LKB1 Is Recruited to the p21/WAF1 Promoter by p53 to Mediate Transcriptional Activation

LKB1 (also known as Stk11) is a serine/threonine protein kinase, of which the inactivation from mutation or deletion causes Peutz-Jeghers syndrome (1–3). Approximately 70% of Peutz-Jeghers syndrome individuals harbor germ-line mutations of LKB1 (4), and somatic mutations lead to predisposition to sporadic cancers such as gastrointestinal, lung, breast, and ovary (5–7). Most alterations are present in the LKB1 catalytic kinase domain and a minority occur within its COOH-terminal noncatalytic region (8). These mutagenic changes result in LKB1 protein reduction, loss, or inactivation. Consistent with results from genetic characterization of clinical samples, gene knock-out studies show that Lkb1^+/− mice develop gastrointestinal polyposis and hepatocellular carcinomas, whereas Lkb1^−/− embryos die during midgestation (9–13). Thus, the mouse studies support an important role for LKB1 during tumor suppression and development.

LKB1 has a broad range of cellular functions. Studies of LKB1 orthologues Par-4 in C. elegans and dLKB1 in Drosophila reveal a critical role in generation of cell polarity (14, 15). Similarly, in mammals, LKB1 induces complete polarity in single intestinal epithelial cells (16). LKB1 is also an essential regulator of energy metabolism through the AMP-dependent kinase (AMPK) signaling pathway (17, 18). In response to energy stress, including low glucose or nutrient deprivation, LKB1 phosphorylates AMPK to block energy consuming processes and induce catabolic pathways that generate ATP (18, 19). LKB1 was recently shown, also via the AMPK pathway, to activate Tuberous sclerosis complex protein TSC2, leading to down-regulation of mammalian target of rapamycin and protein synthesis (20, 21). The end result of these effects on the AMPK pathway is LKB1-mediated inhibition of cell growth. In addition, phosphatase and tensin homologue, a tumor suppressor linked to disease phenotypes similar to Peutz-Jeghers syndrome, interacts with and is phosphorylated by LKB1 (22). Taken together, these observations suggest that LKB1 possesses wide ranging cellular functions involved in its tumor suppressor role.

Mouse LKB1 is predominantly nuclear (23) and human LKB1 is both nuclear and cytoplasmic (24–26). LKB1 associates with STRAD and MO25 to form a trimeric complex in the cell (27–30). Pseudokinase STRAD binds to and activates LKB1 catalytic kinase activity, which leads to transport of LKB1 to the cytoplasm (27). It is clear that LKB1 has a cytoplasmic role in cell growth regulation and generation of cell polarity (16, 27, 31). However, the nuclear role for LKB1 remains obscure. LKB1 has been shown to regulate cell growth and cell death (24, 26, 27, 31), and these functions have been linked to the tumor suppressor p53. Forced LKB1 expression in G361 cells, a melanoma cell line that lacks endogenous LKB1, induces p21/WAF1 expression leading to cell cycle arrest in the G₁ phase in a p53-dependent manner (26, 31). In fibrosarcoma HT1080 cells, LKB1 mediates p53-dependent programmed cell death in response to treatment with microtubule-disrupting agents, and LKB1 physically interacts with p53 in these cells (24). Recently, a knock-out mouse study showed that Lkb1^+/−/p53^+/− mice display earlier tumor formation, increased tumor incidence, and a dramatically reduced life span compared with mice with either Lkb1 or p53 single gene knockout, suggesting cooperation between LKB1 and p53 in tumor suppression in vivo (32). However, the molecular mechanism underlying LKB1 activation of p21/WAF1 transcription remains to be elucidated, as does the basis of the link between LKB1 and p53. In this study, we investigated a nuclear function of LKB1 in association with p53 and whether LKB1 has a direct physical role at p53-regulated genes.

Plasmid constructs. Human LKB1 cDNA was amplified from an IMAGE clone and then ligated into pFLAG-CMV2 (Sigma, St. Louis, MO) to prepare pFLAG-LKB1. pMSCVpuro-LKB1 was constructed by cloning FLAG-LKB1 into pMSCVpuro (Clontech, Palo Alto, CA). pFLAG-LKB1-KDM and pMSCVpuro-LKB1-KDM were generated by site-directed substitution to K78M and D176Y. pCDNA3-p53 plasmid was a gift from Zhimin Yuan (Harvard University, Boston, MA), into which the S392A substitution was site directed. Human Δp53(amino acids 64-393) was subcloned to pFLAG-CMV2, pFLAG-LKB1, and pFLAG-LKB1-KDM to construct pFLAG-p53ΔAD, pFLAG-LKB1-p53ΔAD, and pFLAG-KDM-p53ΔAD plasmids.

Cell culture, stable cell lines, UV treatment, and LKB1 complex purification. HCT116 cells and p53 knock-out derivative 379.2 (HCT116-p53^−/−) cells (gift from B. Vogelstein, Johns Hopkins University, Baltimore, MD) were maintained in McCoy's 5A medium supplemented with 10% fetal bovine serum (FBS). U2OS cells and Lkb1 null mouse embryonic fibroblast (MEF; gift from R. DePinho, Harvard University, Boston, MA) were maintained in DMEM supplemented with 10% FBS. Stable cell lines were generated by transfection of pMSCVpuro-LKB1 and pMSCVpuro-LKB1-KDM into “phoenix” packaging cells with calphos mammalian transfection kit (Clontech). Supernatant containing retroviral FLAG-LKB1 or FLAG-KDM was harvested after 48 hours and infected into HCT116 cells and MEF Lkb1 null cells. Infected cells were selected with 3 μg/mL puromycin (Sigma) for 2 weeks and individual colonies were isolated and analyzed for FLAG-LKB1 expression. Cells were plated in 100- or 150-mm dish. Twenty-four hours later, at 70% to 80% confluence, cells were treated (UV Stratalinker 2400, Stratagene, La Jolla, CA) with or without UVC at 20 J/m², incubated for 6 hours, and harvested. F-LKB1 complex was purified from HCT116 nuclear and cytoplasmic extract using anti-FLAG conjugated M2 agarose beads (Sigma) and protein identification by liquid chtromatography-tandem mass spectrometry and Western blot.

In vitro kinase assay. Glutathione S-transferase (GST)-p53 or substitution mutants (1-2 μg) were mixed with Flag-LKB1 (0.5 μg), CK2 (10 milliunits; Upstate, Lake Placid, NY), or Rsk2 (5 units; Upstate) in kinase buffer [HEPES 100 mmol/L, KCl 200 mmol/L, MgCl₂ 40 mmol/L, DTT 4 mmol/L, phenylmethylsulfonyl fluoride (PMSF) 5 mmol/L, NaF 20 mmol/L] containing 1 μCi [γ-³²P]ATP in a total volume of 20 μL. The reaction was maintained at 30°C for 20 minutes and electrophoresed in 10% SDS-PAGE.

Whole-cell or subcellular fraction extraction and Western blot. Whole-cell extract was prepared by lysing cells with standard radioimmunoprecipitation assay lysis buffer [50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 1 mmol/L PMSF, 1 mmol/L EDTA, 5 μg/mL aprotinin, 5 μg/mL leupeptin, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS]. Nuclear and cytoplasmic extractions were prepared with NE-PER nuclear and cytoplasmic extraction reagents (Pierce, Rockford, IL). Western blots were done with α-mouse-p53 (Cell Signaling, Beverly, MA), α-human-p53 [Oncogene, Cambridge, MA and Santa Cruz Biotechnology (Santa Cruz, CA)], α-phospho-p53 Ser392 (Cell Signaling and gift from T. Hupp, University of Edinburgh, Edinburgh, United Kingdom), α-LKB1 (Upstate), α-STRAD (gift from H. Clevers, Hubrecht Laboratory, Utrecht, the Netherlands), α-Rb (Abcam, Cambridge, MA), and α-β-actin (Sigma).

siRNA transfection and mRNA quantification. LKB1 siRNA oligo was 5′-CUGGUGGAUGUGUUAUACA-3′ and control siRNA was 5′-ACACGUUCAUCCACCGCAU-3′ (Dharmacon, Chicago, IL). Transfection was done with Oligofectamine (Invitrogen, Carlsbad, CA) at 30% to 40% cell confluence in 10-cm plates. siRNA oligos (1.2 nmol) were mixed with 30 μL of Oligofectamine in 2 mL of Opti-MEM medium in 6 mL for 6 hours. A second transfection was done with Lipofectamine 2000 (Invitrogen) 24 hours later. Whole-cell extract or total RNA was prepared after 2 days. mRNA was extracted from siRNA or control treated cells using RNeasy kit (Qiagen, Valencia, CA) and treated with DNase 1 kit (Invitrogen). Purified mRNA was transcribed to cDNA by using cloned AMV reverse trancriptase kit (Invitrogen). Quantitative PCR was done with ABI 7000 sequence detection system (Applied Biosystems, Foster City, CA).

Chromatin immunoprecipitation assay. Cells were cross-linked at 80% to 90% confluence. Conventional cross-linking was done with 1% formaldehyde incubated at room temperature for 10 minutes. Cross-linking reaction was quenched with 0.1 mol/L glycine-PBS for 5 minutes. Sequential cross-linking was done with a second cross-linking reagent, ethyleneglycol bis(succinimidyl)succinate (Pierce), which we have found to be effective for chromatin immunoprecipitation of proteins that do not directly bind to DNA (33), incubation at room temperature for 25 minutes, followed by 1% formaldehyde for 10 minutes. The reaction was quenched with 50 mmol/L glycine-PBS for 10 minutes. Cross-linked cells were harvested using a cell scraper and stored at −80°C. Cells were lysed with lysis buffer I (10 mmol/L Tris-HCl, 10 mmol/L EDTA, 0.5 mmol/L EGTA, 0.25% Triton X-100, 1 mmol/L PMSF) followed by lysis buffer II (10 mmol/L Tris-HCl, 10 mmol/L EDTA, 0.5 mmol/L EGTA, 200 mmol/L NaCl, 1 mmol/L PMSF) at room temperature for 15 minutes each. Extracted nuclei were resuspended in prechilled immunoprecipitation buffer [20 mmol/L Tris-HCl (pH 8.0), 200 mmol/L NaCl, 0.5% Triton X-100, 0.1% NP40, 1 mmol/L PMSF] and sonicated with Bioruptor-UCD200 (Diagenode, Liege, Belgium) or sonicator (Misonix, Inc., Farmingdale, NY). Ten percent of extract before immunoprecipitation was saved for input. The sonicated nuclear extracts were immunoprecipitated with α-phospho-p53 S392 (gift of T. Hupp), α-p53 (Cell Signaling), α-FLAG (Sigma), or α-LKB1 (gift of T. Makela, University of Helsinki, Helsinki, Finland) and salmon sperm DNA/protein A/G agarose (Upstate) beads at 4°C for overnight. Immunoprecipitated beads were sequentially washed with immunoprecipitation buffer and LiCl wash buffer [10 mmol/L Tris-HCl (pH 8.0), 0.25M LiCl, 0.5% NP40, 1% deoxycholic acid, 1 mmol/L EDTA, 1 mmol/L PMSF]. After cross-linking was reversed at 65°C, the DNA was precipitated by phenol/chloroform and ethanol.

Quantitative real-time PCR primers. PCR primers were designed by primer express software (Applied Biosystems). Primer sequences for reverse transcription-PCR (RT-PCR) were mouse p21/WAF1, 5′-TCTTCTGCTGTGGGTCAGGAG-3′/5′-GAGGGCTAAGGCCGAAGATG-3′; human LKB1, 5′-GTGGCATGCAGGAAATGCT-3′/5′-GCACACTGGGAAACGCTTCT-3′; and human p21/WAF1, 5′-CTGTGATGCGCTAATGGCG-3′/5′-AAGTCGAAGTTCCATCGCTCA-3′. Primer sequences for chromatin immunoprecipitation-PCR were mouse p21/WAF1 upstream promoter region. 5′-AAATATTTGTTGAGTACTTTTGTGGTGC-3′/5′-TTAAACAACTTCTGGCTTCCCA-3′; mouse p21/WAF1 TATA region:, ′-CTTGAATGCCTATTTCCCCCT-3′/5′-TGTAATAACAGCGCCCAGTGG-3; human p21/WAF1 upstream promoter region, 5′-GGCTGGTGGCTATTTTGTCCT-3′/5′-CCCCTTCCTCACCTGAAAACA-3′; and human p21/WAF1 gene TATA-5′ untranslated region (UTR), 5′-AGCTGCGCCAGCTGAGG-3′/5′-GCTCCACAAGGAACTGACTTCG-3′. Quantitative PCR was done as described (34) with statistical evaluation as recommended by the manufacturer (Applied Biosystems).

Cell cycle analysis. Cells were maintained in complete culture medium at <50% confluence for 1 week. Wild-type p53 and substitution mutants were transfected into cells by Lipofectamine reagent. After 24 hours, cells were synchronized with nocodazole at 400 μg/mL for 22 hours and released to complete culture medium and treated with UVC. After 14-hour incubation, cells were harvested, fixed in 70% ethanol, and cellular DNA was stained with propidium iodide (Roche, Indianapolis, IN) and analyzed by flow cytometry (Wistar Institute Core Facility, J. Faust, Director).

LKB1 binds to p53 in the nucleus. Previous studies have shown that LKB1 interacts with p53 in vivo (24) but it is not known whether this occurs in the nucleus. We examined whether LKB1 is physically associated with p53 in the nucleus. MEFs obtained from Lkb1 null embryos (9) were transfected with FLAG-tagged human LKB1 (F-LKB1), FLAG-tagged LKB1 kinase-deficient mutant (K78→M and D176→Y; F-KDM; refs. 24, 35), or a vector control to obtain clonal populations of stably expressing cell lines. Cells were treated with UVC to induce p53 protein expression (γ-irradiation was unable to induce p53 in this Lkb1 null MEF line; data not shown) because p53 is undetectable in these cells without irradiation (9). Equal protein amounts of extracts prepared from F-LKB, F-KDM, or vector control cell lines were fractionated into cytoplasm and nuclei (Fig. 1). pRb was used as a marker for the nuclear fraction (Fig. 1, input). Using equal input protein amounts, the level of LKB1 in the cytoplasmic fraction is higher than in the nuclear fraction (Fig. 1, input), as expected (31). In contrast, the level of p53 is much higher in the nuclear than cytoplasmic fraction (Fig. 1, input). Notably, the level of nuclear p53 is higher in cells expressing F-LKB1 than F-KDM, or no LKB1 (Fig. 1, input), consistent with previous observations that p53 is destabilized in MEFs derived from Lkb1 null mice (9).

Equal amounts of protein from each fraction were immunoprecipitated with α-FLAG to pull down F-LKB1 (and similar results were obtained with α-LKB1 immunoprecipitation; data not shown). Both wild-type F-LKB1 and F-KDM associated with STRAD in the cytoplasmic and nuclear fractions (Fig. 1, IP). This association of F-KDM with STRAD indicates that the catalytic mutant we used in this study makes normal protein associations despite the absence of kinase activity, as previously observed for other kinase-defective LKB1 proteins (29).

Wild-type F-LKB1 coprecipitates p53 in the nuclear fraction, and p53 exhibits phosphorylation at Ser15 (Fig. 1, IP), a modification involved in p53 stabilization during transcriptional activation (36, 37). In contrast, FLAG immunoprecipitation of protein derived from vector-transfected cells or from cells expressing F-KDM does not coprecipitate detectable levels of p53 or the Ser15ph form of p53 (Fig. 1, IP). Note that, although there is less p53 in the absence of LKB1 or with the KDM mutant (Fig. 1, input), the amount of input p53 is sufficient to detect a signal in the immunoprecipitated fraction if there were indeed an interaction at the level of wild-type LKB1 with p53. The key observation is that nuclear interaction exists between p53 and LKB1.

LKB1-dependent phosphorylation of p53 in vitro and in vivo. Posttranslational modifications of p53, including phosphorylation, play a crucial role in both stabilization and activation of the p53 protein (38, 39). AMPK, an LKB1 substrate as described above, has recently been shown to phosphorylate p53 at Ser15 (S15ph; ref. 40). p53 S15ph increases the stability and stimulates its trans-activation function through reduction of p53 interaction with murine double minute-2, a ubiquitin ligase that promotes p53 proteolytic degradation (41). Based on these observations, we examined whether LKB1 may also be involved in p53 phosphorylation. We immunoprecipitated FLAG-LKB1 in association with STRAD and MO25 from HCT116 cells, as revealed both by mass spectrometry and by Western blot analysis (Fig. 1B and data not shown). The immunoprecipitated FLAG-LKB1 phosphorylates GST-p53 more efficiently than Rsk2 kinase (Fig. 1D).

We mapped the phosphorylation within p53. p53 has a central DNA binding domain, two transcriptional activation domains at the NH₂ terminus and regulatory and oligomerization domains at the COOH terminus (Fig. 1C). GST fusions to these individual domains are phosphorylated by FLAG-LKB1 more weakly than is the GST fusion to full-length p53 (Fig. 1D). The strongest phosphorylation is within the COOH-terminal regulatory region (amino acids 358-393; Fig. 1D). Weaker activity is detected in the NH₂-terminal 80 residues and in the nuclear localization signal/oligomerization regions (amino acids 300-369), whereas the DNA binding domain is poorly phosphorylated (Fig. 1D). Individual potential phosphorylation sites within the GST-p53(358-393) COOH-terminal domain (S362, S366, S367, S371, S376, T377, S378, T387, and S392) were substituted with alanine either in combinations or as single residues. Only the combination of T387A/S392A or single S392A, but not T387A, is reduced in LKB1-dependent phosphorylation (Fig. 1E,, top). The single S392A substitution was site directed into full-length GST-p53, and the level of phosphorylation is reduced but not eliminated (Fig. 1E,, bottom right). Additional sites in the activation and tetramerization domains (S6, S15, S303, S313, S314, and S315) were altered to alanine in the context of full-length GST-p53, and of these only S15A lowers phosphorylation (Fig. 1E,, bottom and data not shown). The S15/S392A substitution mutant combination was site directed into full-length GST-p53 and the phosphorylation by LKB1 was nearly completely eliminated (Fig. 1E , bottom right). Thus, two major sites of phosphorylation in p53 by immunoprecipitated FLAG-LKB1 are S15 and S392.

LKB1-dependent phosphorylation of p53 was examined in vivo in HCT116 p53-null cells by cotransfecting p53 (or p53 bearing substitution mutations at S15 or S392) along with F-LKB1 [or F-LKB1(KDM)]. Using equivalent amounts of p53 derived from wild-type LKB1 and LKB1(KDM) cells, we determined that the levels of p53 S15ph and p53 S392ph forms are higher in cells expressing F-LKB1 compared with F-LKB1(KDM) (Fig. 1F). Thus, the LKB1 pathway leads to phosphorylation of p53 at S15 and S392. As noted above, p53 S15 phosphorylation is also dependent on AMPK (40); thus, as discussed below, our results may reflect an LKB1-dependent pathway through collaboration with AMPK, leading to phosphorylation of p53 at S15 and S392.

p53-dependent G₁ cell cycle arrest is augmented by LKB1 and requires LKB1-dependent p53 phosphorylation sites. An LKB1-dependent pathway leading to phosphorylation/stabilization of p53 may, in part, underlie previous observations linking LKB1 to cell cycle arrest (26, 31). We analyzed the cell cycle in MEF cells stably expressing LKB1, LKB1(KDM), or vector alone. Fluorescence-activated cell sorting (FACS) analysis of DNA content was done on samples from synchronized cells after treatment with nocodazole to block cells in G₂-M phase followed by growth in normal media for 14 hours. Expression of LKB1 compared with vector leads to increased G₀-G₁ accumulation, and this effect is not observed in cells expressing LKB1(KDM), similar to previously reported results [refs. 26, 31; Fig. 2A, (top) and B]. Cells were transfected with plasmids expressing either wild-type p53 or p53 bearing substitution mutants in the phosphorylation sites. Additional p53 increases G₀-G₁ arrest only in cells expressing LKB1 compared with vector [Fig. 2A, (bottom) and B]. This effect is greatly attenuated in the presence of substitution mutants of p53, at either S15A or S392A [Fig. 2A  (bottom) and B]. Thus, the increased G₁ arrest resulting from coexpression of kinase-active LKB1 and p53 requires intact S15 and S392 in p53, consistent with the idea that there is an LKB1-dependent pathway leading to p53 phosphorylation to augment cell cycle arrest.

LKB1 is associated with the promoter of p21/WAF1, a p53-dependent gene. LKB1 was shown to induce transcription of the p21/WAF1 gene, resulting in cell cycle arrest in the G₁ phase in a p53-dependent manner (31). The preceding results led us to test whether LKB1-dependent induction of the p21/WAF1 gene reflects a direct role of LKB1 in p53-dependent genes. We employed chromatin immunoprecipitation assays to test interaction of LKB1 with p21/WAF1 gene promoter. We used either exogenous F-LKB1 in Lkb1 null cells (Fig. 3), which allowed us to examine a role for Lkb1 kinase activity, or endogenous LKB1 (Fig. 4), which provide normal levels of the protein.

Binding of exogenous LKB1 was analyzed in the MEF cell lines described above, stably expressing F-LKB1 or F-KDM, compared with the control Lkb1 null cell line (vector transfected). Chromatin immunoprecipitation was done using either α-LKB1 or α-FLAG antibodies to immunoprecipitate F-LKB1. We analyzed F-LKB1 association with the p21/WAF1 upstream promoter to examine association in the region of p53 binding sites (42), and compared these results with association at the proximal promoter in the TATA box/5′ UTR region (Fig. 3A). Results from both α-LKB1 and α-FLAG antibody chromatin immunoprecipitation indicate that F-LKB1 is present specifically in the promoter region of the p21/WAF1 gene (Fig. 3B) and not in the TATA box/5′ UTR region (Fig. 3C). Moreover, F-KDM is not associated with the gene promoter (Fig. 3B), although F-KDM is present in the nucleus (Fig. 1). As a control, chromatin immunoprecipitation shows that p53 is present at the promoter in response to UV treatment and increases 4-fold in F-LKB1 wild-type cells compared with vector or F-KDM cells (Fig. 3D). The binding of p53 in the presence of wild-type F-LKB1 reflects p53 stabilization in the presence of kinase-active LKB1 (Fig. 3E; ref. 9). These interactions of F-LKB1 and p53 correlate with increased transcription of p21/WAF1 [normalized to the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene; Fig. 3F]. Thus, the key finding of this series of experiments is that F-LKB1 is associated with the p21/WAF1 gene, specifically at the upstream promoter in the region containing p53 binding sites. We also examined interaction of LKB1 with the Cyclin G1 gene (Fig. 3B) and with other p53-regulated promoters (GADD45 and BAX; data not shown) and found similar chromatin immunoprecipitation signals for LKB1 at these genes. Similarly, LKB1 is required for transcription of Cyclin G1 (Fig. 3F) and other p53-regulated genes (GADD45 and BAX; data not shown).

These results indicate that exogenously expressed LKB1 associates with p53 in the nucleus and is bound to a p53-dependent gene at p53 binding sites. It is critical to show that endogenous levels of LKB1 also associate with the p21/WAF1 promoter. We carried out chromatin immunoprecipitation in U2OS cells, which express both LKB1 and p53, and used siRNA against LKB1 to test the specificity of α-LKB1 antibody. Compared with control siRNA–treated cells, LKB1 siRNA reduces the level of LKB1 mRNA ∼3-fold in UV- or non-UV-irradiated cells (Fig. 4A,, left). LKB1 protein is also reduced compared with control siRNA–treated cells (Fig. 4A,, right). As expected, the p53 level is increased in UV, and its level is reduced in the LKB1 siRNA–treated cells (Fig. 4A,, right). The critical experiment is the chromatin immunoprecipitation assay, which shows endogenous LKB1 increasing at the p21/WAF1 upstream promoter, correlating with p21/WAF1 gene induction in UV-treated cells, but, again, not at the TATA box/5′UTR region (Fig. 4B). Importantly, the LKB1 chromatin immunoprecipitation signal is strongly reduced when cells are treated with LKB1 siRNA relative to treatment with control siRNA, which shows that the α-LKB1 antibody is specific in the chromatin immunoprecipitation assay. Finally, as an additional control, siRNA to LKB1 lowers p21/WAF1 mRNA levels and protein levels compared with control siRNA (Fig. 4A and C). Because UV induces p21/WAF1 degradation (43), the p21/WAF1 protein levels are decreased in UV treatment (Fig. 4A,, right), although their mRNA levels are increased (Fig. 4C). Thus, endogenous LKB1 binds to the p21/WAF1 promoter and the level of binding correlates with higher levels of p21/WAF1 mRNA.

LKB1 binding to p21/WAF1 promoter requires p53. The interaction of LKB1 with p53 in the nucleus and LKB1 association with the p21/WAF1 promoter specifically in the regions of p53 binding sites led us to test whether LKB1 binding to the p21/WAF1 promoter requires p53. We used HCT116-p53^−/− cells, which maintain endogenous LKB1. The cells were transfected either with vector alone, wild-type p53, or a p53 substitution mutant (S392A) in a phosphorylation site that is required for full DNA binding (44). As a control, we found that the p53 S392 mutant is not phosphorylated although the p53 protein is present in comparable levels to wild-type p53 (Fig. 5A). The chromatin immunoprecipitation used α-LKB1 antibody to immunoprecipitate the endogenous LKB1, as in a parallel experiment with p53 wild-type cells (Fig. 4). The result shows that wild-type p53 promotes 3-fold higher LKB1 recruitment to p21/WAF1 promoter in response to UV compared with vector-transfected cells that do not express p53 (Fig. 5B). LKB1 recruitment was lowered ∼2-fold in the presence of the p53 S392A substitution mutant (Fig. 5A). We conclude that p53 is required for LKB1 recruitment to the p21/WAF1 promoter.

Fusion of LKB1 to activation-defective p53 results in p21/WAF1 transcriptional activation. These data suggest that LKB1 is recruited by direct association with p53 to activate p53-dependent gene p21/WAF1 transcription. We then investigated whether LKB1 may have a direct role in activating transcription at p21/WAF1 gene. We examined the ability of LKB1 to substitute for p53 activation domains, which is used to test whether enzymatic activity is involved in transcriptional activation (45, 46). FLAG-tagged wild-type p53 and three derivatives of the p53 were engineered. The two tandem p53 activation domains (47) were deleted (p53ΔAD) and either LKB1 (LKB1-p53ΔAD) or KDM (KDM-p53ΔAD) was genetically fused in place of the activation domains (Fig. 6A). Each of the p53 constructs is stable, as shown by α-FLAG Western blotting (Fig. 6B). The ability of each construct to activate transcription of p21/WAF1 was assayed by quantitative RT-PCR. Transfection of vector indicates the basal mRNA level of p21/WAF1; this level is stimulated only slightly by p53ΔAD after UV treatment (<2-fold compared with no UV treatment; Fig. 6C), showing that p53 is largely defective in the absence of its activation domains. Intact p53 stimulates p21/WAF1 transcription in UV ∼5-fold, compared with p53ΔAD (Fig. 6C). Transfection of LKB1-p53ΔAD stimulates transcription in UV >3-fold, compared with p53ΔAD, whereas KDM- p53ΔAD stimulates transcription more poorly than p53ΔAD (Fig. 6C). Thus, LKB1 can partially complement for deletion of p53 activation domains during transcriptional activation.

We wished to rule out that poor activation by p53ΔAD or KDM-p53ΔAD resulted from inability of these proteins to bind to DNA. We tested whether each of the constructs binds to the p21/WAF1 promoter using chromatin immunoprecipitation assay with α-FLAG antibody. All four proteins, p53, p53ΔAD, LKB1- p53ΔAD, and KDM- p53ΔAD, are present at the promoter in UV conditions (Fig. 6D). Most notably, p53ΔAD, LKB1- p53ΔAD, and KDM- p53ΔAD are present in comparable amounts in UV conditions (Fig. 6D) although only LKB1- p53ΔAD activates transcription strongly. Thus, LKB1, but not enzymatically compromised KDM, can activate transcription when directed to bind to the p21/WAF1 promoter via fusion to p53.

LKB1 kinase was initially identified as a tumor suppressor and is a key determinant in Peutz-Jeghers syndrome, an inherited predisposition to gastrointestinal and other cancers (1–3). In the cytoplasm, LKB1 is involved in a broad range of cellular processes, including regulation of cell polarity, energy balance, protein synthesis, and cell cycle arrest. In contrast, little is known about the function of LKB1 in the nucleus (31). One potential pathway that would fit the role of LKB1 as a tumor suppressor and as a key responder to cellular stress is the association of LKB1 with p53. Previous data have linked LKB1 and p53: LKB1 and p53 physically associate in the cell (24); disruptions of LKB1 in MEFs reduce p53 levels (9); LKB1 arrests the cell cycle in p53-dependent manner; activation of p21/WAF1 promoter by LKB1 requires wild-type p53 (31). However, a clear molecular mechanism linking these proteins has been lacking.

Our study suggests such a potential mechanism. Our results confirm and extend previous observations positively linking LKB1 to stability and activity of p53. First, the presence of either exogenous LKB1 compared with Lkb1 null, or endogenous LKB1 compared with siRNA-reduced levels of LKB1, correlates with p53 stability and p21/WAF1 gene expression (Figs. 1A, 3, and 4). Second, we find that LKB1 protein binds to p53, and this interaction is detectable in the nucleus (Fig. 1A), consistent with a nuclear role for LKB1/p53 interaction. Third, either exogenous or endogenous LKB1 associates with regions upstream of the p21/WAF1 gene, specifically near to sites of p53 binding (Figs. 3 and 4). Significantly, LKB1 association with the promoter requires p53 binding (Fig. 5), suggesting that p53 may recruit LKB1. In a standard test of enzyme activity involved in transcriptional activation (45, 46), LKB1 fused to a transcriptionally crippled p53 is capable of activating transcription (Fig. 6). These data suggest that LKB1 binds to and stabilizes p53 and that p53 recruits LKB1 to the promoter to activate the p21/WAF1 gene.

We queried the involvement of LKB1 kinase activity in these functions in association with p53. We found that LKB1 kinase activity is required to stabilize p53 (Figs. 1A, 3E, and 4A), which is reflected in assays measuring LKB1-p53 interaction (Fig. 1), p53 binding to p21/WAF1 promoter (Fig. 2), and LKB1 recruitment to the promoter (Fig. 3). In contrast, we find that LKB1 kinase activity is not required for stable interaction with STRAD, as was previously found (29).

We also investigated whether p53 is phosphorylated in an LKB1-dependent fashion and found that p53 S15 and S392 are both substrates in vitro and in vivo, and these p53 residues are required for the LKB1-dependent cell cycle arrest in G₁. Because S15ph is required for stability of p53 (37), this likely explains the importance of LKB1 kinase activity for stable expression of p53. One important question is whether LKB1 directly phosphorylates p53. Related to this question is the recent report that, in response to glucose deprivation, AMPK (an LKB1 downstream substrate as described above) induces cell-cycle G₁ arrest through p53 Ser15 phosphorylation (40). Thus, there seems to be a LKB1-dependent pathway, possibly via AMPK, to phosphorylate p53. We are currently examining this critical question to determine whether and how the LKB1 complex collaborates with the AMPK complex to phosphorylate p53, leading to the transcriptional effects we detect in response to genotoxic stress.

Further, we found that LKB1 kinase activity is required in the LKB1/p53 fusion protein assay (i.e., LKB1 activation of transcription when fused to transcriptionally inactivated p53). We show that proteins, consisting of either LKB1 or KDM fused to p53 DNA binding domain/tetramerization domains, are equally recruited to p21/WAF1 (Fig. 6D). However, because the KDM fusion to p53 DNA binding domain/tetramerization domains is unable to activate transcription (Fig. 6C), we conclude that LKB1 has an additional function beyond stabilizing p53. One possible interpretation is that LKB1 may phosphorylate a component of chromatin or the general transcriptional machinery.

LKB1 was previously implicated in p53 functions in arresting the cell cycle and promoting apoptosis; however, it was unclear whether this was an indirect effect of LKB1 through its cytoplasmic functions or LKB1 directly regulates p53 during transcription. Chromatin immunoprecipitation assays, showing that both overexpressed and endogenous LKB1 associate with p21/WAF1 gene promoter in a p53-dependent fashion (Figs. 3-Figs. 5), indicate that LKB1 apparently functions directly in the p53 gene transcription regulation pathways. Thus, it seems that direct LKB1/p53 function in transcriptional regulation is, at least in part, a mechanism underlying previous findings that LKB1 is involved in cell cycle regulation by p53.

These are unexpected results given the recent evidence that LKB1 functions to regulate cell polarity and energy balance and the previous suggestions that the role of LKB1 as a tumor suppressor involves its cytoplasmic functions. Thus, our results suggest that, in addition to its cytoplasmic roles, LKB1 functions within the nucleus in association with p53 to activate genes involved in growth regulation.

Grant support: NIH grant CA078831 (S.L. Berger).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank R. DePinho for the Lkb1 null MEFs; B. Vogelstein for p53 null HCT116 cells; T. Hupp, T. Makela, and H. Clevers for gifts of antibodies; R. Shiekhattar, T. Halazonetis, N. Barlev, G. Moore, and L-Q. Yin for helpful reagents, discussions, and technical assistance; and D. Bungard and members of the Berger laboratory for valuable discussions and critical reading of this manuscript.