Mechanism of von Hippel-Lindau Protein-Mediated Suppression of Nuclear Factor kappa B Activity

Abstract

Biallelic inactivating mutations of the von Hippel-Lindau tumor suppressor gene (VHL) are a hallmark of clear cell renal cell carcinoma (CCRCC), the most common histologic subtype of RCC. Biallelic VHL loss results in accumulation of hypoxia-inducible factor alpha (HIFα). Restoring expression of the wild-type protein encoded by VHL (pVHL) in tumors with biallelic VHL inactivation (VHL^−/−) suppresses tumorigenesis, and pVHL-mediated degradation of HIFα is necessary and sufficient for VHL-mediated tumor suppression. The downstream targets of HIFα that promote renal carcinogenesis have not been completely elucidated. Recently, VHL loss was shown to activate nuclear factor kappa B (NF-κB), a family of transcription factors that promotes tumor growth. Here we show that VHL loss drives NF-κB activation by resulting in HIFα accumulation, which induces expression of transforming growth factor alpha, with consequent activation of an epidermal growth factor receptor/phosphatidylinositol-3-OH kinase/protein kinase B (AKT)/IκB-kinase alpha/NF-κB signaling cascade. We also show that components of this signaling pathway promote the growth of VHL^−/− tumor cells. Members of this pathway represent viable drug targets in VHL^−/− tumors, such as those associated with CCRCC.

One of the genetic hallmarks of clear cell renal cell carcinoma (CCRCC) is the mutation of the von Hippel-Lindau tumor suppressor gene (VHL), located on chromosome 3p25. Hereditary CCRCC cases that occur as a manifestation of the autosomal dominant von Hippel-Lindau syndrome are uniformly associated with germ line VHL gene mutations that affect one of the two VHL alleles. At the molecular level, VHL disease arises from somatic loss or inactivation of the remaining wild-type allele and thus conforms to the Knudson two-hit model. The importance of VHL mutations in the pathophysiology of CCRCC, which accounts for up to 85% of all cases of renal cell carcinoma, is underscored by the fact that up to 80% of sporadic cases manifest biallelic loss/inactivation at the VHL locus as a consequence of gross genetic loss, nonsense and missense point mutations, and hypermethylation of the VHL promoter (20, 28). In contrast to CCRCCs, non-clear-cell RCCs invariably express wild-type pVHL protein (34, 47).

The pVHL protein functions as a ubiquitin ligase that targets various proteins for degradation by the 26S proteasome. A key pVHL target is hypoxia-inducible factor alpha (HIFα), a central regulator of cellular responses to hypoxia that functions as a transcription factor that stimulates expression of genes involved in angiogenesis, anaerobic metabolism, and numerous other cellular functions (28, 34). Under normoxic conditions, HIFα is hydroxylated by a family of oxygen-dependent prolyl hydroxylases (62). As a consequence, pVHL binds to and ubiquitinates HIFα, which results in its degradation by the proteasome (49). Under hypoxic conditions or when pVHL expression is lost or is functionally inactivated, HIFα accumulates. One of the two isoforms of HIFα (HIF1α or HIF2α) then translocates to the nucleus and dimerizes with the constitutively expressed HIFβ. The HIFα/HIFβ complex binds to hypoxia response elements within the promoters of target genes and thereby regulates gene transcription. Genes regulated in this fashion include growth factors (such as the gene coding for transforming growth factor α [TGF-α]), angiogenesis factors (e.g., vascular endothelial growth factor [VEGF]), and genes involved in anaerobic glucose metabolism (e.g., Glut1), as well as a multitude of other genes that modulate various cellular functions (55).

Importantly, a pathophysiologic role for VHL loss in CCRCC has been established. Restoring VHL in VHL^−/− CCRCC cells suppresses tumorigenesis in animal models of CCRCC (24), and pVHL-mediated degradation of HIFα is necessary and sufficient for the tumor-suppressive effects of pVHL (29, 30). In the context of CCRCC, biallelic VHL loss likely represents an early event in carcinogenesis. This is suggested by the observation that in von Hippel-Lindau syndrome, characterized by an inherited loss of one VHL allele and a predisposition to CCRCC, the loss of the remaining VHL allele occurs in premalignant renal lesions (37, 38). Although some of the genes regulated by HIFα, including the VEGF and TGF-α genes, have been shown to play a critical role in RCC oncogenesis, a comprehensive understanding and knowledge of the HIFα transcriptional targets and their downstream biochemistry is lacking (55).

NF-κB is a family of transcription factors that has been associated with diverse cellular functions (5). NF-κB transcriptional activity results in inhibition of apoptosis in most cell systems via induced expression of antiapoptotic proteins, such as Bcl-X_L (6, 17, 42). NF-κB activation has also been associated with proliferative responses mediated by induction of expression of cyclin D1, which drives the transition from the G₁ to the S phase of the cell cycle (19). In addition, NF-κB upregulates expression of proangiogenic factors such as interleukin-8 (IL-8) and adhesion molecules and metalloproteinases that are involved in metastasis development and plays a critical role in drug resistance (6, 16, 63).

An increasing body of evidence has implicated a specific role for heightened NF-κB activation in the oncogenesis of many hematologic malignancies and solid tumors, including RCC (5, 26, 51). The evidence for NF-κB activation in RCC is as follows. First, constitutive NF-κB activation has been observed in many RCC cell lines (44, 50). Moreover, inhibition of NF-κB sensitizes RCC cells to tumor necrosis factor alpha (TNF-α) (50), TNF-α-related apoptosis-inducing ligand (44), and the proteasome inhibitor bortezomib (1). The in vivo evidence for the role of NF-κB in RCC is highlighted by a recent study demonstrating that heightened NF-κB activation is associated with the development and progression of RCC in actual patients (45).

Recently, we and others demonstrated that VHL loss induces heightened activity of NF-κB (1, 50), although the biochemical mechanism that underlies pVHL-mediated suppression of NF-κB has not been elucidated. Given the central role of biallelic inactivating VHL mutations in hereditary and sporadic CCRCC, we sought to identify the biochemical link between VHL loss and increased NF-κB activity.

MATERIALS AND METHODS

Cell lines and cell culture.

All versions of 786-0 cells have been described previously (29, 30) and are as follows: 786-0-v (v) represents stable transfection of empty control vector, 786-0-VHL (VHL) represents stable transfection of wild-type VHL, 786-0-v/v (v/v) represents stable transfection of two empty control vectors, 786-0-VHL/v (VHL/v) represents stable transfection of wild-type VHL and one control vector, and 786-0-VHL/HIF2α^M (VHL/HIF2α^M) represents stable transfection of VHL and a HIF2α mutant that is resistant to pVHL-mediated ubiquitination (29). VHL/HIF2α^M contains double-proline mutations in the oxygen-dependent degradation domain of HIF2α (29). Cells were maintained in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum (FBS) and antibiotics. The isogenic pair of UMRC6 cells with (UMRC6-VHL) and without (UMRC6-v) stable transfection of wild-type VHL were also cultured in Dulbecco's modified Eagle's medium plus 10% FBS, as described previously (10).

Treatment of cells with chemical inhibitors.

The epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI; PD153035) (43), the Raf1 kinase inhibitor {5-iodo-3-[(3,5-dibromo-4-hydroxyphenyl)methylene]-2-indolinone} (31), the AKT inhibitor [1L-6-hydroxymethyl-chiro-inositol 2-(R)-2-O-methyl-2-O-octadecylcarbonate] (23), and the phosphotidylinositol-3-OH kinase (PI3K) inhibitors (wortmannin and LY294002) were purchased from Calbiochem (La Jolla, CA) and dissolved in dimethyl sulfoxide. The final concentration of dimethyl sulfoxide for all experiments was 0.1%. Cells were exposed to inhibitors for 48 h at various concentrations.

Plasmids.

The NF-κB- and cyclic AMP response element (CRE)-driven reporter constructs (pκB-luc and pCRE-luc) were from BD Sciences, Clontech, as was the pRL-SV40 plasmid. To inhibit AKT, we employed a plasmid with a kinase-dead AKT transgene (AKT-DN) (58). To activate AKT, we used a dominant-active AKT (AKT-DA) in which a point mutation of the pleckstrin homology domain confers increased affinity for phospholipid second messengers (3).

Measurement of NF-κB activity.

NF-κB activity was measured by both electrophoretic mobility shift assay (EMSA) and NF-κB-driven reporter gene expression as previously described (2). For transient transfections, the NF-κB- and CRE-driven reporter constructs (pκB-luc and pCRE-luc; BD Sciences, Clontech) were transfected with Lipofectamine Plus (Invitrogen) according to the manufacturer's instructions. The pRL-SV40 plasmid was cotransfected to normalize for transfection efficiency. Protein was harvested 48 h posttransfection. Transfections were performed in triplicate, and the overall experiments were repeated at least twice to confirm the repeatability of the results. Total DNA was held constant with empty control vectors in all transfection experiments. Normalized values are reported as the mean ± standard deviation from triplicate transfections.

Immunoblotting and antibodies.

Immunoblotting was performed as previously described (1). Antibodies against pAKT (Thr308), pAKT (Ser473), total AKT, pJNK, total JNK, phospho-extracellular signal-regulated kinase (pERK) and total ERK, phospho- and total signal transducers and activators of transcription 1 (STAT1), -3, and -5, I kappa B kinase alpha (IKKα), pEGFR, and total EGFR were obtained from Cell Signaling Technology. The monoclonal HIF2α antibody was from Novus Biologicals, the VHL antibody was purchased from BD Pharmingen, and the actin and γ-tubulin antibodies were from Sigma.

In vitro kinase assay.

The AKT in vitro kinase assay was performed with a kit from Cell Signaling Technology. Whole-cell extracts were immunoprecipitated with a total AKT antibody, and phosphorylation of a GSK-3β recombinant substrate was detected by immunoblotting with a phospho-specific GSK-3β antibody. To determine if AKT could phosphorylate IKKα, we employed a variation of this assay in which incorporation of ³²P into a recombinant glutathione S-transferase (GST)-IKKα substrate (Cell Signaling Technology) was assayed as previously described (46).

RNA interference.

Oligofectamine (Invitrogen) was used to transfect small interfering RNA (siRNA) (1,800 pmol per 10-cm dish), according to the manufacturer's instructions. In reporter experiments, reporter plasmids (100 ng) and siRNA (60 pmol) were cotransfected with Lipofectamine Plus (Invitrogen) in a 24-well format.

The double-stranded siRNAs for HIF2α (sense, 5′-CAGCAUCUUUGAUAGCAGUdTdT-3′) and HIF1α (sense, 5′-CUGAUGACCAGCAACUUGAdTdT-3′) were purchased from Dharmacon Research and designed as previously described (59). A Raf1 siRNA (sense, 5′-GGUAAAAAAGCACGCUUAdTdT-3′) was purchased from Ambion, Inc. (Austin, TX) (43). A scrambled control sequence was used as a negative control (Dharmacon).

Ad vectors.

Adenoviral (Ad) vectors containing the phosphatase and tensin homolog (Ad-PTEN) (57), the IκB super repressor (Ad-IκB-SR) (7), the AKT-DA transgene (22), and the control vector (adenovirus-cytomegalovirus construct [Ad-CMV]) were amplified and the titer was determined in 293 cells. The IκB-SR transgene contains mutations at the IκB phosphorylation sites (Ser³² to Ala and Ser³⁶ to Ala), which prevent its phosphorylation, dissociation from NF-κB, and subsequent degradation by the ubiquitin-proteasome pathway, and thereby block NF-κB activity. Pilot experiments performed with an adenoviral vector containing the enhanced green fluorescent protein (EGFP) transgene (Ad-EGFP) demonstrated that >90% of 786-0 and UMRC6 cells expressed EGFP at a multiplicity of infection of >5.

Cell viability assay.

786-0 cells were transferred to serum-free medium supplemented with insulin-transferrin-selenium (ITS), as previously described (13, 18). After overnight growth in ITS medium, cells were exposed to either 10% FBS or various chemical inhibitors or transduced with adenoviral constructs at a multiplicity of infection of 5. After 48 h, viable cells were enumerated by trypan blue exclusion on a hemocytometer. Similar experiments were performed for UMRC6 cells, except that UMRC6 cells were maintained in 10% FBS throughout the experiments.

RESULTS

A principal ubiquitination target of pVHL is either of the two HIFα isoforms, HIF1α and HIF2α (49). Thus, we first tested whether pVHL-mediated regulation of NF-κB activity was HIFα dependent. The CCRCC cell line, 786-0, manifests biallelic VHL mutations and high expression of HIF2α (Fig. 1A) but not HIF1α (30). We used a panel of 786-0 isogenic lines that were engineered to stably express pVHL and/or HIF2α (Fig. 1A). As we and others have previously reported (1, 50), 786-0 cells stably transfected with vector only (786-0-v) demonstrate high constitutive NF-κB activity, and upon VHL restoration (786-0-VHL) with consequent loss of HIF2α expression, NF-κB activation is markedly reduced (Fig. 1D, lanes 1 and 2). To prove that the decreased NF-κB activity observed with VHL restoration is due to HIF2α degradation, we silenced expression of HIF2α in 786-0-v cells by transient transfection of siRNA that specifically targets HIF2α. HIF2α siRNA but not scrambled control siRNA markedly reduced HIF2α protein expression (Fig. 1B, top panels), which was associated with reduction in NF-κB activation as measured by EMSA and NF-κB-driven reporter gene expression (Fig. 1B, bottom panel, and C, respectively). As a control for the specificity of HIF2α gene silencing, we found that HIF2α siRNA had no effect on a CRE-driven reporter construct (data not shown). When HIF2α was expressed in 786-0-VHL cells by stable transfection of a HIF2α mutant that is resistant to pVHL-mediated ubiquitination and yet maintains transcriptional activity (786-0-VHL/HIF2α^M cells, controls for which are 786-0-v/v and 786-0-VHL/v) (29, 30), high NF-κB activity was reestablished (Fig. 1D, compare lanes 3 to 5). Thus, HIF2α expression is both necessary and sufficient for the heightened NF-κB activity observed in a VHL^−/− background.

FIG. 1.

HIFα is necessary and sufficient for pVHL-mediated modulation of NF-κB activity. (A) Immunoblots for pVHL and HIF2α (nuclear protein). 786-0-v (v) represents stable transfection of empty control vector; 786-0-VHL (VHL) represents stable transfection of wild-type VHL; 786-0-v/v (v/v) represents stable transfection of two empty control vectors; 786-0-VHL/v (VHL/v) represents stable transfection of wild-type VHL and one control vector; and 786-0-VHL/HIF2α^M (VHL/HIF2α^M) represents stable transfection of VHL and a HIF2α mutant that is resistant to pVHL-mediated ubiquitination. Western blots for actin and γ-tubulin (nuclear protein) are protein loading controls. (B) Effects of gene silencing of HIF2α on NF-κB activity. (Top panel) Immunoblots demonstrate that HIF-2α siRNA selectively silences HIF2α protein expression. (Bottom panel) HIF2α siRNA reduces NF-κB activity as measured by EMSA. Oct-1 EMSA is a negative control (cont) (C) Transient cotransfection of HIF2α siRNA and an NF-κB-driven reporter. Transient transfection experiments were performed in triplicate, and data represent the means of the results and are reported as relative luciferase units (RLU) ± standard deviation. (D) Effects of restoration of pVHL and HIF2α expression on NF-κB activity. (Top and bottom panels) EMSA for NF-κB activity (top) and Oct-1 (bottom) as a negative control. (E) Effects of HIF1α siRNA on HIF1α protein expression and NF-κB activity in UMRC6 cells. (Top and middle panels) HIF1α siRNA reduces HIF1α but not γ-tubulin expression. (Bottom panel) EMSA for NF-κB activity shows that silencing of HIF1α expression reduces constitutive NF-κB activity. At the far right of the EMSAs are cold competition experiments with molar excess of cold wild-type (WT) and mutant (M) κB and Oct-1 probes, respectively.

We have previously shown that suppression of NF-κB activity by pVHL occurs in RCC cells irrespective of the HIFα isoform that is expressed (1). Thus, to extend our findings to VHL^−/− RCC cells that express HIF1α, we studied the effects of gene silencing of HIF1α on the constitutive activity of NF-κB in UMRC6 cells. VHL^−/− UMRC6 cells manifest HIF1α expression and NF-κB activation, which is markedly reduced when pVHL expression is restored (1). Transfection of siRNA specific for HIF1α into UMRC6-v cells abolished HIF1α expression and NF-κB activity (Fig. 1E) and thus confirms that the phenomenon of HIFα-mediated regulation of NF-κB activity is applicable to both HIFα isoforms (i.e., HIF1α and HIF2α).

When HIFα isoforms accumulate due to VHL loss or hypoxia, they translocate to the nucleus and bind to the constitutively expressed dimerization partner of HIFα, HIFβ, and induce transcription of a wide array of genes (55). Of the many transcriptional targets of HIFα, the growth factor TGF-α has been shown to be overexpressed in VHL^−/− CCRCC patient samples (4, 27, 32, 41, 48, 52). Overexpression of TGF-α is driven by HIFα in VHL^−/− CCRCC cells and promotes the growth of these cells by serving as a ligand for EGFR, which is also overexpressed in CCRCC (4, 13, 18, 27, 32, 41, 48, 52). Moreover, TGF-α overexpression in 786-0-v cells is sufficient to account for heightened EGFR activation in these cells as compared to that in 786-0-VHL cells (13, 18), and inhibition of the EGFR has equivalent tumor-suppressive effects to reintroduction of VHL in VHL^−/− CCRCC murine xenograft models (11, 24). Because ligation of the EGFR can result in stimulation of multiple signaling cascades (25), including the PI3K/AKT and Ras/Raf/ ERK pathways that can activate NF-κB (40, 46), we examined whether HIFα-induced activation of NF-κB was mediated through the EGFR.

In the absence of pVHL expression, EGFR activation was elevated, and upon restoration of pVHL expression, EGFR activation was reduced (Fig. 2A, lanes 1 to 4). Introduction of the VHL-resistant HIF2α mutant resulted in the reestablishment of heightened EGFR activation (Fig. 2A, compare lanes 3 to 5). Inhibition of EGFR activation with the EGFR-TKI, PD153035 (15), not only inhibited EGFR phosphorylation in 786-0-v and 786-0-VHL cells (Fig. 2B) but also effectively blocked NF-κB activation in a dose-dependent fashion, as measured by EMSA and reporter assays (Fig. 2B and C). The effects of EGFR on reducing NF-κB activity were observed not only in 786-0-v cells but also in 786-0-VHL cells, because the latter, despite absent HIF2α expression, still manifests TGF-α expression and EGFR activation, albeit at lower levels than in 786-0-v cells (13, 18). As a negative control, we showed that PD153035 had no effect on a CRE-driven reporter construct (data not shown). These results, along with the well-established fact that HIF2α-dependent TGF-α expression activates the EGFR in 786-0 cells (13, 18), indicate that HIF2α augments NF-κB activation through a TGF-α/EGFR-dependent pathway.

FIG. 2.

EGFR-mediated activation of NF-κB. (A) Effects of restoration of pVHL and HIF2α expression on EGFR activity. Whole-cell extracts were immunoprecipitated (IP) with an anti-EGFR antibody followed by immunoblotting (Western blotting [WB]) with phospho-specific and total EGFR antibodies. (B and C) Inhibition of EGFR activity by PD153035 (top panels) results in reduced NF-κB activity as measured by EMSA (bottom) and NF-κB-driven reporter gene expression (C) (mean ± standard deviation). RLU, relative luciferase units.

We next sought to identify the signaling pathway(s) responsible for HIFα-dependent, EGFR-induced augmentation of NF-κB activity. We screened the Ras/Raf/ERK, PI3K/AKT, Janus kinase (JAK)/STAT, and c-Jun N-terminal kinase (JNK) pathways for differential activation in relation to pVHL and HIF2α expression. Expression of pVHL and HIF2α did not affect the phosphorylation status of JNK or STAT1, -3, or -5 (Fig. 3A and B). In contrast, phosphorylation of ERK was increased in VHL^−/− cells, suppressed upon VHL restoration, and reinduced when the pVHL-resistant HIF2α was coexpressed with pVHL (Fig. 3C, top panels). Moreover, gene silencing of HIF2α in 786-0-v cells resulted in a reduction in ERK phosphorylation to an equivalent level as 786-0-VHL cells (Fig. 3C, bottom panels).

FIG. 3.

VHL- and HIF2α-dependent differential activation of the Ras/Raf/ERK and AKT but not JAK/STAT or JNK pathways. (A) Activation status of STAT1, -3, and -5. There was no detectable phosphorylation of STAT1 or -5 (top and bottom panels, respectively). STAT3 is constitutively but not differentially activated in all 786-0 clones (middle panels). Treatment of HeLa cells with alpha interferon (IFN-α) for 30 min was a positive control for STAT activation. (B) Immunoblotting with a phospho-specific JNK antibody. Treatment of human embryonic kidney 293 cells with tetradecanoyl phorbol acetate (TPA ester; 20 ng/ml for 30 min) served as a positive control for JNK activation. (C) Immunoblotting for phospho-ERK as a measurement of Ras/Raf/ERK pathway activation (top panels). Gene silencing of HIF2α reduces ERK phosphorylation (bottom panels). (D) Assessment of constitutive AKT activation. (Top panels) Immunoblotting (Western blotting) [WB] with phospho-specific AKT antibodies. (Bottom panels) In vitro kinase assays for AKT activation. Densitometry readings are provided above the autoradiograph; densitometric readings from 786-0-VHL/v cells were assigned a value of 1.0. IP, immunoprecipitation. (E) Suppression of HIF2α by siRNA reduces AKT activation as measured by immunoblotting with phospho-specific AKT antibodies. (F) Inhibition of EGFR activation with PD153035 reduces phosphorylation of AKT and ERK in 786-0-v and 786-0-VHL cells.

Like ERK, heightened activation of AKT was observed in VHL^−/− cells compared to that in 786-0-VHL cells, and introduction of the pVHL-resistant HIF2α mutant into 786-0-VHL cells restored the heightened AKT activation, as shown by immunoblotting with phospho-AKT antibodies and confirmed by an AKT in vitro kinase assay (Fig. 3D). In addition, HIF2α gene silencing by transfection of HIF2α-specific siRNA into 786-0-v cells resulted in decreased AKT activation to a level similar to that observed in 786-0-VHL cells (Fig. 3E). Thus, HIF2α expression is necessary and sufficient to explain the difference in constitutive ERK and AKT activation between VHL^−/−- and pVHL-expressing RCC cells. Furthermore, to establish a causative link between EGFR activation and activation of the Ras/Raf/ERK and PI3K/AKT pathways, we showed that inhibition of EGFR activation with PD153035 inhibited the phosphorylation of both ERK and AKT in 786-0-v cells and 786-0-VHL cells (Fig. 3F).

We subsequently determined if the differentially increased HIF2α-dependent activation of the Ras/Raf/ERK and/or the PI3K/AKT pathways was causally associated with the heightened NF-κB activity observed in VHL^−/− cells. The Raf1 kinase inhibitor 5-iodo-3-[(3,5-dibromo-4-hydroxyphenyl)methylene]-2-indolinone (31) failed to reduce NF-κB-driven reporter gene expression, although the inhibitor successfully blocked ERK phosphorylation of both 786-0-v and 786-0-VHL cells (Fig. 4A). Similarly, transfection of Raf1-specific siRNA selectively abrogated Raf1 expression but did not reduce NF-κB activity (Fig. 4B). These findings effectively exclude the Ras/Raf/ERK pathway as a mediator of HIFα-dependent activation of NF-κB in this cell model.

FIG. 4.

The Ras/Raf/ERK pathway does not regulate NF-κB activity in 786-0 cells. (A) A Raf1 inhibitor has no effect on an NF-κB-driven reporter (top panel) but does inhibit ERK phosphorylation in 786-0-v and 786-0-VHL cells (bottom panels). RLU, relative luciferase units. (B) Raf1 siRNA does not affect NF-κB activity. (Top panels) Western blots for Raf1 and actin (as a specificity/loading control). (Bottom panels) EMSAs for NF-κB and Oct-1. Cold competition experiments are shown in the rightmost two lanes of the EMSAs.

In contrast, the PI3K inhibitors wortmannin and Ly294002 inhibited both AKT phosphorylation and NF-κB activation in a dose-dependent fashion (Fig. 5). As a control for the specificity of this effect, we showed that the PI3K inhibitors had no effect on a CRE-driven reporter (not shown). Similarly, the AKT inhibitor 1L-6-hydroxymethyl-chiro-inositol 2-(R)-2-O-methyl-2-O-octadecylcarbonate (23) inhibited AKT phosphorylation and NF-κB activity in 786-0-v and 786-0-VHL cells (Fig. 6A), yet had no effect on CRE-driven reporter gene expression (not shown). A kinase-dead AKT dominant-negative (AKT-DN) construct (58) likewise inhibited NF-κB-driven reporter gene expression in 786-0-v and 786-0-VHL cells (Fig. 6B), whereas a dominant-active AKT (AKT-DA) construct (3) transfected into 786-0-VHL cells increased NF-κB-driven reporter gene expression to a level similar to that of 786-0-v cells (Fig. 6C). Neither the AKT-DN construct nor the AKT-DA construct affected the CRE-driven reporter (not shown). Thus, heightened activation of the PI3K/AKT pathway that is attributable to HIF2α expression is causally associated with augmentation of NF-κB activity.

FIG. 5.

PI3K-dependent regulation of NF-κB activity. (A) The PI3K inhibitor wortmannin inhibits NF-κB activity. (Top panels) Wortmannin reduces phosphorylation of AKT in 786-0-v and 786-0-VHL cells as measured by immunoblotting with phospho-specific AKT antibodies. (Middle panel) EMSA demonstrating that wortmannin inhibits NF-κB. In the far right two lanes are cold competition experiments with cold wild-type (WT) and mutant (M) NF-κB probes. (Bottom panel) Wortmannin inhibits NF-κB-driven reporter gene expression. (B) Same as panel A, but with the PI3K inhibitor Ly294002. Relative luciferase units (RLU) are means of three experiments ± standard deviation.

FIG. 6.

AKT regulates NF-κB activity. (A) Effects of a chemical AKT inhibitor on NF-κB activity. The AKT chemical inhibitor blocks phosphorylation of AKT (top panels) and NF-κB activity as measured by reporter gene expression (middle panel) and EMSA (bottom panel). (B) Transient transfection of a kinase-dead AKT-DN inhibits NF-κB-driven reporter gene expression in 786-0-v and 786-0-VHL cells. RLU, relative luciferase units. (C) Selective activation of AKT by transient transfection of an AKT-DA plasmid into 786-0-VHL cells is sufficient to increase NF-κB-driven reporter gene expression. Data were normalized to those of 786-0-v cells transfected with an empty control vector. (D) AKT phosphorylates IKKα. (Top panels) In vitro kinase assay of immunoprecipitated (IP) AKT with recombinant GST-IKKα as a substrate. Densitometry readings are provided above the autoradiograph; densitometric readings from 786-0-VHL cells were assigned a value of 1.0. WB, Western blotting. (Bottom panels) Inhibition of AKT phosphorylation by the AKT inhibitor (10 μM) was confirmed in the cellular extracts used for the in vitro kinase assays. All reporter gene data represent means of three experiments ± standard deviation. For all reporter assays, total DNA was held constant with a control vector.

To directly link AKT activation to the heightened NF-κB activity observed in VHL^−/− 786-0 cells, we employed an in vitro kinase assay to determine whether AKT that was immunoprecipitated from VHL^−/−- versus VHL-reconstituted cells could differentially phosphorylate recombinant IKKα, a critical kinase involved in NF-κB activation that has been previously shown to be a substrate for AKT activation (46). As shown in Fig. 6D, equal amounts of total AKT were immunoprecipitated from 786-0-v and 786-0-VHL cells, yet the immunoprecipitated AKT from VHL^−/− 786-0-v cells resulted in higher in vitro phosphorylation of recombinant GST-IKKα, which was abrogated by AKT inhibition (Fig. 6D, top panels). Inhibition of AKT phosphorylation by the AKT inhibitor was confirmed in the cellular extracts used for these kinase assays (Fig. 6D, bottom panels). Thus, increased AKT activity plays a direct role in inducing greater NF-κB activation in VHL^−/− RCC cells by phosphorylating IKKα.

TGF-α-mediated activation of EGFR has been shown to promote the growth of 786-0 cells and likely plays a contributory role in renal carcinogenesis (13, 14, 18). 786-0-v and 786-0-VHL cells grow equally in serum-replete medium. However, under serum-free conditions, 786-0-v cells exhibit increased proliferation compared to 786-0-VHL cells that is dependent upon increased TGF-α expression and EGFR activation (13, 18). We postulated that heightened activation of the EGFR/PI3K/AKT/NF-κB signaling cascade observed in VHL^−/− 786-0 cells accounts for the growth-promoting effects due to TGF-α engagement of the EGFR. To test this hypothesis, we sequentially inhibited the activity of the components of the EGFR/PI3K/AKT/NF-κB pathway and assayed overall growth of 786-0-v cells in serum-depleted medium containing ITS, as described previously (13, 18). When the PI3K/AKT pathway was inhibited in 786-0-v cells by exposure to PD153035, wortmannin, or the AKT inhibitor, the growth of these cells in ITS medium was reduced to a level similar to that in 786-0-VHL cells (Fig. 7A). Similarly, when we transduced 786-0-v cells with an adenoviral vector containing the phosphatase and tensin homolog gene (Ad-PTEN), a lipid phosphatase inhibitor of PI3K, cell viability of 786-0-v cells was reduced in 786-0-v cells as compared to cells transduced with a control virus (Ad-CMV; Fig. 7A). The cell viability of 786-0-v cells transduced with Ad-PTEN approximated that of 786-0-VHL cells transduced with control virus (Fig. 7A).

FIG. 7.

Effects of modulating components of the PI3K/AKT/NF-κB pathway on RCC cell viability. (A) 786-0 cells. Cells were incubated in ITS medium overnight prior to the addition of either 10% FBS or ITS ± the indicated inhibitor/activator (wortmannin [Wort], 100 nM; AKT inhibitor, 10 μM; and Raf1 inhibitor, 10 nM) for 48 h prior to cell counting by trypan blue exclusion. (Left panel) Relative cell number. (Right panel) Absolute number of dead (trypan blue positive) cells. (B) UMRC6 cells. Cells were treated and analyzed as in panel A, but UMRC6 cells were maintained in serum-replete medium (10% FBS) throughout the experiment. Relative cell number values represent means of three experiments ± standard deviation.

Next, we selectively inhibited NF-κB activity by transduction of an adenoviral vector that expresses the I kappa B super repressor (Ad-IκB-SR); the IκB-SR transgene contains mutations at the IκB phosphorylation sites (Ser³² to Ala and Ser³⁶ to Ala), which prevent its phosphorylation, dissociation from NF-κB, and subsequent degradation by the ubiquitin-proteasome pathway, and thereby block NF-κB activity. Transduction of the Ad-IκB-SR inhibited the growth of 786-0-v cells to a similar extent as observed with the Ad-PTEN vector (Fig. 7A, left panel). Importantly, inhibition of the Ras/Raf/ERK pathway with the Raf1 inhibitor had no effect on 786-0-v growth (Fig. 7A, left panel). To demonstrate that increased AKT activation is sufficient for growth of 786-0-VHL in serum-free conditions, we showed that selective activation of AKT by adenoviral transduction of an AKT-DA (22) into 786-0-VHL cells restored the growth of these cells to that of 786-0-v cells (Fig. 7A, left panel). Each of the inhibitory treatments of 786-0-v cells not only decreased the total number of viable cells but also increased the number of dead cells (Fig. 7A, right panel), which is consistent with the antiapoptotic effects of the AKT and NF-κB pathways. Activity of the Ad-PTEN, Ad-AKT-DA, and Ad-IκB-SR and vectors was confirmed by demonstrating appropriate changes of AKT phosphorylation by Western blotting and NF-κB activity by EMSA (not shown). Thus, activation of the EGFR/PI3K/AKT/NF-κB cascade is necessary and sufficient for the enhanced growth of VHL^−/− cells in serum-free medium.

To confirm these results, we performed similar studies in the isogenic pair of UMRC6 cells. Because UMRC6-v cells grow more robustly in serum-replete medium than the UMRC6-VHL counterparts (10), we investigated the role of the PI3K/AKT/NF-κB pathway in promoting the in vitro growth of UMRC6-v cells. Exposure of UMRC6-v cells to wortmannin, the AKT inhibitor, or the Ad-PTEN vector reduced their growth to a level similar to that of UMRC6-VHL cells (Fig. 7B, left panel). Activation of AKT by transduction of the Ad-AKT-DA into UMRC6-VHL cells augmented the growth of these cells to that of the VHL^−/− counterpart (Fig. 7B, left panel). In contrast to 786-0-v cells, UMRC6-v cells did manifest reduced growth in response to Raf1 inhibition (Fig. 7B, left panel). Thus, UMRC6-v cells are dependent upon activation of both the PI3K/AKT and Ras/Raf/ERK pathways for optimal growth. Consistent with this finding is the profound reduction in overall cell number upon inhibition of the EGFR with consequent blockade of both the PI3K/AKT and Ras/Raf/ERK pathways (Fig. 7B, left panel). Induction of cell death partially accounts for alterations in the relative cell numbers of UMRC6-v cells upon modulation of the EGFR and its downstream signaling pathways (Fig. 7B, right panel).

DISCUSSION

Here, we have identified the biochemical link between VHL loss and heightened NF-κB activity. Specifically, we have shown that biallelic inactivating mutations of VHL induce NF-κB activity through the accumulation of HIFα expression. In turn, HIFα drives expression of TGF-α, which consequently activates an EGFR/PI3K/AKT/IKKα/NF-κB signaling cascade that is critical to the growth of CCRCC cells. We documented HIFα-mediated activation of NF-κB in RCC cells that express HIF1α or HIF2α, which is particularly germane to RCC, because RCC patient samples frequently express HIF1α and HIF2α, either alone or in combination (60, 61).

The biochemical connection between VHL/HIFα and NF-κB may be potentially relevant not only to the majority of CCRCCs, which manifest biallelic VHL inactivation, but also to any tumor that undergoes hypoxia with consequent accumulation of HIFα. For example, hypoxia induces the development of a clinically aggressive tumor phenotype as well as resistance to chemotherapy and radiation (21, 53), but the mechanism that relates hypoxia to these tumor characteristics has remained unknown. Interestingly, hypoxia also activates NF-κB by enhancing DNA binding and increasing transcriptional activity of NF-κB in other cell types, including pancreatic carcinoma cells (53, 54, 56). Given that NF-κB activation is associated with increased proliferation, tissue invasion and angiogenesis, inhibition of apoptosis, and the development of drug resistance (5), it stands to reason that HIFα-dependent increased NF-κB activity may contribute to the effects of hypoxia on tumor progression and the development of a drug-resistant phenotype.

Hypoxia-induced NF-κB activation may also be operative in nonmalignant cells. For example, hypoxia has been shown to activate NF-κB in vascular endothelial cells (54) and macrophages (33, 35). Interestingly, tumor infiltration by macrophages is associated with a poor prognosis in many human malignancies, including cancers of the breast, prostate, ovary, cervix, lung, and bladder (8), and it is conceivable that factors elaborated by macrophages in response to NF-κB, such as IL-6, IL-8, and VEGF, may drive tumor expansion through induction of angiogenesis or perhaps direct effects on tumor cell growth (12). Given the importance of macrophages and angiogenesis in physiologic responses to wound healing, ischemia, and other disease states characterized by hypoxia (12, 33), the role of hypoxia-induced NF-κB may extend beyond the scope of tumor pathogenesis.

In the context of CCRCC, activation of the TGF-α/EGFR/PI3K/AKT/NF-κB pathway may represent an early event in carcinogenesis. This is suggested by the observation that in hereditary von Hippel-Lindau syndrome, characterized by an inherited loss of one VHL allele and a predisposition to CCRCC, the loss of the remaining VHL allele occurs in premalignant renal lesions (37, 38), which, as a consequence of HIFα accumulation, would be expected to exhibit increased activation of the TGF-α/EGFR/PI3K/AKT/NF-κB pathway. Because the growth and survival of cancers tend to be particularly dependent on oncogenic events that arise earlier in the transformation process (36), targeting the TGF-α/EGFR/PI3K/AKT/NF-κB pathway in conjunction with other cytotoxic agents may serve as a particularly attractive therapeutic approach to the management of CCRCC. For example, heightened activation of NF-κB renders RCC cells resistant to tumor necrosis factor receptor apoptosis-inducing ligand and the proteasome inhibitor bortezomib, and inhibition of NF-κB restores sensitivity to these agents (1, 44). Thus, combinations of these or other drugs along with agents that target components of the TGF-α/EGFR/PI3K/AKT/NF-κB pathway may have potential clinical activity in CCRCC patients.

Because AKT is upstream of NF-κB in the EGFR/PI3K/AKT/NF-κB signaling cascade, it is possible that modulating AKT compared to NF-κB may produce different biochemical and consequently cellular effects. For example, AKT promotes proliferation by inhibiting the activity of p21 and p27 (G₁/S-phase regulatory proteins) and glycogen synthase kinase 3-β, which phosphorylates and destabilizes cyclin D (39). AKT also fosters cellular survival by inhibiting proapoptotic bcl family members (i.e., Bad and Bim) and enhances translation of cell cycle regulatory mRNAs such as cyclin D, transcription factors (e.g., c-myc), and cytokines (e.g., fibroblast growth factor) (39).

In many cell models, both the PI3K/AKT and Ras/Raf/ERK pathways can promote proliferation and inhibit apoptosis (9, 39), yet in 786-0-v cells, Raf inhibition did not affect cell growth. In contrast, Raf inhibition did reduce the growth of UMRC6-v cells. Thus, the fate of RCC cells in response to inhibition of the Ras/Raf/ERK pathway is likely cell specific. Currently, Raf inhibitors are under active investigation in early phase clinical trials in various tumor models, including RCC. A further understanding of the biochemical mechanisms that result in sensitivity to such drugs may prove critical in predicting responses to Raf inhibition.