TSC2 Deficiency Increases PTEN via HIF1α*

Substantial evidence suggests roles of TSC2 and PTEN in the development of cancer predisposition syndromes. Loss of TSC2 results in benign tumors, neurological disorders, and angiomyolipomas. We found that PTEN mRNA and protein levels are elevated in Tsc2−/− mouse embryo fibroblasts with concomitant reduction in Akt phosphorylation. Reconstitution of TSC2 in Tsc2−/− mouse embryo fibroblasts decreases PTEN levels. Interestingly, increased HIF1α activity present in Tsc2 null cells is required for PTEN transcription and protein expression. We identified a canonical hypoxia-responsive element in the PTEN promoter, which regulates the transcription of this tumor suppressor protein in a TSC2-dependent manner. Finally, we demonstrate a positive correlation between expression of HIF1α and PTEN in renal angiomyolipomas from TSC patients. Our results reveal a unique function of HIF1α in up-regulation of PTEN and provide a new mechanism of reduced Akt phosphorylation in Tsc2 null cells. These data suggest that PTEN may safeguard against developing malignant tumors in patients with TSC deficiency.

Mutations in the tuberous sclerosis tumor suppressor genes (TSC1 and TSC2) result in autosomal dominant diseases characterized by benign tumors (hamartomas) in the kidney, heart, brain, and other organs (1). Approximately 1 in 6000 live births is linked to a germ line-inactivating mutation of either of the two genes, TSC1 or TSC2. Thus, each mutated gene accounts for ϳ50% of the cases (2). TSC2 associates with TSC1 to form the active signaling complex (3). The C terminus of TSC2 contains a GTPase-activating protein domain, which blocks mTOR activity by increasing hydrolysis of GTP associated with the small G-protein Rheb (Ras homolog enriched in the brain) (4,5). Growth factor-stimulated Akt and other kinases phosphorylate TSC2 at specific sites, resulting in its dissociation from the TSC1/2 complex (6 -9).
The lipid phosphatase activity of the tumor suppressor PTEN dephosphorylates the second messenger phosphatidylinositol (PI) 5 3,4,5-trisphosphate, thus negatively regulating the PI 3-kinase signaling pathway (10). Germ line mutations in PTEN cause cancer predisposition syndromes, such as Cowden disease (11)(12)(13). Also, loss of PTEN is common in many tumors, including sporadic glioblastoma, endometrial carcinoma, melanoma, meningioma, and renal, breast, prostate, and small cell lung cancer (11)(12)(13). Whereas PTEN deficiency predisposes to malignancy, it is rare in TSC patients (11,13,14). In PTENdeficient cancer cells, even in the absence of growth factors, Akt is constitutively active, which results in phosphorylation and inactivation of TSC2 and activation of mTOR (8,10,15). Similarly, the disruption of TSC2 results in significantly increased mTOR activity (16 -18). In the latter case, the mTOR activation leads to inactivation of Akt through a negative feedback loop involving IRS (insulin receptor substrate) proteins (17, 19 -21). However, additional signaling pathway(s) probably contribute to reduced Akt activation in TSC2 deficiency.
In this report, we demonstrate that TSC2 deficiency results in increased expression of PTEN. As a mechanism, we show that HIF1␣ positively regulates the transcription of PTEN, using a canonical HIF-responsive element. Furthermore, we demonstrate that renal angiomyolipomas in TSC patients express elevated levels of HIF1␣ and PTEN protein. Thus, increased levels of PTEN in renal angiomyolipomas of patients with TSC may mute the malignant potential of these tumors by decreasing Akt activation.

MATERIALS AND METHODS
Cell Culture and Adenovirus Infection-Tsc2 ϩ/ϩ and Tsc2 Ϫ/Ϫ MEFs, generously provided by Dr. D. J. Kwiatkowski (Harvard University), were grown in DMEM with low glucose with 10% fetal bovine serum (22,23). 293 cells were grown in DMEM with high glucose-containing 10% serum. All cell stocks were maintained in the presence of plasmocin and primocin. Tsc2 Ϫ/Ϫ MEFs were infected with adenovirus vector expressing TSC2. This viral vector also expressed green fluorescence protein to detect efficient infection. As a control, an adenovirus vector expressing ␤-galactosidase was used.
Tissue Samples-Kidney angiomyolipoma tissues from TSC patients with angiomyolipoma and normal kidney tissues were obtained from the Brain and Tissue Bank for Development Disorders (University of Maryland). This study has been approved by the Institutional Review Board of the University of Texas Health Science Center at San Antonio.

RNA Extraction and Reverse Transcription (RT)-PCR-Total
RNAs were isolated using the RNAzol kit according to the vendor's protocol. RT-PCR was performed using the Onestep RT-PCR kit according to the manufacturer's instructions.
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared using the kit according to the vendor's protocol. The HRE-3 element from the PTEN promoter was made by annealing the oligonucleotide 5Ј-GAG CAG CGT GGT CA-3Ј with its complementary strand. The probe was labeled with [␥-32 P]ATP using T4 polynucleotide kinase. The electrophoretic mobility shift assay was performed using 10 g of nuclear extracts, as described (27,29,31). To determine the specificity of interaction, the nuclear extracts were incubated with cold doublestranded oligonucleotide prior to incubation with the radiolabeled probe. For antibody reaction, the nuclear extracts were incubated with Hif1␣ antibody, followed by incubation with the labeled probe, as described (28). The DNA protein complex was resolved by 5% polyacrylamide gel electrophoresis.
Site-directed Mutagenesis-The HRE-3 sequence was mutated using the QuikChange II site-directed mutagenesis kit, as described (32). The mutation was verified by sequencing the DNA. The mutated bases in the HRE-3 core sequence are shown in Fig. 5C.
Transient Transfection and Luciferase Activity-The cells were transfected with the reporter plasmid, along with the indicated vector and the indicated expression vectors or siRNAs. Luciferase activity was determined in the cell lysate using a luciferase assay kit according to the manufacturer's protocol (27, 29 -31, 33).
Statistics-Statistical significance of the data was calculated using analysis of variance followed by Student-Newman-Keuls analysis, as described previously (27,29,30). Significance was considered at a p value of Ͻ0.05.

TSC2 Inhibits Expression of PTEN-
Previous studies have established that a feedback mechanism in cells lacking TSC2 inhibits PI 3-kinase activity, which results in decreased activation of Akt (20,21,34). The level of PI 3-kinase product PIP 3 is regulated by PTEN. Inactivating mutation or deficiency of PTEN leads to Akt activation and tumorigenesis (12). Therefore, we investigated a potential role for PTEN in regulating Akt phosphorylation in murine embryonic fibroblasts (MEFs) lacking Tsc2. Immunoblot analysis showed significantly increased abundance of Pten in Tsc2 Ϫ/Ϫ MEFs compared with wild type cells (Fig. 1A). Increased Pten resulted in reduced phosphorylation of Akt (Fig. 1A). Transfection with Pten siRNA, however, showed an increase in Akt phosphorylation ( Fig. 1B, left). Transfection of a pool of three siRNAs from a different source targeting Pten produced the same results (Fig. 1B, right). To demonstrate that the increased expression of Pten was a consequence of the lack of the Tsc2 gene, we reconstituted Tsc2 Ϫ/Ϫ MEFs with human TSC2 using an adenovirus vector. Expression of TSC2 significantly reduced the level of Pten protein in these cells with a concomitant increase in phosphorylation of Akt (Fig. 1C). Expression of PTEN has been shown to be regulated by transcriptional mechanisms (35,36). RT-PCR analysis showed increased expression of Pten mRNA in Tsc2 Ϫ/Ϫ MEFs relative to the levels expressed in wild type MEFs (Fig. 1D). Furthermore, reconstitution of TSC2 in Tsc2 Ϫ/Ϫ MEFs lowered the Pten mRNA level (Fig. 1E). To evaluate whether TSC2 regulates PTEN transcription, we used a reporter plasmid in which the luciferase gene is driven by the PTEN promoter (PTEN-Luc) (36). Transient transfection assays with this reporter plasmid showed significantly elevated transcription of PTEN in Tsc2 Ϫ/Ϫ MEFs as compared with that in wild type cells (Fig. 1F). Cotransfection of TSC2 with PTEN- Luc into Tsc2 Ϫ/Ϫ MEFs inhibited transcription of the PTEN reporter in a concentration-dependent manner (Fig. 1G).
The above results from Tsc2 Ϫ/Ϫ MEFs indicate that Tsc2 inhibits expression of Pten. To examine whether the increase in the PTEN expression is a direct effect of TSC2 mutation or a result of other mutations present in the Tsc2 Ϫ/Ϫ MEFs, we disrupted TSC2 signaling in human embryonic kidney epithelial 293 cells. Transfection of TSC2 siRNA significantly lowered TSC2 protein expression ( Fig. 2A). Compared with scrambled siRNAtreated cells, PTEN expression was increased following knockdown of TSC2 expression, which resulted in reduced Akt phosphorylation (Fig.  2B, left). Transfection of a different set of siRNAs targeting TSC2 showed similar results (Fig. 2B,  right). Down-regulation of PTEN in 293 cells using two different siRNAs increased Akt phosphorylation (Fig.  2C, left and right). Down-regulation of TSC2 with two independent sets of siRNAs also significantly increased transcription of PTEN, as judged by reporter transfection assays using PTEN-Luc (Fig. 2D, left  and right). Conversely, expression of TSC2 in 293 cells significantly reduced PTEN transcription (Fig.  2E). These results provide evidence that TSC2 negatively regulates expression of the PTEN tumor suppressor gene.
HIF1␣ Regulates PTEN Expression-The results described above demonstrate transcriptional regulation of PTEN in a TSC2-sensitive manner. A recent report demonstrated increased expression of Hif1␣ transcription factor in Tsc2 Ϫ/Ϫ MEFs (37). We also confirmed the elevated levels of Hif1␣ in these cells (supplemental Fig. S1) Incubation of Tsc2 Ϫ/Ϫ MEFs with two independent inhibitors of Hif1␣ (24 -26), emetine and YC-1, significantly reduced the expression of Pten protein and resulted in increased Akt phosphorylation (Fig.  3, A-D). To confirm these results genetically, we used plasmid-derived siRNA expression targeting Hif1␣. Transfection of Hif1␣ siRNA plasmid in Tsc2 Ϫ/Ϫ MEFs inhibited Hif1␣ protein expression, resulting in significantly reduced expression of Pten protein (Fig. 3E, left). Using a second set of siRNAs containing a pool of three siRNAs also showed the same results (Fig. 3E, right). Down-regulation of HIF1␣ with plasmid-derived or oligonucleotide pool siRNAs increased phosphorylation of Akt (Fig. 3F, left). After transfection of siRNAs from two different sources to knock down Hif1␣, RT-PCR analysis showed reduced Pten mRNA expression in Tsc2 Ϫ/Ϫ MEFs (Fig. 3G). Since VHL tumor suppressor  OCTOBER 9, 2009 • VOLUME 284 • NUMBER 41 protein induces degradation of Hif1␣ through the proteasomal pathway, we used a plasmid expressing FLAG-tagged VHL. Expression of VHL in Tsc2 Ϫ/Ϫ MEFs inhibited Pten protein expression (supplemental Fig. S2). Furthermore, in 293 cells, ectopic expression of HIF1␣ significantly increased the expression of PTEN mRNA and protein (Fig. 3, H and I). Collectively, these results demonstrate that up-regulated Hif1␣ in Tsc2deficient cells positively regulates PTEN expression.

HIF1␣ Increases PTEN
HIF1␣ Regulates PTEN Transcription-We next examined the mechanism by which HIF1␣ regulates PTEN. Analysis of the 5Ј-flanking sequence of the PTEN gene upstream of the start codon revealed the presence of three putative HIF1␣-responsive elements (HREs), which share homology with VEGF, endothelin-1, and Nur-77 HREs (Fig. 4A) (38 -42). Transfection of Tsc2 Ϫ/Ϫ MEFs with Hif1␣ siRNA plasmid or oligonucleotide pool recognizing Hif1␣ reduced transcription of PTEN in PTEN-Luc reporter assays (Fig. 4B, left and  right). In contrast, expression of HIF1␣ increased transcription of PTEN in 293 cells (Fig. 4C). In addition, VHL, which induces degradation of HIF1␣, significantly inhibited both basal and HIF1␣-induced PTEN transcription (Fig. 4D). Since HRE-1 and HRE-2 are present in the untranslated region of the PTEN mRNA (35), we used 3Ј deletion of the PTEN promoter (from Ϫ2 to Ϫ778), which retains the transcription start site at Ϫ1031 bp and HRE-3 (Fig. 4E). The effect of HIF1␣ on transcription of this reporter was tested in 293 cells. As expected, HIF1␣ increased the transcription of PTEN from PTEN-Luc containing all three HREs (Fig. 4F). But no significant difference in response to HIF1␣ was found in transcription of PTEN using HRE-3-Luc reporter plasmid (Fig. 4F). These data suggest that HRE-3 at Ϫ1391 bp (Fig.  4A) from the start codon regulates the transcription of PTEN.
In order to test the involvement of HRE-3 specifically, we employed electrophoretic mobility shift assays using nuclear extracts from Tsc2 ϩ/ϩ and Tsc2 Ϫ/Ϫ MEFs. Double-stranded oligonucleotide representing the HRE-3 from the PTEN promoter was used as a probe. Formation of protein-DNA complexes was detected with nuclear extract isolated from Tsc2 Ϫ/Ϫ MEFs as compared with those from Tsc2 ϩ/ϩ cells (Fig. 5A,  compare lane 3 with lane 2). Incubation of nuclear extracts with cold HRE-3 oligonucleotide inhibited the DNA-protein complex formation, demonstrating the specificity of the interaction (Fig. 5A, lanes 4 and 5). To identify Hif1␣ in this protein-DNA complex, we performed an electrophoretic mobility shift assay in the presence of a Hif1␣-specific antibody. Incubation of nuclear extracts from Tsc2 Ϫ/Ϫ MEFs with Hif1␣ antibody completely blocked DNA binding relative to that in the presence or absence of control IgG (Fig. 5B, lane 4 versus lanes 3 and 5). Next, we examined the requirement of HRE-3 for the transcription of PTEN. We mutated the HRE-3 in the HRE-3-Luc reporter plasmid containing the 3Ј deletion (Fig. 5C). Transient transfection assays using this plasmid in Tsc2 Ϫ/Ϫ MEFs showed significantly low reporter activity, compared with the wild type HRE-3-containing promoter (Fig. 5D). These results suggest that increased Hif1␣ present in Tsc2 Ϫ/Ϫ MEFs is not sufficient to induce transcription of PTEN if HRE-3 is mutated. We also examined the activity of this mutant HRE-3 reporter in 293 cells in response to HIF1␣, which mimics the increased levels of HIF1␣ protein in Tsc2 Ϫ/Ϫ MEFs. The mutant HRE-3-containing reporter was significantly less responsive to HIF1␣ relative to the wild type HRE-3 reporter (Fig. 5E). To further confirm the role of HRE-3, we mutated this site in the context of the full-length promoter (PTEN-Luc-HRE-3 Mut) (supplemental Fig.  S3A) Transfection of this reporter plasmid into 293 cells showed significantly reduced response to Hif1␣ as compared with PTEN-Luc reporter (supplemental Fig. S3B). Thus, HRE-3 is necessary and sufficient for regulation of PTEN transcription by HIF1␣.
Expression of PTEN and Hif1␣ in Human Renal Angiomyolipomas-The results described above demonstrate a correlation between PTEN expression and TSC2 deficiency. TSC patients commonly develop renal angiomyolipomas, relatively benign tumors of the kidney, which consist of adipose tissue, muscle cells, and blood vessels (43,44). We examined expression of PTEN in renal angiomyolipoma samples from six TSC patients. RT-PCR analysis showed significantly increased expression of PTEN mRNA in the angiomyolipoma samples relative to the healthy kidney tissues (Fig. 6, A and B). Since our data showed that Hif1␣ in Tsc2 Ϫ/Ϫ MEFs up-regulates PTEN, we determined the expression of these two proteins in the angiomyolipoma samples. Expression of PTEN was markedly elevated in the renal angiomyolipomas (Figs. 6, C (top) and D). These increased PTEN levels were directly correlated with increased HIF1␣ expression (Figs. 6, C (middle) and E). Co-elevation of HIF1␣ and PTEN in angiomyolipomas and in Tsc2 Ϫ/Ϫ MEFs suggests an antagonistic relationship between TSC2 and PTEN, which may contribute to the relatively benign nature of angiomyolipomas in patients with TSC.

DISCUSSION
The current study demonstrates a previously unrecognized role of TSC2 deficiency for attenuation of Akt phosphorylation by increased expression of PTEN expression. Moreover, we have delineated a molecular pathway by which the increased level of HIF1␣ in TSC2 deficiency regulates PTEN expression (Fig. 7). We identified a HIF1␣-responsive element in the PTEN promoter, which regulates PTEN transcription. Finally, using renal angiomyolipoma tumor samples from TSC patients, we provide the first evidence for the presence of a positive correlation between HIF1␣ and PTEN expression.

HIF1␣ Increases PTEN
The requirement of increased Akt phosphorylation due to inactivating mutations of PTEN or constitutive activation of growth factor receptors and PI 3-kinase for cell growth in different cancer and cancer predisposition syndromes is well established (11)(12)(13). In support of this, Akt1 has been shown to be necessary for the development of tumors in Pten ϩ/Ϫ mice (47). The observation that Akt phosphorylation is dramatically reduced in Tsc2 Ϫ/Ϫ cells may represent the major mechanism explaining why inactivating lesions in TSC2 do not lead to malignancy. Although TSC2 represents a tumor suppressor gene, its loss is not seen in highly proliferative and invasive cancers. In this study, we show that TSC2 deficiency increases expression of PTEN, thus contributing to the inhibition of phosphorylation of Akt (Figs. 1 and 7). Our results provide a further mechanism explaining why Tsc2 Ϫ/Ϫ MEFs are more sensitive to apoptosis, since PTEN induces apoptosis in many cells (12,19,21,48,49).
Recently, Manning and co-workers (50) described a novel positive regulation of TORC2-mediated Akt Ser-473 phosphorylation by TSC2. Direct input from PI 3-kinase may not be necessary for TORC2-mediated Ser-473 phosphorylation of Akt. Specific inhibition of TORC1, which regulates the negative feedback loop integrating IRS-1 into the lipid kinase, has been shown to partially increase Akt Ser-473 phosphorylation in Tsc2 Ϫ/Ϫ MEFs in response to insulin (50). Furthermore, direct inhibition of PI 3-kinase by wortmannin partially blocks TORC2 kinase activity, suggesting a role for PI 3-kinase in TORC2-mediated phosphorylation of Akt (50). Our results also support this notion. Increased PTEN expression would inhibit PI 3-kinase signaling, leading to attenuation of Akt phosphorylation in TSC2 deficiency.
Pten haploinsufficiency is required for increased Akt activation and virulent tumorigenesis in Tsc2 ϩ/Ϫ mice (51,52). This demonstration indicates that PTEN may contribute to regulation of Akt phosphorylation in the event of TSC2 loss. Our results in TSC2-deficient cells support this notion that PTEN contributes to reduced Akt phosphorylation. Furthermore, our results provide a positive correlation between elevated PTEN levels and the benign nature and relative limited proliferative capacity of the TSC tumors.
Elevated expression of angiogenic factors has been detected in tumors from TSC patients (53). In humans, loss of VHL tumor suppressor is associated with highly vascular renal cell carcinomas and hemangiomas similar to those found in animal models of TSC (54,55). Increased angiogenic activity found in the TSC tumors is mediated by VEGF (53,56). Recently, it was shown that increased levels of HIF1␣ present in Tsc2 Ϫ/Ϫ MEFs are necessary for the production of VEGF in these cells as well as in the TSC mouse model (37,56). Similarly, we show a requirement of Hif1␣ in inducing the expression of Pten in Tsc2 Ϫ/Ϫ MEFs (Fig. 3). These results indicate that HIF1␣ may produce opposing activities by inducing angiogenic/mitogenic growth factors (such as VEGF/TGF␣) and antiproliferative signals (PTEN) in the tumor microenvironment of TSC patients.
In TSC-deficient cells, Lee et al. (57) recently reported significantly elevated levels of p53 tumor suppressor protein. Other authors have shown that PTEN is transcriptionally regulated by p53 (35). HIF1␣-deficient embryonic stem cells exhibit low p53 protein expression, suggesting a role for HIF1␣ in p53-dependent gene expression (58). Although these results fit with our observation of increased PTEN expression by p53, it should be noted that Tsc2 Ϫ/Ϫ MEFS used in our study are p53-deficient (23). Therefore, this rules out the possibility of PTEN up-regulation due to HIF1␣-dependent increase in p53 abundance in TSC2-deficient cells. Rather, our data provide the first evidence for a direct role of Hif1␣ in regulating PTEN expression (Fig. 3). We identified a HIF1␣ DNA binding element, HRE-3, in the PTEN promoter, which is necessary and sufficient to increase the expression of PTEN (Figs. 4 and 5 and supplemental Fig.  S3). Furthermore, in 293 cells, induction of hypoxia increased transcription of PTEN similar to that obtained with a reporter construct containing three copies of the hypoxia response element (3ϫHRE) in the erythropoietin gene (supplemental Fig.  S4, A and B). Taken together, our data provide a mechanism by which elevated expression of HIF1␣ in TSC2 deficiency increases the expression of PTEN, which in turn inactivates PI 3-kinase signaling to reduce Akt phosphorylation observed in Tsc2 Ϫ/Ϫ MEFs and in TSC patients.
Angiomyolipomas represent benign tumors with loss of heterozygosity of one or both TSC genes (43,59,60). Loss of TSC2 is also found in sporadic angiomyolipomas (59,60). More than half of the TSC patients develop angiomyolipomas of the kidney, which contribute to morbidity secondary to destruction of renal parenchyma and hemorrhage. Although patients lacking TSC1 develop angiomyolipomas, TSC2 deficiency-associated pathology is more severe than TSC1 cases (2). In this study, we demonstrate increased HIF1␣ protein levels, which correlate with increased PTEN mRNA and protein expression in renal angiomyolipomas (Fig. 6). Our findings in TSC2-deficient cells that HIF1␣ transcriptionally regulates expression of PTEN (Figs. 4 and 5) are also consistent with our data in tissue samples from patients with renal angiomyolipomas. Since PTEN inhibits PI 3-kinase signal transduction, leading to reduced Akt activation (Fig. 7), our results demonstrating increased PTEN expression in TSC2 deficiency may represent a mechanism for the presence of less malignant tumors in TSC patients.