Biochemical Journal

Research article

Akt phosphorylates and suppresses the transactivation of retinoic acid receptor α

Harish Srinivas, Dianren Xia, Nicole L. Moore, Ivan P. Uray, Heetae Kim, Long Ma, Nancy L. Weigel, Powel H. Brown, Jonathan M. Kurie

Abstract

The transactivation of nuclear receptors is regulated by both ligand binding and phosphorylation. We previously showed that RARα (retinoic acid receptor α) phosphorylation by c-Jun N-terminal kinase contributes to retinoid resistance in a subset of NSCLC cells (non-small cell lung cancer cells), but the aetiology of this resistance in the remainder has not been fully elucidated [Srinivas, Juroske, Kalyankrishna, Cody, Price, Xu, Narayanan, Weigel and Kurie (2005) Mol. Cell. Biol. 25, 1054–1069]. In the present study, we report that Akt, which is constitutively activated in NSCLC cells, phosphorylates RARα and inhibits its transactivation. Biochemical and functional analyses showed that Akt interacts with RARα and phosphorylates the Ser96 residue of its DNA-binding domain. Mutation of Ser96 to alanine abrogated the suppressive effect of Akt. Overexpression of a dominant-negative form of Akt in an NSCLC cell line decreased RAR phosphorylation, increased RAR transactivation and enhanced the growth-inhibitory effects of an RAR ligand. The findings presented here show that Akt inhibits RAR transactivation and contributes to retinoid resistance in a subset of NSCLC cells.

  • Akt
  • non-small cell lung cancer cell (NSCLC cell)
  • phosphorylation
  • retinoic acid receptor α (RARα)
  • retinoid resistance
  • transactivation

INTRODUCTION

RARs [RA (retinoic acid) receptors] are ligand-dependent transcription factors that regulate genes involved in various biological processes, such as pattern formation, reproduction, cell differentiation and maintenance of tissue homoeostasis [1,2]. RARs and RXRs (retinoid X receptors) bind to their target response elements as heterodimers and regulate gene expression in response to their ligands, all-trans RA and 9-cis RA [3]. In the absence of ligand, RXR–RAR heterodimers associate with a multiprotein complex containing transcriptional co-repressors that induce histone deacetylation, chromatin condensation and transcriptional suppression. Ligand binding causes the receptors to dissociate from the co-repressors and associate with co-activators that have histone acetyltransferase activity and induce local chromatin decondensation, recruitment of the RNA polymerase II holoenzyme and activation of target gene transcription [4,5].

Three major classes of co-activators have been studied in detail regarding their role in co-activating nuclear receptors. Members of the Swi/Snf/Brg class are involved in ATP-dependent remodelling of chromatin [6]. Members of the second class, which includes CBP [CREB (cAMP-response-element-binding protein)-binding protein]/p300, P/CAF (p300/CBP-associated factor) and p160/SRC (steroid receptor co-activator) have intrinsic histone acetyltransferase activity and have a crucial role in histone modification of transcriptionally active genes [7]. Members of the third class belong to TRAP (thyroid hormone receptor-associated protein)/DRIP (vitamin D receptor-interacting protein)/ARC (activator-recruited cofactor) complexes, interact with ligand-bound nuclear receptors and stimulate transcription of in vitro DNA templates [8]. Co-activators contain one to several copies of leucine-rich LXXLL motifs in their receptor interaction domains. Crystallographic and biochemical studies have revealed that these LXXLL motifs interact with the ligand-activated AF-2 (activation function-2) domain of nuclear receptors, thereby providing a molecular basis for nuclear receptor-co-activator recruitment [911].

In addition to being activated by RA, RAR transactivation is regulated by phosphorylation. RARs are substrates for a variety of serine/threonine kinases, including PKA (protein kinase A), PKC, CDK7 (cyclin-dependent kinase 7) and p38 [1216]. PKC phosphorylates RARα and strongly reduces its transactivation by inhibiting its dimerization with RXRα [13], whereas CDK7 increases RAR transactivation by phosphorylating its A/B domain [14]. We have shown that activation of JNK (c-Jun N-terminal kinase) by stress signals leads to phosphorylation of RARα, resulting in ubiquitin-mediated proteasomal degradation of the receptor [17]. Another serine/threonine kinase previously shown to phosphorylate nuclear receptors is PKB/Akt, a downstream mediator of class I PI3K (phosphoinositide 3-kinase) [18,19]. Akt phosphorylates the AF-1 domain of oestrogen receptor-α, thereby increasing its activity in a ligand-independent manner [20]. In contrast, Akt suppresses the transactivation of the androgen receptor [21]. Beyond its role in regulating nuclear receptor function, Akt is the product of a proto-oncogene and phosphorylates a number of substrates that are important regulators of cell survival and transformation [22].

We previously showed that RARα is aberrantly phosphorylated in lung cancer cells by JNK and that RARα phosphorylation contributes to the blockade of retinoid signalling observed in a subset of lung cancers [17]. However, JNK is not responsible for the retinoid signalling defects observed in most of the retinoid-resistant NSCLC (non-small cell lung cancer) cell lines. Retinoids have demonstrated efficacy in the treatment of certain cancers but are ineffective in the treatment of lung cancer patients in part due to a defect in the transactivation properties of RARs in lung cancer cells, the aetiology of which has not been fully defined [23]. In the present study, we examined whether this retinoid resistance is instead mediated by Akt, which is constitutively active in NSCLC cell lines through amplification of genes encoding the p110α catalytic subunit of PI3K (PI3KCA) and ErbB family members, epigenetic silencing of the PTEN gene and activating mutations in the genes PI3KCA, KRAS and EGFR [2427]. We found that Akt phosphorylates RARα at Ser96 and inhibits its transactivation. We also found that inhibition of Akt enhances RAR activity and the growth-inhibitory effects of retinoids on NSCLC cells. Together, our findings support the hypothesis that Akt contributes to the retinoid resistance commonly found in NSCLC cells.

EXPERIMENTAL

Reagents

We purchased all-trans RA, TTNPB {E-4-[2-(5,5,8,8-tetramethyl-5,6,7,8-tetrahydro-2-naphthalenyl)-1-propenyl]benzoic acid}, 9-cis RA and IGF-I (insulin-like growth factor-I) from Sigma–Aldrich (St. Louis, MO, U.S.A.); [γ-32P]ATP, [32P]Pi and [32P]dCTP from ICN Biomedicals (Costa Mesa, CA, U.S.A.); antibodies against human RARα and PARP [poly(ADP-ribose) polymerase] from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.); polyclonal antibodies against Akt, Ser473-phosphorylated Akt (P-Akt), Ser21/Ser9-phosphorylated GSK3 (glycogen synthase kinase 3), GSK3, Thr389-phosphorylated p70S6K (p70 S6 kinase), p70S6K and PTEN (phosphatase and tensin homologue deleted from chromosome 10) from Cell Signaling Technology (Beverly, MA, U.S.A.); monoclonal antibodies against FLAG and tubulin from Sigma–Aldrich; anti-HA (haemagglutinin) antibody from Roche Molecular Biochemicals (Indianapolis, IN, U.S.A.); and purified recombinant phosphorylated and unphosphorylated Akt from Upstate Biotechnology (Lake Placid, NY, U.S.A.).

Cell culture

COS-1, HEK-293T (human embryonic kidney 293T), HeLa and NSCLC cells (H1299, Calu-1 and H460) were purchased from the American Type Culture Collection (Manassas, VA, U.S.A.). COS-1, HEK-293T and HeLa cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum at 37 °C. NSCLC cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum at 37 °C.

cDNA constructs

HA–Akt, HA–Akt (T308D/S473D or DD mutant) and HA–Akt (K179A/T308A/S473A or AAA mutant) expression vectors were gifts from Dr Gordon Mills (M. D. Anderson Cancer Center). The Myr–Akt (where Myr stands for myristoylation) vector was constructed by inserting the myristoylation/palmitoylation signal of Lck tyrosine kinase [28] into the N-terminus of Akt. The FLAG–RARα, RARβ-luc, DR5TK-luc (where DR5 is direct repeat separated by 5 nt, and TK is thymidine kinase), TK-luc, GALTK-luc (where GAL is GAL4 DNA-binding sequence) and GST (glutathione S-transferase)–RARα deletion constructs have been described elsewhere [17]. The GAL–RARα (where GAL is yeast GAL4 protein) expression plasmids were constructed by cloning the cDNA encoding RARα into the pFA-CMV vector (Stratagene, La Jolla, CA, U.S.A.). The S95A, S96A, S115A, S154A and S157A mutations were introduced into the GAL–RARα by site-directed mutagenesis using the QuikChange® mutagenesis kit (Stratagene).

Purification of GST-tagged proteins

GST-tagged wild-type RARα and deletion mutants were expressed and purified from the BL21 bacterial strain (Stratagene) according to manufacturer's instructions. The integrity of the purified proteins was assayed by Coomassie Blue staining and Western blotting with anti-GST antibodies.

In vivo labelling and in vitro kinase assay

For the in vivo labelling experiments in HeLa cells, Lipofectamine™ 2000 (Invitrogen Life Technologies, Carlsbad, CA, U.S.A.) was used to co-transfect HeLa cells (1.5×106 cells per 10 cm dish) with plasmids expressing HA–Akt or Myr–Akt and FLAG-tagged, wild-type RARα. After 24 h, the cells were serum-starved overnight in phosphate-free medium and then incubated in the same medium for 4 h in the presence of 250 μCi/ml [32P]Pi. Where indicated in the Figure legends, the cells were exposed to IGF-I (100 ng/ml) 1 h before harvesting. Cells were lysed in radio-immunoprecipitation assay buffer containing 100 nM okadaic acid, 1 mM orthovanadate and protease inhibitor cocktail (Sigma–Aldrich). FLAG-tagged RARα was immunoprecipitated with anti-FLAG antibodies and the immunoprecipitates were subjected to electrophoresis and autoradiography. For in vivo labelling experiments in Calu-1 cells, cells were incubated with 250 μCi/ml [32P]Pi for 8 h and RARα was immunoprecipitated with anti-RARα antibodies.

For the in vitro kinase assays, recombinant active Akt was incubated with 2 μg of substrate proteins in kinase buffer containing 25 mM Hepes, 25 mM MgCl2, 25 mM β-glycerophosphate, 20 μM ATP, 0.5 mM dithiothreitol and 10 μCi of [γ-32P]ATP for 30 min at 37 °C. The labelled proteins were separated by SDS/PAGE and detected by autoradiography.

For stoichiometric analysis of phosphorylation, 0.75 μg of purified GST–RARα was incubated with increasing amounts of recombinant active Akt in kinase buffer containing 5 μCi of [γ-32P]ATP for 30 min. The labelled proteins were separated by SDS/PAGE. A GST–RARα band that exhibited saturated levels of phosphorylation was excised from the gel, and radioactivity in the protein was quantified by liquid-scintillation counting. The amount (mol) of phosphate incorporated per mol of GST–RARα was calculated.

Phospho amino acid analysis

GST–RARα DBD (DNA-binding domain) (5 μg) was subjected to in vitro Akt kinase reaction and the labelled protein was separated by SDS/PAGE. The phosphorylated band corresponding to GST–RARα DBD, visualized by autoradiography, was excised and phospho amino acid analysis was performed as described previously [29].

Luciferase assays

COS-1, HEK-293T and NSCLC cells (8×104 cells/well) were transfected in 24-well plates with the DR5TK-luc (200 ng), RARβ-luc (200 ng) or GALTK-luc (50 ng) reporter constructs, along with plasmids expressing Akt or GAL–RAR fusion proteins (50 ng). pRLTK-luc (50 ng) was transfected as an internal control. The total amount of plasmid was kept to 1 μg/well. Cells were serum-starved overnight 24 h later and treated with all-trans RA or TTNPB (10−6 M) for 15 h, and cell lysates were assayed for luciferase activity using a dual-luciferase reporter assay system (Promega, Madison, WI, U.S.A.). Results were expressed as relative luciferase activity, the ratio of firefly to Renilla luciferase activity. Each experiment was performed three times, and results from a representative experiment are shown as the means±S.D. (indicated by error bars) from three identical wells.

EMSA (electrophoretic mobility-shift assay)

DR5 oligonucleotides (5′-AGCTTAAGGGTTCACCGAAAG-TTCACTCGCAT-3′) and an NF-κB (nuclear factor κB) binding site used as non-specific oligos (5′-TTGGGGGAAGGG-GGAATCTCTAGGCAAAGG-3′) were used in the present study. Recombinant GST–RARα (40 ng) and in vitro-translated RXRα were incubated with 32P-labelled DR5 oligos for 30 min at 4 °C in a binding buffer (10 mM Tris/HCl, 40 mM KCl, 6%, v/v, glycerol, 1 mM dithiothreitol and 0.05% Nonidet P40). Where indicated in the Figure legends, GST–RARα was pre-incubated with active Akt kinase for 20 min in the presence of 1 mM ATP at 30 °C. The RXR–RAR heterodimers bound to an RARE (RA response element) were resolved on 6% (w/v) polyacrylamide gels and subjected to autoradiography.

GST pull-down assays

GST and wild-type GST–RARα (3 μg) were immobilized on glutathione–agarose beads and incubated with purified Akt for 30 min on ice in binding buffer [buffer A (25 mM Tris/HCl, 150 mM NaCl and 0.1% Triton X-100)]. Protein complexes were washed twice with the same buffer, and bound materials were resolved by SDS/PAGE and subjected to Western blotting with anti-Akt antibodies.

Co-immunoprecipitation experiments

To study Akt–RAR interactions, COS-1 cells were transfected with plasmids expressing HA–Akt and FLAG–RARα. After 48 h, the cells were lysed in buffer A, and the lysates were incubated with anti-FLAG or control IgG antibodies at 4 °C. The immunoprecipitates were washed twice with buffer A and subjected to Western blotting with anti-HA antibodies.

Generation of stably transfected cell lines

H1299 and Calu-1 NSCLC cells were transfected with HA–Akt(DD) and HA–Akt(AAA) constructs respectively using Lipofectamine™ (Invitrogen). Transfectants were selected in G418 (600 μg/ml) for 2 weeks, and individual clones were isolated and later analysed for expression of transfected gene products.

Quantitative PCR

H1299 cells stably transfected with empty vector or Akt(DD) were treated with RA (10−6 M) for 5 h. The cells were harvested, and total RNA was extracted using an RNA isolation kit (Qiagen, Valencia, CA, U.S.A.). Reverse-transcriptase reactions were performed with 1 μg of total cellular RNA using specific reverse primers for RARβ and cyclophilin. The reverse-transcriptase reaction (10 μl) was added to 40 μl of PCR mixture containing 1×PCR buffer (20 mM Tris, pH 8.4, and 50 mM KCl), 400 nM primers, 5 mM MgCl2, ROX reference dye (Invitrogen), Taq polymerase and 100 nM fluorogenic probe labelled with 6-carboxyfluorescein and a quencher. Amplification and quantification were based on real-time monitoring of probe degradation and were carried out using an ABI Prism 7700 performing 40 cycles at 94 °C for 12 s and 60 °C for 30 s. RARβ transcript levels were normalized to those of cyclophilin mRNA, which was used as a control.

Cell proliferation assays

Calu-1 cells (1×103) stably transfected with empty vector or Akt(AAA) construct were grown in RPMI 1640 medium containing 1% fetal bovine serum in 96-well plates. After 24 h, the cells were treated for 8 days with different concentrations of TTNPB as indicated in the Figure. Ligand was replenished every 48 h. Relative cell density was measured by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide) assay, as described elsewhere [30].

Statistical analysis

Statistical analyses were performed using one-way ANOVA test for luciferase assays and two-way ANOVA test for cell proliferation assays.

RESULTS

Akt represses ligand-dependent activation of retinoid receptors

We began investigating the effect of Akt on retinoid signalling with HEK-293T cells, because they express functional RAR and RXR proteins that transcriptionally activate RAREs in response to ligand [31]. We transiently transfected HEK-293T cells with a reporter construct containing the DR5 response element to which the RXR–RAR heterodimeric complex binds. Treatment with the RAR ligand all-trans RA led to a 5-fold increase in DR5TK-luc activity, whereas the activity of the control TK-luc reporter was unchanged (Figure 1A). Co-transfection of increasing amounts of Myr–Akt, an activated form of Akt [28], inhibited RA-induced DR5TK-luc activity in a dose-dependent manner (Figure 1A). To control for the possibility that PH domains suppressed RAR transactivation, we transfected HEK-293T cells with dominant-negative Akt (AAA mutant), which had no effect on DR5TK activity (Figure 1A).

Figure 1 Akt down-regulates ligand-dependent activation of RXR–RAR heterodimers

(A) Left panel: DR5TK-luc reporter activity in HEK-293T cells transiently co-transfected with increasing amounts of Myr–Akt or Akt(AAA). Right panel: control TK-luc activity in HEK-293T cells, with or without Myr–Akt. (B) RARβ-luc activity in HEK-293T cells, with or without Akt constructs. Relative luciferase activity in (A, B) is the ratio of firefly to Renilla luciferase activity. *P<0.01 (n=3, one-way ANOVA test) for empty vector transfected versus Myr–Akt cells, in the presence of ligand. Expression levels of HA-tagged Akt constructs are shown by Western blotting with anti-HA antibodies. (C) Western-blot analysis of lysates of H1299 cells stably expressing empty vector (EV) or Akt(DD), using antibodies to P-GSK3 and GSK3. Results from two separate clones of Akt(DD) transfectants are illustrated. (D) Quantitative PCR analysis of RARβ mRNA levels in H1299 cells stably expressing empty vector (EV) or Akt(DD) construct (two clones).

We also co-transfected HEK-293T cells with Akt and a reporter construct driven by the RARB gene promoter, which contains the DR5 response element. Myr–Akt, but not Akt(AAA), suppressed RA-induced activation in those cells, suggesting that Akt represses the DR5 response element in the context of a natural promoter (Figure 1B). Stable transfection of H1299 NSCLC cells with Akt(DD), another constitutively active form of Akt, stimulated Akt-dependent signalling, as shown by increased expression of P-GSK3 (phosphorylated GSK3) (Figure 1C), and suppressed RA-induced RARB expression (Figure 1D). Together, these findings indicate that activation of Akt inhibits ligand-dependent activation of retinoid receptors.

Akt phosphorylates DBD of RARα

We performed in vivo labelling experiments with [32P]Pi to determine whether Akt activation increases RARα phosphorylation in cells. RARα phosphorylation increased in cells transfected with Myr–Akt (Figure 2A). To determine whether RARα is a substrate of Akt, we looked for evidence of physical interactions between the two. In COS-1 cells, HA-tagged Akt associated with FLAG-tagged RARα (Figure 2B). In vitro kinase assays showed that purified, active Akt phosphorylates GST–RARα (Figure 2C) and purified Akt kinase interacted with GST–RARα in vitro (Figure 2D). Stoichiometric analysis of the in vitro kinase reaction revealed that 33% of the purified RARα protein was phosphorylated by Akt.

Figure 2 RARα is a substrate of Akt

(A) Phosphorylation of RARα in HeLa cells. Top panel: FLAG–RARα was immunoprecipitated (IP) using anti-FLAG antibodies from lysates of cells transfected with empty vector (−) or Myr–Akt, and the immunoprecipitate was visualized by autoradiography. Bottom panel: FLAG–RARα levels were measured by Western blotting (WB) with anti-FLAG antibodies. (B) Co-immunoprecipitation of FLAG–RARα and HA–Akt in COS-1 cells. Upper panel: cell lysates were subjected to immunoprecipitation (IP) using mouse IgG or anti-FLAG antibodies followed by Western blotting with anti-HA antibodies. Lower panel: HA–Akt protein levels in cell extracts were measured by Western blotting with anti-HA antibodies. (C) In vitro kinase assay of Akt, using recombinant wild-type GST–RARα or deletion mutants as substrates. Upper panel: phosphorylated proteins were resolved by SDS/PAGE followed by autoradiography. Phosphorylated bands of the appropriate molecular masses are indicated with arrows. A non-specific, 60 kDa band was seen in all lanes. Lower panel: gel loading was verified by Coomassie Blue staining. (D) Western-blot analysis of purified Akt protein incubated with GST or GST–RARα immobilized on glutathione beads. (E) Phospho amino acid analysis of GST–RARα DBD (82–167 amino acids) following in vitro kinase reaction with Akt.

We next used deletion mutants to determine the regions of RARα that are phosphorylated by Akt. Akt phosphorylated the N-terminus (residues 1–187) and DBD, but not the LBD (ligand-binding domain), of RARα (Figure 2C). Phospho amino acid analysis of RARα DBD revealed phosphoserine residues, consistent with Ser/Thr-specific kinase activity of Akt (Figure 2E).

DBD of RARα is required for Akt-mediated repression

The inhibition of nuclear receptor function by phosphorylation may involve mechanisms that reduce DNA-binding activity, nuclear export, or stability of the receptor [3234]. We investigated the effect of Akt on the DNA-binding activity of RARs by performing EMSAs. Recombinant RARα was subjected to in vitro kinase reactions with Akt and then incubated with in vitro-translated RXRα and 32P-labelled DR5 probe. We did not observe any change in the DNA-binding activity of the RXR–RAR heterodimer after the kinase reaction (Figure 3A).

Figure 3 The DBD of RARα is required for Akt-mediated repression

(A) Analysis of DNA-binding activity of RXR–RAR heterodimers after GST–RARα was subjected to in vitro kinase reactions, with or without active Akt (P-Akt). DNA–protein complexes were resolved by SDS/PAGE and autoradiography. (B) Western blots of HEK-293T cells transiently transfected with empty vector (−) or Myr–Akt construct. Cells were lysed 48 h after transfection, and nuclear (N) and cytosolic (C) fractions were analysed for the indicated proteins. PARP and tubulin were used as loading controls for the nuclear and cytosolic fractions respectively. (C) Schematic representation of wild-type and mutant GAL–RARα constructs. (D) Effect of Myr–Akt or Akt(AAA) co-transfection on transactivation of wild-type (WT) and mutant GAL–RARα constructs. Relative luciferase activity is the ratio of firefly to Renilla luciferase activity. *P<0.01 (n=3, one-way ANOVA test) for empty vector-transfected versus Myr–Akt cells, in the presence of ligand. (E) Effect of Myr–Akt co-transfection on transactivation of GAL–RXRα. Relative luciferase activity is the ratio of firefly to Renilla luciferase activity. (F) Expression levels of GAL–RAR constructs are shown by Western blotting with anti-GAL4 antibodies. Bands of expected molecular masses are indicated with arrows. Lower-molecular-mass band in each lane may represent a truncated form of GAL chimaeras.

To examine the effect of Akt activation on RARα protein stability and localization, we transiently transfected HEK-293T cells with Myr–Akt. Nuclear and cytosolic fractions of the cells were prepared, and RARα protein levels were assayed. Myr–Akt expression did not alter the protein levels in the nuclear fraction or increase the cytoplasmic localization of RARα (Figure 3B). PARP and α-tubulin were used as loading controls for nuclear and cytosolic fractions respectively. Taken together, these results suggest that the down-regulation of RARα function by Akt does not involve changes in DNA-binding activity, stability, or localization of RARα.

We next examined whether Akt would repress RARα transactivation when fused to a heterologous DBD. We transfected HEK-293T cells with a GAL4 DBD–RARα expression plasmid, along with a reporter construct containing a GAL4 response element (GALTK-luc). Because RXR is not required, we used TTNPB, a synthetic RAR ligand, to transactivate GAL–RAR. Expression of Myr–Akt substantially inhibited ligand-induced transactivation of GAL–RARα but not GAL4 (Figure 3D). Myr–Akt had no effect on GAL–RXRα function (Figure 3E), suggesting that Akt represses retinoid signalling through RAR.

RARα contains two activation domains (AF-1 and AF-2) that are required for maximal activation by its ligand. AF-1 and AF-2 are within the A/B region and LBD of RARα respectively. We investigated which of these domains is required for the suppression of RARα transactivation by Akt. We transiently co-transfected HEK-293T cells with Myr–Akt along with GAL–RAR ΔAB (which lacks the AF-1 domain) or GAL–RAR ΔAB/DBD (which lacks both DBD and AF-1 domain). In the absence of Myr–Akt, compared with wild-type RARα, the transactivation of GAL–RAR ΔAB was decreased in the presence of ligand. This was expected because of the known synergy between the AF-1 and AF-2 domains [35,36]. The transactivation of GAL–RAR ΔAB was repressed by Myr–Akt, whereas the transactivation of the ΔAB/DBD RAR construct was resistant to down-regulation by Akt (Figure 3D), suggesting that the DBD is required for repression by Akt. To confirm this further, we transfected HEK-293T cells with GAL–RAR ΔDBD, a construct that lacks the DBD but retains both AF-1 and AF-2 domains. Myr–Akt failed to repress the transactivation of GAL–RAR ΔDBD (Figure 3D). To control for the possibility that PH domains suppressed GAL4 chimaeric protein transactivation, we transfected HEK-293T cells with Akt(AAA), which had no effect on wild-type GAL–RARα or GAL–RAR ΔDBD (Figure 3D).

Ser96 is an Akt phosphorylation site

Sequence analysis of the DBD revealed no consensus Akt phosphorylation sites. We therefore created a series of GAL–RARα constructs in which all the serine residues in this region were individually mutated to alanine residues, and we analysed the constructs for transactivation function. Myr–Akt suppressed ligand-induced transactivation of all GAL–RARα mutant constructs except the S96A mutant (Figure 4A). In the in vitro kinase assays using RARα DBD as the substrate, Akt phosphorylated wild-type GST–DBD but not the S96A mutant (Figure 4C), indicating that Ser96 is an Akt phosphorylation site.

Figure 4 Akt phosphorylates Ser96 of RARα

(A) GALTK-luc activity in HEK-293T cells transiently co-transfected with equal amounts of wild-type (WT) GAL–RARα or various mutants, in the presence or absence of the Myr–Akt or Akt(AAA) expression plasmid. (B) Expression levels of mutant RARα constructs are shown by Western blotting using anti-GAL4 antibodies. Bands of expected molecular masses are indicated with an arrow. (C) In vitro kinase assays of Akt, using GST-tagged RARα DBD and DBD S96A as substrates. Upper panel: phosphorylated proteins were resolved by SDS/PAGE followed by autoradiography. Lower panel: gel loading was verified by Coomassie Blue staining. (D) DR5TK-luc activity in COS-1 cells transiently co-transfected with wild-type RARα or RARα S96A, with or without Myr–Akt. Expression levels of FLAG-tagged wild-type and mutant RARα proteins are shown by Western blotting using anti-FLAG antibody. (E) EMSA to analyse DR5 DNA-binding activity of RAR wild-type and S96A mutant proteins in the presence of RXR. In vitro-translated proteins were incubated with radioactive DR5 probe in the presence or absence of competitor, including 10-fold excess specific or non-specific non-radioactive oligonucleotides as shown in the Figure. (F) Sequence alignment of the conserved Ser96 region of mouse (m), rat (r) and human (h) RARα, and human RARβ and RARγ subtypes. Ser96 of hRARα and corresponding serine residue of other members are shown in boldface. (G) GALTK-luc activity in HEK-293T cells transiently co-transfected with human GAL–RARβ, with or without Myr–Akt or Akt(AAA). Relative luciferase activity in (A, D, G) is the ratio of firefly to Renilla luciferase activity.

To investigate the functional significance of Ser96 phosphorylation in the context of RXR–RAR heterodimers, we examined the ability of the S96A mutation to block Akt-mediated repression of DR5 transactivation by RXR–RAR heterodimers. We observed that, in the context of a DR5 element, ligand-induced transactivation of RARα S96A mutant was less than that of wild-type RARα (Figure 4D), which was not the case in the setting of GAL chimaeric constructs (Figure 4A). Furthermore, Myr–Akt repressed ligand-induced DR5 transactivation in COS-1 cells co-transfected with wild-type RARα but not those co-transfected with RARα S96A (Figure 4D). EMSA revealed that RARα S96A bound to a DR5 element and heterodimerized with RXRα as well as wild-type RARα did (Figure 4E), indicating that the reduced DR5 transactivation by RARα S96A was not due to impaired DNA-binding or heterodimerization properties. Sequence analysis of RARβ and RARγ revealed that Ser96 is a conserved residue (Figure 4F). We therefore postulated that Akt also inhibits these RAR subtypes. Indeed, Myr–Akt, but not Akt(AAA), repressed transactivation of GAL–RARβ in HEK-293T cells (Figure 4G). Together, these findings indicate that RAR Ser96 phosphorylation is required for the suppression of RXR–RAR transactivation by Akt.

Akt contributes to retinoid resistance in NSCLC cells

Although many NSCLC cells express normal levels of retinoid receptors, they are not sensitive to the biological effects of RA in part because of a defect in RAR transactivation function [23]. To test our hypothesis that high Akt activity contributes to RAR dysfunction in NSCLC cells, we inhibited Akt signalling in two NSCLC cell lines, Calu-1 and H460, which differed with respect to basal Akt activation, as shown by Western blotting of P-Akt (Figure 5A). Akt is constitutively phosphorylated in Calu-1 cells in part due to its low expression of the PTEN lipid phosphatase [37]. Inhibition of Akt by transient transfection of the dominant-negative mutant Akt(AAA) increased DR5TK activity in Calu-1 cells but not in H460 cells (Figure 5B), supporting a role for Akt in the suppression of RAR function in Calu-1 cells.

Figure 5 Akt contributes to retinoid resistance in NSCLC cells

(A) Western-blot analysis of lysates of Calu-1 and H460 cells, using antibodies to P-Akt, Akt and PTEN. (B) DR5TK-luc activity in Calu-1 and H460 cells, in the presence or absence of the dominant-negative Akt(AAA) expression construct. Relative luciferase activity is the ratio of firefly to Renilla luciferase activity. (C) Western-blot analysis of lysates of Calu-1 cells stably expressing empty vector (EV) or Akt(AAA) construct, using antibodies to P-p70S6K and p70S6K. Results are illustrative of two separate Akt(AAA) transfectants. (D) Phosphorylation of RARα in Calu-1 cells. Upper panel: after metabolic labelling, FLAG–RARα was immunoprecipitated from cell lysates with anti-RARα antibodies, and the immunoprecipitate was visualized by autoradiography. Lower panel: RARα levels were measured by Western blotting with anti-RARα antibodies. (E) MTT assay of Calu-1 cells stably expressing empty vector or Akt(AAA), in the absence or presence of different concentrations of TTNPB. Results are illustrative of two separate Akt(AAA) transfectants. Values are expressed as means±S.D. from five identical wells and statistical analysis was performed by two-way ANOVA test.

Finally, we examined whether the high Akt activity in Calu-1 cells contributes to the resistance of those cells to growth inhibition by the RAR ligand. Calu-1 cells stably transfected with Akt(AAA) displayed a block in Akt signalling, as shown by Western blotting for P-p70S6K (Figure 5C). Metabolic labelling studies revealed that RARα was phosphorylated to a lesser extent in Calu-1 Akt(AAA) transfectants than controls (empty vector transfectants) (Figure 5D). Compared with its effect on controls, TTNPB more profoundly inhibited the proliferation of Akt(AAA) transfectants (Figure 5E). Together, these findings support the hypothesis that Akt contributes to retinoid resistance in a subset of NSCLC cells.

DISCUSSION

In the present study, we explored the role of Akt in the regulation of RAR function and found that Akt phosphorylates RARα and suppresses its transactivation. We identified an Akt phosphorylation site on Ser96 within the DBD of RARα. Furthermore, Akt inhibition enhanced retinoid sensitivity in an NSCLC cell line. These findings indicate that RAR is a novel Akt substrate, that Akt is important in the regulation of RAR function, and that constitutive Akt activation may contribute to retinoid resistance in a subset of NSCLC cells.

Several lines of evidence suggest that RARα is a direct substrate of Akt. We showed that RARα physically associates with Akt in vitro (Figure 2D) and in vivo (Figure 2B). Akt also increased the phosphorylation of RARα in vitro (Figure 2C) and in vivo (Figure 2A). Mutation of Ser96 of the DBD to an alanine residue prevented Akt-mediated phosphorylation (Figure 4C) and restored transactivation (Figure 4A). Finally, Ser96 is conserved across species and RAR subtypes (Figure 4F). These results suggest that Ser96 is a bona fide Akt phosphorylation site. However, Ser96 does not resemble the consensus Akt phosphorylation site (RXRXXS/T) [38], and we do not know the mechanism by which Akt recognizes this region. There are other examples of Akt substrates that do not have the consensus motif, such as CREB and YB-1 (Y-box binding protein-1) [39,40]. Furthermore, although mitogen-activated protein kinases usually phosphorylate serine or threonine residues that are followed by proline, there are several physiological substrates such as TAB1 [TAK-1 (transforming growth factor-β-activated kinase-1) binding protein], cdc25 and microtubule-associated protein Tau where this is not observed [4143].

We showed that the DBD of RARα is important for mediating repression by Akt (Figure 3D), indicating that the DBD has functions other than DNA binding. Although nuclear receptors usually interact with co-regulatory proteins through their LBDs, there are several reports of DBDs being used as docking sites for co-activator recruitment. For example, Jun dimerization protein-2 interacts with the DBD of the progesterone receptor and stimulates AF-1 function by recruiting other co-activators such as CBP and P/CAF [44]. Similarly, the PPARγ (peroxisome-proliferator-activated receptor γ) co-activator PGC-1 (PPARγ co-activator-1) interacts with the DBD of PPARγ. This interaction induces a conformational change in PGC-1 that enables its recruitment of other co-activators, such as SRC-1 and CBP/p300 [45]. Along similar lines, P/CAF has also been shown to interact with the DBDs of RXR and RAR in vitro. This RAR–P/CAF interaction potentiates the ligand-dependent transactivation of RA-responsive reporter constructs [46,47]. Thus we have identified a novel role of the DBD in RAR transactivation. While the mechanism has not been fully defined, one possibility is that RARα phosphorylation by Akt inhibits the ability of RAR to recruit essential co-activators to the receptor complex.

Whereas normal human bronchial epithelial cells are sensitive to the growth-inhibitory effects of retinoids, NSCLC cells are typically resistant. Multiple mechanisms contribute to retinoid resistance in NSCLC cells. First, expression of RARB, which has tumour-suppressing properties [48,49], is silenced in certain NSCLC cell lines through DNA methylation and aberrant histone acetylation on the gene's promoter [50,51]. Secondly, we previously showed that RAR phosphorylation by JNK decreases RAR levels by inhibiting RAR stability in H322 NSCLC cells [17]. Thirdly, certain NSCLC cell lines are resistant to retinoids despite having high levels of retinoid receptors, but the receptors in those cells have transactivation defects [23]. Fourthly, we previously showed that certain retinoid-resistant NSCLC cells, including the Calu-1 cells studied in the present study, express co-activators at levels similar to those in retinoid-sensitive cells, suggesting that aberrant expression of co-activators does not have a role in retinoid resistance [51]. In the present study, we show that co-activator function may be aberrant in NSCLC cells with high Akt activity, thereby contributing to retinoid resistance.

Acknowledgments

This work was supported in part by National Institutes of Health grant number R01CA80686.

Abbreviations: AF, activation function; CREB, cAMP-response-element-binding protein; CBP, CREB-binding protein; CDK7, cyclin-dependent kinase 7; DBD, DNA-binding domain; DR5, direct repeat separated by 5 nt; EMSA, electrophoretic mobility-shift assay; GSK3, glycogen synthase kinase 3; GST, glutathione S-transferase; HA, haemagglutinin; HEK-293T, cells, human embryonic kidney 293T cells; IGF-I, insulin-like growth factor-I; JNK, c-Jun N-terminal kinase; LBD, ligand-binding domain; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazolium bromide; NSCLC, cell, non-small cell lung cancer cell; p70S6K, p70 S6 kinase; PARP, poly(ADP-ribose) polymerase; P/CAF, p300/CBP-associated factor; P-GSK3, phosphorylated GSK3; PPARγ, peroxisome-proliferator-activated receptor γ; PGC-1, PPARγ co-activator-1; PI3K, phosphoinositide 3-kinase; PKA, protein kinase A; PTEN, phosphatase and tensin homologue deleted from chromosome 10; RA, retinoic acid; RAR, RA receptor; RARE, RA response element; RXR, retinoid X receptor; SRC, steroid receptor co-activator; TK, thymidine kinase; TTNPB, E-4-[2-(5,5,8,8-tetramethyl-5,6,7,8-tetrahydro-2-naphthalenyl)-1-propenyl]benzoic acid

References

View Abstract