression of VEGF, which paralleled the time course of ectopic Id2 (Fig. 6A). Conversely, infection with Ad-GFP control virus did not change VEGF.

Id2 is a transcriptional target of Myc oncoproteins and is highly expressed in human neuroblastoma cells with N-myc gene amplification. In these cells, Id2 accumulates at levels that exceed those of hypophosphorylated, active Rb and correspond to those required by ectopic Id2 to overcome cell cycle arrest by Rb in vitro (17, 24-26, 46). As a further test of the ability of Id2 to induce VEGF, we transfected the human neuroblastoma cell line SK-N-SH-N (SH-N) with a plasmid driving the expression of Id2 and selected polyclonal populations. SH-N-Id2 expressed remarkably larger amounts of VEGF mRNA and protein and secreted twofold more VEGF than did SH-N vector as determined by enzyme-linked immunosorbent assay (Fig. 6B and C and data not shown). Remarkably, the analysis of five independent Flag-Id2-expressing clones established from the neuroblastoma cell line IMR32 showed a clear correlation between the ectopic amounts of Flag-Id2 and the levels of VEGF present in each individual clone (Fig. 6D).

Although Id2 is abundant in N-myc-amplified neuroblastoma, it has not been clear whether deregulated Id2 mediates a specific tumorigenic function(s) of human neuroblastoma cells. Treatment of N-myc-amplified neuroblastoma cells with retinoic acid induces cell cycle arrest and differentiation, and downregulation of N-myc precedes the phenotypic response (52). Recent data suggested that expression of VEGF is modified during cellular differentiation (9, 58). Given the regulatory role of N-Myc and Id2 in multiple differentiation pathways, we asked whether Id2 controls the expression of VEGF in a cellular model of neuronal differentiation. In the N-myc-amplified neuroblastoma cell line LAN-1, addition of retinoic acid led to a rapid (within 1 day) and sustained (up to 5 days) decrease in the levels of N-Myc, Id2 and VEGF (Fig. 6E). Exit from the cell cycle occurred only after 2 days of treatment with retinoic acid, as indicated by inhibition of cyclin A and accumulation of p27^Kip1. These results suggest that inhibition of the expression of VEGF is not consequent to cell cycle arrest and may require early downregulation of Id2. To investigate the effect of Id2 on the retinoic acid-mediated decrease of VEGF, we infected LAN-1 with the Id2 adenovirus. Indeed, transduction of LAN-1 with Ad-Id2 prevented down-regulation of VEGF by retinoic acid (Fig. 6F). As a further test of the role of Id2 in cell proliferation and accumulation of VEGF by N-myc-amplified neuroblastoma, we targeted the expression of endogenous Id2 in LAN-1 with the vector pSuper.retro, which directs the synthesis of siRNAs matching two independent sequences from the Id2 mRNA (pRSId2-112 and pRSId2-171). Transfection of pRSId2-112 and pRSId2-171 resulted in efficient suppression of Id2 expression. Western blot analysis revealed that loss of Id2 was associated with powerful inhibition of VEGF expression. Suppression of Id2 and VEGF took place in the absence of changes of Id1, suggesting that the decreased expression of VEGF was a specific consequence of suppression of Id2 (Fig. 6G). Silencing of Id2 also resulted in almost complete cessation of cell proliferation as shown by the greatly reduced clonogenic ability of LAN-1 cells transfected with each of the two Id2-siRNA plasmids (Fig. 6H). In transient promoter-luciferase reporter assays, Id2 did not activate a −2362/+956 human VEGF promoter/regulatory region, suggesting that Id2 regulates VEGF expression in an indirect fashion (data not shown). Together, these results indicate that Id2 is sufficient to induce VEGF in fibroblasts and neuroblastoma cell lines. They also suggest that accumulation of Id2 in N-myc-amplified neuroblastoma tumor cells is necessary for proliferation and constitutive expression of VEGF.

Id2 in pituitary tumor initiation.

We have shown that Id2 is expressed abundantly in the developing and proliferating intermediate lobe of the mouse pituitary gland but absent in the adult and quiescent organ. Within the time window of Rb LOH, Id2-null melanotropes display early withdrawal from the cell cycle with accumulation of the cdk inhibitor p27^Kip1 and premature expression of differentiation markers of the melanotroph lineage. These effects appear to be initiated by loss of the negative control imposed by Id2 on bHLH transcription factors expressed in the intermediate lobe of the developing pituitary gland. Consequently, Id2-null pituitary cells in the intermediate lobe undergo premature activation of the bHLH transcription cascade, with early differentiation and exit from the cell cycle. Indeed, neurogenic bHLH proteins bind and activate transcription from specific E-boxes in the POMC regulatory sequence and promote the differentiation of POMC-expressing cells (44, 45). Together, these findings identify Id2 as a factor that sets the timing of differentiation and quiescence of cells of the melanotroph lineage. Recently, it was shown that the most critical barrier to malignant transformation mediated by inactivation of the pocket proteins (Rb, p107, and p130) is terminal differentiation rather than apoptosis (8, 55). By showing that expression of Id2 prevents premature differentiation and sustains malignant transformation of Rb^+/− melanotroph precursors, our results provide a molecular explanation to this observation.

In the absence of Id2, premature cell cycle exit and differentiation of developing melanotropes narrow the time window for tumor-initiating events, thus providing resistance to malignant transformation. The observation that Id2^−/−; Rb^+/− pituitaries contain markedly fewer early tumor lesions suggests that the proportion of individual melanotropes undergoing LOH at the residual wild-type Rb allele is lower in Id2-null pituitaries. The notion that efficiency of Rb inactivation in the pituitary determines the latency of tumor development is supported by the accelerated tumorigenesis found in chimeric mice and conditional Rb mutants (34, 60, 61). In these mouse models, melanotropes with complete inactivation of Rb are already present during early pituitary development and a second genetic event is not required for malignant transformation. Recent results have suggested that Rb deficiency leads to higher rates of genomic instability, a phenomenon implicated in both the origin and progression of most malignant tumors (30, 36, 65). It is tempting to speculate that deregulated Id2 activity resulting from loss of Rb may be one of the effecting mechanisms leading to the increased rate of chromosomal alterations.

Id2 in proliferation of Rb mutant cells.

After completing differentiation, mature and quiescent melanotropes in the adult intermediate lobe extinguish the expression of Id2. However, Id2 is the only Id protein expressed by Rb-null pituitary tumor cells, suggesting that either maintenance or reactivation of Id2-dependent signaling is selected for tumor progression in the absence of Rb. The mechanism by which mouse and human Rb-null tumors accumulate high levels of Id2 is unknown, but we suggest that it may be related to loss of the repressor function of Rb on the genomic promoter of c-myc, whose protein product acts as a transcriptional activator for Id2 (16, 26).

The early lesion analysis has demonstrated that Id2-null EAPs are significantly smaller, have a lower proliferation index, and contain half the number of tumor cells than those identified in mice with intact Id2. This suggests that Id2 is also necessary for aberrant proliferation of pituitary tumor cells after the Rb LOH. Thus, the phenotypic consequences of loss of Id2 can be explained by impaired cell cycle progression in the intermediate lobe before and after the Rb LOH initiating event. In support of this notion, the aberrant expression of the G1-S cyclins E and A observed in Rb-null pituitary tumors is reversed in the absence of Id2. However, accumulation of cyclin D1 in Id2-null pituitary tumors may be the event that partially compensates for Id2 deficiency to ensure residual proliferation. This observation is in agreement with earlier studies showing that Id2 and cyclin D1 implement parallel and alternative pathways to enhance cell proliferation (25, 35). These data help to define a broad scope of effects of deregulated Id2 activity in the absence of Rb that extends to aberrant expression of cyclins and uncontrolled progression of the cell cycle.

Id2 in tumor angiogenesis.

For continued growth and survival, a tumor must promote angiogenesis. Our study directly links angiogenesis with Id2. Through gain- and loss-of-function experiments with the Rb^+/− pituitary tumor model and human neuroblastoma cells, we have shown that Id2 is essential for the angiogenic switch during tumor progression. This conclusion is supported by (i) the impaired vascularization and defective production of VEGF of Id2-null pituitary tumors in the Rb mutant background; (ii) the impaired expression of HIF-1α by Id2-null pituitary tumors; (iii) the enhanced production of VEGF by normal and tumor cells expressing ectopic Id2; (iv) the coregulation of Id2 and VEGF in the differentiation response of neuroblastoma tumor cells to retinoic acid and the rescue of VEGF inhibition by expression of Id2; and (v) the reduced synthesis of VEGF in the neuroblastoma cell line LAN-1 in which endogenous Id2 has been suppressed by siRNA.

Several studies have implicated an indispensable role of VEGF in tumor angiogenesis, particularly in tumors of embryonal origin (2, 9, 23, 58). Other studies have reported that multiple oncogenic events combined with inactivation of tumor suppressor genes upregulate the expression of VEGF either directly or through constitutive activation of growth factor-mediated signaling (3, 6, 28, 54, 66). Our data are consistent with a model whereby loss of Rb results in activation of Id2 signaling, which in turn upregulates VEGF. The nonredundant role played by Id2 to promote the expression of VEGF in vivo and in vitro reproduces a similar activity recently attributed to Myc, an oncogenic transcription factor that induces both VEGF and Id2 gene expression, suggesting that Id2 may participate in the proangiogenic functions of Myc. Our data do not exclude the possibility that, as shown for Myc, Id2 promotes angiogenesis by inducing multiple positive cytokines. We also found that, like Myc, Id2 failed to induce luciferase activity driven by the VEGF promoter/enhancer (4). Thus, the mechanism by which the Myc-Id2 pathway activates expression of VEGF remains to be elucidated (4, 11, 14, 26, 37, 42).

The Id2 requirement for proper expression of VEGF and possibly other angiogenic cytokines has profound implications. Loss of the Id family members Id1 and Id3 leads to defective tumor vascularity because of cell-autonomous effects in endothelial cells, which are unable to respond to VEGF (32, 33). We propose a model whereby cooperation between Id proteins supports tumor growth and angiogenesis. Deregulated Id2 activity inactivates the Rb tumor suppressor pathway in tumor cells and promotes the expression of VEGF. Expression of Id1 and Id3 in the tumor endothelium renders the target cells competent for VEGF signaling. Together, these results suggest that the dual accumulation of Id proteins observed in tumor and endothelial cells from the most aggressive neuroectodermal neoplasms may be indispensable for tumor angiogenesis (57).