Tumor suppressor protein (p)53, is a regulator of NF-κB repression by the

Samantha H. Murphya,b, Kotaro Suzukia,1, Michael Downesc, Genevieve L. Welchd, Paul De Jesusd,2, Loren J. Miragliad, Anthony P. Orthd, Sumit K. Chandad,2, Ronald M. Evansc, and Inder M. Vermaa,3

aLaboratory of Genetics, Salk Institute for Biological Studies, La Jolla, CA 92037; bGraduate Program, Division of Biology, University of California, San Diego, CA 92093; cGene Expression Laboratory, Salk Institute for Biological Studies, La Jolla, CA 92037; and dGenomics Institute of the Novartis Research Foundation, San Diego, CA 92121

Contributed by Inder M. Verma, September 6, 2011 (sent for review August 8, 2011) can inhibit inflammation by abrogating the activ- (Dex), a synthetic glucocorticoid that activates ity of NF-κB, a family of factors that regulates the GR upon binding, is a commonly prescribed antiinflammatory production of proinflammatory cytokines. To understand the mo- drug used to provide relief from inflammation, infection, and lecular mechanism of repression of NF-κB activity by glucocorti- immune diseases associated with NF-κB up-regulation (16–20). coids, we performed a high-throughput siRNA oligo screen to Although the precise molecular details of how glucocorticoids identify novel involved in this process. Here, we report that repress NF-κB activity are not fully elucidated, a number of loss of , a tumor suppressor protein, impaired repression of NF- mechanisms have been proposed which include (i) increased ex- κB target transcription by glucocorticoids. Additionally, loss pression of IκBα by GR, which increases the sequestration of p65/ of p53 also impaired transcription of glucocorticoid receptor (GR) p50 heterodimers in the and limits NF-κB translocation target genes, whereas upstream NF-κB and glucocorticoid receptor to the nucleus and subsequent activation of its target genes (21, signaling cascades remained intact. We further demonstrate that 22); (ii) GR binding as a monomer to the promoter regions of NF- p53 loss severely impaired glucocorticoid rescue of death in κB target genes, thus inhibiting NF-κB binding and transcriptional a mouse model of LPS shock. Our findings unveil a new role for activation; (iii) GR and NF-κB competition for Creb-binding p53 in the repression of NF-κB by glucocorticoids and suggest im- protein and p300, members of the transcriptional activation ma- portant implications for treatment of the proinflammatory micro- chinery; (iv) GR recruitment of complexes to environments found in tumors with aberrant p53 activity. the promoter regions of NF-κB target genes; and (v)GRandNF- κB interaction in the nucleus, where GR tethers NF-κB, thereby uclear factor kappa B (NF-κB) is a family of transcription inhibiting transcriptional activation of its target genes (23–27). factors that have an essential regulatory function in in- Some of the proposed mechanisms of repression are likely to be N fi fi flammation, the immune response, cell proliferation, and apo- cell-type speci c or context speci c, and a universal mechanism of ptosis (1). Constitutive activation of the NF-κB pathway is often repression has yet to be clearly established. Discovery of novel genes κ associated with cancer and chronic inflammatory diseases such as involved in NF- B repression by GR, elucidation of their mecha- fl nisms, and identification of universal mechanisms, will be critical for multiple sclerosis, in ammatory bowel disease, rheumatoid ar- κ thritis (RA), and asthma. NF-κB family members include p50 a more comprehensive understanding of how GR inhibits NF- B. To address these issues, we used a RNAi-based screening approach (NF-κB1), p52 (NF-κB2), p65 (RELA), RELB, and c-REL, and to identify novel genes that regulate GR repression of NF-κBand,in these subunits bind to form homo- and heterodimerized com- κ κ follow-up experiments, validate p53 as a novel regulator of NF- B plexes (2, 3). In the classical NF- B signaling pathway, the p65 and repression by GR both in vitro and in vivo. p50 subunits heterodimerize and are sequestered in an inactive complex in the cytoplasm bound to IκBα (2, 4, 5). Upon activation Results fl by proin ammatory stimuli such as tumor necrosis factor-alpha p53 Is Required for Repression of NF-κB Activity. We developed and (TNFα) or lipopolysaccharide (LPS), the IκB kinase (IKK) com- optimized a high-throughput luciferase assay using four siRNA plex phosphorylates IκBα (6), targeting it for ubiquitination and oligo library collections to search for potential genes that play a role proteasomal degradation (7), allowing the p65/50 complex to in NF-κB repression by the glucocorticoid receptor. The number of translocate to the nucleus where p65 can bind to the promoter genes represented and characteristics of two kinase libraries and regions of its target genes. p65 phosphorylation by protein kinase two druggable genome libraries used for screening are summarized A (PKA) or mitogen- and stress-activated protein kinase-1 (MSK- in Table S1. siRNA oligos from each of the libraries were pre- 1) at serine (Ser)276 (8, 9) and also by ζPKC at serine 311 (10) arrayed into 384-well plates and reverse transfected for 48 h into then aids in the recruitment of transcriptional activation machin- 293T cells stably transfected with a NF-κB luciferase reporter (Fig. MEDICAL SCIENCES ery, leading to transcription of NF-κBtargetgenes. 1A). Cells were then treated for 24 h with 10 ng/mL TNF ± 1 μM The glucocorticoid receptor (GR) is a member of the nuclear Dex in duplicate and assayed for luciferase activity (Fig. 1A). superfamily and plays a critical role in meta- The efficacy of the screen was determined by luciferase anal- bolism, development, reproduction, and homeostasis. Intracell- ysis of siGR, sip65, si luciferase, and mock controls added to ular GR resides in the cytoplasm in an inactive state, bound to a low molecular weight protein (p23), immunophilin FKBP51, and a variety of heat shock proteins (hsps), including , , Author contributions: S.H.M., K.S., M.D., S.K.C., R.M.E., and I.M.V. designed research; S.H.M., K.S., and G.L.W. performed research; P.D.J., L.J.M., A.P.O., and S.K.C. contributed new hsp56, and hsp40 (11, 12). Upon passive diffusion and subse- reagents/analytic tools; S.H.M., K.S., and L.J.M. analyzed data; and S.H.M., M.D., R.M.E., quent binding by glucocorticoids, GR can dissociate from its and I.M.V. wrote the paper. inhibitory complex and translocate to the nucleus via active The authors declare no conflict of interest. transport by dynein (13). In the nucleus, GR homodimers bind to 1Present address: Chiba University Graduate School of Medicine, Chiba 260-8670, Japan. glucocorticoid response elements (GREs) in the promoter re- 2Present address: Sanford-Burnham Medical Research Institute, La Jolla, CA 92037. gions of GR target genes to activate their transcription. GR can 3To whom correspondence should be addressed. E-mail: [email protected]. also bind as a monomer to negative (n)GREs in promoter This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. regions of genes to inhibit their transcription (14, 15). 1073/pnas.1114420108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1114420108 PNAS | October 11, 2011 | vol. 108 | no. 41 | 17117–17122 Downloaded by guest on October 1, 2021 3x-NF B-Luc homeostasis, developmental, and metabolic processes, as well as A TNF + puromycin Luciferase Assay apoptotic, cell cycle, and kinase activity categories were among the selection 384 well plates list of overrepresented pathways (Fig. S3A). KEGG pathway analysis showed enrichment for 29 pathways with P values <0.05, TNF + Dex including MAPK signaling (a pathway with known p53 and GR 293T Cells 293T-NF B-Luc cells pre-coated with interactions), apoptosis, cell cycle, and 13 cancer-related pathways siRNA oligos Lipofectamine 2000 (Fig. 1B). We also performed Reactome analysis to determine

Timeline: 48hr Transfection 24hr Treatment which pathways were overrepresented in our list of 290 genes and 72 hours Day 1 Day 3 Day 4 to confirm results from KEGG pathway analysis. A total of 181 genes, of the 290 analyzed, were not represented in the database, Cell cycle 14 Invitrogen Kinase Library suggesting that a number of uncharacterized and/or unstudied B Apoptosis C Renal cell carcinoma 12 genes may play a role in GR repression of NF-κB. Reactome Acute myeloid leukemia 10 < Pathways in cancer 88 analysis revealed that 109 events with P values 0.05 were over- Colorectal cancer 66 Pancreatic cancer represented, including a number of events related to Melanoma 44 κ Glioma 22 and MAPK signaling, as well as NF- B related signaling events of TNF Condition Thyroid cancer 00 (Fig. S3B), confirming results obtained by KEGG pathway analysis Chronic myeloid leukemia 14121086420

Avg. Normalized Raw Value Avg. 0246810 Prostate cancer Avg. Normalized Raw Value (Fig. 1B). Through the STRING database, we analyzed protein– Bladder cancer of TNF + Dex Condition Non-small cell lung cancer protein interactions for genes overrepresented in the MAPK and Endometrial cancer MAPK signaling cancer pathways by KEGG analysis (Fig. S4). We included GR in

01234567 each STRING analysis to map known interactions with GR and -log P-value the genes of interest shown in each pathway. p53, shown in red, was found in both the MAPK signaling pathway (Fig. S4, Upper) and ** 293Ts ** the cancer pathways (Fig. S4, Lower). D 4040 3535 Untreated Sixteen wells containing a siRNA oligo targeting p53 were 3030 TNF included as a control in the Invitrogen kinase library collection.

Fold Change 1 2525 TNF + Dex When siRNAs were rank ordered according to fold change be- 2020 1515 tween the TNF and TNF + Dex conditions, 14 of the 16 wells

Fold Change 10 Fold Change 1010 were near or at the top of the list. To visualize this observation, 55 each siRNA’s average normalized raw value for the TNF and 0 00 sisi controlcontrol No Notransf. sip53sip53 siGR sip65 sip65 siLucsiLuc TNF + Dex condition was plotted and 10 wells containing sip53 transf. Condition were indicated in red, illustrating that wells containing sip53 are Fig. 1. p53 identified as potential gene of interest by high-throughput significant outliers compared with other siRNAs (Fig. 1C). screening. (A) Schematic representation of screen design. siRNA oligos pre- To confirm initial screening results, we transfected 293T cells arrayed into 384-well plates, were reverse transfected into 293T cells stably stably expressing a NF-κB luciferase reporter with a Rous sar- transfected with a NF-κB luciferase reporter for 48 h. Cells were then treated coma virus (RSV)–lacZ construct and either sip53, siGR, sip65, with 10 ng/mL TNF ±1 μM Dex in duplicate for 24 h. (B) KEGG analysis. A total si luciferase, or a siRNA control, and treated cells with 10 ng/mL of 290 genes of interest were analyzed to determine reactions and/or TNF ± 1 μM Dex. p53 knockdown was confirmed in this cell line pathways that were statistically overrepresented. A total of 16 of 29 over- represented categories with P value <0.05 are shown. (C) Graphical repre- by qPCR (Fig. S5A). Untransfected cells and cells transfected sentation of average normalized raw luciferase values of individual wells in with a negative control siRNA exhibit a 38 and 48% respective the Invitrogen kinase library treated with TNF (y axis) and TNF + Dex (x axis). decrease in NF-κB luciferase activity upon Dex addition (Fig. Red points indicate 10 wells containing sip53. Raw values in each 384-well 1D, first two conditions). In contrast, cells transfected with sip53 plate were normalized to that plate’s average luciferase activity value. (D) and siGR exhibit only a 6 and 18% respective decrease in NF-κB NF-κB luciferase assay in 293T cells transfected with sip53, siGR, si luciferase, luciferase repression upon Dex addition (Fig. 1D, third and sip65, or a siRNA negative control for 48 h and treated with 10 ng/mL TNF ± fourth conditions). As expected, a decrease in NF-κB luciferase < Dex for 24 h. **P 0.01. activity is seen upon knockdown of p65 and luciferase. Our preliminary conclusion is that a decrease in p53 expression impairs GR-mediated repression of NF-κB luciferase activity in each of the 384-well plates. Expected luciferase activity trends these cells. were seen upon knockdown of GR, p65, and luciferase in all four siRNA oligo libraries, providing validity that the screen could be Validation of p53 as a Mediator of NF-κB Repression by GR. To val- used to successfully identify potential genes of interest (Fig. S1 A idate p53’s effect on glucocorticoid repression of NF-κB activity and B). Redundant siRNA activity (RSA) analysis is an accept- in a more physiologically relevant cell line, we transduced human able statistical method to interpret data from large-scale RNAi monocytic THP-1 cells with a NF-κB luciferase reporter (Fig. screens, while minimizing off-target effects (28). Of the 8,515 S6A), and either a lentiviral shp53–GFP vector or a lentiviral fi genes tested, we identi ed 290 genes of interest by performing GFP vector as control (Fig. S6B), and treated cells with 10 ng/ three RSA analyses: one analyzing data from the two kinase li- mL TNF ± 1 μM Dex. GFP control cells showed more than a 3.5- braries, one from the two druggable genome libraries, and one fold decrease in NF-κB luciferase activity upon Dex addition from data collected from all four library collections (Fig. S2 A– (Fig. 2A, Left). However, this repression upon Dex addition is C). Each gene of interest was specific to the TNF + Dex con- lost in cells expressing shp53 (Fig. 2A, Right). p53 knockdown dition and did not significantly affect luciferase activity by TNF was confirmed by qPCR (Fig. S5B). alone. Genes in each list were rank ordered according to log P To further confirm the effect of p53 loss on NF-κB repression value, and the number one gene of interest, based on log P value by GR, we transduced WT and p53KO MEF cells with the by RSA analysis using all four library collections, was p53 lentiviral NF-κB luciferase reporter and treated cells with 10 ng/ (log P < −27), a tumor suppressor gene that plays a critical role mL TNF ± 1 μM Dex. WT cells show more than a 1.3-fold de- in cell cycle regulation and apoptosis and is mutated in over 50% crease in NF-κB luciferase activity upon Dex addition (Fig. 2B, of human cancers (Fig. S2A). Left). In contrast, NF-κB activity in p53KO MEF cells is com- (GO) analysis of the 290 genes showed enrich- pletely restored in the presence of Dex (Fig. 2B, Right). p53 loss ment for 114 biological processes, P values <0.001. A number of was confirmed in these cells by qPCR (Fig. S5C). On the basis of

17118 | www.pnas.org/cgi/doi/10.1073/pnas.1114420108 Murphy et al. Downloaded by guest on October 1, 2021 κ – fi A B Untreated expression levels of IP-10, a NF- B speci c target gene. We TNF conclude that p53 loss disrupted repression of NF-κB target gene *** MEFs ** TNF + Dex 50 THP-1s 2 transcription by GR. Interestingly, Dex could still repress LPS- 45 1.8 40 1.6 induced TNF mRNA expression in the absence of p53, sug- 35 1.4 30 1.2 gesting that p53 regulates Dex-mediated transcriptional re- 25 1 20 0.8 κ 0.6 pression of some, but not all NF- B target genes.

15 Fold Change Fold Change 10 0.4 5 0.2 0 0 p53 Does Not Alter Upstream NF-κB Signaling. To understand the EF1 -GFP EF1 -shp53-GFP WT p53KO Condition Cell Type mechanism of involvement of p53 in Dex-induced suppression of NF-κB activity, we next analyzed the effect of p53 loss on the Untreated upstream NF-κB signaling cascade. WT and p53KO MEF cells C *** D LPS 16 MEFs BMDMs were treated with 10 ng/mL TNF ± 1 μM Dex for 10, 20, 30, or LPS + Dex 14 5 10 κ α 12 4 88 60 min, and the effect of p53 loss on I B degradation and 10 3 66 phosphorylation kinetics was determined by immunoblot analy- 8 2 4 4 6 κ α 1 22 sis. In WT cells, I B total protein levels decrease starting at the 4 Fold Change 2 Fold Change 0 00 10-min time point upon TNF addition, and the protein is 0 WT p53KO WT p53KO resynthesized by the 60-min time point (Fig. 3A, Top, lanes 1–5). p53KO p53KO reconstituted with WT p53 Cox2 MCP-1 The degradation observed at the 10-min time point correlates to Cell Type the increase in IκBα phosphorylation (Fig. 3A, Middle, lane 2). Fig. 2. p53 loss impairs glucocorticoid repression of NF-κB target genes. (A) Degradation and phosphorylation remain unchanged in WT cells NF-κB luciferase assay in THP-1 cells transduced with a GFP control (Left)or upon Dex addition (Fig. 3A, Top and Middle, lanes 6–10), and shp53–GFP (Right). Cells were treated for 6 h with 10 ng/mL TNF ± 1 μM Dex. these kinetics are unchanged in p53KO cells (Fig. 3A, Top and ***P < 0.001. (B)NF-κB luciferase assay in WT (Left) and p53KO MEF cells Middle, lanes 11–20). Protein levels were normalized to p65 ± μ < (Right) following 24-h treatment with 10 ng/mL TNF 1 M Dex. **P 0.01. loading control (Fig. 3A, Bottom) and quantified using ImageJ (C)NF-κB luciferase assay in p53KO MEF cells (Left) and in p53KO MEF cells κ α reconstituted with WT p53 (Right) following a 6-h treatment with 10 ng/mL (Fig. S8A). We conclude that p53 loss does not alter either I B TNF ± 1 μM Dex. ***P < 0.001. (D) qPCR analysis measuring mRNA levels of phosphorylation or degradation, two essential steps in NF- NF-κB target genes, Cox2 (Left), and MCP-1 (Right) in BMDMs treated with κB activation. 10 ng/mL LPS ± 1 μM Dex for 2 h. mRNA levels normalized to cyclophilin A.

WT MEF cells p53KO MEF cells A Dex -- -- - + + ------+ + + + these results, we conclude that knockdown or loss of p53 impairs - + + GR repression of NF-κB. TNF -+++ + - + + + + - ++++ - + + + + To determine whether we could rescue Dex repression, we Time 010203060 01020 30 60 0 10 20 30 60 01020 30 60 transduced p53KO MEF cells with a lentiviral–WT p53 construct I B and treated cells with 10 ng/mL TNF ± 1 μM Dex. p53KO MEF PI B

cells, transduced with a lentiviral vector making WT p53, show p65 κ substantial restoration of repression of NF- B luciferase activity Lane 123 4 5 6 7 8 910 11 12 13 14 15 16 17 18 19 20 upon treatment with Dex (Fig. 2C). p53 activation is often regulated by complex formation with B WT p53KO murine double minute-2 (MDM2), an E3 ubiquitin ligase that DAPI p65 Merge DAPI p65 Merge inhibits p53 transcriptional activity (29). Nutlin-3 is a protease

inhibitor that was found to specifically disrupt the MDM2/p53 Untreated interaction, consequently stabilizing p53 (30). To determine the effect of p53 stabilization on GR repression of NF-κB, THP-1 cells stably expressing a NF-κB luciferase reporter were treated TNF for 24 h with 10 μM nutlin-3, then treated for an additional 24 h with 10 ng/mL TNF ± 1 μM Dex. Cells that were not treated with nutlin had an average of less than 1.7-fold decrease in NF-κB

luciferase activity upon Dex treatment (Fig. S7, Left). In contrast, TNF + Dex cells treated with nutlin exhibited an average of more than a 2.3- C D fold decrease in NF-κB luciferase activity upon Dex treatment WT p53KO WT p53KO

(Fig. S7, Right). These results suggest that stabilization of p53 MEDICAL SCIENCES enhances GR repression of NF-κB. NT

To further investigate the role of p53 in GR repression of NF- TNF κB, we isolated bone marrow-derived macrophages (BMDMs) from WT and p53KO mice, treated cells with 10 ng/mL TNF ± 1 μM Dex or 10 ng/mL LPS ± 1 μM Dex for 2 h, and analyzed the TNF+D DAPI p65 S311 Merge DAPI p65 S311MergeDAPI p65 S276 Merge DAPI p65 S276 Merge mRNA levels of NF-κB target genes. p53 loss was confirmed in Fig. 3. p53 loss does not alter the upstream NF-κB signaling cascade. (A) these cells by qPCR (Fig. S5D). An increase in cyclooxygenase Western blot analysis measuring total protein levels of IκBα (Top) and phos- (Cox2) mRNA levels is seen upon LPS stimulation, and a more phorylated IκBα (Middle). p65 total protein levels (Bottom) are shown to in- than twofold repression is seen upon Dex addition (Fig. 2D, dicate loading control. WT MEF cells (Left) and p53KO MEF cells (Right) were Left). However, Dex can no longer regulate Cox2 mRNA levels treated for 10, 20, 30, or 60 min with 10 ng/mL TNF ± 1 μM Dex as indicated. fl in p53KO cells. Cells also exhibited an increase in monocyte (B) Immuno uorescence of endogenous p65 expression in WT (Left)and p53KO (Right) BMDMs treated with 10 ng/mL TNF ± 1 μM Dex for 30 min. (C) chemotactic protein-1 (MCP-1) mRNA levels upon LPS stimu- Immunofluorescence of endogenous p65 phosphorylation at serine 311 in WT lation, and a more than twofold decrease upon Dex addition (Left) and p53KO (Right) BMDMs treated with 10 ng/mL TNF ± 1 μM Dex for (Fig. 2D, Right). However, Dex repression of MCP-1 mRNA 30 min. (D) Immunofluorescence of endogenous p65 phosphorylation at levels was impaired in the absence of p53. In addition, p53 loss serine 276 in WT (Left) and p53KO (Right) BMDMs treated with 10 ng/mL impaired Dex repression of both TNF and LPS-induced mRNA TNF ± 1 μM Dex for 30 min. NT, no treatment; TNF + D, TNF + Dex.

Murphy et al. PNAS | October 11, 2011 | vol. 108 | no. 41 | 17119 Downloaded by guest on October 1, 2021 We next evaluated the effect of p53 loss on p65 translocation to A WT p53KO the nucleus. BMDMs were isolated from WT and p53KO mice, DAPI GR Merge DAPI GR Merge treated for 30 min with 10 ng/mL TNF ± 1 μM Dex, and fixed with

4% paraformaldehyde. p65 nuclear accumulation is seen upon Untreated TNF treatment and remains unchanged upon Dex addition in WT cells (Fig. 3B, Left). Similarly, we can see nuclear accumulation of endogenous p65 after treatment with TNF and after treatment Dex with TNF + Dex in p53KO cells (Fig. 3B, Right). Quantification of the number of cells with nuclear p65 staining in each condition BMDMs Mt2 Fkbp5 showed little or no difference (Fig. S8B). We conclude that p53 B 140 30 BMDMs 120 - Dex 25 - Dex loss does not alter p65 translocation to the nucleus. 100 + Dex 20 + Dex p65 is phosphorylated at serine 276 by PKA and MSK-1 and at 80 15 serine 311 by ζPKC, two sites of phosphorylation that are impor- 60 10 Fold Change κ Fold Change 40

tant for NF- B transcriptional activation. To analyze the effect of Fold Change 5 p53 loss on endogenous p65 phosphorylation at these two sites, we 20 0 0 performed endogenous immunofluorescence on WT and p53KO WT p53KO WT p53KO BMDMs treated with 10 ng/mL TNF ± 1 μM Dex for 30 min. Cell Type Cell Type fi 160 BMDMs GILZ Using either Ser311- or Ser276-speci c phosphoantibodies, the C WT Untreated extent of p65 accumulated in the nucleus upon treatment with 140 WT + Dex TNF or TNF + Dex, in either WT or p53KO BMDMs, remains 120 100 p53KO Untreated unchanged (Fig. 3 C and D). Together, these results suggest that 80 p53KO + Dex p53 loss does not alter GR repression of NF-κB by affecting 60 phosphorylation at these two crucial sites required for p65 40

transcriptional activation. Relative Expression Level 20 0 30min 1hr 2hr 4hr 6hr p53 Does Not Alter Upstream GR Signaling but Is Required for GR- Time Point Mediated Transcription. To determine the effect of p53 loss on Fig. 4. p53 loss does not alter upstream GR signaling but does impair Dex- endogenous GR translocation, BMDMs were isolated from WT fl μ fi mediated transcription. (A) Immuno uorescence of endogenous GR locali- and p53KO mice, treated for 30 min with 1 M Dex, and xed zation in WT (Left) and p53KO (Right) BMDMs treated with or without 1 μM with 4% paraformaldehyde. GR nuclear accumulation is seen Dex for 30 min. (B) qPCR analysis measuring mRNA levels of GR target genes, upon Dex treatment in both WT and p53KO cells (Fig. 4A), Mt2 (Left) and Fkbp5 (Right) in BMDMs treated with or without 1 μM Dex for suggesting that p53 loss does not play a role in GR nuclear 2h.(C) Time course showing GILZ mRNA levels measured by qPCR in BMDMs translocation upon ligand stimulation. isolated from WT and p53KO mice and treated with or without Dex for each We then evaluated the effect of p53 loss on GR-mediated time point as indicated. mRNA levels normalized to cyclophilin A. transcription. WT and p53KO BMDMs were treated for 2 h with Dex and quantitative PCR was performed to analyze changes in GR target gene mRNA levels. In WT cells, there is significant We also analyzed the effect of p53 loss on LPS shock-induced increase in mRNA levels of GR target genes, Mt2 and Fkbp5, reduction in body temperature. Body temperatures from WT upon Dex addition (Fig. 4B). In contrast, the increase in mRNA (n = 3) and p53KO (n = 3) C57BL6 mice sham treated with PBS transcription levels of Mt2 and Fkbp5 upon Dex addition, is by i.p. were analyzed every 6 h over a 48-h period using a rectal severely impaired in p53KO cells (Fig. 4B). thermometer probe. The maximum recorded temperature de- We next wanted to analyze changes in mRNA levels of gluco- crease over that period by one individual mouse was 1.1 °C. corticoid-induced (GILZ), a Temperature measurements were also recorded every 6 h over and transcriptional target of GR known to repress NF-κB upon a 48-h period and for every subsequent 24 h up to 5 d total for ± treatment with Dex (31, 32). WT and p53KO BMDMs were WT and p53KO mice treated with 50 mg/kg LPS 10 mg/kg Dex treated with 1 μM Dex for 30 min and 1, 2, 4, and 6 h, and GILZ by i.p. injection. Eleven of 12 WT mice and all p53KO mice (n = mRNA levels were analyzed. Basal mRNA levels of GILZ were 8) treated with LPS exhibited body temperature decreases of at similar in WT and p53KO cells throughout the 6-h period (Fig. least 12 °C within 12 h (Fig. 5B). Nine of 12 WT mice treated 4C). However, Dex-induced expression of GILZ over time is im- with LPS + Dex exhibited varying degrees of temperature de- paired in p53KO cells compared with relative mRNA levels in WT crease from LPS shock, but these mice recovered to within 1 °C cells (Fig. 4C). From the cumulative results shown in Fig. 4, we of their starting temperature within 5 d (Fig. 5C, Left). In con- conclude that a loss of p53 impairs GR-dependent transcription. trast, only 3 of 14 p53KO mice treated with LPS + Dex were able to recover to their starting temperature (Fig. 5C, Right). From In Vivo Suppression of NF-κB Activity by Glucocorticoids. To in- these results, we conclude that p53 loss impaired GR repression vestigate the effect of p53 loss on GR repression of NF-κB of NF-κB activity in vivo in a mouse model of LPS shock. in vivo, we developed a mouse model of LPS shock. WT and p53KO mice treated with 50 mg/kg LPS ± 10 mg/kg Dex by i.p. Discussion injection were monitored for death by LPS shock every 6 h for The mechanisms explaining GR repression of NF-κB activity to the first 54 h and every 24 h for the subsequent 4 d (7-d moni- suppress inflammation are complex and varied. Our RNAi toring period total). All but one WT (n = 11) and all p53KO screen identified p53 as a regulator of NF-κB repression by the (n = 8) mice treated with LPS alone succumbed to LPS shock glucocorticoid receptor, and we confirmed its biological signifi- within 36 h of treatment (Fig. 5A). The last surviving WT mouse cance using both in vitro and in vivo assays. Current insight into treated with LPS died within 54 h (Fig. 5A). Glucocorticoid the relationship between p53 and GR is unclear. It has been treatment led to the survival of 9 out of 12 WT mice and, in suggested that p53 and GR interact with each other, thereby contrast, only 3 of 14 p53KO mice treated with LPS + Dex leading to the suppression of GR transcriptional activity (33–37). survived (Fig. 5A). The Kaplan–Meier survival curve shows these However, most of these studies were performed using trans- trends in Fig. 5A, as 75% of WT mice and only 21% of p53KO fection and overexpression techniques, and none include data mice treated with Dex survived death by LPS shock. from primary cells or in vivo mouse models as shown here. Other

17120 | www.pnas.org/cgi/doi/10.1073/pnas.1114420108 Murphy et al. Downloaded by guest on October 1, 2021 40 WT LPS Treated Mice genes or might also help to recruit transcriptional activation A B 38 wt-WT- LPS LPS 36 machinery to the promoter regions of GR target genes (Fig. 5D, p53KO-p53KO- LPS LPS 34 wt- LPS wt-WT- LPS LPS + Dex + Dex 32 Right). We cannot rule out the potential that p53 may be in- 100 p53KO- LPS 30 p53KO- LPS + Dex 90 wt- LPS + Dex 28 volved in removal of repressive complexes and/or recruitment of 26 80 24 other coactivators as well. Although we have no formal proof 22 n=11 70 75% 20 that p53 transcriptional activation played a role in GR repression 60 wt- LPS

40p53KO-0 LPS6 κ 12 18 30 36 42 48 72 50 wt- LPS p53KO LPS24 Treated Mice of NF- B, we must consider the possibility that downstream 38wt- LPS + Dex 40 36p53KO- LPS Time Point (Hours) targets of p53 may be involved in repression of NF-κB by GR.

Survival (%) p53KO- LPS + Dex 30 34wt- LPS + Dex 32p53KO- LPS + Dex Elucidating the exact mechanism that describes the role of p53 20 30 21% (Degrees Celcius) Temperature in GR repression of NF-κB is an important lingering question. 10 28 26 fi 0 Over half of the genes we identi ed as potential genes of interest 0 20 40 60 80 100 24 22 n=8 were unrepresented in the STRING and/or Reactome databases, Time (Hours) 20 0 6 12 18 24 30 36 42 48 72 96 suggesting that a number of uncharacterized or unstudied genes Time (Hours) may play a role in GR repression of NF-κB. This possibility is

40 WT LPS + Dex Treated Mice 40 p53KO LPS + Dex Treated Mice interesting, as these unique genes may represent a number of C 38 38 entirely new therapeutic targets and reveal unknown mecha- 36 36 34 34 nisms of GR repression. Further studies are needed to in- 32 32 vestigate precisely how p53 functions in GR repression of NF- 30 30 κ 28 28 B. To better understand the protein interactions of these three 26 26 important transcription factors, mutagenesis studies, as well as 24 24 22 n=12 22 n=14 endogenous immunoprecipitation and ChIP analyses, may help

Temperature (Degrees Celcius) Temperature 20 20 0 6 12 18 24 30 36 42 48 72 96 120 0 6 12 18 24 30 36 42 48 72 96 120 elucidate the domains involved and regions of interaction. In Time (Hours) Time (Hours) addition, it would also be interesting to determine whether the other two p53 protein family members, p63 or , play a similar D role in GR repression of NF-κB, and to investigate whether p53 is also involved in repression of NF-κB by other nuclear hormone Direct or Indirect Role in GR and p65 Complex Formation () Direct or Indirect Role in GR Transcription receptor family members. Our discovery that p53 is a regulator of NF-κB repression by GR may have important therapeutic implications. The p53 tu- mor suppressor protein plays a critical role in the development of NF- B target genes GR target genes many human cancers and is often mutated or deleted, leading to tumor progression (42, 43). NF-κB is often up-regulated in tu- mor cells deficient in p53 and in tumor cells that have other Fig. 5. p53 loss impairs glucocorticoid rescue of death in a mouse model of – mutations, which lead to impaired p53 signaling. There is also LPS shock. (A) Kaplan Meier survival curve analysis measuring death from κ fl LPS shock. WT and p53KO C57BL6 mice treated with 50 mg/kg LPS ± 10 mg/ growing evidence that NF- B induction of in ammatory genes kg Dex by i.p. injection are shown. (B) Temperature analysis of WT and is an important component for tumor progression (44, 45). Our p53KO mice treated with LPS. (C) Temperature analysis of WT and p53KO findings suggest that NF-κB could be up-regulated in tumor cells, mice treated with LPS + Dex. Measurements recorded every 6 h over a 48-h in part, because endogenous glucocorticoids, due to a loss of p53 period and every 24 h up to 5 d posttreatment using a rectal thermometer function, can no longer efficiently repress NF-κB. In addition, probe. (D) Proposed mechanisms of p53 role in GR repression of NF-κB. p53 glucocorticoid treatment, although a potent antiinflammatory may play a direct or indirect role in GR and p65 complex formation in the therapy, has not been shown to significantly reduce inflammation nucleus (Left) and/or in GR transcriptional activity (Right). in tumor cells. Our findings suggest that a possible explanation fl κ groups have reported that suppression of p53 impairs GR for why glucocorticoids cannot repress in ammation and NF- B fi function in the presence of Dex and that glucocorticoid treat- in these cells is, in part, because p53 signaling is de cient. We κ ment enhances p53 transcriptional activity (38, 39). These find- have unveiled a unique role for p53 in the repression of NF- B fi ings reinforce our hypothesis that p53 is playing a modulatory by glucocorticoids in vitro and in vivo, a nding that ultimately role in GR function and transcriptional activation. In addition, has important implications for the study and treatment of the fl patients with RA, an autoimmune disorder in which glucocorti- proin ammatory microenvironments found in tumors with ab- coid treatment is often an ineffective antiinflammatory therapy, errant p53 activity. showed reduced p53 expression levels in blood mononuclear Materials and Methods

cells (40) and were also found to exhibit p53 mutations in syn- MEDICAL SCIENCES oviocytes (41). The observations made in these clinical studies Luciferase Assays for Validation of Screen Results. (i) 293T cells stably trans- fected with a 3× NF-κB luciferase reporter were cotransfected with a RSV– lend further credence to our findings, suggesting that non- lacZ construct and sip53 oligo (Invitrogen), siGR (Invitrogen), sip65 (Invi- functional p53 may correlate to the impaired glucocorticoid re- fl trogen), si luciferase (Invitrogen), or a stealth RNAi negative universal pression of in ammation seen in RA patients. control (Invitrogen). Forty-eight hours following transfection with Lip- Loss of p53 does not affect GR or NF-κB nuclear trans- ofectamine 2000, cells were treated with 10 ng/mL TNF ± 1 μM Dex for 24 h. location (Fig. 3B and Fig. 4A), suggesting that p53 aids in GR Luciferase assay and normalization to lacZ transfection was then performed repression of NF-κB when all three proteins are localized to the using Steady Glo (Promega) and Beta Glo (Promega) reagent. (ii) THP-1 cells nucleus. Once all three proteins are in the nucleus, p53 may were transduced with concentrated lentiviral 5×–NF-κB–luciferase–mPGK– target GR to NF-κB, aiding in the complex formation between mcherry and EF1α–GFP or EF1α–shp53–GFP. Cells were sorted for the these two transcription factors, preventing binding to NF-κB mcherry(+) and GFP(+) cell population. Cells were then treated with 10 ± μ sites (Fig. 5D, Left). It is also possible that p53 interacts with GR ng/mL TNF 1 M Dex for 6 h and luciferase assay was performed using κ Steady Glo reagent. (iii) WT and p53KO MEF cells were transduced with and NF- B directly, and that all three form a complex in the ×– κ – – – κ concentrated lentiviral 5 NF- B luciferase mPGK mcherry and sorted for nucleus, which prevents binding of the NF- B proteins to their the mcherry(+) cell population. Cells were treated with 10 ng/mL TNF ± 1 μM cognate DNA binding site(s). We also confirm that p53 loss Dex for 24 h and luciferase assay was performed using Steady Glo reagent. impairs GR transcription of its target genes (Fig. 4 B and C). p53 (iv) p53KO MEF cells were transduced with concentrated lentiviral 5×–NF-κB– may directly target GR to the promoter regions of its target luciferase–mPGK–mcherry and sorted for the mcherry(+) cell population.

Murphy et al. PNAS | October 11, 2011 | vol. 108 | no. 41 | 17121 Downloaded by guest on October 1, 2021 Cells were then transduced with concentrated lentiviral WT–p53. Cells were translocation study were adjusted in Photoshop to the red channel to con- treated with 10 ng/mL TNF ± 1 μM Dex for 24 h and luciferase assay was trast the green seen in the p65 translocation and phosphorylation experi- performed using Steady Glo reagent. (v) THP-1 cells expressing 5×–NF-κB– ments. Nuclei were stained with the fluorescent dye 4′,6-diamidino-2- luciferase–mPGK–mcherry were pretreated for 24 h with 10 μM nutlin phenylindole (DAPI). Fluorescence microscopy was performed on a Leica TCS ± μ (Sigma) and then treated for another 24 h with 10 ng/mL TNF 1 M Dex. SP2 AOBS confocal microscope. Luciferase assay was performed using Steady Glo reagent. P values for each ’ experiment were calculated by Student s t test. Kaplan–Meier Survival Analysis and Temperature Measurements. Body tem- peratures from WT (n = 3) and p53KO (n = 3) C57BL6 mice sham treated with qPCR. Bone marrow-derived macrophages were isolated from C57BL6 WT and PBS (by i.p. injection) were analyzed every 6 h over a 48-h period using p53KO mice and grown in 20% FBS in DMEM. After differentiation for 5 d with a rectal thermometer probe to measure normal temperature changes macrophage colony stimulating factor (M-CSF) (R&D Systems), cells were throughout a day. WT and p53KO mice treated with 50 mg/kg LPS ± 10 mg/ treated with 10 ng/mL LPS (Sigma) ± 1 μM Dex, 10 ng/mL TNF ± 1 μM Dex, 1 μM Dex, or left untreated for 2 h. Total mRNA was extracted using TRIzol reagent kg Dex by i.p. injection were monitored for death by LPS shock every 6 h for fi (Invitrogen), qPCR was performed using SYBR green PCR mastermix (Applied the rst 54 h and every 24 h for the subsequent 4 d (7-d monitoring period Biosystems), and mRNA expression levels were normalized to cyclophilin A. total). Temperature measurements were also recorded at these time points using a rectal thermometer probe. Mice treated with LPS or LPS + Dex Immunofluorescence. Bone marrow-derived macrophages isolated from WT exhibiting less than a 1.5 °C temperature decrease were excluded. and p53KO mice were grown in 20% FBS in DMEM, differentiated for 5 d with M-CSF, and treated for 30 min with 10 ng/mL TNF, 10 ng/mL TNF + 1 μM Dex, ACKNOWLEDGMENTS. We thank Drs. Oded Singer and Aaron Parker for or 1 μM Dex, then fixed with 4% paraformaldehyde. Cells were incubated their assistance. This work was supported, in part, by the National Institutes of with rabbit polyclonal anti-p65 detecting phospho S311 (ab51059), rabbit Health (AI048034, DK057978, and HL2782081), CA014495 California Institute of Regenerative Medicine, Ipsen/Biomeasure, Sanofi-Aventis, the H. N. and polyclonal anti-p65 detecting phospho S276 (ab30623), mouse monoclonal fi Frances C. Berger Foundation, The Helmsley Charitable Trust, and Howard anti-p65 (sc-8008), or rabbit polyclonal anti-GR (Af nity Bioreagents; PA1- Hughes Medical Institute. I.M.V. is an American Cancer Society Professor of 512). Secondary antibodies were Alexa 488-conjugated goat antimouse IgG Molecular Biology and holds the Irwin and Joan Jacobs Chair in Exemplary or Alexa 488-conjugated goat antirabbit IgG, generating green fluorescence Life Sciences. R.M.E. is a Howard Hughes Medical Institute Investigator and for both the monoclonal and polyclonal antibodies. GFP images in the GR holds the March of Dimes Chair in Molecular and .

1. Li Q, Verma IM (2002) NF-kappaB regulation in the . Nat Rev Immunol 24. Gerritsen ME, et al. (1997) CREB-binding protein/p300 are transcriptional coactivators 2:725–734. of p65. Proc Natl Acad Sci USA 94:2927–2932. 2. Verma IM, Stevenson JK, Schwarz EM, Van Antwerp D, Miyamoto S (1995) Rel/NF-κ B/I 25. McKay LI, Cidlowski JA (2000) CBP (CREB binding protein) integrates NF-κB and glu- κ B family: Intimate tales of association and dissociation. Genes Dev 9:2723–2735. cocorticoid receptor physical interactions and antagonism. Mol Endocrinol 14: 3. Ghosh S, May MJ, Kopp EB (1998) NF-κ B and Rel proteins: Evolutionarily conserved 1222–1234. mediators of immune responses. Annu Rev Immunol 16:225–260. 26. Xu W, et al. (2001) A transcriptional switch mediated by cofactor methylation. Science 4. Baeuerle PA, Henkel T (1994) Function and activation of NF-κ B in the immune system. 294:2507–2511. Annu Rev Immunol 12:141–179. 27. Ray A, Prefontaine KE (1994) Physical association and functional antagonism between 5. Thanos D, Maniatis T (1995) NF-κ B: A lesson in family values. Cell 80:529–532. the p65 subunit of transcription factor NF-κ B and the glucocorticoid receptor. Proc 6. Mercurio F, et al. (1997) IKK-1 and IKK-2: cytokine-activated IkappaB kinases essential Natl Acad Sci USA 91:752–756. for NF-kappaB activation. Science 278:860–866. 28. König R, et al. (2007) A probability-based approach for the analysis of large-scale 7. Spencer E, Jiang J, Chen ZJ (1999) Signal-induced ubiquitination of IkappaBalpha by RNAi screens. Nat Methods 4:847–849. the F-box protein Slimb/beta-TrCP. Genes Dev 13:284–294. 29. Momand J, Zambetti GP, Olson DC, George D, Levine AJ (1992) The mdm-2 oncogene 8. Zhong H, Voll RE, Ghosh S (1998) Phosphorylation of NF-kappa B p65 by PKA stim- product forms a complex with the p53 protein and inhibits p53-mediated trans- ulates transcriptional activity by promoting a novel bivalent interaction with the activation. Cell 69:1237–1245. coactivator CBP/p300. Mol Cell 1:661–671. 30. Vassilev LT, et al. (2004) In vivo activation of the p53 pathway by small-molecule 9. Vermeulen L, De Wilde G, Van Damme P, Vanden Berghe W, Haegeman G (2003) antagonists of MDM2. Science 303:844–848. Transcriptional activation of the NF-kappaB p65 subunit by mitogen- and stress-ac- 31. Yang N, Zhang W, Shi XM (2008) Glucocorticoid-induced leucine zipper (GILZ) me- tivated protein kinase-1 (MSK1). EMBO J 22:1313–1324. diates glucocorticoid action and inhibits inflammatory cytokine-induced COX-2 ex- 10. Duran A, Diaz-Meco MT, Moscat J (2003) Essential role of RelA Ser311 phosphoryla- pression. J Cell Biochem 103:1760–1771. tion by zetaPKC in NF-kappaB transcriptional activation. EMBO J 22:3910–3918. 32. Di Marco B, et al. (2007) Glucocorticoid-induced leucine zipper (GILZ)/NF-kappaB in- 11. Pratt WB, Toft DO (1997) receptor interactions with and teraction: Role of GILZ homo-dimerization and C-terminal domain. Nucleic Acids Res immunophilin chaperones. Endocr Rev 18:306–360. 35:517–528. 12. Dittmar KD, Demady DR, Stancato LF, Krishna P, Pratt WB (1997) Folding of the 33. Zhang L, Nie L, Maki CG (2006) P53 and p73 differ in their ability to inhibit gluco- glucocorticoid receptor by the heat shock protein (hsp) 90-based chaperone ma- corticoid receptor (GR) transcriptional activity. Mol Cancer 5:68. chinery. The role of p23 is to stabilize receptor.hsp90 heterocomplexes formed by 34. Suehiro T, et al. (2004) Regulation of human glucocorticoid receptor gene tran- hsp90.p60.hsp70. J Biol Chem 272:21213–21220. scription by Sp1 and p53. Mol Cell Endocrinol 222:33–40. 13. Davies TH, Ning YM, Sánchez ER (2002) A new first step in activation of steroid re- 35. Sengupta S, Vonesch JL, Waltzinger C, Zheng H, Wasylyk B (2000) Negative cross-talk ceptors: Hormone-induced switching of FKBP51 and FKBP52 immunophilins. J Biol between p53 and the glucocorticoid receptor and its role in neuroblastoma cells. Chem 277:4597–4600. EMBO J 19:6051–6064. 14. Schoneveld OJ, Gaemers IC, Lamers WH (2004) Mechanisms of glucocorticoid signal- 36. Sengupta S, Wasylyk B (2001) Ligand-dependent interaction of the glucocorticoid ling. Biochim Biophys Acta 1680:114–128. receptor with p53 enhances their degradation by Hdm2. Genes Dev 15:2367–2380. 15. Schaaf MJ, Cidlowski JA (2002) Molecular mechanisms of glucocorticoid action and 37. Maiyar AC, Phu PT, Huang AJ, Firestone GL (1997) Repression of glucocorticoid re- resistance. J Steroid Biochem Mol Biol 83:37–48. ceptor and DNA binding of a glucocorticoid response element within 16. McKay LI, Cidlowski JA (1999) Molecular control of immune/inflammatory responses: the serum/glucocorticoid-inducible protein kinase (sgk) gene promoter by the p53 Interactions between nuclear factor-κ B and steroid receptor-signaling pathways. tumor suppressor protein. Mol Endocrinol 11:312–329. Endocr Rev 20:435–459. 38. Urban G, et al. (2003) Identification of a functional link for the p53 tumor suppressor 17. Reichardt HM, Schütz G (1998) Glucocorticoid signalling—multiple variations of protein in dexamethasone-induced growth suppression. J Biol Chem 278:9747–9753. a common theme. Mol Cell Endocrinol 146:1–6. 39. Crochemore C, Michaelidis TM, Fischer D, Loeffler JP, Almeida OF (2002) Enhance- 18. Boumpas DT, Chrousos GP, Wilder RL, Cupps TR, Balow JE (1993) Glucocorticoid therapy for ment of p53 activity and inhibition of neural cell proliferation by glucocorticoid re- immune-mediated diseases: Basic and clinical correlates. AnnInternMed119:1198–1208. ceptor activation. FASEB J 16:761–770. 19. Funder JW (1997) Glucocorticoid and mineralocorticoid receptors: Biology and clinical 40. Maas K, Westfall M, Pietenpol J, Olsen NJ, Aune T (2005) Reduced p53 in peripheral relevance. Annu Rev Med 48:231–240. blood mononuclear cells from patients with rheumatoid arthritis is associated with 20. Neeck G (2002) Fifty years of experience with therapy in the study and loss of radiation-induced apoptosis. Arthritis Rheum 52:1047–1057. treatment of rheumatoid arthritis. Ann N Y Acad Sci 966:28–38. 41. Yamanishi Y, et al. (2005) p53 tumor suppressor gene mutations in fibroblast-like 21. Scheinman RI, Cogswell PC, Lofquist AK, Baldwin AS, Jr. (1995) Role of transcriptional synoviocytes from erosion synovium and non-erosion synovium in rheumatoid ar- activation of I κ B α in mediation of immunosuppression by glucocorticoids. Science thritis. Arthritis Res Ther 7:R12–R18. 270:283–286. 42. Vogelstein B, Lane D, Levine AJ (2000) Surfing the p53 network. Nature 408:307–310. 22. Auphan N, DiDonato JA, Rosette C, Helmberg A, Karin M (1995) Immunosuppression 43. Vousden KH, Lu X (2002) Live or let die: The cell’s response to p53. Nat Rev Cancer 2: by glucocorticoids: Inhibition of NF-κ B activity through induction of I κ B synthesis. 594–604. Science 270:286–290. 44. Viatour P, Merville MP, Bours V, Chariot A (2005) Phosphorylation of NF-kappaB and 23. Horwitz KB, et al. (1996) coactivators and corepressors. Mol Endo- IkappaB proteins: Implications in cancer and inflammation. Trends Biochem Sci 30:43–52. crinol 10:1167–1177. 45. Clevers H (2004) At the crossroads of inflammation and cancer. Cell 118:671–674.

17122 | www.pnas.org/cgi/doi/10.1073/pnas.1114420108 Murphy et al. Downloaded by guest on October 1, 2021