Article15 July 2004free access Concurrent versus individual binding of HuR and AUF1 to common labile target mRNAs Ashish Lal Ashish Lal Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Krystyna Mazan-Mamczarz Krystyna Mazan-Mamczarz Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Tomoko Kawai Tomoko Kawai Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Xiaoling Yang Xiaoling Yang Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Jennifer L Martindale Jennifer L Martindale Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Myriam Gorospe Corresponding Author Myriam Gorospe Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Ashish Lal Ashish Lal Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Krystyna Mazan-Mamczarz Krystyna Mazan-Mamczarz Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Tomoko Kawai Tomoko Kawai Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Xiaoling Yang Xiaoling Yang Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Jennifer L Martindale Jennifer L Martindale Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Myriam Gorospe Corresponding Author Myriam Gorospe Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA Search for more papers by this author Author Information Ashish Lal1, Krystyna Mazan-Mamczarz1, Tomoko Kawai1, Xiaoling Yang1, Jennifer L Martindale1 and Myriam Gorospe 1 1Laboratory of Cellular and Molecular Biology, National Institute on Aging-IRP, National Institutes of Health, Baltimore, MD, USA *Corresponding author. Box 12, LCMB, NIA-IRP, NIH, 5600 Nathan Shock Drive, Baltimore, MD 21224, USA. Tel.: +1 410 558 8443; Fax: +1 410 558 8386; E-mail: [email protected] The EMBO Journal (2004)23:3092-3102https://doi.org/10.1038/sj.emboj.7600305 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info RNA-binding proteins HuR and AUF1 bind to many common AU-rich target mRNAs and exert opposing influence on target mRNA stability, but the functional interactions between HuR and AUF1 have not been systematically studied. Here, using common target RNAs encoding p21 and cyclin D1, we provide evidence that HuR and AUF1 can bind target transcripts on both distinct, nonoverlapping sites, and on common sites in a competitive fashion. In the nucleus, both proteins were found together within stable ribonucleoprotein complexes; in the cytoplasm, HuR and AUF1 were found to bind to target mRNAs individually, HuR colocalizing with the translational apparatus and AUF1 with the exosome. Our results indicate that the composition and fate (stability, translation) of HuR- and/or AUF1-containing ribonucleoprotein complexes depend on the target mRNA of interest, RNA-binding protein abundance, stress condition, and subcellular compartment. Introduction Post-transcriptional processes such as RNA splicing, and mRNA export, stability, and translation are emerging as critical mechanisms of gene regulation in mammalian cells. These regulatory programs are primarily governed by RNA-binding proteins (RBPs) that associate with pre-mRNAs and mRNAs and ensure their proper processing (splicing, 5′ end and 3′ end modifications, export) as well as their subcytoplasmic transit, half-life, and translation rate. Specialized RBPs that bind specific mRNA subsets have received increasing attention in recent years, given their influence on the expression of genes that are essential to the stress response, cell division, immune function, and tumorigenesis. Many such RBPs have been described that selectively recognize and bind to mRNAs bearing U- or AU-rich sequences (collectively termed AU-rich elements (AREs)), and modulate their translation and/or stability (Ross, 1995). ARE-binding proteins include BRF1, AUF1 (hnRNP D), tristetraprolin (TTP), TIAR, TIA-1, KSRP, and the Hu proteins HuB, HuC, HuD, and HuR (Zhang et al, 1993; Antic and Keene, 1997; Min et al, 1997; Carballo et al, 1998; Gueydan et al, 1999; Stoecklin et al, 2002). The influence of RBPs on mRNA turnover has been studied most extensively for two ubiquitous proteins: HuR, which promotes mRNA stabilization, and AUF1, which enhances mRNA decay (Loflin et al, 1999; Brennan and Steitz, 2001). HuR binds with high affinity and specificity to target mRNAs bearing AREs and modifies their expression by either enhancing their stability, altering their translation, or performing both functions (Brennan and Steitz, 2001; Kullmann et al, 2002; Mazan-Mamczarz et al, 2003). HuR is predominantly nuclear (>90%), but a variety of stimuli can trigger its translocation to the cytoplasm (Keene, 1999; Wang et al, 2000a; Gallouzi and Steitz, 2001). While the hypothesis that HuR exports target mRNAs to the cytoplasm awaits experimental demonstration, HuR's influence on target mRNA stabilization and translation is robustly linked to its cytoplasmic presence. Through its post-transcriptional effects on target mRNAs (such as those encoding cyclin A, cyclin B1, c-fos, VEGF, TNF-α, β-catenin, c-myc, cyclooxygenase-2, myogenin, MyoD, GM-CSF, interleukins, p21, p27, p53, and hsp70), cytoplasmic HuR is emerging as a major regulator of various cellular responses, including cell division, carcinogenesis, muscle cell differentiation, replicative senescence, immune cell activation, and the stress response (Atasoy et al, 1998; Wang et al, 2000b, 2001; Kullmann et al, 2002; Figueroa et al, 2003). An ARE-binding protein causing transcript destabilization, AUF1, is expressed in four isoforms (p37, p40, p42, and p45) arising through alternative splicing of a common pre-mRNA (Wilson and Brewer, 1999). Some differences in the activity of the various AUF1 isoforms have been documented, but all isoforms appear to enhance target mRNA decay, a process that is closely linked to the ubiquitination and targeting of AUF1 to the proteasome (Laroia et al, 1999; Loflin et al, 1999). Like HuR, AUF1 target mRNAs also encode mitogenic, immune response, cancer-associated, stress response, and cell cycle regulatory proteins (such as c-fos, c-jun, c-myc, egr-1, interleukins, p21, hsp70, MnSOD, catalase, cyclin D1, and cdc25) (Ross, 1995; Bhattacharya et al, 1999; Lin et al, 2000). Given that HuR and AUF1 overlap in their tissue distribution, preferentially localize in the nucleus, and influence the expression of many common target mRNAs, functional links between the two RBP families have been postulated (Gao et al, 1994; Bhattacharya et al, 1999; Chen et al, 2002; Lu and Schneider, 2004). According to a widely established view, HuR/ELAV proteins can compete with AUF1 (Park et al, 2000; Cok et al, 2004) or with other decay-promoting RBPs (Ming et al, 2001; Stoecklin et al, 2002) for binding to the same ARE on specific target mRNAs, but other regulatory schemes have not been formally tested. Moreover, no studies have directly assessed the endogenous association of HuR and AUF1 with target mRNAs. In the present work, we provide evidence that endogenous HuR and AUF1 are capable of associating with the same target RNAs on nonoverlapping sites, both in intact cells and in vitro. We propose that whether HuR and AUF1 bind concurrently or competitively to a given mRNA is critically influenced by several factors, including the target RNA sequence, the abundance of each RBP, the stress condition of the cell, and the subcellular compartment in which the ribonucleoprotein (RNP) complex is investigated. Results Unique and shared mRNA targets for HuR and AUF1 In order to investigate functional links between RBPs HuR and AUF1, we set out to identify collections of endogenous mRNAs to which each protein associated in human cervical carcinoma HeLa cells. Immunoprecipitation (IP) reactions were performed to isolate mRNA subsets bound to HuR and AUF1 using specific antibodies. The nonspecific association of mRNAs with IP reagents was determined by parallel incubations with IgG1. The identification of mRNAs in each IP material was achieved by reverse transcription (RT) followed by cDNA array hybridization (Tenenbaum et al, 2002; López de Silanes et al, 2004). Representative array fields illustrate the relative abundance and specificity of signals from HuR IPs, AUF1 IPs, as well as control IgG1 IPs (Figure 1A). As shown (Figure 1B), 201 HuR-specific target transcripts, 194 AUF1-specific target transcripts, and 267 transcripts shared by HuR and AUF1 were identified on this array. The association of transcripts with either HuR or AUF1 was deemed specific when Z ratios were ⩾1.00 in comparisons of signals in HuR IPs versus IgG1 IPs, or signals in AUF1 IPs versus IgG1 IPs, respectively. See http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE1361 for complete array results. In each mRNA subset, AREs of either Class I or III were found in ∼80% of the transcripts tested. Figure 1.Identification of HuR and AUF1 target mRNAs. (A) RNAs bound to either HuR or AUF1 in HeLa whole-cell lysates were isolated by IP assay using the corresponding antibodies, and the reverse-transcribed radiolabeled products were used for cDNA array hybridization (Materials and methods). Control IP reactions were performed using IgG1. Representative fields of each array are shown (black arrows, specific signals enriched in samples obtained by either HuR IP or AUF1 IP; white arrows, signals enriched in both HuR and AUF1 IP materials). (B) Partial list of genes encoding transcripts that were significantly enriched in the HuR IP only (left), in the AUF1 IP only (center), and in both the HuR and AUF1 IPs (right, left value corresponds to Z ratio for HuR, right value for AUF1). Parentheses, Z ratios. Complete array data are available (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE1361). (C) For validation of HuR and AUF1 target mRNAs, HeLa whole-cell lysates were prepared and endogenous target transcripts were detected by RT–PCR assay of the IP material. PCR products were visualized on 1% agarose gels. Amplification of housekeeping transcripts encoding GAPDH and SDHA, bound at low levels with the IP material, showed equal loading of IP samples. Download figure Download PowerPoint The validity of the target mRNA identification scheme was tested experimentally by monitoring the presence of several randomly chosen target transcripts in each category (unique HuR targets, unique AUF1 targets, and shared HuR and AUF1 targets) by IP assay followed by RT–PCR amplification (IP+RT–PCR) using sequence-specific primers. As shown, HuR targets were effectively amplified only from HuR IPs, AUF1 targets only from AUF1 IPs, while targets of both proteins were amplified from both IP groups (Figure 1C). Typically, IgG1 IPs showed undetectable or low-level amplification signals in these PCR reactions, performed at ∼25 cycles; control mRNAs encoding GAPDH and SDHA, which were not targets of either RBP, served to monitor background binding of mRNAs to the IP material. According to the Z ratio cutoff of ⩾1.00 chosen for this analysis, transcript enrichments were calculated for several sample mRNAs by quantitative real-time RT–PCR: enrichment was ∼6-fold for a Z ratio of 1.00 (e.g., TP53 mRNA in AUF1 IPs), ∼60-fold for a Z ratio of 1.90 (e.g., CCND1 mRNA in HuR IPs), and ∼1000-fold for a Z ratio of 4.13 (e.g., PTMA mRNA in HuR IPs) (data not shown). This analysis revealed that 42% of AUF1-bound mRNAs were uniquely associated with AUF1 and not with HuR, while 43% of HuR-bound mRNAs were uniquely associated with HuR and not with AUF1. Conversely, 58% of AUF1 target mRNAs were also targets of HuR; a comparable number of HuR targets (57%) were found to be targets of AUF1 also, revealing an extensive and previously unrecognized collection of mRNAs that were common putative targets of both proteins. RNA-dependent association of HuR and AUF1 The discovery that a large proportion of HuR target mRNAs were also targets of AUF1 (and vice versa) led us to formally examine the possibility that both HuR and AUF1 could simultaneously bind to a given RNA molecule. To begin to address this hypothesis, we performed IP reactions to ascertain the joint presence of both proteins in the same RNP complex. As shown in Figure 2A, IP reactions using an anti-AUF1 antibody yielded abundant quantities of HuR (Ctrl., top), while a reciprocal IP assay using an anti-HuR antibody was capable of yielding AUF1 (Ctrl., middle), albeit less efficiently, possibly due to a decrease in AUF1 binding caused by the HuR antibody itself, or to other as yet unidentified reasons. A control antibody (IgG1) failed to show either protein in the IP material. In additional control groups, IP reactions were supplemented with heparin, a polyanion used extensively as a competitor of RNA to dissociate complexes formed by weak electrostatic interaction, and hence possibly in a nonspecific fashion (Piñol-Roma et al, 1988). The presence of heparin (+Heparin) failed to disrupt the association between HuR and AUF1, supporting the notion that the AUF1–HuR presence in common RNP complexes was specific. Moreover, a brief incubation with RNase A (+RNase) caused HuR to dissociate from AUF1, indicating that HuR and AUF1 likely associated through their joint binding to common target RNAs or RNP complexes, and did not form direct protein–protein complexes. Evidence that these RNP interactions were also likely to occur in intact cells was obtained by using either formaldehyde or irradiation with short-wavelength ultraviolet light (UVC), each capable of crosslinking RNP complexes in living cells. The analysis of lysates prepared after exposure of cells to crosslinking agents (Figure 2A, bottom) further supported an association between HuR and AUF1 in intact cells. Additional control IP assays (Figure 2B) revealed that these RNP complexes also encompassed other general RBPs involved in RNA processing, such as hnRNP A1 and hnRNP C1/C2 (Dreyfuss et al, 1993). Together, these data supported the view that HuR and AUF1 do not associate through protein–protein interactions, but can simultaneously associate in common RNA-containing complexes. Figure 2.Joint presence of HuR and AUF1 within the same RNP complexes. IP assays were carried out using whole-cell lysate from HeLa cells and either anti-HuR antibody, anti-AUF1 antibody, or IgG1. IP reactions were performed without further treatment (Ctrl.), or in the presence of either heparin (+Heparin) or RNases (+RNase). (A) The presence of HuR (top) and AUF1 (four isoforms, middle) in the IP materials was monitored by Western blotting. Lys., 10 μg of whole-cell lysate, included as Western blotting control; HC, heavy chain; LC, light chain. Bottom: RNPs were crosslinked in intact cells by using either UVC irradiation or formaldehyde (Supplementary material), then subjected to IP and Western blotting. (B) The levels of hnRNP A1 and hnRNP C1/C2 in the IP materials were assessed by Western blotting. Download figure Download PowerPoint Distinct abundance of HuR and AUF1 in different cellular compartments The association of HuR and AUF1 was further investigated in different subcellular fractions prepared from HeLa cells. As shown in Figure 3A, HuR and AUF1 associated extensively on common RNP complexes in the nucleus (note Western blotting signals in HuR and AUF1 IP materials, Nuc. lanes). By contrast, cytoplasmic HuR (typically 5–10% of cellular HuR) associated with AUF1 to a much lesser extent (note Western blotting signals on HuR and AUF1 IP materials in the Cyto. lanes). Accordingly, HuR and AUF1 appeared much less likely to bind to the same RNAs in the cytoplasm than in the nucleus, prompting further examination of the subcytoplasmic distribution of the two proteins. In keeping with earlier reports, HuR was found to colocalize primarily with low- and high-molecular-weight (LMW and HMW, respectively) polysome fractions, prepared by centrifugation of the cytoplasmic material through sucrose gradients (Figure 3B) (Antic and Keene, 1997; Gallouzi et al, 2000). By contrast, AUF1 was found to be largely restricted to the fractions devoid of all ribosomal components (unbound (Unb.), lanes 1 and 2), where cytosolic RBPs TTP, TIAR, and TIA-1, as well as β-tubulin, were also detected. Together, these data support the view that nuclear AUF1 and HuR associate through their joint binding to common target RNAs and/or RNPs, but this association is disrupted in the cytoplasm, where AUF1 is found in the cytosolic fraction and HuR with the polysome-bound fraction. Figure 3.Subcellular distribution of HuR and AUF1. (A) Left: the relative abundance of HuR and AUF1 in RNP complexes obtained through IP (described in the legend of Figure 2) was assessed using either cytoplasmic (50 μg, Cyto.) or nuclear (10 μg, Nuc.) lysates. The relative abundance of HuR, AUF1, β-tubulin (a cytoplasm-specific protein), and TFIID (a nucleus-specific protein) was assessed by Western blotting using 10 μg of either cytoplasmic or nuclear lysates. (B) Representative polysome distribution profile (top left) and pattern of ethidium bromide-stained RNA (top right) from sucrose gradients (Materials and methods). From left to right: fractions lacked ribosomes or ribosome subunits (Unb.), contained ribosome subunits or single ribosomes (monosomes, Mono.), and spanned low-molecular-weight (LMW) and high-molecular-weight (HMW) polysomes. The levels of HuR, AUF1, hnRNP A1, TTP, TIAR, and β-tubulin were monitored by Western blotting using equal volumes of lysates from each fraction. Download figure Download PowerPoint Binding of HuR and AUF1 to endogenous and synthetic mRNAs encoding p21 and cyclin D1 To obtain additional support for the joint RNA-binding model emerging from these findings, we focused on the analysis of specific mRNAs that were shared targets of both HuR and AUF1: the cyclin D1 mRNA (CCND1; Figure 1B; Lin et al, 2000) and the p21 mRNA (Joseph et al, 1998). The endogenous association of HuR and AUF1 with these two mRNAs was evidenced by IP assays using anti-HuR and anti-AUF1 antibodies followed by measurement of p21 and cyclin D1 mRNAs by RT–PCR. As indicated, both cyclin D1 and p21 mRNAs were found to be targets of HuR and AUF1; very little or no amplification was seen in the IgG1 IP lanes, and control GAPDH mRNA was only found as low-level contaminating material present in all of the IP reactions (Figure 4A). PCR amplification was performed after RT reactions using oligo-d(T), indicating that HuR and AUF1 bound polyadenylated cyclin D1 and p21 mRNAs (Figure 4A, mRNA). Importantly, however, the pre-mRNA forms (and possibly also excised introns) of these transcripts were also specifically bound by HuR and AUF1, as evidenced by the amplification of intron sequences of both transcripts in HuR and AUF1 IPs (in these RT reactions, random hexamers were used instead of oligo-d(T)) (Figure 4A, pre-mRNA). In all cases, samples were digested with DNase before RT and ‘no RT’ control reactions were included to ensure that no genomic sequences were amplified (Figure 4A). Figure 4.Binding of HuR and AUF1 to distinct sites of the p21 and cyclin D1 mRNAs. (A) Following IP reactions using either whole-cell (1.5 mg, left), cytoplasmic (400 μg, Cyto., center), or nuclear (500 μg, Nuc., right) lysates and anti-HuR, anti-AUF1, or IgG1 antibodies, the binding of endogenous HuR and AUF1 to endogenous target mRNAs (mRNA) was detected by RT–PCR as explained above (Figure 1C); pre-mRNAs (pre-mRNA) were detected by IP of nuclear lysate followed by RT using random hexamers and PCR amplification using primers specific to intron sequences of the p21 and cyclin D1 genes; no amplification was seen in ‘No RT’ control samples. (B) Pull-down assays to assess the ability of endogenous HuR and AUF1 to bind to biotinylated transcripts spanning the p21 and cyclin D1 mRNAs. The indicated biotinylated transcripts (1 μg each) were incubated with 40 μg of HeLa whole-cell lysate, whereupon their association with HuR or AUF1 was detected by Western blotting. Lys., 5 μg of whole-cell lysate; CR, coding region. Blackened box, CR; shaded, AREs. Bottom: schematic of proposed binding regions for HuR and AUF1. Download figure Download PowerPoint While the aforementioned results suggested that HuR and AUF1 were each able to form complexes with endogenous p21 and cyclin D1 mRNAs, it was important to assess whether the two proteins were capable of binding to each target on distinct, nonoverlapping sites of the same mRNAs, thereby permitting, at least theoretically, the joint binding of both proteins on each target (as we propose it occurs in the nucleus). Such in vitro analysis was performed by testing the ability of biotinylated transcripts encompassing various regions of the p21 and cyclin D1 mRNAs to associate with endogenous HuR or AUF1 (Figure 4B). Here, complex formation was visualized by ‘pull-down’ of the RNP associations using streptavidin-coated beads and detection of RBPs on Western blots. As shown, various 3′ untranslated region (3′UTR) partial transcripts, but not coding region (CR) transcripts, were capable of pulling down both HuR and AUF1. Importantly, binding was found to occur at multiple discrete sites: for the p21 mRNA, HuR preferentially bound transcript 3′A, proximal to the CR, within an AU-rich stretch (shaded), while AUF1 displayed a preference for the distal region (3′C). Regarding the cyclin D1 3′UTR, HuR bound fragment 3′C, whereas AUF1 bound the proximal fragment 3′A but not the AU-rich stretch, as previously reported (Lin et al, 2000), and also bound the distal fragment 3′D (Figure 4B). The finding that AUF1 and HuR can bind to the same target RNAs on different, nonoverlapping regions supports the notion that the two proteins could, in principle, bind target transcripts simultaneously. Exposure to UVC has opposing effects on the post-transcriptional regulation of p21 and cyclin D1 Given the ability of p21 and cyclin D1 mRNAs to associate with both HuR and AUF1, two RBPs exerting opposite effects on the stability of target transcripts, we sought to test if the steady-state levels of the two mRNAs were regulated in the same fashion in response to a given stimulus. UVC irradiation increased HuR–p21 mRNA associations in the cytoplasm, and consequently increased p21 mRNA stability and abundance in colon cancer cells (Wang et al, 2000a). In HeLa cells, UVC irradiation (15 J/m2) also induced p21 mRNA levels, and suppressed cyclin D1 mRNA levels (Figure 5A). Likewise, UVC elevated HuR cytoplasmic abundance, but the cytoplasmic association of HuR with AUF1 did not increase proportionately (Figure 5B). Interestingly, the binding of HuR and AUF1 to target mRNAs was influenced in opposite directions by UVC irradiation, as determined by IP+RT–(real-time)PCR assays: after UVC, HuR–p21 mRNA associations increased and HuR–cyclin D1 mRNA associations decreased, whereas AUF1–p21 mRNA associations were reduced and AUF1–cyclin D1 mRNA complexes were markedly elevated (Figure 5C). Figure 5.Effect of UVC irradiation on the expression of p21 and cyclin D1 mRNAs. At 5 h after exposure of HeLa cells to 15 J/m2 UVC, p21 and cyclin D1 mRNA abundance was assessed by Northern blotting (20 μg total RNA per lane) (A), the presence of HuR and AUF1 on common RNP complexes was monitored by IP and Western blot assays using cytoplasmic lysate, as described above (Figure 2) (B), and the presence of endogenous p21 and cyclin D1 mRNAs with each protein was assayed in the IP material from cytoplasmic lysates by real-time RT–PCR (C); fold differences in abundance were calculated after estimating the CT values (representing the number of PCR cycles required to reach a threshold set arbitrarily at 0.8) for each amplification curve (Materials and methods). Low-level amplification of GAPDH and SDHA served to monitor equal addition of RNA from IP materials. (D) Cytoplasmic lysates from either untreated (Ctrl.) or UVC-treated HeLa cultures were fractionated through sucrose gradients, whereupon aliquots from each fraction were used for Western blotting to detect HuR, AUF1, the exosome protein hRrp4p, and the control protein β-tubulin. (E) The relative presence of p21 and cyclin D1 mRNAs in each fraction from the sucrose gradient was determined by Northern blot analysis. (F) The relative abundance of p21 and cyclin D1 mRNAs bound to either HuR or AUF1 in pooled unbound fractions (1 and 2, Unb.) or pooled polysomal fractions (6–10, Polysome) from either untreated or UVC-treated cells was determined by IP+RT–(real-time)PCR. Download figure Download PowerPoint UVC treatment reduced the polysomal content of the cell, but it actually caused HuR to increase its presence in the polysome-bound material (Figure 5D). By contrast, AUF1 remained almost exclusively localized in fractions 1 and 2, where other cytosolic components were identified, including the exosome protein hRrp4p. Northern blotting confirmed that p21 mRNA abundance increased in LMW and HMW fractions after UVC, while cyclin D1 mRNA levels were overall lower (Figure 5E). Experimental evidence for UVC-triggered changes in the composition of HuR–RNA complexes was obtained by IP+RT–(real-time)PCR analysis of pooled polysome fractions 6–10. Following UVC treatment, HuR–p21 mRNA complexes were found to be ∼4-fold more abundant in the polysomal fractions, while HuR–cyclin D1 mRNA complexes were ∼3.5-fold less abundant (Figure 5F). Testing of the unbound fractions (1 and 2) by IP+RT–(real-time)PCR using anti-HuR antibodies failed to amplify p21 or cyclin D1 products, likely due to the low abundance of HuR in these fractions (Figure 5F). While we cannot exclude the possibility that AUF1–RNA associations were disrupted by the fractionation process, no AUF1-bound RNAs were found in the polysome fractions (Figure 5F), where AUF1 was basically undetectable (Figure 5D), or in the unbound material (fractions 1 and 2, Figure 5F), where AUF1 reportedly associates with the exosome; such interaction has been linked to a rapid decay of AUF1-bound mRNAs (Laroia et al, 1999; Chen et al, 2001), and may explain our inability to amplify bound mRNAs. siRNA-mediated reduction of HuR or AUF1 levels alters p21 and cyclin D1 expression In order to determine the relative influence of HuR on the expression of p21 and cyclin D1, an RNA interference (RNAi)-based approach was devised. By 3 days after transfection of HeLa cells using an HuR-targeting siRNA (HuR siRNA), HuR abundance was reduced to ∼20% of the levels in the control siRNA-transfected group (C, Figure 6A). Cultures expressing reduced HuR showed markedly lower p21 and cyclin D1 expression levels than did control siRNA-transfected cells, as detected by both Western and Northern blotting (Figure 6A and B). As shown in Figure 6C, the decreased p21 and cyclin D1 mRNA steady-state levels were due, at least in part, to the lower stability of each mRNA in cells with reduced HuR expression, as determined using actinomycin D-based mRNA half-life measurements. When the control and HuR siRNA treatment groups were compared, p21 mRNA half-life declined from ∼4.6 to 2.7 h, cyclin D1 mRNA from ∼8 to 3.3 h, respectively (Figure 6C); by contrast, the stability of the long-lived GAPDH mRNA (inset) was >18 h in all transfection groups. These results underscore the stabilizing influence of HuR on target mRNAs, including those that encode p21 and cyclin D1. Interestingly, in lysates obtained from HuR siRNA-transfected cells, AUF1 prominently associated with the p21 3′A transcript and the cyclin D1 3′C transcript (Figure 6D), two sites that were preferentially bound by HuR in control lysates (Figure 4B), suggesting that HuR and AUF1 may compete for binding to these sites. Figure 6.Effect of siRNA-mediated suppression of HuR levels on the expression and stability of p21 and cyclin D1 mRNAs. At 3 days after transfection with an siRNA that suppressed HuR protein levels by RNAi (HuR siRNA) or a control siRNA (C), the expression levels of HuR, cyclin D1, p21, and GAPDH (a control protein serving to monitor the equal loading and transfer of samples) were examined by Western blotting (A), and the abundance of p21 and cyclin D1 mRNAs (and control 18S rRNA) was assessed by Northern blotting (B). (C) A three days after transfection, the half-lives of p21 and cyclin D1 mRNAs in each siRNA group were assessed by using actinomycin D (2 μg/ml); mRNA half-lives (parentheses) were calculated from Northe