|Home | About | Journals | Submit | Contact Us | Français|
The p53 tumor suppressor utilizes multiple mechanisms to selectively regulate its myriad target genes, which in turn mediate diverse cellular processes. Here, using conventional and single molecule mRNA analyses, we demonstrate that the nucleoporin Nup98 is required for full expression of p21, a key effector of the p53 pathway, but not several other p53 target genes. Nup98 regulates p21 mRNA levels by a post-transcriptional mechanism in which a complex containing Nup98 and the p21 mRNA 3′-UTR protects p21 mRNA from degradation by the exosome. An in silico approach revealed another p53 target (14-3-3σ) to be similarly regulated by Nup98. The expression of Nup98 is reduced in murine and human hepatocellular carcinomas (HCC) and correlates with p21 expression in HCC patients. Our study elucidates a previously unrecognized function of wild-type Nup98 in regulating select p53 target genes that is distinct from the well-characterized oncogenic properties of Nup98 fusion proteins.
The p53 tumor suppressor functions as a sequence-specific transcriptional regulator for a great variety of target genes. Different subsets of these target genes mediate various cellular outcomes such as cell cycle arrest, senescence, programmed cell death or others (reviewed in (Vousden and Prives, 2009). The underlying molecular processes that dictate p53 target gene preference are complex and still only partially understood. Previously we found that the nuclear transport factor hCAS/Cse1L, binds to a subset of p53 target gene promoters and is required for their full activation upon stress-mediated p53 induction (Tanaka et al., 2007).
An RNAi screen against other nuclear transport factors, nuclear pore components in particular, suggested that the nucleoporin Nup98 may also contribute to target gene selectivity of p53. Nucleoporins (Nups), essential components of the nuclear transport machinery, form the nuclear pore complex (NPC), which is embedded in the nuclear envelope and facilitates nucleocytoplasmic trafficking (D'Angelo and Hetzer, 2008). An increasing body of evidence (largely from studies in yeast) suggests that in addition to their well-characterized role in transport, some Nups are also involved in regulating gene expression (Kohler and Hurt, 2010; Strambio-De-Castillia et al., 2010). In particular, Nup98, a major component of the NPC (Cronshaw et al., 2002) was reported to move from the NPC into the nucleoplasm in a transcription-dependent manner (Griffis et al., 2002) and also to be involved in mRNA export (Enninga et al., 2002; Powers et al., 1997), Further, Drosophila Nup98 is required for activation of genes related to stress response, cell cycle progression, and development (Capelson et al., 2010; Kalverda et al., 2010). Nup98 has also been linked to cancer, yet only within the context of Nup98 oncogenic fusion proteins that occur as a consequence of chromosomal translocations in leukemias (e.g. Nup98-HOXA9 t(7;11)(p15;p15); Moore et al., 2007; Xu and Powers, 2009)). The function of wild-type Nup98 in tumor development is largely unknown.
Since a Nup siRNA screen identified Nup98 as a likely determinant of p53 target gene selectivity (Figure S1A), we transfected HepG2 liver cancer cells (wild-type p53) with Nup98 siRNAs and then treated the cells with Camptothecin (CPT) to activate p53. The induction of p53 target gene mRNAs including p21, PUMA, GADD45, NOXA and ATF3 was measured by quantitative real-time PCR (qRT-PCR) (Figure 1A). p21 mRNA (Figure 1A) or protein (Figure 1B) was unique in being affected by Nup98 ablation. Similar results were obtained in HepG2 cells treated with Daunorubicin (data not shown) as well as in Huh-6 and Sk-Hep1 cells treated with CPT (Figure S1B). Nup98 knockdown also led to decreased p21 (but not PUMA) protein upon treatment with Nutlin, a compound that disrupts the p53-Mdm2 complex. This indicates that the effects of Nup98 depletion were independent of DNA damage signaling (Figure 1C). Further, Nup98 depletion selectively reduced p21 mRNA induction in Hep3B-4Bv cells that harbor a temperature-sensitive p53 (Friedman et al., 1997) at 32°C when p53 is in wild-type conformation (Figure S1C). The basal levels of p21 mRNA in the presence or absence of Nup98 were largely unchanged in three different liver cancer cell lines, suggesting that only induced p21 expression (here in response to up-regulated levels of p53) relies on full expression of Nup98 (Figure S1D). Taken together, our data suggest that Nup98 is required for induction of p21 by p53 in a DNA damage-independent manner.
The predominant biogenesis of Nup98 occurs by autocleavage of a Nup98/Nup96 precursor protein (Fontoura et al., 1999; Rosenblum and Blobel, 1999). Arguing strongly against the possibility that Nup96 is also involved in p21 expression, is that, upon transfection of two Nup98 siRNAs, despite decreased amounts of precursor Nup98/96 mRNA, Nup96 protein levels remained unaltered. (Figure S1E: left panels, top and bottom; see right panel for Nup96 antibody specificity). Having excluded Nup96 as a factor in controlling p21 gene expression, we went on to examine the mode by which Nup98 regulates p21 mRNA accumulation.
To determine the stage of p21 gene expression that is regulated by Nup98 we compared induction of nascent unspliced and mature spliced p21 mRNA in HepG2 cells, using primers that anneal to intronic regions of unspliced p21 mRNA (1st and 2nd intron, respectively) or exon-spanning primers that detect only spliced, mature p21 mRNA (Figure 2A, upper left panel). Interestingly, while induction of nascent p21 mRNA (upon CPT treatment) was not significantly impacted by Nup98 knockdown (Figure 2A, upper right and lower left panels), levels of mature p21 mRNA were significantly decreased upon Nup98 depletion compared to the control siRNA treated cells (Figure 2A, lower right panel).
To confirm that the lowered p21 mRNA levels upon Nup98 knockdown were independent of transcription initiation, and also to exclude a splicing defect, we utilized an H1299 derivative p53-null cell line (H24-p21) in which p21 mRNA is transcribed from a p21 cDNA construct under the control of an artificial “tet-off” promoter (Niculescu et al., 1998). This construct lacks intronic regions, but does contain part of the p21 3’-UTR. In these cells Nup98 knockdown again resulted in impaired accumulation of p21 mRNA (Figure 2B left panel) and protein (Figure 2B right panel). As before, Nup96 protein levels were unaltered by Nup98 siRNAs (Nup98#1 and Nup9898#2) despite a decrease in the Nup98/Nup96 precursor mRNA (Figure S2A). In a related H1299 cell derivative, inducible ectopic PUMA expression was unaltered by Nup98 knockdown (Figure 2C).
Based on these findings we hypothesized that Nup98 stabilizes mature p21 mRNA. To test this, we performed mRNA half-life experiments by blocking transcription with Actinomycin-D in the presence or absence of Nup98 siRNAs. As predicted, p21 mRNA induced either by Nutlin (Figure 2D) or CPT (Figure S2B) treatment in HepG2 cells had a shorter half-life (approximately 45 minutes) when Nup98 was ablated by siRNA (siNup9898#2) compared to the control condition (approximately 90 minutes). Similar findings were obtained with another Nup98 siRNA (Figure S2C). Consistent with the observation that uninduced p21 mRNA levels do not change in response to Nup98 ablation in HepG2 cells, we did not observe differences in basal p21 mRNA half-life under these conditions (Figure S2D).
Taken together, Nup98-mediated regulation of p21 mRNA occurs by a splicing-independent post-transcriptional mechanism that involves stabilization of the mature p21 mRNA transcript.
Our results indicated that Nup98 knockdown is associated with a higher rate of p21 mRNA turnover. Since mRNA degradation can occur at each step from the site of transcription to the site of translation (Houseley and Tollervey, 2009), we sought to determine in which cellular compartment Nup98 affected p21 mRNA stability. Furthermore, since Nup98 is reported to be involved in mRNA export (Enninga et al., 2002) we wanted to address whether or not Nup98 knockdown leads to nuclear p21 mRNA accumulation. We used Single Molecule Sensitive Fluorescence in situ Hybridization (mRNA FISH) staining (Femino et al., 1998), to obtain a detailed and quantitative view of nuclear and cytoplasmic p21 mRNA molecules before and after p53 activation in the presence or absence of Nup98. As expected, Nutlin treatment caused an increase in the overall number of single p21 mRNA transcripts in HepG2 cells (Figure 3A-B and summarized in Table S1) which was significantly decreased upon Nup98 knockdown (Figure 3C-D and Table S1). Nup98 depletion led to a marked reduction in cytoplasmic and more modest diminution of nuclear p21 mRNA molecules in Nutlin treated cells (Figure 3C-D and Table S1). As an additional control, in cells containing p21 siRNA even fewer cytoplasmic p21 mRNA molecules were detected (Figure 3E and Table S1). Although p21 mRNA did not accumulate in the nucleus upon Nup98 depletion (but actually decreased), a role for Nup98 in p21 mRNA export remains plausible (see Discussion section).
mRNA degradation in yeast strains with mutated Nup116 (a homologue of mammalian Nup98) is mediated by the exosome (Das et al., 2003). We therefore determined whether the reduced levels of p21 mRNA could be rescued by co-depletion of Exosc3 (Rrp40), a core component of the exosome. As shown in Figure 3F, co-depletion of Exosc3 with Nup98 in the H24-p21 cell line prevented the decrease of p21 mRNA levels observed after Nup98 knockdown alone. These results implicate a role of Nup98 in stabilizing p21 mRNA transcripts in a manner that precludes their turnover by the exosome.
Nup98 localizes primarily at the NPC, reflected in a punctate nuclear rim staining, and to a minor extent within the nucleus, detected as intranuclear foci previously described as GLFG bodies (Enninga et al., 2002; Griffis et al., 2002; and Figure S3A). We hypothesized that Nup98 may associate with p21 mRNA and thereby affect its stability in the nucleus (and the cytoplasm as well). To address this, RNA-IP experiments were performed in which Nup98 was immunoprecipitated from formaldehyde cross-linked samples of HepG2 cells that were untreated or CPT-treated (Figure 4A) or treated with Nutlin (data not shown). For amplification by qRT-PCR, primer pairs specific for different regions of p21 mRNA or PUMA mRNA (as a negative control) were used to localize the potential regions of interaction with Nup98 (or Nup98-associated protein). Remarkably, Nup98 could be co-immunoprecipitated with p21 mRNA with the most significant interaction being detected in the 3′-UTR. By contrast, an anti-Nup98 RNA-IP revealed almost no interaction within the PUMA 3’-UTR. Nup98 also co-immunoprecipitated with p21 mRNA in H24 p21 tet-off cells, again with the strongest interaction being detected in the 3’-UTR (Figure S3B).
These results were extended by experiments showing that ectopic Nup98 was able to significantly increase the levels of co-expressed full-length p21 cDNA (p21 FL) but not p21 cDNA lacking the 3′-UTR (p21 del 3′-UTR) (Figure 4B; see Figure S3C for corresponding protein levels). Approximately 10-fold more of the transfected p21 FL construct than of the p21 del 3′-UTR construct was required to reach equivalent p21 mRNA levels, suggesting that the p21 3′-UTR per se is a negative regulator of p21 mRNA stability. We also observed that the p21 3’UTR fused to the coding DNA sequence (CDS) of Luciferase (Luc/p21 3’UTR) mediates a similar response to Nup98 (Figure 4C). Finally, Nup98 overexpression led to an increase of a hybrid construct (CDS of PUMA attached to the 3’UTR of p21; Pu/p21) while full length Puma 3’UTR (Pu FL) was unaffected (Figure 4D). Taken together our data show that the p21 3’ UTR is necessary and sufficient for the ability of Nup98 to augment the accumulation of p21 mRNA.
Two domains of Nup98 have been implicated for its role in mRNA export. One is the glycine-leucine-phenylalanine-glycine (GLFG) repeat domain that is important for transcription-dependent mobility of Nup98 and which serves as a docking site for the mRNA export receptor TAP/NXF1(Blevins et al., 2003; Griffis et al., 2004). The other is the Gle-2-binding domain (GBD) mediating the interaction of Nup98 with the mRNA export factor Rae1/Gle2 that also interacts with TAP (Blevins et al., 2003). Different truncated versions of Nup98 were expressed in H1299 cells along with co-transfected p21 (p21 FL). Overexpression of full length Nup98 (construct #1) maximally increased p21 protein expression in a concentration-dependent manner, whereas the C-terminus alone (construct #2; 506–920) only produced a modest increase (Figure 4E). Overexpression of the full length GLFG repeat domain plus the C-terminus (construct #3; 221–920) but lacking the Gle2-binding-domain (GBD) was sufficient to increase p21 protein to similar levels as did wild type Nup98. The C-terminal half of the GLFG repeat domain plus the C-terminus (construct #4; 362–920) was not able to fully stabilize p21. Thus the intact GLFG-repeat but not the GBD of Nup98 is necessary for full stabilization of p21. Nup98-dependent regulation of p21 may require the mobility of Nup98 and/or its interaction with TAP but is independent of its interaction with Rae1/Gle2. In line with this, in our siRNA screen the depletion of Rae1/Gle2 did not lead to reduced p21 mRNA accumulation (Figure S1A).
Since the strongest association of Nup98 with p21 mRNA occurs in its 3′-UTR, we sought to determine whether other mRNAs within the p53 transcriptional regulon might also be regulated post-transcriptionally by Nup98 (Keene, 2007). We adapted a computational method previously used to infer mechanisms of post-transcriptional messenger RNA stability regulation from genome-wide changes in steady-state transcript abundance in yeast (Foat et al., 2005). The REDUCE software suite (see Figure 5A and Experimental Procedures) was used to discover RNA sequence motifs whose occurrence in 3′-UTRs is predictive of the difference in gene expression 72 h after doxorubicin treatment, i.e. between cells in which p53 is active vs. inactive (Chau et al., 2009). This yielded a single, highly significant C-rich motif (Figure 5A), occurrences of which in the 3′-UTR are predictive of mRNA stabilization upon p53 activation (p=value = 5x10−26). To determine whether the motif indicates a potential mRNA or genomic DNA binding element, we analyzed how the correlation with differential gene expression differed when counting motif occurrences in 400 nt windows immediately upstream or downstream of poly-adenylation sites in genomic DNA (Figure 5B). The correlation was significantly higher for the upstream window, suggesting that the C-rich motif represents the specificity of an RNA-binding factor. The C-rich motif is also robust in the sense that using other differential steady-state expression data yields consistent results. First, occurrences in the 3′-UTR correlated significantly with the difference in mRNA expression 72 h vs. 48 h after doxorubicin treatment in p53-active cells (p = 3*10−49) (Chau et al., 2009). Second, they correlate with the difference in mRNA expression 24 h after Nutlin vs. mock treatment (p = 3*10−32) (Goldstein et al.).
The above results suggested that the C-rich motif mediates a post-transcriptional and p53-dependent regulatory signal. We used this motif to rank a list of 120 bona fide p53 target genes (Riley et al., 2008) in terms of the total motif score in their 3′-UTR (Table S2). We found that 14-3-3σ scored the highest and p21 ranked 6th. Indeed, in HepG2 cells Nup98 was required for full 14-3-3σ expression (Figure 5C). Furthermore, as with p21, we were able to co-immunoprecipitate the 14-3-3σ mRNA 3′-UTR with Nup98 (Figure 5D). The finding that the highest-scoring candidate behaved similarly to p21 supports the efficacy of this in silico screen, and increases further the likelihood that Nup98 interaction with 3’UTR regions of select p53 target genes contributes to their stability.
Although p21 is well-known for its ability to regulate the cell cycle, it also functions to inhibit cell death under stress conditions (reviewed in (Abbas and Dutta, 2009). Decreased p21 levels observed after Nup98 knockdown may sensitize cells to drug induced apoptosis. To test this, HepG2 cells were treated with siRNAs specific for Nup98 or p21 and harvested after 24 h treatment with CPT followed by FACS analysis. While there were no major changes in cell cycle distribution, there was an increase in cell death after Nup98 knockdown (Figure 6A). This corresponded well with the results of p21 siRNA-mediated knockdown where we observed an even higher percentage of sub-G1 cells, possibly due to an even greater reduction of p21 protein levels (Figure 6B). Decreased p21 induction associated with Nup98 depletion is very likely to be a contributing factor to the increased cell death that occurs upon Nup98 knockdown.
Cellular senescence is another important p53-mediated and p21-related cellular outcome (Abbas and Dutta, 2009). After 5 days of Nutlin treatment of control siRNA transfected Sk-Hep1 cells, there was a strong increase of senescence-associated beta-galactosidase activity (SA-beta-Gal) (43±15% positive cells) compared to corresponding DMSO treated controls (<0.01% positive cells) which was significantly diminished when Nup98 levels were reduced (12±9% and 16±8%, in cells containing Nup98 siRNA #1 and siRNA98#2, respectively) (Figure 6C). Thus Nup98 offers partial protection from DNA damage mediated apoptosis, but cells require Nup98 to undergo senescence mediated by p53.
Other than the H1299 lung carcinoma-derived cell line engineered to inducibly express ectopic p21 (H24 p21 cells), the only cell lines in which Nup98 reduction repeatedly produced a strong effect on p21 levels were derived from liver cancers (data not shown) suggesting that Nup98 levels may be altered in hepatocellular carcinoma (HCC). We tested this possibility in both a murine model of HCC and in tissues from HCC patients (Figure 7). The multidrug resistance 2 (MDR-2) knock-out mouse is a model for inflammation-associated HCC development and resembles the human clinical progression of the disease (Mauad et al., 1994). We analyzed Nup98 expression in tumorous (T) and non-tumorous (NT) mouse liver tissue by immunohistochemistry (IHC). Strong Nup98 IHC staining was detected at the nuclear membrane and within the nucleus in hepatocytes (black arrows) and in the sinusoidal cells (blue arrows) of NT-liver tissue (Figure 7A lower left panel). Strikingly, while the sinusoidal cells and the infiltrating inflammatory cells (blue arrows) retained strong Nup98 IHC staining in the T-liver tissue (Figure 7A; lower right panel) there was a significant decrease in Nup98 immunoreactivity (p<0.01) of hepatocellular tumor cells (black arrows) compared to NT-tissue (Figure 7A; lower right panel and see semi-quantitative analysis of hepatocellular Nup98 staining intensity in NT-tissue and T-tissue of MDR-2 −/− mice in Figure 7B). Thus, Nup98 is substantially down-regulated in murine HCC compared to adjacent non-tumorous tissue.
To gain insight into a possible role for Nup98 in human hepatocellular carcinoma, we analyzed Nup98 mRNA expression in a cohort of HCC patients comparing HCC tissue samples with NT-liver tissue or with primary hepatocytes (Figure 7C). Nup98 expression was significantly reduced (by a factor of ≥ 2 compared to controls) in 25% of the cases (7/27) and was overexpressed in only 1 out of 27 cases (by ≥ 2-fold compared to control levels). Moreover, Nup98 and p21 expression were associated with a correlation coefficient of r=0.56 (p<0.01) (Figure 7D) suggesting that Nup98-dependent regulation of p21 may also be relevant in vivo. Noteworthy, in a second cohort of HCC patients we found a similar expression pattern with downregulation of Nup98 expression compared to non-tumorous liver in 39% (13/33) and upregulation of Nup98 expression in 12% (4/33) of the cases based on the same thresholds (Figure S4). Again we found a correlation between Nup98 and p21 expression (r=0.40; p<0.05, data not shown). Taken together, these data suggest that Nup98-dependent regulation of p21 is likely to be generalizable to a significant fraction of human HCCs.
An RNAi screen revealed that interference with the NPC causes pleiotropic effects on p53 target gene induction. While ablation of a subset of Nups such as Nup98, TPR (translocated promoter region), and Nup93 selectively reduced p21 mRNA accumulation, Aladin knockdown selectively increased p21 mRNA induction. Despite the limitations of this screen (e.g. differences in knockdown efficiency, secondary structural alterations of the NPC upon decreased expression of some Nups, etc) we can assume that selectively reduced p21 mRNA accumulation upon Nup98 ablation is not a general phenomenon observed by depleting any nuclear pore component. Since components of the NPC modulate the expression of p53 target genes in different directions, stabilization of p21 mRNA by Nup98 is unlikely to be the only mechanism that can be envisioned (others include transcriptional regulation, altered protein import/export, etc). This adds another layer of complexity to p53 target gene selection in general and to p21 expression in particular.
Nup98 prevents p21 mRNA degradation by the exosome. This might be related to a role for Nup98 in p21 mRNA export. In fact, a yeast Nup116 deletion mutant exhibits an mRNA export block and accumulation of nuclear mRNAs (Das et al., 2003). Yet, in our single molecule FISH analysis of p21 mRNA, Nup98 knockdown does not lead to increased accumulation of nuclear p21 mRNA. Thus, if knockdown of human Nup98 causes a p21 mRNA export block, this must be tightly coupled to degradation of p21 mRNA transcripts most likely by the nuclear exosome. Indeed, in the Nup116 delta yeast strain, accumulation of nuclear mRNAs is followed by rapid degradation that can be prevented by disruption of the nuclear exosome (Das et al., 2003). In our system, an analogous immediate degradation of unexported p21 mRNA could account for the observed lowered levels of both nuclear and cytoplasmic p21 mRNA. A second scenario can be imagined in which, in addition to its putative effects on p21 mRNA nuclear stability, Nup98 might independently regulate the accumulation of cytoplasmic p21 mRNA. Given the association of Nup98 with the 3’-UTR of p21 mRNA, it may recruit additional factors that stabilize p21 mRNA in the cytoplasm. In either case the exosome likely recognizes and degrades p21mRNA upon Nup98 depletion as a process of mRNA surveillance either related to impaired export or defects in RNA-protein complex formation in the 3’UTR region. Supporting this hypothesis the exosome has been documented to participate in control mechanisms that remove aberrant RNAs in the nucleus and in the cytoplasm (Houseley et al., 2006). Future studies will hopefully elucidate the particular stability defect and subsequent degradation of p21mRNA that occurs in the absence of Nup98.
Bioinformatics analysis identified a C-rich motif that correlates with the stabilizing effect of Nup98 on p21 mRNA and 14-3-3σ mRNA. It will be important eventually to determine which other p53 targets that have high (or low) scores with the C-rich motif are similarly regulated (or not) by Nup98. Particularly relevant to our studies, RNA binding proteins including different members of the poly-C binding protein family (PCBP1, 2 and 4; the latter being a p53 target gene) are implicated in p21 mRNA destabilization via binding to its 3′-UTR (Scoumanne et al., 2010; Waggoner et al., 2009; Zhu and Chen, 2000). Based on the observation that basal p21 mRNA levels are not significantly affected by Nup98 depletion, perhaps Nup98 competes with a p21 mRNA destabilizing factor that is induced by p53, such as MCG-10 (=poly C binding protein 4) (Zhu and Chen, 2000). Alternatively, p21 protein induction may repress expression of a stabilizing factor of its own mRNA. Under basal conditions this factor could compensate for the loss of Nup98. However, under induced conditions (e.g. CPT), this factor would be unavailable to stabilize p21 mRNA due to its depletion via p21-dependent mechanisms. Finally, posttranslational modifications of Nup98 upon p53/p21 activation may be required for the stabilizing effect on the p21mRNA transcript. Exploring aspects of these proposed scenarios will be of considerable interest for further studies.
Data from the murine HCC model, combined with data from both cohorts of HCC patients (Figure 7 and S4) suggest that it may be advantageous for tumors to lose Nup98 expression i.e. that Nup98 acts under certain circumstances as a tumor suppressor in hepatocarcinogenesis. In line with this is the observation that p53-mediated induction of senescence, an important liver tumor-suppressive response (Xue et al., 2007), is impaired upon Nup98 depletion (Figure 6). However, more studies including mouse models with a liver-specific conditional Nup98 knock-out are required to draw firm conclusions about tumor-suppressive functions of Nup98 that might be context- and stress-specific. In fact, since up to 12% of HCC patients showed overexpression of Nup98 it is quite possible that Nup98 may play a dual role in cancer formation as does p21 or, even more intriguingly, through p21. The latter could be supported by the correlation of Nup98 and p21 expression in both patient data sets. We expect that future studies will elucidate the specific circumstances under which Nup98 may act as barrier to cancer formation.
HepG2 cells were maintained in RPMI medium supplemented with 10% fetal bovine serum (FBS). Hep3B cells and the derivatives Hep3B-4Bv (expressing the temperature sensitive p53val135 mutation) were kindly provided by M. Oren (Friedman et al., 1997) and maintained in MEM with 10% FBS (and 1 µg/mL puromycin for Hep3B-4Bv). H1299 cells and their derivative ‘tet-off’ H24-p21 inducible cell line were previously described (Niculescu et al., 1998) and maintained in DME medium with 10% FBS and tetracycline (5 µg/ml). p21 expression was induced by removing tetracycline 48 h before harvesting the cells. H1299 PUMA-inducible cells (‘tet-on’) were generously provided by K. Vousden (Nakano and Vousden, 2001) and grown in DME with 10% FBS. PUMA expression was induced by adding Doxycycline (2 µg/mL) 24 h before harvesting. Camptothecin (CPT), Nutlin-3a, and Actinomycin D (Sigma) were used as indicated.
siRNA transfection assays were performed with DharmaFECT 1 (Dharmacon) or Oligofectamin (Invitrogen). siRNA duplexes, designed and synthesized by Qiagen were used individually at 50 nM (sequences are listed in Table S3). The Qiagen All-Stars duplex was used as the negative control siRNA for all knockdown experiments besides the p21 mRNA half-life experiment, where Luc siRNA was used. HA-Nup98 was generously provided by Dr. N.R. Yaseen (Washington University School of Medicine, USA). GFP-Nup98 full length, GFP-Nup98 C-terminus (506–920), GFP-Nup98 GLFG+C-terminus (221–920), GFP-Nup98 GLFG(C)+C-terminus (362–920) were as described (Griffis et al., 2004).. HA-Nup96 was kindly provided by Dr. B. Fontoura (UT Southwestern Medical Center, USA). cDNAs encoding p21 full length (FL), PUMA full length (Pu FL) including the respective 5’ and 3’ UTRs and p21 lacking the 3’-UTR were amplified by PCR with a Platinum Pfx DNA polymerase (Invitrogen) using cDNA from HepG2 cells as template. The resulting p21 FL and p21 delta 3’-UTR cDNAs were then cloned into a pcDNA3.1 vector (Invitrogen). The Pu/p21 hybrid construct was generated by amplifying the PUMA CDS (including the 5’UTR) and p21 3’UTR from the respective FL constructs, that were sequentially cloned into a pcDNA3.1 vector (Invitrogen). The Luc/p21 3’UTR and Luc control were as described (Giles et al., 2003). Transient plasmid transfections were done with Lipofectamin2000 (Invitrogen) according to the manufacturer’s protocol.
RNA was isolated from cells with the Qiagen RNeasy Mini kit, and was DNAse treated and reverse transcribed using the Qiagen Quantitect reverse transcription kit. Samples were analyzed on an ABI 7300 real-time PCR instrument with default settings using the ΔΔCT method. Expression levels were normalized to those of RPL32 or 18s RNA (patient samples). Primers used were designed with Primer Express (Applied Biosystems) and sequences are listed in Table S4.
Cells were harvested and subjected to immunoblotting as previously described (Ohkubo et al., 2006). Commercial antibodies used in the study are listed in table S5. Nup98 rabbit polyclonal antibody was kindly provided by Dr. G. Blobel (Rockefeller University, USA) and Nup96 rabbit polyclonal antibody was a kind gift from Dr. Joseph Glavy (Stevens Institute of Technology, USA). Signal detection was performed by the using the Odyssey Infrared Imaging System (LI-COR).
HepG2 and H24 p21 cells were grown on coverslips and treated with drugs as indicated. Immunofluorescent staining was performed as previously described (Karni-Schmidt et al., 2007). Nup98 (C-5; Santa-Cruz) was used as primary antibody followed by Alexa Fluor 488 rabbit anti-mouse secondary antibody (Molecular Probes, Eugene, OR). Images were analyzed by confocal laser scanning microscopy (Olympus model 1381) using Fluoview software (Canter Valley, PA).
Cells were processed with slight modifications as described before (Ohkubo et al., 2006). Cell cycle stages were analyzed by using the ModFit LT version 3.0 program and subG1 fractions were analyzed using Cell Quest Pro.
To detect p21 mRNA 14 oligodeoxynucleotide probes were designed, synthesized and labeled as previously described (Femino et al., 1998). Each probe had a length of 50 nt and contained five amino-modified nucleotides (amino-allyl T). The free amines were chemically coupled to cyanine 5 fluorescent dye after synthesis. The sequences of probes used to detect p21 mRNA are available upon request. mRNA-FISH was performed based on a published protocol (http://www.singerlab.org/protocols). For further information see the Supplemental Experimental Procedures.
MDR-2 −/− animals were described previously (Mauad et al., 1994). Liver tissue of 15 month old male MDR-2 −/− mice was immunostained based on a published protocol (Singer et al., 2007) by using a polyclonal Nup98 antibody. For additional information see the Supplemental Experimental Procedures.
Total RNA isolation of HCC samples (n=27 and n=33) and healthy liver tissues (n=2 and n=5) of both cohorts for semiquantitative real-time PCR was performed using the NucleoSpin RNA II kit according to the manufacturers' protocol (Macherey-Nagel, Duren, Germany). The study was approved by the institutional ethics committee of the Medical Faculty at Heidelberg University.
We used the MatrixREDUCE program from the REDUCE software suite (http://bussemakerlab.org/software/REDUCE) to perform a genomewide fit of a position-specific affinity matrix (PSAM) to the logarithm of mRNA expression fold-differences between wild-type cells and p53 knock-down cells (after 72 h). For the sequence associated with each log-ratio, we used the March 2006 version of human 3′-UTRs from the UCSC genome website (http://genome.ucsc.edu). Genes with overlapping 3′-UTRs were filtered out. In the PSAM search, we considered motif lengths from 1 to 10 and used a p-value threshold of 0.001. To rank the p53 target genes, we calculated the total affinity of the 3′-UTR using the AffinityProfile program from the REDUCE suite to score the enrichment of each 3′-UTR for the discovered C-rich motif. Total relative affinities were calculated as the sum of PSAM-based predicted affinities in a sliding window over the length of the 3′-UTR. We controlled for variation in 3′-UTR length by dividing the total affinity by the UTR length and used the resulting affinity density to rank the list of published p53 targets (Riley et al., 2008).
We thank Ella Freulich, Melissa Lopez-Jones, Juliane Winkler, Helen Mueller, Michaela Bissinger, Andrea Hain and Eva Eiteneuer for excellent technical assistance, Oleg Laptenko, Will Freed-Pastor and other members of the Prives Laboratory for helpful suggestions, Gabor Halasz for compiling a non-redundant set of 3′-UTR sequences, and Xiang-Jun Lu for implementing and maintaining the REDUCE software suite. S.S. was supported by a fellowship from the German Research Foundation Si-1487/1-1 and SFB/TRR77-C5. J.H. and T.P. were supported by SFB/TRR-A7, K.B. by SFB/TRR-B7, T.L. and O.N. by SFB/TRR-B5. This research was supported by NIH grant CA77742 to C.P. DG and the imaging services were partially supported by NIH GM86217 to RHS. HJB was supported by NIH grants R01HG003008 and U54CA121852, as well as a John Simon Guggenheim Foundation Fellowship. MF was supported by NIH training grant T32GM082797.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.