Destabilization of vascular endothelial growth factor mRNA by ... - Nature

Oncogene (2004) 23, 8673–8680

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc

Destabilization of vascular endothelial growth factor mRNA by the zinc-finger protein TIS11b Delphine Ciais1,3, Nadia Cherradi1,3, Sabine Bailly1, Emilie Grenier1, Edurne Berra2, Jacques Pouyssegur2, Jonathan LaMarre1,4 and Jean-Jacques Feige*,1 1

INSERM EMI 01-05, Department of Cellular Responses and Dynamics, Commissariat a` l’Energie Atomique, 17 Rue des Martyrs, F-38054 Grenoble Cedex 9, France; 2Institute of Signalling, Developmental Biology and Cancer Research, CNRS UMR 6543, Centre Antoine Lacassagne, 33 Avenue Lacassagne, F-06189 Nice, France

Vascular endothelial growth factor (VEGF) is an angiogenic cytokine, which plays a major role in tumor angiogenesis. VEGF mRNA expression is controlled by hypoxia, growth factors and hormones through both transcriptional and post-transcriptional mechanisms. VEGF mRNA has a short half-life and its abundance is regulated by the binding of stabilizing (HuR, hRNP-L) and still uncharacterised destabilizing proteins to its 30 -untranslated region. Here, we report that the ACTHregulated zinc-finger protein TIS11b and its homologs TIS11 and TIS11d interact with the 30 -untranslated region of VEGF mRNA and decrease its stability (halflife reduced from 130 to 60 min). Within the 2201 bp 30 untranslated region of VEGF mRNA, we identified a 75 bp domain, containing two consensus AU-rich motifs, which binds TIS11b and mediates its destabilizing activity. Ribonucleoprotein (RNP) complex immunoprecipitation experiments allowed us to demonstrate that the interaction between TIS11b and VEGF 30 -untranslated region occurs in live cells. Knocking down TIS11b expression in primary adrenocortical cells with small interfering (si)RNAs clearly indicated that TIS11b participates in the control of both basal and, to a larger extent, ACTH-induced VEGF mRNA expression levels. TIS11b is the first VEGF mRNA-destabilizing protein identified so far and therefore appears as a new potential target in antiangiogenic therapies. Oncogene (2004) 23, 8673–8680. doi:10.1038/sj.onc.1207939 Published online 27 September 2004 Keywords: VEGF; TIS11b; mRNA stability; AU-rich element

*Correspondence: JJ Feige, INSERM EMI 01-05, DRDC/ANGIO, CEA, 17 Rue des Martyrs, F-38054 Grenoble Cedex 9, France; E-mail: [email protected] 3 These authors contributed equally to this work. 4 Current address: Department of Biomedical Sciences, Ontario Veterinary College, University of Guelph, Ontario N1G 2W1, Canada Received 30 April 2004; accepted 28 May 2004; published online 27 September 2004

Introduction Angiogenesis, the process by which new blood vessels are formed from a pre-existing vascular network, is essential during embryonic tissue development and is implicated in a number of pathological situations including cancer. Among the dozen of angiogenic factors characterized to date, vascular endothelial growth factor (VEGF) has emerged as the most important regulator of physiological and pathological angiogenesis (Ferrara et al., 2003). Hypoxia, growth factors and hormones control VEGF expression through both transcriptional and post-transcriptional mechanisms. VEGF mRNA has a constitutively short half-life (60–180 min). It is stabilized under hypoxic conditions through the binding of proteins (HuR, hRNP-L) to AUrich elements (AREs) present in its 30 -untranslated region (30 -UTR) (Goldberg-Cohen et al., 2002; Shih and Claffey, 1999). AREs, which have been identified in the 30 -UTRs of several short-lived mRNAs as important determinants of RNA instability, are composed of several consensus pentameric AUUUA, nonameric UUAUUUA(A/U)(A/U) or simply U-rich motifs (Chen and Shyu, 1995). To date, the mechanisms of VEGF mRNA destabilization remain unknown. In a previous work, we observed that ACTH treatment of primary adrenocortical cells results in a rapid and transient increase in VEGF mRNA levels (peaking at 2–3 h) and a delayed but sustained induction of TIS11b mRNA (reaching a maximum after 3–5 h) (Chinn et al., 2002). As the increase of TIS11b expression was concomitant with VEGF mRNA decay, we postulated that the product of this gene might negatively control VEGF mRNA expression. This hypothesis was supported by the following pieces of information. TIS11b belongs to a family of RNA-binding proteins (comprising also TIS11/TTP/tristetraprolin and TIS11d) that share characteristic tandem CCCH-type zinc-finger domains. The canonical family member, TTP, has been shown to bind to AREs in the 30 -UTR of tumour necrosis factor a (TNFa) mRNA through these zinc-finger domains, and to decrease TNFa mRNA stability by promoting its deadenylation and degradation (Carballo et al., 1998; Lai et al., 2003). All members of the family are involved

Destabilization of VEGF mRNA by TIS11b D Ciais et al

8674

in regulation of several short-lived cytokine mRNAs including granulocyte macrophage colony-stimulating factor (GM-CSF), and interleukin 3 (Carballo et al., 2000; Stoecklin et al., 2000) and appear to share similar ARE-binding and mRNA-destabilizing activities. However, their tissue-specific expression and regulation presumably contribute to their specificity of action. Here, we studied the effects of TIS11b and its related family members on VEGF mRNA stability and characterized the molecular interaction between TIS11b and VEGF mRNA 30 -UTR. Our original data show that (i) TIS11b binds to a 75 bp-long domain of VEGF mRNA that contains two AU-rich motifs, (ii) TIS11b is complexed with VEGF mRNA in living cells as determined by ribonucleoprotein (RNP) complex immunoprecipitation, (iii) binding of TIS11b to VEGF mRNA results in its destabilization, (iv) TIS11b is a physiological inhibitor of VEGF mRNA stability in adrenocortical cells since siRNA-mediated downregulation of TIS11b results in increased basal and ACTHstimulated VEGF mRNA levels. Results Effect of TIS11b and related proteins on VEGF mRNA 30 -UTR We previously observed that expression of TIS11b in 3T3 cells expressing the luciferase reporter gene cloned upstream of the rat VEGF 30 -UTR (a construct hereafter named Luc-V30 ) dose-dependently decreased luciferase activity (Chinn et al., 2002). We first investigated the effects of tristetraprolin (TTP) family members on the activity of this same reporter gene. As shown in Figure 1a, all three members (TIS11, TIS11b and TIS11d) induced a comparable dose-dependent inhibitory effect on luciferase activity, reaching 50–60% inhibition when 50 ng of plasmid were transfected. Transfection of a control b-galactosidase expression plasmid induced a slight dose-dependent inhibition of reporter gene activity, which did not exceed 25%. As the effects were similar with all family members, we pursued the study with TIS11b. We investigated whether decreased luciferase activity resulted from decreased mRNA levels. Quantitative real-time RT–PCR was used to measure Luc-V30 mRNA levels following transfection of increasing concentrations of either a TIS11b- or a b-galactosidase-expression plasmid. We observed that Luc-V30 mRNA levels were dose-dependently reduced in TIS11b-expressing cells similarly to the reporter gene activity (Figure 1b). We then examined whether this decrease in mRNA levels resulted from changes in the stability of Luc-V30 mRNA. The half-life of this mRNA was measured in b-galactosidase- and TIS11b-transfected cells, using the transcription inhibitor DRB. In the presence of TIS11b, the half-life of Luc-V30 mRNA was reduced from 130 to 60 min (Figure 1c). This result prompted us to analyse the physical interaction between TIS11b and a radiolabeled fragment encoding the fulllength 30 -UTR of VEGF mRNA by EMSA (Figure 1d). Escherichi coli-expressed TIS11b delayed migration of Oncogene

the radiolabeled VEGF 30 -UTR probe (lane 3), and addition of a polyclonal anti-TIS11b antibody (anti N-ter) resulted in a supershift of the probe–protein complex (lane 6). The complex was specific since it was disrupted in the presence of an excess of unlabeled probe (lanes 4 & 5) and was not observed in the presence of a control bacterial extract (lane 1). These results suggested that the regulatory effects observed in Figure 1a–c resulted from the formation of a molecular complex between TIS11b and the 30 UTR of VEGF mRNA. Identification of the TIS11b-binding site within VEGF mRNA 30 -UTR Our next goal was to identify the TIS11b-binding sequence within the VEGF 30 UTR. In silico analysis of the 2201 bp-long rat VEGF 30 -UTR revealed the presence of two nonameric UUAUUUA(A/U)(A/U) and seven pentameric AUUUA motifs (Figure 2a). We constructed various deletion mutants of the Luc-V30 expression plasmid (shown in Figure 2a) and cotransfected them with a TIS11b expression plasmid into 3T3 fibroblasts. The inhibition of luciferase activity induced by TIS11b was determined under each condition (Figure 2b). Deletion of a large fragment of the VEGF 30 UTR containing all ARE motifs (DARE) abrogated the inhibitory effect of TIS11b on reporter gene activity by 80%. Deletion of sequences 30 or 50 to the NsiI site at position 1255 (Nsi-50 and Nsi-30 , respectively) clearly indicated that the TIS11b-binding element was located in Nsi-50 , that is, between nt 0–1255. Further truncation of this sequence allowed us to localize a 75 bp sequence (nt 1161–1235) that responded to TIS11b similarly although slightly less than the full-length construct. This sequence is subsequently referred to as TBE (TIS11b-binding element). TBE contains one nonameric (N) and one pentameric (P) consensus motifs that are both conserved in the human VEGF sequence. We prepared single and double mutants of these motifs. Each mutation (N, P and N/P) prevented TIS11b inhibitory effect similarly to the DARE construct. This strongly suggested that each of the two AREs present within this 75 bp-long sequence is involved in TIS11bbinding and regulatory activity. We next examined the association between TIS11b and the 2201 bp-long VEGF 30 -UTR by UV crosslinking of the radiolabeled RNA probe to the recombinant protein. As shown in Figure 2c, after RNAse digestion, a covalent RNP complex with an apparent molecular size of 38 kDa could be detected by PAGE–SDS. An excess of unlabeled 75 bp TBE sequence (Figure 2c) totally inhibited the interaction. This confirmed that TBE is likely to be the primary TIS11b-binding domain within the VEGF 30 UTR. We next attempted to crosslink the radiolabeled 75 bp-long TBE RNA and the recombinant TIS11b protein. The results shown in Figure 2d clearly demonstrate the formation of a molecular complex between TBE and TIS11b that is dissociated by excess unlabeled wild-type probe but resists to an excess of any of the mutant TBE RNA probes (N, P and N/P). The failure of the double mutant TBE probe to associate


8675

Figure 1 Effect of TIS11b and related proteins on VEGF mRNA 30 -UTR. (a) TTP family proteins decrease reporter gene activity through VEGF 30 -UTR. 3T3 cells were cotransfected with pLuc-V30 and indicated amounts of CMV-driven expression plasmids encoding b-galactosidase (b-gal), TIS11b, TIS11 or TIS11d. Firefly luciferase activity was measured and normalized to renilla luciferase activity as described in Materials and methods. Each condition was performed in triplicate and the bars represent the mean values7s.d. of the results from three independent experiments. The values from TIS- and b-galactosidase-expressing cells were compared using the unpaired t-test. *, Po0.05; **, Po0.01. (b) TIS11b expression decreases reporter gene mRNA levels through VEGF 30 -UTR. 3T3 cells were cotransfected with pLuc-V30 and indicated amounts of expression plasmids encoding either bgalactosidase (b-gal, black bars) or TIS11b (gray bars). After 24 h of expression, the amount of luciferase mRNA was quantitated by real-time RT–PCR. The bars represent the mean values7s.d. of triplicate determinations. The asterisks indicate statistical significance (*, Po0.05; **, Po0.01) of comparisons with respective control b-galactosidase transfections. (c) TIS11b expression decreases reporter gene mRNA half-life through VEGF 30 -UTR. 3T3 cells were cotransfected with pLuc-V30 and 25 ng of expression plasmids encoding either b-galactosidase (b-gal) or TIS11b. After 24 h, transcription was inhibited by addition of DRB, and the luciferase mRNA decay was measured over 90 min by real-time RT–PCR. Each time point is represented as the mean value7s.d. of triplicate determinations. (d) TIS11b protein physically interacts with VEGF mRNA 30 -UTR. Control (CTL) and TIS11b-expressing (TIS11b) bacterial extracts were prepared as described in Materials and methods. They were incubated with radiolabeled full-length VEGF 30 -UTR RNA probe (FL*) in the absence or presence of a 2.5-fold (lane 4) or fivefold (lanes 2 and 5) excess of unlabeled competitor probe (Comp.). Then, 0.5 mg of anti-TIS11b IgGs was added to the incubation mixture in lane 6. The ribonucleoprotein complexes were then analysed by EMSA as described in Materials and methods. The bracket encompasses the retarded probe and the arrow points to the suspershifted (SS) antibody–TIS11b–probe complex

with TIS11b was further demonstrated by a direct UV crosslinking experiment (Figure 2e). Taken together, the results of the UV crosslinking experiments confirmed that each of the two ARE motifs present in the TBE sequence is essential for TIS11b binding. In vivo interaction between TIS11b and VEGF 30 UTR After characterizing the interaction between TIS11b and VEGF 30 UTR in vitro, our next goal was to demonstrate that this interaction could also occur in live cells. For this, we transfected NIH 3T3 cells with pCMV-TIS11b and pLuc-V30 plasmids and subsequently subjected them to in vivo crosslinking. TIS11b-containing RNP complexes were then immunoprecipitated from cell

lysates using a specific anti-TIS11b antibody (anti Cter) and the immunoprecipitates were subjected to RT– PCR amplification of the firefly luciferase mRNA. As shown in Figure 3, luciferase mRNA was specifically detected in anti-TIS11b immunoprecipitates, whereas it was only detectable in the supernatants of the control immunoprecipitations. These results demonstrate that the TIS11b–VEGF 30 UTR interaction occurs not only in reconstituted in vitro systems but also in living cells. Effect of TIS11b ‘knockdown’ on VEGF mRNA expression We have previously shown that ACTH treatment of primary adrenocortical cells results in a rapid and Oncogene


8676

Figure 2 Identification of the TIS11b-binding site within VEGF mRNA 30 -UTR. (a) Restriction map of the deletion constructs derived from the 2201 bp-long 30 -UTR of VEGF mRNA. Deletion constructs were generated from the full-length 30 -UTR of VEGF mRNA using the restriction sites indicated in the upper part of the figure. The TBE fragment (nt 1161–1235) was mutated to invalidate either one or both AREs. Flags with white circles represent the nonameric ARE motif UUAUUUA(A/U)(A/U) and those with black circles represent the pentameric motif AUUUA. Flags with gray squares represent mutated motifs. (b) Mapping of the sequence in VEGF mRNA 30 -UTR required for TIS11b inhibition of reporter gene activity. 3T3 cells were cotransfected with 10 ng of a TIS11b (or b-galactosidase as a control) expression plasmid, 5 ng of a plasmid encoding renilla luciferase and 500 ng of either one of the plasmids encoding firefly luciferase cloned upstream of the constructs described in (a). The TIS11b-induced inhibition of luciferase activity observed with each construct was compared to that obtained with the full-length (FL) 30 -UTR. Results are expressed as the mean7s.e. of duplicate determinations. Values with identical superscript letters are not statistically different (P>0.05). This is a representative experiment of three independent experiments that produced similar results. (c-e) Mapping of the TIS11b-binding sequence in VEGF mRNA 30 -UTR. Radiolabeled polyadenylated RNA probes encoding some of the constructs shown in (a) were UV crosslinked to control (CTL) or TIS11b-expressing (TIS11b) bacterial extracts. Either full-length (FL) 30 -UTR (c), intact TBE (TBE) (d, e) or doublemutated TBE (TBE mutN/P) (e) was used as probes. A fivefold excess of unlabeled competitor RNA (Comp.) was added to some assay conditions, as indicated. After RNAse T1 digestion of the unprotected probes, the complexes were analysed by 10% PAGE–SDS. The position and size (in kDa) of MW standards are indicated on the left of panel c. Arrowheads point to the position of the 38 kDa TIS11b–probe complexes. The upper bands correspond to complexes between probes and bacterial proteins unrelated to TIS11b

transient increase in VEGF mRNA levels and a delayed but sustained induction of TIS11b mRNA (Chinn et al., 2002). To determine the contribution of TIS11b in the regulation of VEGF mRNA expression by ACTH, we used the RNA interference methodology. Following Oncogene

transfection with siRNA duplexes targeting four regions (Ts1–Ts4) in TIS11b gene, bovine adrenocortical cells were incubated in the presence or in the absence of ACTH for the indicated periods of time as described in ‘Materials and methods’ section. As shown in Figure 4a,


8677

the pool of four siRNA duplexes (Ts1 þ Ts2 þ Ts3 þ Ts4) efficiently reduced TIS11b protein levels in adrenocortical cells, whereas scrambled siRNA duplexes (Scr) of these targeting sequences (serving as nonspecific controls) were inefficient. Quantification of TIS11b mRNA level in four independent experiments revealed

Figure 3 In vivo interaction between TIS11b and VEGF 30 UTR. 3T3 cells were cotransfected with a TIS11b expression plasmid and pLuc-V30 . At 24 h after transfection, RNP complexes were crosslinked in living cells, as described in Materials and methods. The complexes containing TIS11b were then immunoprecipitated with anti-TIS11b IgGs and protein A-agarose. Controls included the addition of nonimmune IgGs or no IgG. The supernatants (SUP) and immunoprecipitates (IP) were collected separately and the luciferase mRNA was amplified by RT–PCR and analysed, as described in Materials and methods. The Input lane shows the results of a parallel reaction performed with total RNP complexes that were not subjected to immunoprecipitation

that TIS11b gene expression was knocked down by 75% (data not shown). ACTH-stimulated increase in TIS11b mRNA level, peaking at 5 h post-treatment (Figure 4b and c), was completely prevented in siRNA-treated cells, indicating a sustained and efficient action of siRNAs on TIS11b nascent transcripts. Remarkably, TIS11b knockdown resulted in an upregulation of VEGF expression by approximately 25–50% in unstimulated cells (Figure 4b and d). In ACTH-stimulated cells, siRNA treatment enhanced the hormone-induced expression of VEGF by 60–150%. Altogether, these results indicate that TIS11b is an important negative regulator of VEGF mRNA levels under basal condition and to a larger extent under ACTH treatment. Discussion Post-transcriptional regulation of VEGF mRNA is a key process in the control of VEGF expression. VEGF mRNA has a short half-life and its stabilization has been shown to strongly contribute to the induction of VEGF expression that is observed under hypoxic conditions (Levy et al., 1996). Several domains of VEGF mRNA 30 UTR are involved in hypoxic stabilization of

Figure 4 siRNA-induced TIS11b gene repression leads to upregulation of VEGF expression. Bovine adrenocortical cells in primary culture were transfected with or without a 10 nM solution of four 21-mer siRNA duplexes as described in the ‘Materials and methods’ section. After 48 h, cells were either lysed to evaluate TIS11b knockdown or incubated with fresh medium in the absence or in the presence of 10 nM ACTH for the indicated period of time. (a) Effect of siRNA treatment on TIS11b expression. The content of TIS11b protein in total cell lysates from nontransfected (Cont) or siRNA-treated cells was analysed by immunoblotting as described in Materials and methods section. Scrambled siRNA (Scr) served as nonspecific control siRNAs. siRNA duplexes Ts1, Ts2, Ts3, Ts4 were tested either separately or altogether. The combination of the four siRNA was more efficient in knocking down TIS11b gene expression. Immunodetection of a-Tubulin was used to check for protein loading in each lane. (b-d) Effect of TIS11b knockdown on VEGF expression. At each time point of stimulation by ACTH, total RNA was isolated and RT–PCR analysis was performed to determine TIS11b, VEGF or HPRT mRNA expression levels (b). Amplification products were quantitated using a Fluorimager and normalized to HPRT (c, d) Oncogene


8678

VEGF mRNA through the recruitment of several specific AREs-binding proteins. These include HuR and hRNP-L (Levy et al., 1998; Shih and Claffey, 1999). The existence of antagonistic proteins, favoring VEGF mRNA degradation, has been postulated but never demonstrated. To our knowledge, our present work is the first to report the identification of such an AREbinding protein that negatively regulates the stability of VEGF mRNA. Three major conclusions can be drawn from the results of the present work: (i) a functional interaction between TIS11b and VEGF mRNA 30 UTR results in decreased VEGF mRNA stability and mRNA levels; (ii) the site of TIS11b interaction with the VEGF mRNA 30 UTR is restricted to a 75 bp-long sequence that contains one nonameric and one pentameric AU-rich motifs; (iii) TIS11b participates to the physiological regulation of basal as well as ACTH-induced VEGF expression. Interestingly, the TBE identified in the present study is located very close to a recently described HuRbinding element (Goldberg-Cohen et al., 2002). Proximity of HuR- and TIS11b-binding elements in VEGF 30 UTR suggests that stabilizing and destabilizing AREbinding proteins may compete for binding to similar regions of the 30 UTR as it has been recently observed for several cytokine mRNAs (Raghavan et al., 2001). Interestingly, this TBE sequence has previously been shown to be involved in the stabilization of VEGF mRNA after JNK and p38/HOG activation (Pages et al., 2000). In this study, the stimulation of stress-activated protein kinases induced the formation of a RNP complex that bound to a region containing the TBE and triggered stabilization of VEGF mRNA. Furthermore, it was reported that phosphorylation of TIS11 (TTP) by p38 kinase could reduce its binding affinity for ARE motifs (Carballo et al., 2001). One could then extrapolate that p38 activation might disrupt the TIS11b–TBE complex and allow recruitment of a stabilizing ARE-binding protein which remains to be identified. The second important finding of this work relates to the modulation of VEGF expression by TIS11b in vivo. We bring evidence that TIS11b and VEGF 30 UTR directly interact in living cells, and that TIS11b participates in the physiological regulation of VEGF expression by accelerating the turnover of VEGF mRNA. Using RNA interference in primary adrenocortical cells, we confirmed our previous hypothesis postulating that the mRNA destabilizing activity of TIS11b is involved in the decay phase of ACTH-elicited VEGF mRNA levels (Chinn et al., 2002). Reducing endogenous TIS11b expression by 75% caused a small increase in basal and a marked increase in ACTHinduced VEGF mRNA levels. This indicates that TIS11b plays an essential role as a limiting factor for VEGF expression. One could speculate that a complete knockdown of TIS11b gene may result in a sustained hormone-induced expression of VEGF mRNA. It is also possible that TIS11b participates in the control of the stability of many other ACTH target genes. For Oncogene

example, the gene encoding StAR (Steroidogenesis Acute Regulatory protein), a mitochondrial protein essential for cholesterol transfer to the inner mitochondrial membrane, is a likely candidate as its expression is rapidly and transiently induced by ACTH in adrenocortical cells (Brand et al., 1998) and since several AREs are present in its mRNA 30 -UTR (Ariyoshi et al., 1998). Alterations of mRNA stabilization have been reported to occur during tumour growth (Ross et al., 1991). Very recently, it has been reported that TTP acts as a potent tumour suppressor when overexpressed in a v-H-ras-dependent mast cell tumour model (Stoecklin et al., 2003). Our results would suggest that this effect could result from decreased VEGF mRNA levels and decreased angiogenesis. Unfortunately, tumor vascularization was not documented in this study. In conclusion, our results suggest that TIS11b might represent a novel antiangiogenic and antitumoral agent acting through its destabilizing activity on VEGF mRNA. Future studies from our laboratory will address this question.

Materials and methods Plasmid constructs Plasmids containing the firefly luciferase cDNA cloned upstream of the rat VEGF 30 -UTR and downstream of the thymidine kinase promoter (pLuc-V30 and truncations) were derived from the pSp64 plasmids described in Levy et al., 1996. The sequences of wild-type and mutant TBE are shown in Table 1. Human TIS11b cDNA was obtained by PCR from H295R cells and cloned into pTarget (Promega, Madison, WI, USA) to generate pCMV-TIS11b and into pGEX4-T to generate pGEX-TIS11b. The pCMV-TIS11 and pCMVTIS11d plasmids were generated by cloning mouse TIS11 and TIS11d cDNAs (kindly provided by Dr KD Brown, Babraham Institute, Cambridge, UK) into pTarget. PRK7bgal is a control plasmid in which b-galactosidase expression is driven by the CMV promoter. pRL-TK encoding renilla luciferase was obtained from Promega. Quantitative RT–PCR Quantitative RT–PCR was performed on a Light Cycler apparatus using SYBR green PCR core reagents (Roche Diagnostics, Mannheim, Germany). The following primers and conditions were used for luciferase: CGCCAAAAGCAC TCTGATTGA and CCTTGTCGTATCCCTGGAAGATG (951C for 15 s, 551C for 5 s, 721C for 10 s; 45 cycles in 3 mM MgCl2) and for GAPDH: AACGACCCCTTCATTGAC and TTCACGACATACTCAGCA (951C for 15 s, 571C for 5 s, 721C for 10 s; for 45 cycles in 4 mM MgCl2). DNA transfection and dual luciferase activity assay NIH-3T3 fibroblasts were grown in DMEM supplemented with 10% fetal calf serum and transfected using lipofectamine (Invitrogen, Cergy-Pontoise, France). Various amounts (1–50 ng) of either pCMV-TIS11b, pCMV-TIS11, pCMVTIS11d or PRK7-bgal were cotransfected with 500 ng pLucV30 , 5 ng of pRL-TK and pUC19 up to a total of 1 mg plasmid DNA into 3 105 cells. Renilla and firefly luciferase activities were measured sequentially 48 h after transfection with the


8679 Table 1 Wild-type and mutant TBE sequences TBE TBEMutN TBE MutP TBE MutN/P

GGUACUUAUUUAAUAGCCCUUUUUAAUUAGAAAUUAAAACAGUUAAUUUAAUUAAAGAGUAGG GUUUUUUCAGUA GGUACUUAGGUAAUAGCCCUUUUUAAUUAGAAAUUAAAACAGUUAAUUUAAUUAAAGAGUAGG GUUUUUUCAGUA GGUACUUAUUUAAUAGCCCUUUUUAAUUAGAAAUUAAAACAGUUAAGGUAAUUAAAGAGUAGG GUUUUUUCAGUA GGUACUUAGGUAAUAGCCCUUUUUAAUUAGAAAUUAAAACAGUUAAGGUAAUUAAAGAGUAGG GUUUUUUCAGUA

Mutations (bold characters) were introduced in the nonameric (N) and pentameric (P) motifs (framed sequences) of the TBE sequence (nt 1161– 1235) as shown

Dual-Luciferase reporter assay system (Promega). Results are expressed as relative light units of firefly luciferase activity over relative light units of renilla luciferase activity. For RNP complex immunoprecipitation experiments, 3.5 mg of pLuc-V30 and 1.5 mg of pCMV-TIS11b were transfected into 1.7 106 NIH-3T3 cells using lipofectamine. Production of anti-TIS11b antibodies Rabbit polyclonal antibodies to TIS11b were produced by CovalAb (Lyon, France) using two synthetic peptides from the NH2- and COOH-termini as immunogens. The anti-C ter and anti-N ter IgGs were then purified from the antiserum by affinity chromatography.

temperature. The reaction was stopped by 0.25 M glycine and cells were lysed in RIPA buffer containing protease inhibitors. Protein A-agarose preadsorbed cell lysates were immunoprecipitated using protein A-agarose beads preincubated with 3 mg of anti-TIS11b antibody (anti-Cter), 3 mg of nonimmune IgG, or no protein. After crosslinking reversion at 701C for 45 min, RNA was isolated from supernatants and immunoprecipitates, treated with DNAse I (Invitrogen), and reverse transcribed with Superscript II (Invitrogen). A PCR amplification of the luciferase transcripts was then carried out using Taq polymerase (QBiogen, Illkirch, France) and the same primers used in quantitative PCR, under the following conditions: 941C for 1 min, 551C for 1 min and 721C for 1 min for 40 cycles. The PCR products were analysed by 2% agarose gel electrophoresis.

Bacterial expression of TIS11b Overnight cultures of E. coli transformed with pGEX-TIS11b or pGEX4-T (control) were diluted with LB and grown at 301C to limit the formation of inclusion bodies. IPTG (0.5 mM) was added for 4 h to induce protein synthesis. Western blotting of the bacterial extract unexpectedly revealed the presence of a small proportion of GST-TIS11b fusion protein and a larger amount of untagged TIS11b presenting an expected MW of 38 kDa. The fusion protein was eliminated from the extracts by adsorption on glutathione–agarose beads. EMSA and UV crosslinking assay 32 P-labeled and unlabeled riboprobes were synthesized in vitro using the riboprobe SP6 system kit (Promega). Integrity of RNA transcripts was determined using a lab-on-chip analyser (Bioanalyzer 2100, Agilent Technologies, Palo Alto, CA, USA). In all, 5 105 c.p.m. of RNA transcripts were incubated for 20 min at room temperature with 10 mg of bacterial extracts in 10 mM HEPES pH 7.6, 3 mM MgCl2, 40 mM KCl, 5% glycerol, 0.5% NP40 and 2 mM DTT, in the presence or absence of specific competitors. Yeast tRNA (50 ng/ml) and heparin (2 mg/ml) were then added for 10 min and 0.5 mg of antiTIS11b Nter antibody was eventually added for another 10 min. Protein–RNA complexes were separated by electrophoresis on 4% polyacrylamide gels and visualized with a b-imager. For UV crosslinking studies, mixtures were prepared as for EMSA and exposed to UV light for 30 min at 41C. Then, 100 U of RNase T1 (Invitrogen) was added for 20 min and RNA–protein complexes were analysed by 10% PAGE–SDS.

RNP complex immunoprecipitation and analysis by RT–PCR RNP complexes were immunoprecipitated after reversible crosslinking between RNA and proteins as previously described (Niranjanakumari et al., 2002). Briefly, cell suspensions were incubated in 1% formaldehyde for 10 min at room

RNA interference Expression of TIS11b was inhibited by transfection of siRNAs (Elbashir et al., 2001). Briefly, primary cultures of bovine adrenocortical fasciculata cells were prepared by enzymatic digestion and cultured at 371C in Ham’s F12 medium supplemented with 10% horse serum, 2.5% FCS and antibiotics as previously reported (Negoescu et al., 1994). At 1 day after plating, adrenocortical cells were transfected with or without 10 nM of siRNA duplexes, using siPORT lipid reagent (Ambion, Austin, TX, USA). siRNA duplexes (21nucleotide) were designed based on four 19 bp TIS11b cDNA regions (positions 128–147, 462–481, 848–867 and 1086–1105), and generated using Silencer siRNA construction kit from Ambion. A Blast search on the NCBI database ensured specific targeting to TIS11b mRNA. Tis11b siRNA oligonucleotide templates were the following: (1) antisense 30 -AAGGGTAACAAGATGCTCAACCCTG TCTC-30 , sense 30 -AAGTTGAGCATCTTGTTACCCCCTG TCTC-30 , (2) antisense 30 -AAAACGGTGCCTGTAAGTAC GCCTGTCTC-30 , sense 30 -AACGTACTTACAGGCACCGT TCCTGTCTC-30 , (3) antisense 30 -AATAACCCCTTCGCCT TTTCCCCTGTCTC-30 , sense 30 -AAGGAAAAGGCGAAG GGGTTACCTGTCTC-30 , (4) antisense 30 -AACTCAAGAC GCCTGCCCATTCCTGTCTC-30 , sense 30 -AAAATGGGCA GGCGTCTTGAGCCTGTCTC-30 . Scrambled siRNA duplexes of these targeting sequences served as nonspecific control siRNAs. Typically, cells were analysed for the loss of TIS11b mRNA and protein expression 48 h after transfection using either RT–PCR or immunoblotting. At this time point, culture medium was changed and cells were treated for the indicated periods of time with or without 10 nM ACTH. For RT–PCR analysis of TIS11b, VEGF or HPRT gene expression in siRNA-transfected cells, 2 mg of total RNA was reverse transcribed with Superscript II (Life Technologies) and PCR amplified using Taq polymerase (QBiogen, Illkirch, France) Oncogene


8680 using the same primers and amplification conditions as described by Chinn et al. (2002). Amplification products were analysed by 2% agarose gel electrophoresis and visualized under UV light. SDS–PAGE analysis and immunoblotting SDS–polyacrylamide (0.1%; 10%) gel electrophoresis was performed according to Laemmli (1970). In total, 3 106 primary adrenocortical cells, plated in 6-cm Petri dishes, were lysed in RIPA buffer (in the presence of protease inhibitors) following 48 h treatment by TIS11b siRNA. The lysate was cleared by centrifugation for 10 min at 10 000 g at 4 C. Aliquots of 10 mg were resolved by SDS–PAGE and transferred onto a polyvinylidene fluoride (PVDF) membrane which was incubated in blocking buffer (PBS, 0.1% Tween 20, 5% nonfat dry milk) for 1 h, then exposed to 0.1 mg/ml of rabbit polyclonal anti-TIS11b antibody at room temperature for 2 h. The membrane was thoroughly washed with PBS/Tween buffer (3 10 min), and then incubated for 1 h with horseradish peroxidase-labeled goat anti-rabbit IgG. The PVDF sheet was then washed for 4 10 min and the antigen–antibody complex was revealed by Enhanced Chemiluminescence using the Western blotting detection kit from Amersham Biosciences (Orsay, France). The same membrane was stripped and reprobed with a monoclonal antibody for a-tubulin (a generous gift from Dr D Job, CEA-Grenoble) to check for loading of protein in each lane.

Statistical analysis Results are expressed as the mean7s.d. Statistical comparisons were made using one-way ANOVA test (StatView, Abacus Concepts Inc., Berkeley, CA, USA). A value of Po0.05 was considered as statistically significant.

Abbreviations VEGF, vascular endothelial growth factor; TIS11, TPA-induced sequence 11; TTP, tristetraprolin; 30 -UTR, 30 -untranslated region; ARE, AU-rich element; DRB, 5,6-dichloro-1-d-Dribofuranosyl benzimidazole; TNFa, tumour necrosis factor a; GM-CSF, granulocyte macrophage colony-stimulating factor; PVDF, polyvinylidene fluoride.

Acknowledgements We are indebted to Dr G Pages for providing us the pSp64-Sty plasmid. This work was supported by INSERM (EMI 01-05), University Joseph Fourier, CEA (DSV/DRDC/ANGIO), GEFLUC, Fondation pour la Recherche Me´dicale (grant to DC), and Association pour la Recherche sur le Cancer (project no 7626). We are indebted to the Ligue Nationale contre le Cancer for its financial support to DC (Doctoral grant, Comite´ de l’Ise`re) and NC (post-doctoral grant) and to INSERM for its support to JL (poste orange).

References Ariyoshi N, Kim YC, Artemenko I, Bhattacharyya KK and Jefcoate CR. (1998). J. Biol. Chem., 273, 7610–7619. Brand C, Cherradi N, Defaye G, Chinn A, Chambaz EM, Feige JJ and Bailly S. (1998). J. Biol. Chem., 273, 6410–6416. Carballo E, Cao H, Lai WS, Kennington EA, Campbell D and Blackshear PJ. (2001). J. Biol. Chem., 276, 42580– 42587. Carballo E, Lai WS and Blackshear PJ. (1998). Science, 281, 1001–1005. Carballo E, Lai WS and Blackshear PJ. (2000). Blood, 95, 1891–1899. Chen CY and Shyu AB. (1995). Trends Biochem. Sci., 20, 465–470. Chinn AM, Ciais D, Bailly S, Chambaz E, LaMarre J and Feige JJ. (2002). Mol. Endocrinol., 16, 1417–1427. Elbashir SM, Harborth J, Lendeckel W, Yalcin A, Weber K and Tuschl T. (2001). Nature, 411, 494–498. Ferrara N, Gerber HP and LeCouter J. (2003). Nat. Med., 9, 669–676. Goldberg-Cohen I, Furneaux H and Levy AP. (2002). J. Biol. Chem., 277, 13635–13640. Laemmli UK. (1970). Nature, 227, 680–685.

Oncogene

Lai WS, Kennington EA and Blackshear PJ. (2003). Mol. Cell. Biol., 23, 3798–3812. Levy AP, Levy NS and Goldberg MA. (1996). J. Biol. Chem., 271, 2746–2753. Levy NS, Chung S, Furneaux H and Levy AP. (1998). J. Biol. Chem., 273, 6417–6423. Negoescu A, Labat-Moleur F, Brambilla E, Chambaz EM and Feige JJ. (1994). Mol. Cell. Endocrinol., 105, 155–163. Niranjanakumari S, Lasda E, Brazas R and Garcia-Blanco MA. (2002). Methods, 26, 182–190. Pages G, Berra E, Milanini J, Levy AP and Pouyssegur J. (2000). J. Biol. Chem., 275, 26484–26491. Raghavan A, Robison RL, McNabb J, Miller CR, Williams DA and Bohjanen PR. (2001). J. Biol. Chem., 276, 47958– 47965. Ross HJ, Sato N, Ueyama Y and Koeffler HP. (1991). Blood, 77, 1787–1795. Shih SC and Claffey KP. (1999). J. Biol. Chem., 274, 1359–1365. Stoecklin G, Gross B, Ming XF and Moroni C. (2003). Oncogene, 22, 3554–3561. Stoecklin G, Ming XF, Looser R and Moroni C. (2000). Mol. Cell. Biol., 20, 3753–3763.