Positive and Negative Feedback Loops in the P53 and Mrna 3′ Processing Pathways

Positive and negative feedback loops in the p53 and mRNA 3′ processing pathways

Emral Devanya,1, Xiaokan Zhanga,1, Ji Yeon Parkb, Bin Tianb, and Frida Esther Kleimana,2

aChemistry Department, Hunter College and Graduate Center, City University of New York, New York, NY 10065; and bDepartment of Biochemistry and Molecular Biology, New Jersey Medical School, University of Medicine and Dentistry of New Jersey, Newark, NJ 07101

Edited by Carol Prives, Columbia University, New York, NY, and approved January 17, 2013 (received for review July 20, 2012)

Although the p53 network has been intensively studied, genetic signal AREs in their 3′ untranslated regions (UTRs), which have analyses long hinted at the existence of components that remained been shown to play significant roles in mRNA stability regulation elusive. Recent studies have shown regulation of p53 at the mRNA (12). PARN has been shown to be involved in ARE-mediated level mediated via both the 5′ and the 3′ untranslated regions and deadenylation and to promote tristetraprolin (TTP)-directed affecting the stability and translation efficiency of the p53 mRNA. deadenylation in vitro (13). KH-type splicing regulatory protein Here, we provide evidence of a feedback loop between p53 and the recruits PARN to ARE-containing mRNAs to initiate the poly(A) tail shortening that is followed by exosome-mediated degradation poly(A)-specific ribonuclease (PARN), in which PARN deadenylase (14). Interestingly, tumor suppressors, such as breast cancer type keeps p53 levels low in nonstress conditions by destabilizing p53 1 susceptibility protein (BRCA1) and BRCA1-associated RING mRNA, and the UV-induced increase in p53 activates PARN dead- domain protein (BARD1), associated to the polyadenylation enylase, regulating gene expression during DNA damage response factor CstF1 have been shown to regulate deadenylation by in a transactivation-independent manner. This model is innovative functional interactions with PARN deadenylase (5). because it provides insights into p53 function and the mechanisms Extending those studies, here we are investigating the possi- behind the regulation of mRNA 3′ end processing in different cel- bility that p53 might regulate not only mRNA 3′ cleavage (4) but lular conditions. also PARN-dependent deadenylation in different cellular conditions. As part of these studies, we also identified the mRNA deadenylation | mRNA 3′ processing | mRNA steady state levels | targets of PARN in nonstress conditions. Our results provide gene expression control evidence of a unique feedback loop between p53 and PARN, in which PARN deadenylase keeps p53 levels low in nonstress conditions by destabilizing p53 mRNA through its 3′ UTR, and ownstream signaling in the p53 pathway includes several cel- the UV-induced increase in p53 activates PARN, representing Dlular responses. The expression of a large number of genes a mechanism of gene expression regulation in a transactivation- involved in DNA repair, cell cycle arrest, and/or apoptosis is reg- independent manner. ulated by transactivating properties of p53. This occurs via specific DNA binding of the p53 protein to a p53 response element that is Results found either in promoters or introns of target genes (1). Trans- activation-independent functions of p53 have also been described To investigate the role of p53 in deadenylation, we used a group of BIOCHEMISTRY isogenic cell lines that express different levels of p53 (Fig. 1A and (2). For example, certain microRNAs (miRNAs) are regulated by Fig. S1): the colon cancer HCT116 and p53-null HCT116 cell lines, p53, and these miRNAs cause dramatic changes in gene expres- the colon carcinoma RKO and RKO-E6 (low p53 levels) cell lines, sion, offering an indirect p53-mediated control of gene expression and the mouse embryonic fibroblasts (MEFs) and p53-null MEFs. at the posttranscriptional level (3). Recently, we showed that p53 Nuclear extracts (NEs) from those cells were assayed for dead- can inhibit mRNA 3′ cleavage through its interaction with the enylation activity using a radiolabeled L3(A ) RNA substrate as cleavage stimulation factor 1 (CstF1) (4). CstF1 can also interact 30 fi described (5). As described in previous studies (5), Fig. 1A shows with poly(A)-speci c ribonuclease (PARN) deadenylase, and the that deadenylation activity in NEs of RKO cells treated with con- CstF1/PARN complex formation has a role in the regulation of fi ′ trol siRNA increased signi cantly after UV treatment. Interestingly, gene expression by inhibition of mRNA 3 cleavage and activation siRNA-mediated knockdown of p53 in RKO cells abolished the of deadenylation upon DNA damage (5). PARN, an mRNA decay UV-induced activation of deadenylation, suggesting that p53 enzyme, has been studied extensively in vitro at the biochemical levels might activate deadenylation. Consistent with this, RKO- levels but very little is known of its biological targets and its role in E6, p53-null HCT116, and p53-null MEFs cells did not show UV- different cellular conditions. Recently, it has been shown that induced activation of deadenylation, which was observed in RKO, PARN regulates the expression of genes involved in mRNA me- HCT116, and MEFs cells. Importantly, only increasing amounts tabolism, transcription, and cell motility in mouse myoblasts (6). of either full-length His-p53 (Fig. 1 B and C) or the C-terminal Our studies indicate that the CstF/PARN complex can decrease fragment of p53 (Fig. 1C) can induce deadenylation in a reaction the mRNA levels of housekeeping genes under DNA-damaging using a limited amount of His-PARN in a cell-free assay, sug- conditions and of genes involved in cell growth and differentiation gesting that this is a transactivation-independent function of under nonstress conditions (5). p53. However, neither the two p53 derivatives that lacked the Almost all eukaryotic mRNA precursors, with the exception of C-terminal region of p53 nor GST alone had an effect on the histones, undergo a cotranscriptional cleavage followed by poly- ′ fi deadenylation reaction (Fig. 1C). None of the His-p53 derivatives adenylation at the 3 end. This rst round of polyadenylation is were able to deadenylate the substrate in the absence of His- considered a default modification for most mRNAs and confers stability. In contrast, activation of deadenylation alters the length of poly(A) tails, affecting mRNA stability, transport, or translation initiation, and hence gene expression (7). Thus, mecha- Author contributions: E.D., X.Z., B.T., and F.E.K. designed research; E.D., X.Z., and J.Y.P. performed research; E.D., X.Z., and J.Y.P. contributed new reagents/analytic tools; E.D., nisms controlling deadenylation are highly regulated and play key X.Z., J.Y.P., B.T., and F.E.K. analyzed data; and E.D., X.Z., B.T., and F.E.K. wrote the paper. roles in cellular responses, such as mRNA surveillance, DNA fl damage response (DDR), and tumor progression, as well as cell The authors declare no con ict of interest. development and differentiation (5, 8–11). Deadenylation of This article is a PNAS Direct Submission. mRNA is regulated by miRNAs, adenylate-uridylate–rich element 1E.D. and X.Z. contributed equally to this work. (ARE) binding proteins, polyadenylation factors, and RNA bind- 2To whom correspondence should be addressed. E-mail: [email protected]. ing (RB) factors that recognize cis-acting sequences in the target. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. About 12% of mammalian mRNAs bear important regulatory 1073/pnas.1212533110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1212533110 PNAS | February 26, 2013 | vol. 110 | no. 9 | 3351–3356 Downloaded by guest on September 26, 2021 A RKO RKO-E6 siRNA: control p53 --- HCT116 HCT116 p53 -/- MEF MEF p53 -/- UV: - + - + - + - UV: - + - + UV: - + - +

-L3(A30) -L3(A ) 30 -L3(A30)

-L3 -L3 -L3

RD: 34 58 28 27 21 22 RD: 14 42 0.2 0.2 RD: 18 37 16 14 B C His-p53: - - --- His-p53: - - - 0.1 0.25 0.5 0.5 ng PARN: 1 0.2 0.2 0.2 0.2 0.2 - - - - 0.2 0.2 0.2 ng PARN: 1 0.5 0.2 0.2 0.2 0.2 - ng GST: ------0.1 0.25 0.5 ng

-L3(A30) -L3(A30)

-L3 -L3

RD: 12 9 4 32 57 76 0 RD: 26 14 52 12 35 9 3.5 2 2 2 2 2 2.5

Fig. 1. The levels of deadenylation correlate with expression levels of p53. (A) Samples from cells expressing different levels of p53 show different levels of deadenylation. A representative deadenylation reaction from three independent assays is shown. siRNA-mediated knockdown of p53 abolishes UV-induced ac-

tivation of deadenylation in RKO cells. NEs of the indicated cells treated with UV irradiation and allowed to recover for 2 h were analyzed for radiolabeled L3(A30) deadenylation as described (4, 5). NEs from RKO cells treated with p53/control siRNA and UV irradiation were also analyzed. Positions of the polyadenylated RNA

L3(A30) and the L3 deadenylated product are indicated. Numbers beneath gel lanes indicate relative deadenylation (RD). RD was calculated as [L3 fragment/ (L3 fragment + L3 (A30)] × 100. Quantiﬁcations were done with ImageJ software (http://rsb.info.nih.gov/ij/). (B) p53 can activate PARN-dependent deadenylation in vitro. Deadenylation assays using different concentrations of His-PARN were performed in the presence of capped L3(A30)RNAsubstrateasde- scribed (5) and increasing amounts of His-p53. Reactions were analyzed as in A. (C) The C-terminal domain of p53 activates PARN in vitro. Deadenylation assays were performed as in B with the addition of either full-length p53 (FL) or His-p53 derivatives. The truncated forms of p53 used in this assay were described before (4) and include p53 amino acids 1–293, p53 amino acids 94–293 (DNA binding domain), and p53 amino acids 94–393.

PARN (Fig. 1C). To further analyze the role of p53 in dead- interactions were probably not due to an RNA tethering effect. enylation, we examined the physical association of p53 with PARN. These results indicate that the same region of p53 required for Our pull-down (Fig. 2 A–C) and coimmunoprecipitation assays binding PARN (Fig. 2B) is necessary for activating PARN dead- (Fig. 2D) indicate that p53 can form (a) protein complex(es) with enylase (Fig. 1C). Together these results indicate that p53 can in- PARN in NEs from RKO cells in nonstress conditions and after teract with PARN to form (a) complex(es) and that p53 expression UV treatment. The results showed that the C-terminal domain of levels can activate PARN deadenylase and, therefore, might reg- PARN (Fig. 2B), which has been described to interact with CstF1 ulate gene expression. and cap-binding protein 80 (CBP80) (5), and the C-terminal do- Because PARN is involved in ARE-mediated deadenylation main of p53 (Fig. 2C), which has been described to have regulatory (13), promotes TTP-directed deadenylation (16), and decreases functions in DDR (15), are important for the complex formation. mRNA levels of ARE-containing genes under nonstress conditions Because samples were treated with RNase A, the observed (5), we decided to extend these studies and determine which

A Fig. 2. p53 interacts with PARN to form a protein complex. (A) Input PD SN Input PD SN Immobilized His-PARN (Left) or His-p53 (Right) on nickel beads UV: -+ -+ -+ UV: - + - + - + were incubated with NEs from untreated or UV-treated RKO cells. -Topo II -Topo II A representative pull-down reaction from three independent assays is shown. Equivalent amounts of the pulldowns (PD) and -His-PARN -His-p53 supernatants (SN) were analyzed and proteins were detected by -p53 -PARN immunoblotting with the indicated antibodies. His-p53 and His- PARN constructs were previously described (4). Twenty percent of 1 .0 2.7 0.6 1.3 0.6 0.9 1.0 1.1 0.9 1.0 1.3 1.2 B C the NE used in the pull-down reactions is shown as input. The basal FL NTD CTD FL 1-293 94-393 94-293 level of the proteins was arbitrarily set at 1.0 in the ﬁrst lane, and PD SN PD SN PD SN PD SN PD SN PD SN PD SN relative fold change of each protein level is shown below each -Topo II -Topo II lane. (B) C-terminal domain of PARN interacts with p53. Pull-down assays were performed as in A using full-length, N-terminal do- -p53 -PARN main (NTD) or C-terminal domain (CTD) of His-PARN. (C)C-terminal 1.0 0.8 1.0 0.1 1.0 0.7 1.0 1.0 0.9 0.1 0 0.9 0.6 0.4 0 0.9 domain of p53 interacts with PARN. Pull-down assays were performed as in A with full-length or the His-p53 derivatives described p53 PARN D Input IP SN IP SN Input IP SN IP SN in Fig. 1D.(D) PARN and p53 coimmunoprecipitation from NEs UV: - + --++ UV: - + --++ of HeLa cells. A representative immunoprecipitation (IP) reaction -Topo II -Topo II from three independent assays is shown. The NEs were immunoprecipitated with anti-p53 and anti-PARN. Equivalent amounts of - -PARN -p53 the SN and the pellets (IP) were resolved by SDS/PAGE and proteins -p53 were detected by immunoblotting using antibodies against PARN, -PARN p53, and topoisomerase II (Topo II) as described (4, 5). Twenty 1 .0 1.3 0.2 1.4 0.2 0.8 1.0 1.2 0.4 1.1 0.9 1.1 percent of the NE used in the IP reaction is shown as input.

3352 | www.pnas.org/cgi/doi/10.1073/pnas.1212533110 Devany et al. Downloaded by guest on September 26, 2021 mRNAs might be regulated by PARN using microarray assays. and mRNA instability of FBJ osteosarcoma oncogene and mye- Nuclear RNA samples were isolated from HeLa cells treated with locytomatosis oncogene (c-myc), keeping their expression levels control or PARN siRNAs under nonstress conditions and analyzed low under nonstress conditions (5). Because PARN is a dead- by microarray. Pathway analysis of regulated genes using Ingenuity enylase, we expected an increase in the steady-state levels of its Systems applications indicated that the p53 signaling pathway is the targets by PARN knockdown. However, our results show both up- most significantly affected by PARN knockdown in nonstress and down-regulation of transcripts, suggesting complex effects of conditions (Fig. 3A and Tables S1 and S2), suggesting that PARN PARN knockdown on the expression of these mRNAs. Impor- expression has a specific effect on the expression of genes associ- tantly, consistent with the pathway analysis results, PARN knock- ated with p53-mediated signaling in those conditions. In addition, down resulted in a significant increase not only of p53 mRNA but the p53-related gene network was found to be the most significantly also of p53 protein levels (compare lanes 1 and 2 in Fig. 3BRight), regulated by network analysis and transcription factor analysis (Fig. reaching expression levels similar to that observed after UV S2). We further confirmed the effect of PARN knockdown on the treatment (compare lanes 2 and 3 in Fig. 3BRight). After UV abundance of several transcripts in the p53 signaling pathway by treatment, the changes in the levels of p53 mRNA and p53 protein quantitative (q) RT-PCR (Fig. S3). Although there was a good were PARN-independent, indicating there is (are) other mecha- correlation between the change observed in the microarray and nism(s) involved in the regulation of p53 expression during DDR. that determined by qRT-PCR, the qRT-PCR showed changes of To determine the effect of PARN expression on the stability of a greater magnitude than the array. These results support our p53 mRNA, we compared mRNA decay rates of p53 transcript in previous study that showed that PARN can promote deadenylation cells treated with control- or PARN-siRNA. The half-life of the

A B 0.6

0.5 siRNA: UV: --+ + 0.4 -Topo II

0.3 -p53 1 3.1 3.6 3.5 0.2 -PARN (relative to GAPDH) (relative

p53 mRNA expression 0.1 1 0 1.5 0.3

0 siRNA: cont PARN cont PARN UV: - - + + C D E 4 4 1.2 p53 cont) 3 3 BIOCHEMISTRY

2 2 0.8

fragmentation 1 1 RKO HCT116 -200 0 0 DNA fragmentation DNA 102 103 104 102 103 104 0.4 t1/2=4.74h

cells/well (Fold change compared to cells/well

CTRL (Fold change compared to cont) PARN KD t1/2=17.11h

6 cont) 4 Fraction mRNAFraction remaining polyA tail 0 -100 length 3 0 60 120 180 240 4 Time (min) 2 2 CTRL PARN KD fragmentation 1 Act-D: 0 0.5 1 2 4 0 0.5 1 2 4 h RKO-E6 p53 -/- HCT116 -0 0 DNA fragmentation DNA 0 -Topo II 102 103 104 102 103 104 (Fold change compared to (Fold change compared to cont) cells/well cells/well -PARN -Actin exon 3-4 Control control+UV PARN KD

Fig. 3. PARN deadenylase significantly affects the cellular expression of genes of the p53 pathway under nonstress conditions. (A) Pathway analysis of significantly regulated genes by PARN. Nuclear RNA samples isolated from HeLa cells, treated with siRNAs targeting PARN or control, were analyzed using the Human Gene 1.0 ST GeneChip (Affymetrix) array. Significant genes were selected by t test (P value < 0.05). Analysis of canonical pathways was conducted by using data from Ingenuity Systems (www.ingenuity.com). The bar graph shows significance of pathway for regulated genes. P values were calculated using the Fisher’s exact test, and the −log (P value) values are displayed. Only the top five pathways are shown. Data represent three independent experiments. Refer also to Fig. S2.(B) p53 mRNA and protein levels are affected by PARN expression. qRT-PCR and Western blot analysis of p53 expression after UV treatment using RNA or protein samples, respectively, from cells treated with control/PARN siRNA. A representative Western blot from three independent assays is shown. Topo II was used as loading control. The basal level of the proteins was arbitrarily set at 1.0 in the first lane, and relative fold change of each protein level is shown below each lane. (C) PARN regulates p53 mRNA half-life. mRNA decay rates for p53 and ACTIN, a non-PARN target gene, were determined by qRT-PCR at different time points following PARN/control siRNA- and Act-D treatment. The relative half-life of the p53 transcript was calculated from three independent samples. Errors represent the SD derived from three independent experiments. Western blot analysis of PARN expression after PARN/ control siRNA and Act-D treatment is also shown. (D) PARN regulates p53 mRNA poly(A) tail length. Nuclear RNA from PARN/control siRNA-treated cells was reverse-transcribed using an oligo(dT)-anchor primer and amplified using an oligonucleotide that hybridizes within the 3′ UTR of p53 mRNA. The products were separated on a nondenaturing PAGE and detected by ethidium bromide staining. An RT-PCR product from a non-PARN target gene (ACTIN exon 3–4) was used as a loading control. A representative PAGE from three independent assays is shown. Molecular weight standard (MWS, 100-bp ladder from Promega) is also included. (E) PARN knockdown and UV treatment induce similar apoptotic responses in a p53-dependent manner. The DNA fragmentation was calculated from three independent samples.

Devany et al. PNAS | February 26, 2013 | vol. 110 | no. 9 | 3353 Downloaded by guest on September 26, 2021 p53 transcript was analyzed by qRT-PCR of nuclear RNA samples expression on cell death is p53-dependent. Together these results taken at different time points from PARN/control siRNA- and indicate that PARN plays a role in controlling p53 mRNA steady- actinomycin D (Act-D)-treated cells. Previous observations have state levels and p53 expression in nonstress conditions. Although shown a similar half-life for the p53 transcript (17). Our results many reports have been published about control of p53 protein indicate that PARN knockdown significantly stabilized the p53 expression and its effect on downstream pathways, very little is transcript (Fig. 3C). Because PARN is a deadenylase, the stabili- known of the mechanisms behind the control of p53 mRNA zation of p53 mRNA by its knockdown might be due to changes in steady-state levels in different conditions (20). the poly(A) tail length. Importantly, as shown in Fig. 3D, siRNA- Because most of the regulatory elements involved in PARN- mediated knockdown of PARN elongated the poly(A) tail length mediated regulation of mRNA stability are located in the 3′ UTR of p53 mRNA (quantification is shown in Fig. S4). Together, our of the genes, we decided to determine whether the PARN-induced results indicate that the PARN deadenylase affects p53 expression decrease of p53 mRNA levels under nonstressed conditions is by regulating poly(A) length and hence mRNA stability. The bi- through this region of p53. The firefly luciferase assay was used ological relevance of this observation is supported by the fact that with constructs under the control of either the p53 3′ UTR or the PARN knockdown induces an apoptotic response similar to that vector 3′ UTR (Fig. 4A). A significant increase in firefly/Renilla observed after UV treatment in a p53-dependent manner (Fig. ratio for the construct with the p53 3′ UTR relative to the control 3E). Our results indicate that UV treatment and PARN knock- construct was detected in RKO/RKO-E6 and HCT116/HCT116 − − down have a similar effect on cell death in cells expressing normal p53 / cells treated with PARN siRNA (Fig. 4B). RNA immu- levels of p53 (HCT116 and RKO). RKO-E6 cells showed levels of noprecipitation (RIP) assays using antibodies against PARN showed apoptosis similar to those in RKO cells because the disruption that p53 mRNA can form a complex with PARN in samples from of wild-type p53 function by E6 expression results in loss of p53- cross-linked RKO cells (Fig. 4C and Fig. S5), indicating that dependent DNA repair but not UV-induced apoptosis (18). Con- PARN can regulate p53 mRNA stability by, most probably, an sistent with Ford and Hanawalt (19), p53-null HCT116 cells indirect association to the 3′ UTR. RIP assays also showed that did not show an induction of apoptosis after either UV treat- PARN can form a complex with c-myc RNA, which is another ment or PARN knockdown, suggesting that the effect of PARN target of PARN deadenylase (5). Recently, a G-quadruplex

A D 2 ARE PAS PAS 1.5 SV40 Firefly Renilla p53 3’UTR CMV enhancer Luciferase Luciferase Luc-p53 renilla ratio

PAS 1 PAS SV40 Firefly Renilla vector CMV enhancer Luciferase Luciferase vect 0.5 Firefly/ noARE PAS PAS SV40 Firefly Renilla p53 3’UTR CMV Luc- noARE enhancer Luciferase Luciferase 0 Luc: p53 noARE p53 noARE p53 noARE p53 noARE RKO RKO-E6 HCT116 HCT116 p53-/-

B 2.5 E 2.5

2 2

1.5 1.5

1 1 siRNA/control ratio 0.5 0.5 p53 3’UTR/vector ratio p53 3’UTR/vector 0

0 PARN siRNA cont PARN cont PARN cont PARN cont PARN Luc: p53 noARE p53 noARE p53 noARE p53 noARE RKO RKO-E6 HCT116 HCT116 p53-/- RKO RKO-E6 HCT116 HCT116 p53-/-

C 30 F 9 G p53 noARE

20 6 -Topo II in IP 10 3 in PARN IP -p53 old change of RNA old change F

p53/noARE RNA ratio -p53* 0 0 IP: PARN IgG PARN IgG PARN IgG PARN IP: p53 noARE p53 noARE p53 noARE p53 noARE p53 c-myc actin RKO RKO-E6 HCT116 HCT116 p53-/- -PARN 1 0.8 1.2 0.3 1.3

Fig. 4. PARN regulates p53 expression through ARE sequence present in the 3′ UTR of p53 mRNA. (A)Diagramoffirefly luciferase reporter constructs with different 3′ UTR sequences from the p53 gene. Polyadenylation signals (PAS) are indicated. (B) Constructs carrying the p53 3′ UTR (p53) or not (vector) were transfected in cells treated with PARN or control siRNAs. The ratios of the firefly/Renilla values for the p53 construct relative to the vector construct are shown. The firefly/Renilla values were calculated from three independent samples. Errors represent the SD derived from three independent experiments. (C)PARNcaninteract with p53 and c-myc mRNAs under nonstress conditions. The extracts were immunoprecipitated with either anti-PARN or IgG antibodies. The endogenous nuclear RNA immunoprecipitated with the antibodies was quantified by qRT-PCR using primers specific for each gene. The qRT-PCR values were calculated from three independent samples. (D) Constructs carrying the p53 3′ UTR (p53) or ARE-replaced p53 3′ UTR (noARE) were transfected in cells. The firefly/Renilla values were as in A. (E) Luciferase assay done as in D using cells treated with control or PARN siRNA. The ratio of the firefly/Renilla values obtained for each construct in PARN knock-down cells relative to control siRNA-treated cells are shown. (F) ARE sequences present in the 3′ UTR are involved in the interaction of PARN with p53 mRNA under nonstress conditions. RIP analysis of samples cells transfected with luciferase constructs carrying either the p53 3′ UTR (p53) or ARE-replaced p53 3′ UTR (noARE) was performed as in C. The ratio of the fold change for p53/no-ARE RNA values obtained for each construct is shown. (G) PARN interacts with the p53 3′ UTR in an ARE-dependent manner. RNA pull-down experiments were performed using biotinylated RNA of the p53 3′ UTR or ARE-replaced p53 3′ UTR and NEs from RKO cells. *An overexposed film is included to show the weak nonspecific interaction of p53 with the 3′ UTR of its own mRNA. A representative pull-down reaction from three independent assays is shown.

3354 | www.pnas.org/cgi/doi/10.1073/pnas.1212533110 Devany et al. Downloaded by guest on September 26, 2021 structure that protects the p53 mRNA from degradation upon The study presented here shows an alternative mechanism to stress by binding to heterogeneous nuclear ribonucleoprotein H/F regulate the expression levels of p53 based on the control of the has been described (21). This structure, which is located down- steady-state levels of p53 mRNA by PARN deadenylase under stream of the 3′ cleavage site, was not included in our luciferase nonstress conditions (Fig. 5A). Supporting this, our previous work construct (Fig. 4A). Interestingly, the 3′ UTR of p53 mRNA also indicates that PARN has a role in decreasing the levels of short- contains ARE that associates with ARE-binding proteins, such as lived mRNAs involved in the control of cell growth, DDR, and wild-type p53-induced gene 1 (22) and human antigen R (23) and differentiation, keeping their expression levels low under non- regulates p53 mRNA steady-state levels. Importantly, the re- stress conditions (5). Under stress conditions, the induction of p53 placement of the ARE sequence from the p53 3′ UTR (noARE expression is associated with a decrease in the levels of total poly construct) significantly increases the firefly/Renilla ratio com- (A) mRNA (24). Because mRNA poly(A) tails are important for pared with the WT p53 3′ UTR construct (Fig. 4D), showing that the regulation of mRNA stability, it is possible that these changes the AREs can decrease mRNA stability and hence expression of of poly(A) mRNA levels might represent another mechanism of the luciferase-p53 3′ UTR construct. Interestingly, the siRNA- p53-mediated control of gene expression. In fact, our studies in- mediated knockdown of PARN only increases the expression dicate that an increase in the expression of p53 inhibits the mRNA ratio of firefly/Renilla luciferase from the constructs carrying the 3′ cleavage step of polyadenylation (4) and induces PARN dead- AREs but not from the constructs without the AREs (Fig. 4E), enylase activity (Fig. 1), suggesting that the p53 associated to the indicating that the AREs in the p53 3′ UTR are necessary for PARN/CstF/BARD1 complex might regulate gene expression by PARN-mediated regulation of p53 expression. Supporting this controlling the steady-state levels of mRNAs (Fig. 5B). Consider- idea, our RIP assays indicate that PARN can form a complex with ing that the p53 pathway is tightly controlled in cells (reviewed in the luciferase mRNA carrying the 3′ UTR of p53 and this is ref. 25), the p53-associated control of mRNA 3′ processing ma- abolished when AREs are replaced by other sequences (Fig. 4F). − − chinery could represent an indirect mechanism to repress target RKO/RKO-E6 and HCT116/HCT116 p53 / cells showed similar gene expression at the posttranscriptional level. The anti- ratios for the expression of firefly/Renilla luciferase (Fig. 4 D proliferative factor BTG2 represents another example of a general and E) and for the RIP assays with different constructs (Fig. 4F), activator of mRNA deadenylation by its direct interaction with the indicating that the PARN-mediated regulation of p53 expression Pop2–Caf1 and Ccr4 deadenylases (26). This model is consistent and PARN binding to AREs are p53-independent (Fig. S6). This with the idea proposed by Singh et al. (10) that the interaction of was confirmed using RNA pull-down assays with in vitro tran- the 3′ processing machinery and factors involved in the DDR/tu- scribed biotinylated RNAs, either with sequences of the p53 3′ mor suppression might result in cell-specific3′ processing profiles. UTR or ARE-replaced p53 3′ UTR, and NEs from RKO cells. Control of deadenylation could represent a mechanism to reg- Our results indicate that the RNA with the p53 3′ UTR pulled ulate gene expression in different cellular conditions, such as de- down PARN from the NEs, and this RNA–PARN interaction was velopment, stress treatment, or different metabolic conditions. lost in the absence of the ARE sequence (Fig. 4G). A weak non- Supporting this idea, recently it has been shown that PARN reg- specific interaction of p53 with the 3′ UTR of its own RNA was ulates the expression of genes involved in mRNA metabolism, detected (Fig. 4G and Fig. S7). Together, these results indicate transcription, and cell motility in mouse myoblasts, resulting in that the AREs in the p53 3′ UTR are important for the PARN- PARN-dependent regulation of cell motility and wound healing in mediated regulation of p53 mRNA steady-state levels and that this those cells (6). Indeed, our Gene Ontology analysis also revealed regulation is p53-independent. significant down-regulation of genes involved in similar pathways

(Table S3), such as structure morphogenesis, cell adhesion, cell BIOCHEMISTRY Discussion migration, and so on. Consistently, our microarray data showed In this study we provide evidence of a unique feedback loop be- a decrease in the abundance of mRNA for several genes involved tween p53 and PARN deadenylase, in which PARN keeps p53 in cell motility, such as adenosine A2b receptor, ankyrin repeat- levels low in nonstress conditions by destabilizing p53 mRNA, and containing domain 54, and collagen alpha-2(I) chain in PARN the UV-induced increase in p53 activates PARN deadenylase knock-down cells. However, the p53 signaling pathway was not regulating gene expression during DDR in a transactivation-in- reported by Lee et al. (6), suggesting cell-specific functions of dependent manner. Several lines of evidence support this model. PARN. Like Lee et al. (6), we also observed a decrease in the First, the C-terminal domain of p53 can activate PARN-dependent steady-state levels of some transcripts by PARN knockdown. deadenylation in vitro and p53 expression levels correlate with However, it is not clear whether this reflects the function of PARN levels of mRNA deadenylation (Fig. 1). Second, our results show per se or is the indirect consequence of PARN’s effect on genes the direct interaction of the C-terminal domain of p53 with the involved in other mRNA metabolic pathways, such as transcrip- C-terminal domain of PARN and the existence of protein com- tion and RNA processing factors. plexes of these factors in cellular NEs (Fig. 2). Third, PARN sig- The characterization of the regulatory elements in the 3′ UTR nificantly affects the cellular expression of genes in the p53 of p53 and the factors involved in this PARN-dependent regula- pathway under nonstress conditions and the stability and poly tory pathway may allow us to better understand the mechanisms (A) length of the p53 mRNA (Fig. 3 A–D). Fourth, PARN that control p53 expression and to find alternative strategies for knockdown and UV treatment induce a similar increase in p53 treating tumorigenesis and metastasis in various cancers. expression and apoptotic responses (Fig. 3 B and E). Finally, PARN regulates p53 expression through the ARE sequence present in the Materials and Methods 3′ UTR of p53 mRNA (Fig. 4). Taken together, our results provide Tissue Culture Methods and DNA Damaging Agents. HeLa, RKO, RKO-E6, − − − − insights into p53 function and the mechanisms behind the regu- HCT116, HCT116 p53 / , MEF, and MEF p53 / cell lines were cultured and lation of mRNA 3′ end processing in different cellular conditions. UV-treated as described (4, 5, 27).

A B Fig. 5. Model for the regulation of expression of nucleus nucleus genes in the p53 pathway by PARN deadenylase- p53 PARN associated p53 in different cellular conditions. (A) CstF BARD1 ? PAS p53 p53 PARN deadenylase decreases the stability of the mRNA ARE AAAAAAAAAA p53 p53 ORF PARN PAS ? p53 mRNA in nonstress conditions. The AREs in the mRNA PARN target mRNAs AAAAA A ′ deadenylation activation A 3 UTR of the p53 mRNA have an important role in p53 p53 this regulatory process. (B) Under DNA damage degradation ? PAS conditions, p53 protein accumulates, allowing its p53 mRNA p53 ORF ARE AAAAA A association to and activation of PARN deadenylase A degradation resulting in the decrease levels of target mRNAs in the p53-dependent DDR pathway.

Devany et al. PNAS | February 26, 2013 | vol. 110 | no. 9 | 3355 Downloaded by guest on September 26, 2021 Preparation of NEs. After UV treatment, NEs were prepared from harvested RIP Assays. IP of nuclear RNA–protein complexes was performed as de- cells essentially as described (4, 5). scribed (28) using antibodies against PARN (H-105) or control rabbit IgG (Sigma). 32 Deadenylation Assays. P-labeled L3(A30) substrates were prepared and analyzed as in ref. 5. Protein concentrations of the extracts were equalized by RT-qPCR Assays. As described before (4, 5), equivalent amounts (2 μg) of Bradford assays (Bio-Rad) before use in deadenylation reactions. purified RNA were used as a template to synthesize cDNA using random hexamer primers, oligo-d(T) primers, and GoScript Reverse Transcriptase Pull-Down Assays. One microgram of His-PARN bound to Ni magnetic beads (Promega). Relative levels were calculated using the ΔCτ method. was incubated with 30 μL of NEs from RKO cells and then analyzed as described (4, 5). His-p53 and His-PARN constructs were previously described (4). Cell Death ELISA Assay. Fragmentation of DNA after induction of apoptosis was determined by photometric enzyme immunoassay (Cell Death Detection IP Analysis. Total protein (100 μg) from the indicated NEs was immunopre- ELISAPLUS; Roche Applied Science) as recommended by the manufacturer. cipitated with the polyclonal antibody against PARN (H-105; Santa Cruz Biotechnology) and p53 (SC-126; Santa Cruz Biotechnology) as described (4, 5). RNA Isolation and qRT-PCR Analysis of mRNA Half-Lives. Control and PARN knockdown RKO cells (see above) were treated with Act-D (8 μg/mL) for 30 PARN- and p53-siRNA Knockdown. The siRNAs specific for human p53, PARN, min before the beginning of the time course. Nuclear RNA was purified at and the control RNA duplex were synthesized by Dharmacon RNA Tech- nologies. siRNA and UV treatments were as described (4, 5). different time points using RNeasy Mini Kit (QIAGEN) according to the manufacturer’s directions. RNA Purification and Microarray Analysis. Nuclear RNA was purified from HeLa cells using the RNeasy kit (Qiagen) following manufacturer’s protocol. The RNA Pull-Down. Biotin-labeled RNAs were in vitro transcribed with the biotin RNA concentrations of the RNA samples obtained under different conditions RNA labeling mix (Roche) and T7 RNA polymerase (Promega) following were equalized. Equivalent amounts of purified RNA were used in micro- manufacturer’s instructions. Biotinylated RNA was incubated with 1 mg of array analysis. The GeneChip Human Gene 1.0 ST (Affymetrix) expression NEs and then analyzed as described (17). array was used. Microarray data were normalized using the Robust Multi- chip Average method. RACE-poly(A) Test Assays. Nuclear RNA from RKO cells treated with PARN/control siRNA was reverse-transcribed using oligo (dT)-anchor primer (5′-GGGGAT- Plasmid Constructs. Luciferase vector pEZX-MT01 with TP53 miTarget micro- CCGCGGTTTTTTTTTT-3′) and GoScript Reverse Transcriptase (Promega). One ′ RNA 3 UTR target clones (product ID HmiT054283) was purchased from microliter of each cDNA was used for PCR amplification by GoTaq PCR mix ′ GeneCopoeia. Mutations in the ARE sequence of p53 3 UTR were intro- (Promega) using p53 3′ UTR-specificprimer[5′-CTGCATTTTCACCCCACCCTTCC-3′ duced with the QuikChange Lightning Site-Directed Mutagenesis Kit (Agilent located 90 bp upstream of the poly(A) site] and oligo(dT)-anchor primer. Technologies) and the following primers 5′-GGGTCAATTTCC GTTCGCGAATTC- TGTTCTGATCTGCTTTTTCTTTGAGACTGGG-3′ and 5′-CCCAGTCT CAAAGAAAA- RNA Electrophoretic Mobility Shift Assay (REMSA) Supershift. NE fractions (10 AGCAGATCAGAACAGAATTCGCGAACGGAAATTGACCC-3′, following the μg) were incubated with 4 μg of the indicated antibody for 1 h on ice before manufacturer’s instructions. Plasmids were sequenced to confirm the presence of the mutation. Twenty-four micrograms of the different luciferase constructs the addition of the radiolabeled RNA as described elsewhere (17). were transfected into cells using Lipofectamine 2000 reagent (Invitrogen). ACKNOWLEDGMENTS. We thank Dr. C. Prives for p53-encoding plasmids, Dr. A. Virtanen for PARN-encoding plasmids, Dr. B. Vogelstein for cell Luciferase Assay. Cells were cotransfected with the luciferase constructs in- lines HCT116 and p53-null HCT116, Mirjana Persaud and Sana Khan dicated and either siRNA-targeting PARN or control siRNA. Forty-eight hours for their technical contribution, and Drs. M. Cevher and A. Saxena for after transfection cells were harvested and a dual luciferase assay was per- advice and discussion. This work is supported by National Institute of formed using a Luc-pair miR Luciferase kit from GeneCopoeia following General Medicine Science Grants SC1GM083806 (to F.E.K.) and GM084089 manufacturer’s instructions. (to B.T.).

1. Tokino T, Nakamura Y (2000) The role of p53-target genes in human cancer. Crit Rev 16. Lai WS, Kennington EA, Blackshear PJ (2003) Tristetraprolin and its family members Oncol Hematol 33(1):1–6. can promote the cell-free deadenylation of AU-rich element-containing mRNAs by 2. Takwi A, Li Y (2009) The p53 pathway encounters the microRNA world. Curr Genomics poly(A) ribonuclease. Mol Cell Biol 23(11):3798–3812. 10(3):194–197. 17. Mazan-Mamczarz K, et al. (2003) RNA-binding protein HuR enhances p53 translation 3. Chang TC, et al. (2007) Transactivation of miR-34a by p53 broadly influences gene in response to ultraviolet light irradiation. Proc Natl Acad Sci USA 100(14):8354–8359. expression and promotes apoptosis. Mol Cell 26(5):745–752. 18. Ford JM, Baron EL, Hanawalt PC (1998) Human fibroblasts expressing the human ′ 4. Nazeer FI, et al. (2011) p53 inhibits mRNA 3 processing through its interaction with papillomavirus E6 gene are deficient in global genomic nucleotide excision repair and – the CstF/BARD1 complex. Oncogene 30(27):3073 3083. sensitive to ultraviolet irradiation. Cancer Res 58(4):599–603. 5. Cevher MA, et al. (2010) Nuclear deadenylation/polyadenylation factors regulate 19. Ford JM, Hanawalt PC (1995) Li-Fraumeni syndrome fibroblasts homozygous for p53 ′ – 3 processing in response to DNA damage. EMBO J 29(10):1674 1687. mutations are deficient in global DNA repair but exhibit normal transcription-coupled 6. Lee JE, et al. (2012) The PARN deadenylase targets a discrete set of mRNAs for decay repair and enhanced UV resistance. Proc Natl Acad Sci USA 92(19):8876–8880. and regulates cell motility in mouse myoblasts. PLoS Genet 8(8):e1002901. 20. Vilborg A, Wilhelm MT, Wiman KG (2010) Regulation of tumor suppressor p53 at the 7. Zhang X, Virtanen A, Kleiman FE (2010) To polyadenylate or to deadenylate: That is RNA level. J Mol Med (Berl) 88(7):645–652. the question. Cell Cycle 9(22):4437–4449. 21. Decorsière A, Cayrel A, Vagner S, Millevoi S (2011) Essential role for the interaction 8. Ji Z, Lee JY, Pan Z, Jiang B, Tian B (2009) Progressive lengthening of 3′ untranslated between hnRNP H/F and a G quadruplex in maintaining p53 pre-mRNA 3′-end pro- regions of mRNAs by alternative polyadenylation during mouse embryonic de- cessing and function during DNA damage. Genes Dev 25(3):220–225. velopment. Proc Natl Acad Sci USA 106(17):7028–7033. 22. Vilborg A, et al. (2009) The p53 target Wig-1 regulates p53 mRNA stability through an 9. Mayr C, Bartel DP (2009) Widespread shortening of 3′UTRs by alternative cleavage AU-rich element. Proc Natl Acad Sci USA 106(37):15756–15761. and polyadenylation activates oncogenes in cancer cells. Cell 138(4):673–684. 23. Zou T, et al. (2006) Polyamine depletion increases cytoplasmic levels of RNA-binding 10. Singh P, et al. (2009) Global changes in processing of mRNA 3′ untranslated regions characterize clinically distinct cancer subtypes. Cancer Res 69(24):9422–9430. protein HuR leading to stabilization of nucleophosmin and p53 mRNAs. J Biol Chem – 11. Cevher MA, Kleiman FE (2010) Connections between 3′-end processing and DNA 281(28):19387 19394. damage response. Wiley Interdiscip Rev RNA 1(1):193–199. 24. Ljungman M, Zhang F, Chen F, Rainbow AJ, McKay BC (1999) Inhibition of RNA – 12. Guhaniyogi J, Brewer G (2001) Regulation of mRNA stability in mammalian cells. Gene polymerase II as a trigger for the p53 response. Oncogene 18(3):583 592. — 265(1-2):11–23. 25. Vousden KH (2006) Outcomes of p53 activation Spoilt for choice. J Cell Sci 119(Pt 13. Lin WJ, Duffy A, Chen CY (2007) Localization of AU-rich element-containing mRNA in 24):5015–5020. cytoplasmic granules containing exosome subunits. J Biol Chem 282(27):19958–19968. 26. Mauxion F, Faux C, Séraphin B (2008) The BTG2 protein is a general activator of mRNA 14. Gherzi R, et al. (2004) A KH domain RNA binding protein, KSRP, promotes ARE- deadenylation. EMBO J 27(7):1039–1048. directed mRNA turnover by recruiting the degradation machinery. Mol Cell 14(5): 27. Sur S, et al. (2009) A panel of isogenic human cancer cells suggests a therapeutic 571–583. approach for cancers with inactivated p53. Proc Natl Acad Sci USA 106(10):3964–3969. 15. Sauer M, et al. (2008) C-terminal diversity within the p53 family accounts for differ- 28. Selth LA, Close P, Svejstrup JQ (2011) Studying RNA-protein interactions in vivo by ences in DNA binding and transcriptional activity. Nucleic Acids Res 36(6):1900–1912. RNA immunoprecipitation. Methods Mol Biol 791:253–264.

3356 | www.pnas.org/cgi/doi/10.1073/pnas.1212533110 Devany et al. Downloaded by guest on September 26, 2021