Stress-activated protein kinases (JNK and p38/HOG) are essential for vascular endothelial growth factor mRNA stability.

Stability of the vascular endothelial growth factor (VEGF) mRNA is tightly regulated through its 3'-untranslated region (3'-UTR). Here, we demonstrate that VEGF mRNA levels are increased by anisomycin, a strong activator of stress-activated protein kinases. Hence, VEGF mRNA induction is inhibited by SB202190, an inhibitor of JNK and p38/HOG kinase. Furthermore, VEGF mRNA expression is increased in cells that overexpress JNK and p38/HOG by an increase in its stability. We show by two different approaches that anisomycin exerts its effect on the VEGF mRNA 3'-UTR. First, by using an in vitro mRNA degradation assay, the half-life of the VEGF mRNA 3'-UTR region transcript was found to be increased when incubated with extracts from anisomycin-treated cells; and second, the 3'-UTR was also sufficient to confer mRNA instability to the Nhe3 (Na(+)/H(+) exchanger 3) heterologous reporter gene, and anisomycin treatment stabilized the chimeric mRNA (Nhe3 fused to the VEGF mRNA 3'-UTR). This chimeric mRNA is also more stable in cells overexpressing p38/HOG and JNK that have been stimulated by anisomycin. We show that such regulation is mediated through an AU-rich region of the 3'-UTR contained within a stable hairpin structure. By RNA electrophoretic mobility shift assays, we show that this region binds proteins specifically induced by anisomycin treatment. These findings clearly demonstrate a major role of stress-activated protein kinases in the post-transcriptional regulation of VEGF.

In this study, we have analyzed the transduction pathways implicated in the increased expression of VEGF mRNA by anisomycin, a strong activator of stress-activated protein kinases (SAPKs). We demonstrate that anisomycin increases VEGF mRNA stabilization through the activation of p38/HOG kinase and JNK. This action is presumably mediated by specific recruitment of proteins, which remain to be identified, on various parts of the 3Ј-untranslated region of VEGF mRNA.

EXPERIMENTAL PROCEDURES
Cell Culture-The established Chinese hamster lung fibroblast line CCL39 (American Type Culture Collection), its derivatives PS120 and PS200 (which lack Nhe1 antiporter activity) (27), and human embryonic kidney 293 and corresponding transfected cells were cultivated in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 7.5% fetal calf serum, 50 units/ml penicillin, and 50 g/ml streptomycin sulfate. For the p38/HOG and JNK stable transfectants, we have used two CCL39 derivatives, PS200 and PS120, in which we cotransfected hemagglutinin-tagged JNK and p38/HOG, respectively (28,29). The Na ϩ /H ϩ exchanger expression vectors (Nhe1 isoform) were cotransfected as a selection gene (30) with a 20:1 ratio. Nhe1-expressing clones were selected by the acid load test (27), and p38/HOG or JNK expression was detected by using anti-hemagglutinin antibody in Western blot experiments. For the selection of cells stably transfected with Nhe3 chimeras, we used the acid load test in the presence of amiloride derivatives, which specifically inhibit Nhe1, but not Nhe3 (31).
Materials-Restriction and DNA-modifying enzymes were obtained from New England Biolabs Inc. or from Eurogentec (Liège, Belgium).
Cloning of the Human VEGF 3Ј-UTR-First-strand cDNA was synthesized from 1 g of human embryonic kidney 293 poly(A) ϩ RNA using avian myeloblastosis virus reverse transcriptase with an oligo(dT) primer. This material was used as a template for polymerase chain reaction amplification. The following oligonucleotides derived from the human VEGF 3Ј-UTR (7,34) were synthesized and used as primers for the polymerase chain reaction: oligonucleotide 1, 5Ј-GATGTGACAAGC-CAAGGCGGTG-3Ј; and oligonucleotide 2, 5Ј-GAGGAGCTTTGAGAT-CAGAAT-3Ј. Intermediate oligonucleotides were also chosen to amplify smaller fragments. These oligonucleotides overlapped the StuI restriction site situated in the middle of the 3Ј-UTR sequence. The sequences of these oligonucleotides are as follows: oligonucleotide 3, 5Ј-GCAGAT-GTCCCGGCGAAGAG-3Ј; and oligonucleotide 4, 5Ј-CTCTTCGCCGG-GACATCTGC-3Ј. An aliquot of cDNA was amplified in a 50-l reaction volume with 200 ng of each primer, 200 M dNTPs, and 2.5 units of Goldstar Taq DNA polymerase (Eurogentec) or Ampli-Taq (Roche Molecular Biochemicals) in buffer containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl 2 , and 0.001% gelatin. Polymerase chain reaction amplification was performed in a DNA thermal cycler (Perkin-Elmer) using the following parameters: 30 s at 95°C, 1 min at 55°C, and 1 min at 72°C for 30 cycles, followed by an extra cycle with a 10-min extension step at 72°C. Expected fragments were purified on agarose gels and cloned in the pTAG vector (R&D Systems Europe Ltd.) following the manufacturer's protocol. The different inserts were analyzed by restriction enzymes and partially sequenced using the Universal, T7 primer, or specific oligonucleotides.
Plasmid Constructs-The coding region corresponding to Nhe3 (35) was subcloned in the pECE vector (36) at the KpnI site to obtained the pECE/Nhe3 vector. An EcoRI fragment of 1700 base pairs obtained from the pTAG vector containing the entire 3Ј-UTR was subcloned within the pECE/Nhe3 vector at the 3Ј-end of the corresponding Nhe3 region. Then, this construct was cut with BstXI and XbaI, and a fragment of 1130 base pairs obtained from the pTAG vector cut with the same enzymes was ligated to obtain the Nhe3 WT/TOT 3Ј-UTR. The Nhe3 3Ј-UTR TOT vector was obtained by removal of the polyadenylation signal of SV40 provided by the pECE vector by cutting the Nhe3 WT/ TOT 3Ј-UTR with XbaI and NaeI, blunting with Klenow, and ligating. The 3Ј-UTR SspI construct was obtained by cutting the Nhe3 WT/TOT 3Ј-UTR with SspI and NaeI, blunting with Klenow, and religating. The 3Ј-UTR SpeI vector was obtained by cutting the Nhe3 3Ј-UTR TOT vector with SpeI and SmaI, blunting with Klenow, and ligating. The 3Ј-UTR PmlI/SspI vector was obtained by cutting the 3Ј-UTR SspI construct with PmlI and SmaI and ligating. The plasmids containing the VEGF 3Ј-UTR used in EMSA and RNA degradation assays contained the rat VEGF 3Ј-UTR and have been previously described (12). For generating the StyI construct, the vector was digested with HindIII (polylinker of the pSP64 vector) and StyI, blunted with Klenow, and ligated.
Preparation of RNA-Cells were washed in ice-cold PBS and lysed in RNA Insta-Pure buffer (Eurogentec). The supernatant was cleared by centrifugation, isopropyl alcohol-precipitated, and resuspended in sterile water. In mRNA stability experiments, cells were treated with 5 g/ml actinomycin D for the times indicated on the figures. For determination of mRNA half-life, 18 S ribosomal RNA was used as an internal standard. The entire cDNA for mouse VEGF (kindly provided by Dr. Werner Risau) or Nhe3 was used as a probe in Northern blot experiments.
Preparation of Cytosolic Extracts, RNA Electromobility Shift Assays, and in Vitro mRNA Stability Assay-CCL39 cells and PS200 hemagglutinin-tagged JNK-and p38/HOG-containing clones were grown to confluence and stimulated with 500 ng/ml anisomycin for different periods of time. Cells were then rinsed with ice-cold PBS and lysed as described previously (34). RNA transcripts were synthesized from linearized pSP64 poly(A) 3Ј-UTR template (Promega) (12) using the SP6 bacteriophage RNA polymerase with the RNA transcription kit (Promega) according to the manufacturer's protocol. Transcription reactions were treated with RNase-free DNase (Stratagene) for 15 min at 37°C, extracted with phenol/chloroform, and isopropyl alcohol-precipitated. In vitro RNA stability assays were performed as described previously (12). Briefly, capped, polyadenylated, and radiolabeled RNA transcripts were incubated with 130 g of cytoplasmic extracts. At each time point, the reaction was stopped by the addition of ammonium acetate and EDTA. mRNA were extracted with phenol/chloroform and precipitated with isopropyl alcohol. Samples were electrophoresed on a formaldehyde-agarose gel and transferred to nylon membrane. Quantification of the undegraded transcript was analyzed with the use of a Fuji phosphoimaging system. For EMSA experiments, capped, polyadenylated, and radiolabeled RNA transcripts were combined with protein extracts (10 g) in binding buffer as described (12). The reaction mixture was incubated for 15 min at 30°C and treated for 15 min at room temperature with 40 units of ribonuclease T1. When competitors were used, they were incubated for 15 min with the proteins in binding buffer before the addition of the radiolabeled transcripts. The reaction mixtures were resolved on native polyacrylamide gels (30% acrylamide and 0.8% bisacrylamide) in 0.5ϫ Tris borate/EDTA buffer. Gels were dried and autoradiographed.
Precipitation with GST Fusion Protein, in Vitro Kinase Assays, and Immunoblotting-Cell cultures were washed with ice-cold PBS and lysed in Triton lysis buffer (37). For GST fusion protein precipitation, 5 g of GST-Jun was added per sample, followed by 20 l of glutathione-Sepharose. For kinase assays, 4 l of total antiserum directed against murine p38/HOG (38) 2 was preincubated with 20 l of protein A-Sepharose before the addition of the extract. Precipitates were washed four times with lysis buffer plus an additional wash with kinase buffer (37) and then resuspended in 40 l of kinase reaction buffer for 15 min. Samples were boiled in Laemmli buffer and resolved on SDS-polyacrylamide gels. The radioactivity incorporated into GST-Jun or GST-ATF2 (the p38/HOG substrate) was analyzed with the use of the Fuji phosphoimaging system. For immunoblotting, 60 g of proteins was run on a 10% SDS-polyacrylamide gel and transferred to an Immobilon membrane (Millipore Corp.), and the membrane was incubated overnight with the appropriate primary antibody at 4°C. The protein was decorated with an anti-rabbit horseradish peroxidase-conjugated secondary antibody and developed using the ECL system (Amersham Pharmacia Biotech).
Determination of Protein Synthesis-Exponentially growing CCL39 cells were seeded in 12-well plates and incubated with [ 3 H]leucine (L-[2,3,4,5-3 H]leucine; 2 mCi/ml, 0.8 mM final concentration) in the absence or presence of anisomycin or cycloheximide at different concentrations. After 6 h of stimulation, cells were fixed and washed twice with ice-cold trichloroacetic acid (5%). The precipitated material was solubilized with 0.1 N NaOH, and the radioactivity was subjected to liquid scintillation counting.

RESULTS
Anisomycin Induces an Increase in VEGF mRNA Expression-To identify if SAPKs play a role in the signaling pathway leading to VEGF mRNA expression in CCL39 cells, we studied the effects of a strong activator of these kinases, anisomycin. As shown in Fig. 1, VEGF mRNA was rapidly induced by anisomycin and reached a maximum after 7 h of stimulation. Such an induction could be obtained with a low concentration of anisomycin (5 ng/ml) for the CCL39 cells (Fig. 2) and PS200 cells (data not shown). However, the maximal induction was obtained at higher concentrations in all cell lines tested (CCL39 and its derivatives PS120 and PS200, HeLa cells, 9L cells, and pro-B cells) ( Fig. 2 and data not shown). As anisomy-2 A. Brunet, unpublished data.
FIG. 1. Anisomycin induces VEGF mRNA expression. 30 g of total RNA isolated from exponentially growing or anisomycin-stimulated (200 ng/ml) CCL39 cells for the times indicated was analyzed by Northern blotting for expression of VEGF. 28 S ribosomal RNA is shown as a loading control. The specific signal for VEGF was quantified with the use of a phosphoimaging system (lower panel). This experiment is representative of three independent experiments. cin is a strong inhibitor of protein synthesis (Fig. 2b), we have compared the effects of anisomycin and cycloheximide, which was previously shown to induce VEGF mRNA (39,40). Fig. 2a shows that in our cell system, only very high concentrations of cycloheximide (total protein synthesis inhibition was obtained at 10 g/ml) could slightly induce VEGF mRNA. This result shows that the anisomycin-dependent induction of VEGF cannot be completely attributable to its effect on protein synthesis inhibition.
Anisomycin Increases VEGF mRNA Stability-As anisomycin has no effect on VEGF gene transcription (26), we suspected that it could act at the post-transcriptional level by increasing VEGF mRNA stability. To demonstrate this effect, we performed a decay assay in the presence of actinomycin D. Consistent with our hypothesis, removal of anisomycin resulted in a rapid decay of VEGF transcripts (Fig. 3a, lanes 1-3), whereas the maintenance of anisomycin induced a persistence in VEGF transcript levels (lanes 5-7). Under this last condition, anisomycin increased the VEGF mRNA half-life from 30 min to Ͼ4 h. Thus, anisomycin increases VEGF mRNA stability by blocking its degradation.
SAPKs p38/HOG and JNK Play a Key Role in the Stabilization of VEGF mRNA by Anisomycin-To analyze the implication of SAPKs in the induction of VEGF mRNA by anisomycin, we used a specific inhibitor of p38/HOG and JNK, SB202190 (32,33). At a concentration of 20 M, this drug inhibited the phosphorylation of c-Jun and ATF2, which are the direct targets of JNK and p38/HOG, respectively (Fig. 4a). We then evaluated the ability of SB202190 to block VEGF mRNA stabilization by anisomycin. Indeed, pretreatment of the cells with SB202190 totally abolished the effect of anisomycin (Fig. 4b). This result demonstrates that anisomycin stabilizes VEGF mRNA through the activation of SAPKs. To confirm the implication of the SAPKs in VEGF mRNA stabilization, we produced cells overexpressing p38/HOG or JNK. Fig. 5a shows that basal and anisomycin-stimulated JNK and p38/HOG activities were significantly increased in these cells. We next analyzed the level of VEGF mRNA in these cells. Fig. 5b shows that not only the basal, but also the anisomycin-induced VEGF mRNA levels were increased in stably transfected cells compared with control cells. VEGF mRNA levels were 5-and 10fold higher in these cells than in their transfected controls, respectively, after anisomycin stimulation (200 ng/ml). This induction was particularly elevated in the JNK-expressing clone tested, which expressed 5-10-fold more JNK than the control cells. Such induction is directly dependent on the JNK expression level (data not shown). Then, we measured the VEGF mRNA half-life in these cells in the absence of anisomycin. After 5 h of incubation with anisomycin to reach a high level of VEGF mRNA, we substituted anisomycin with actinomycin D for the indicated periods of time. As expected, overexpression of SAPK was sufficient to slow down the VEGF mRNA degradation (Fig. 6), a result reflecting the increased basal levels of p38/HOG and JNK. Quantification of the transcript levels showed that the VEGF mRNA half-life was 30 min in control cells, 1 h and 30 min in p38/HOG-transfected cells, and 2 h and 30 min in JNK-transfected cells (Fig. 6). These results are in agreement with those of Li et al. (41), who observed a similar increase in the VEGF mRNA half-life in rat aortic smooth muscle cells after interleukin-1␤ stimulation, another p38/HOG and JNK activator. Anisomycin Exerts Its Stabilization Effect through the VEGF 3Ј-UTR-Most of the elements implicated in VEGF mRNA turnover have been shown to be located in the 3Ј-UTR (12,16,34). Thus, we wondered whether anisomycin is able to induce VEGF 3Ј-UTR stabilization in an in vitro degradation assay. Fig. 7 shows that [ 32 P]UTP-labeled, capped, and polyadenylated in vitro transcribed RNA containing the full-length VEGF 3Ј-UTR had a significantly longer half-life when incubated with anisomycin-treated cell extracts versus unstimulated cell extracts (ratio of 1.7 Ϯ 0.2, n ϭ 12). This protection is even more important with cellular extracts issued from p38/HOG-and JNK-expressing cells either treated or not with anisomycin (data not shown). These results highlight the importance of these protein kinases in anisomycin-mediated VEGF mRNA stabilization. These results also suggest that the effects observed in vivo are mediated through the 3Ј-UTR.
VEGF mRNA 3Ј-UTR Confers Instability to a Heterologous Transcript-We next wanted to study whether the VEGF 3Ј-UTR will confer anisomycin sensitivity to a stable heterologous transcript. The entire VEGF 3Ј-UTR or an appropriate deletion of this 3Ј-UTR was fused to the Na ϩ /H ϩ exchanger reporter gene (Nhe3 isoform) (Fig. 8a). Since transcription of all the chimeric constructs was driven by the vector containing the SV40 promoter, differences at the chimeric mRNA levels could be attributed to fused VEGF 3Ј-UTR sequences. Fig. 8a recapitulates the different constructs obtained by deletion on the 3Ј-UTR. The chimeric constructs WT, 3Ј-UTR TOT, 3Ј-UTR SspI, 3Ј-UTR SpeI, and 3Ј-UTR PmlI/SspI were stably transfected into CCL39 cells. Expression of transfected constructs was assayed by the ability of Nhe3 to confer acid load cell resistance in the presence of amiloride derivatives (31). This method of selection is particularly interesting in our study since low amounts of Nhe3 are sufficient to confer resistance to the acid load test. Therefore, it allows selection of cells expressing very low levels of chimeric mRNA, a result anticipated by grafting the VEGF 3Ј-UTR. As shown in Fig. 8b, in the absence of any stimulus, Nhe3/VEGF 3Ј-UTR TOT chimeric mRNA amounts were strongly reduced compared with the Nhe3 mRNA. This result is in favor of a constitutive degradation mechanism through the instability elements contained within the 3Ј-UTR (12). This result is best illustrated after treatment with actinomycin D. As was the case for the endogenous VEGF mRNA, the chimeric Nhe3/VEGF 3Ј-UTR TOT mRNA was rapidly degraded. However, control Nhe3 mRNA was very stable as judged by the absence of degradation in the presence of actinomycin D (Fig. 8b). It is important to note that we could also detect different sizes of chimeric RNA after hybridization with the Nhe3 probe. This is probably due to the utilization of different polyadenylation signals present in the VEGF 3Ј-UTR. We have also measured the half-lives of the chimeric RNAs containing different regions of the VEGF mRNA 3Ј-UTR. SspI and SpeI constructs have half-lives that are similar to chimeric RNA containing the entire 3Ј-UTR (Fig. 8c). These results are in agreement with recent reports that have shown that sequences present in both constructs contain instability elements similar to those previously found in c-fos mRNA (12,13,34).
Anisomycin Increases Chimeric mRNA Stability by Stimulating SAPK-To evaluate whether the activation of SAPK by anisomycin can protect chimeric mRNA from degradation, we were transfected with p38/HOG and JNK cDNAs, respectively. Functional expression of transfected constructs was examined regarding their ability to phosphorylate ATF2 for p38/HOG or c-Jun for JNK (see "Experimental Procedures"). Kinase activity measured by 32 P incorporation in specific substrates was quantified with the use of a phosphoimaging system. Presented are the means of three independent experiments. b, the effects of both kinases on VEGF mRNA expression were analyzed by Northern blotting. Cells were either untreated or treated with anisomycin (A, ANISO; 200 ng/ml) for 5 h. 18 S ribosomal RNA is shown as a loading control (CONT.). Note that the time of exposition of the blot was 7 h instead of overnight as for the blots in Fig. 1 and 2. transfected the 3Ј-UTR TOT construct in control, p38/HOG. and JNK stable transfectants. Fig. 9 shows that anisomycin increased chimeric mRNA stability in control cells. This effect was more pronounced in the p38/HOG-and JNK-transfected cells. As described above for endogenous VEGF mRNA, these cells conferred mRNA stabilization in the absence of any stimulation, which reflects the basal p38/HOG or JNK activity. Interestingly, the stability of chimeric mRNA was more intense in cells overexpressing JNK, as was the case for endogenous VEGF mRNA.
Target Region for SAPK Can Form a Hairpin Structure-It has been shown that JNK exerts its regulatory effect on interleukin-3 mRNA stabilization through an AUUUA-rich region (33). We therefore investigated the effect of anisomycin on a region of the VEGF 3Ј-UTR containing such sequences. We found five AUUUA regions in the VEGF 3Ј-UTR, but the one comprised between the PmlI and SspI sites (Fig. 8a) seems to be particularly interesting. By using an RNA-folding algorithm, MUFOLD (42,43), we showed that this region can form a very stable hairpin structure (Fig. 10a). The folding energy of the stem-loop formed was Ϫ12.8 kcal/mol. Within this region, there exists a "bubble" containing the UUAUUUA(U/A)(U/A) consensus sequence that corresponds to an instability domain equivalent to those described in c-fos, c-myc, and granulocyte/ macrophage colony-stimulating factor mRNAs (44). This region also contains another AUUUA sequence, but it does not fit the exact consensus described above. Furthermore, Ming et al. (33) and Winzen et al. (45) have shown that such sequences are the target of JNK-and p38/HOG-mediated stabilization of interleukin-3, -6, and -8 mRNAs. Since the presence of secondary structure was previously shown to be required for protein recognition (34,46), we tested whether this domain can confer anisomycin sensitivity to the Nhe3 heterologous transcript. Fig. 10b shows that the chimeric mRNA containing this short domain was almost undetectable under basal conditions. This result suggests that the PmlI/SspI region is sufficient to confer mRNA instability. However, following anisomycin stimulation, it was highly induced and even more stable than a chimeric mRNA containing the total VEGF 3Ј-UTR (half-life of 2 h and 30 min instead of 1 h). These results allow us to identify at least a minimal target region for anisomycin-induced VEGF mRNA stabilization.
Anisomycin Stimulation Induces Protein Binding to the Target Region-The total in vitro transcribed VEGF 3Ј-UTR and a subdomain containing the stem-loop described above were incubated with S100 extracts to allow identification by EMSA of anisomycin-induced VEGF mRNA-binding proteins. Fig. 11 shows that anisomycin-inducible protein complexes bound to the full-length (FL) and StyI probes. These complexes cannot FIG. 6. Anisomycin mediates VEGF mRNA stability by stimulating p38/HOG and JNK transduction pathways in vivo. a, control (CONT) and p38/HOG-and JNK-transfected cells were stimulated with anisomycin (ANISO; 200 ng/ml) for 5 h, washed with PBS, and then incubated with actinomycin D (ActD; 5 g/ml) for the times indicated. 18 S ribosomal RNA is shown as a loading control. This experiment is representative of three independent experiments. b, shown is the quantification of the signals shown in a by phosphoimaging, where the 18 S ribosomal RNA-normalized values at time 0 were taken as 100%.

FIG. 7. Anisomycin exerts its effect on VEGF mRNA stability through its 3-UTR: importance of the stress kinase pathways.
[ 32 P]UTP-labeled, capped, and polyadenylated VEGF mRNA 3Ј-UTR transcript was generated in vitro as described under "Experimental Procedures." It was incubated with S100 cytoplasmic extracts from control CCL39 cells stimulated or not with anisomycin (ANISO; 500 ng/ml) for the times indicated. After incubation, the remaining mRNA was treated as described under "Experimental Procedures," run on a denaturing formaldehyde-agarose gel, transferred to nylon membrane, and exposed on x-ray film. This experiment is the mean of four independent experiments. a, representative autoradiograph of products from a cell-free degradation assay. Times refer to the times after the addition of the extracts to the RNA. The arrows show the positions of the undegraded transcript. b, log linear regression lines of VEGF RNA 3Ј-UTR degradation quantitated by a phosphoimaging system. The remaining total mRNA amounts were quantified using a phosphoimaging system. mRNA half-lives was deduced from the regression line (t1 ⁄2 for control unstimulated cells (ϪANISO) ϭ 7 min; t1 ⁄2 for control stimulated cells (ϩANISO) ϭ 11.6 min). be visualized in the presence of an excess of unlabeled probe (data not shown). The region delimited by the StyI site in the rat VEGF mRNA 3Ј-UTR corresponds to the same region delimited by the PmlI and SspI sites in the human VEGF mRNA 3Ј-UTR. This domain presents 93% homology between the two species and can form a hairpin structure in the human as well rat VEGF 3Ј-UTRs. This result points out the importance of the stem-loop structure in the anisomycin-mediated mRNA stabilization.

DISCUSSION
In this work, we have identified the signal transduction pathways implicated in the stabilization of VEGF mRNA in response to anisomycin. Anisomycin has been shown to be an inhibitor of protein synthesis as well as a strong activator of the SAPKs JNK and p38/HOG (28,38,47,48). In this study, we show that anisomycin induces VEGF mRNA independently of an effect on protein synthesis. Therefore, the increase in VEGF mRNA must be mediated through p38/HOG and JNK activation. Indeed, VEGF mRNA amounts are higher in cells overexpressing p38/HOG and JNK, and this effect is enhanced following anisomycin stimulation. The fact that SB202190, a strong inhibitor of p38/HOG and JNK (32,33), inhibits anisomycindependent VEGF mRNA induction confirms the essential role played by both kinases in the regulation of VEGF mRNA stabilization. We also provide evidence that anisomycin exerts its effects by increasing the stability of VEGF mRNA. Our data are FIG. 8. VEGF mRNA 3-UTR confers rapid decay to a reporter construct. a, the different constructs used are shown. The Nhe3 gene is inserted in the pECE vector containing the SV40 polyadenylation signal (SV) to obtain the Nhe3 WT vector. The SV40 polyadenylation signal was replaced by the VEGF mRNA 3Ј-UTR (3Ј-UTR TOT) or two overlapping regions by using two different restriction sites (SspI or SpeI) to obtain the 3Ј-UTR TOT, 3Ј-UTR SspI, and 3Ј-UTR SpeI constructs, respectively. The PmlI/SspI construct was obtained as indicated under "Experimental Procedures." The sequence forming a hairpin structure is indicated. b, the Nhe3 WT or 3Ј-UTR TOT constructs were transfected into CCL39 cells. 20 g of total mRNA isolated from transfected cells treated for the times indicated with actinomycin D (Act D; 5 g/ml) was analyzed for the expression of the chimeric mRNAs in a Northern experiment. Ectopic gene expression was tested using an Nhe3 probe. 18 S ribosomal RNA is shown as a loading control. c, 20 g of total RNA isolated from cells transfected with Nhe3 WT, 3Ј-UTR TOT, SspI, or SpeI and treated or not with actinomycin D for 3 h was analyzed in a Northern experiment. Expression of ectopic genes was tested using an Nhe3 probe. 18 S ribosomal RNA is shown as a loading control.
FIG. 9. Expression of p38/HOG and JNK increases stability of the chimeric Nhe3 mRNA 3-UTR. a, control (CONT) and p38/HOGor JNK-expressing cells were transfected with the Nhe3 3Ј-UTR construct. They were treated (ϩ) or not (Ϫ) for 5 h with anisomycin (ANISO; 200 ng/ml), rinsed twice with PBS, and then incubated for the times indicated with actinomycin D (Act D). 20 g or total mRNA isolated from the different cells tested was analyzed for the expression of the chimeric mRNA in a Northern experiment. Expression of ectopic genes was tested using an Nhe3 probe. 18 S ribosomal RNA is shown as a loading control. b, shown is the quantification by phosphoimaging of the signals shown in a for the anisomycin-treated cells, where the 18 S ribosomal RNA-normalized values at time 0 were taken as 100%.
compatible with those of White et al. (49), who have shown that VEGF mRNA is stabilized by UV radiation, one of the strongest activators of JNK.
We then focused our attention on the 3Ј-UTR of VEGF mRNA, which contains cis-active sequences previously shown to be important for mRNA stabilization in response to hypoxia (12,13,34), and asked whether the 3Ј-UTR is the target for regulation by SAPKs. By in vitro degradation assays, we demonstrated that anisomycin also exerts its effect via sequences present in the VEGF 3Ј-UTR region. We also demonstrated that the anisomycin-as well as SAPK-mediated VEGF mRNA regulation through the 3Ј-UTR can be conferred to a heterologous gene. In this context, we have shown that a chimeric Nhe3 mRNA containing the entire VEGF mRNA 3Ј-UTR is rapidly degraded, as was the case for the VEGF mRNA. We also confirmed the presence of sequences implicated in rapid degradation of VEGF mRNA by using overlapping chimeric constructs. Such data are compatible with the findings of Levy et al. (12,13), who have shown by an independent approach the presence of such cis-active sequences within the 3Ј-UTR of VEGF mRNA. As was the case for the endogenous VEGF mRNA, p38/HOG and JNK participated in chimeric mRNA stabilization upon anisomycin stimulation. Our results are compatible with those of Chen et al. in the VEGF mRNA 3Ј-UTR, and one of them is situated in a region of the 3Ј-UTR that can form a very stable hairpin structure. Using the chimeric constructs, we show that this domain alone is sufficient to confer remarkable instability to the heterologous Nhe3 transcript as well as anisomycin-induced stabilization. Finally, we have shown by EMSA experiments that this region recruits protein(s) upon anisomycin stimulation. This domain is highly conserved between the rat and human sequences, whereas a lot of divergence usually exists in the 3Ј-UTR of mRNAs. The fact that this sequence is conserved in both species points out its importance in the post-transcriptional regulation of VEGF. However, it is not excluded that other regions participated in the regulation of VEGF mRNA stabilization by anisomycin. The AUUUA sequence is the target of the already described AUF1 proteins, which are responsible for rapid mRNA degradation. Nakamaki et al. (50) have shown that their binding is independent of any stimulation of the granulocyte/macrophage colony-stimulating factor mRNA 3Ј-UTR. Recently, Sirenko et al. (51) showed that their binding to the GRO␣ mRNA is a phosphorylation-dependent mechanism sensitive to protein kinase inhibitors. It will be interesting to test whether AUF1 proteins can interact with the hairpin structure and if they are direct targets for phosphorylation by p38/HOG or JNK. Identification of the in vivo VEGF 3Ј-UTR-interacting proteins will add to the knowledge of the regulation of VEGF expression during stress situations. FIG. 10. Localization of an anisomycin response element in the VEGF 3-UTR. a, shown is a representation of the stem-loop structure contained within the anisomycin response element as determined by MUFOLD (42,43). The potential RNA loop structure energy is indicated. b, cells stably transfected with the 3Ј-UTR PmlI/SspI construct were untreated (lanes 1-3) or treated (lanes 4 -8) for 5 h with anisomycin (ANISO; 200 ng/ml), rinsed twice with PBS, and then incubated for the times indicated with actinomycin D (Act D). 20 g or total mRNA isolated from the different cells tested was analyzed for the expression of the chimeric mRNA in a Northern experiment. Ectopic gene expression was tested using an Nhe3 probe. 18 S ribosomal RNA is shown as a loading control.
FIG. 11. Identification of anisomycin-inducible RNA-protein complexes by EMSA. RNA EMSA using the full-length (FL) or StyI fragment as a template was performed as described under "Experimental Procedures." The plus and minus signs indicate extracts from cells stimulated or not with anisomycin (ANISO), respectively. The arrows point to anisomycin-inducible complexes. The brackets encompass the free and degraded probe.