Accumulation of sequence-specific RNA-binding proteins in the cytosol of activated T cells undergoing RNA degradation and apoptosis.

Engagement of the T cell antigen receptor (TCR) causes transformed T cell hybridomas to produce lymphokines and then die by an apoptotic mechanism. Here we show that these functional effects of TCR-mediated signaling are associated with the accumulation of several cytosolic mRNA-binding proteins including the previously described AU-A, -B, and -C proteins as well as a novel 70-kDa protein. The results indicate that the 70-kDa protein derived from a 90-kDa precursor present in unactivated T cells and that both proteins bound to two independent sites at the 3′-end of the interleukin-2 mRNA, one in the coding region and the other in an AU-rich segment of the 3′-untranslated region. Glucocorticoids, TCR engagement by monoclonal antibodies, pharmocologic mimics of TCR signaling, or high concentrations of protein or RNA synthesis inhibitors all induced apoptosis, the cytosolic appearance of the RNA-binding proteins, and an increased rate of RNA turnover. Moreover, drugs that interfere with TCR-mediated signals, such as cyclosporin A and staurosporin, prevented both apoptosis and the appearance of the RNA-binding factors. The fact that the accumulation of these factors occurred in the presence of inhibitors of transcription and translation suggested that these proteins are present in an inactive form in unstimulated T cells and are activated when apoptosis is induced.

Several hormonal and developmental signals have been shown to modulate gene expression by altering mRNA stability or mRNA degradation (1). Post-transcriptional regulation of mRNA expression has been shown to play an important role in controlling the level of rapidly induced mRNAs. This level of regulation is particularly important for proteins that are active for brief periods, such as growth factors, transcription factors, and proteins that control cell cycle progression (1). mRNA stability is controlled by specific cis-acting elements including an AU-rich sequence that is found in the 3Ј-untranslated region (UTR) 1 of a variety of unstable mRNAs (2) and has been shown to confer instability to otherwise stable RNAs (3). The AU-rich element (ARE) consists of multiple copies of the AUUUA pentanucleotide and/or uridine-rich segments. These elements are thought to target mRNA to rapid degradation by binding proteins involved in RNA turnover (4), and several AU-specific, RNA-binding proteins, have been described previously (5,6,7). In human T cells, Bohjanen et al. (8,9), reported that T cell antigen receptor (TCR)-mediated signals induced three different AU-specific RNA-binding proteins: AU-A (34 kDa), AU-B (30 kDa), and AU-C (43 kDa). Phorbol myristate acetate, known to increase lymphokine mRNA stability (10), inhibited the induction of AU-B, providing a correlation between the binding of this protein and increased mRNA turnover (8).
Apoptotic cells have an increased rate of RNA turnover, and it has been suggested that RNA degradation could play a role in the progression to cell death (11,12). In the immune system, apoptosis is thought to be involved in the elimination of potentially destructive, self-reactive immature lymphocytes (14) and excess effector lymphocytes that are generated in primary immune responses (13,14,15). Apoptosis can be induced experimentally in T lymphocytes by withdrawal of growth factors, glucocorticoid treatment, heat shock, and UV-induced DNA damage or, in dividing cells, as a consequence of antigen receptor-mediated activation (13,16). An early marker of the death program is the activation of an endonuclease that cleaves chromosomal DNA between nucleosomes, yielding DNA multimers of 200 base pairs. DNA fragmentation is then followed by a variety of morphological changes that include condensation of the cytoplasm, swelling of the endoplasmic reticulum, plasma membrane blebbing, nuclear shrinkage, and the formation of dense chromatin masses known as apoptotic bodies. This process eventually leads to cell lysis.
Apoptosis can be induced in vitro in actively proliferating murine T cell hybridomas by exposure to antigen or mAb directed against the TCR or the CD3 components of the TCR (17). In this system, TCR signaling initially results in a G 1 /S cell cycle block followed by genomic DNA fragmentation (18). At low concentrations, drugs that inhibit transcription and translation prevent the TCR-induced DNA fragmentation, suggesting that the expression of new gene products is required for TCR-induced apoptosis (19). However, at higher concentrations, inhibitors of transcription and translation cause DNA fragmentation directly (18) through an unknown mechanism.
Here we show that in T cell hybridomas, TCR-generated signals result in the cytosolic accumulation of several previously described AU-specific RNA-binding proteins and a novel 70-kDa protein that is capable of binding the AU-rich element within the IL-2 3Ј-UTR as well a second site within the IL-2 coding region. The coordinate appearance of these RNA-binding proteins correlated with a shorter half-life of IL-2 mRNA, an increased rate of RNA degradation, and the induction of apoptotic cell death. It is therefore possible that RNA-binding proteins are involved in apoptosis-related RNA degradation.

EXPERIMENTAL PROCEDURES
Cells and Reagents-The A.AE7.2 T cell hybridoma was obtained by fusing the pigeon cytochrome c-specific T cell clone A.E7 to the TCR Ϫ BW 1100 thymoma as described previously (20). The cells were maintained in Dulbecco's modified Eagle's medium (Celox, Hopkins, MN) supplemented with 10% heat-inactivated fetal calf serum, 1 mM HEPES, nonessential amino acids, 2 mM sodium pyruvate, 2 mM Lglutamine, 100 mg/ml streptomycin, 100 units/ml penicillin, 20 mg/ml gentamicin, and 50 M ␤-mercaptoethanol at 37°C in a 10% CO 2 atmosphere. The nontransformed A.E7 T cell clone was maintained in Eagle's Hank's amino acid (EHAA) medium (Celox, Hopkins, MN) supplemented with 10% fetal calf serum by periodic stimulation with pigeon cytochrome c and splenic antigen-presenting cells as described previously (21). A.E7 cells were used in experiments at least 10 days after their last exposure to antigen and were in the G 0 /G 1 stage of the cell cycle at this time.
Northern Blot and mRNA Decay-Total RNA was isolated from 5 ϫ 10 6 cells using RNAzol B (Tel-Test, Inc., Friendswood, TX). Typically 20 g of RNA was run on a 1.2% formaldehyde-formamide gel, transferred to a nylon membrane, and UV cross-linked to the membrane using a Stratalinker device (Stratagene, La Jolla, CA). cDNA probes were labeled with [␣-32 P]dCTP with the Prime-It II random primer kit (Stratagene). Unincorporated nucleotides were removed on a Sephadex G-50 column (Nuc Trap purification columns, Stratagene). Denatured probe was hybridized to the filter-bound RNA according to the QuickHyb protocol (Stratagene). Following washing, the membrane was exposed to x-ray film with an intensifying screen, and the specific bands were detected by autoradiography. To measure the half-life of IL-2 mRNA, cells were incubated for 3, 6, or 8 h on immobilized anti-CD3 mAb, and then transcription was blocked by the addition of 0.2 g/ml of cyclosporin A (22). Total RNA was then isolated after 20, 40, 80, and 120 min and analyzed by Northern blot. The intensity of the bands present on autoradiograms was determined by densitometry and image analysis using the Image 1.47 program. The values for IL-2 mRNA were normalized to values for ␤-actin mRNA, the expression of which did not change throughout the experiment.
Preparation of Cytosolic Extracts-Cells (1-2 ϫ 10 6 /well) were incubated in 1 ml of medium at 37°C for 2-24 h in 24-well plates that were coated (5 g/ml of purified antibody for 1 h at 37°C) with a mAb (clone 145-2C11) specific for the CD3-⑀ component of the TCR. In some ex-periments, actinomycin D, DRB, cycloheximide, emetine (all obtained from Sigma), dexamethasone sodium phosphate (American Reagents Laboratories, Shirley, NY), or cyclosporin A were added at the time the cultures were initiated. After the stimulation period, the cells were collected, washed with ice-cold phosphate-buffered saline, and lysed in 50 l of 10 mM HEPES (pH 7.9), 40 mM KCl, 3 mM MgCl 2 , 1 mM dithiotreitol, 5% glycerol, 0.2% Nonidet P-40, 1 g/ml aprotinin, 1 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride for 5-10 min on ice. Nuclei were removed by centrifugation at 14,000 rpm for 20 s in an Eppendorf microcentrifuge. The nuclei were then lysed in 25 l of a hypertonic buffer containing 20 mM HEPES (pH 7.9), 420 mM KCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, 25% glycerol, 1 g/ml aprotinin, 1 g/ml leupeptin, and 0.5 mM phenylmethylsulfonyl fluoride for 30 min in ice and then centrifuged for 15 min at 14,000 rpm at 4°C to remove the insoluble fraction. Cytosolic and nuclear extracts were then frozen and stored at Ϫ70°C. The protein concentration of each extract was determined by the Bradford assay (Bio-Rad, Hercules, CA).
RNA Probes and in Vitro Transcription-An RNA oligonucleotide encompassing the ARE of the mouse IL-2 3Ј-UTR was chemically synthesized and labeled by an in vitro kinase reaction in the presence of [␥-32 P]ATP (specific activity, 7,000 Ci/mmol, ICN). The 32 P-labeled oligonucleotide was separated from unincorporated nucleotides by chromatography on a Nuctrap G-50 column (Stratagene). The mouse c-myc cDNA template was obtained by annealing two single-stranded oligonucleotides encompassing the previously described AU-A-binding region (8) and subcloning the resulting double-stranded oligonucleotide into the pGEM-3 vector. The oligonucleotides were designed with a PstI site at the 3Ј-end that was used to linearize the construct for in vitro transcription. The HindIII-PstI fragment of the mouse IL-2 cDNA, or polymerase chain reaction products containing progressive deletions at the 5Ј-or 3Ј-ends of this fragment, were subcloned into the pGEM-3 vector. Recombinant IL-2 plasmids were linearized with SpeI, SspI, RsaI, TaqI, and DdeI for in vitro run-off transcription reactions that were performed using the SP6 RNA polymerase according to the Promega (Madison, WI) protocol. Labeled RNA transcripts with a specific activity of 0.2-0.6 ϫ 10 9 cpm/g were generated by inclusion of [␣-32 P]UTP (800 Ci/mmol, Amersham Corp.) in the transcription reaction. Unlabeled RNAs used in the competition assays were transcribed with the RiboMax Promega kit, and the concentration was determined by spectrophotometric analysis. The RNA probes used in these FIG. 1. Schematic representation of the RNA probes. Transcripts spanning different regions of IL-2 mRNA were obtained by SP6 polymerase-mediated in vitro run-off transcription from recombinant plasmids (described under "Experimental Procedures") linearized by the enzymes indicated in the figure. Whenever the IL-2 plasmid was linearized with SpeI or SspI, RNAs containing 4 or 2 AUUUA sequences were transcribed. RNA probes with no AUUUA sequences were obtained when the recombinant plasmids were linearized with TaqI, DdeI, or RsaI. The RNA IL-2 oligonucleotide was chemically synthesized and is restricted to the 3Ј-UTR ARE (indicated as the shadowed box). The c-myc cDNA was obtained by subcloning a double-stranded DNA oligonucleotide corresponding to the sequence that was previously shown to bind AU-A studies are shown in Fig. 1.
RNA-Protein UV Cross-linking Assay-5 g of cytosolic extracts were incubated with 32 P-labeled RNA (10 5 cpm) in the presence of 0.5 mg/ml heparan sulfate (to reduce nonspecific binding) in a total volume of 10 l in a 1.5-ml microcentrifuge tube at room temperature for 10 min. The IL-2 RNA probe encompassing the HindIII-SpeI sites was used in UV cross-linking experiments unless otherwise indicated. Subsequently, the RNA-protein complexes were cross-linked in ice for 5 min with 254-nm UV radiation in a Stratalinker cross-linking apparatus (Stratagene). RNase A was then added to a final concentration of 1 mg/ml, and the reaction mixtures were incubated for 30 min at 37°C to allow digestion of unbound RNA. The samples were boiled in the presence of an equal volume of 2ϫ Laemli buffer (23) for 5 min and separated by SDS-polyacrylamide gel electrophoresis on a 10% polyacrylamide gel. The gel was run for 3-4 h at 30 mA, dried, and exposed to Kodak film at Ϫ70°C with an intensifying screen for 12-24 h.
V8 Protease Digestion-The RNA-protein complexes were covalently linked by UV cross-linking as described above. The specific bands were identified by autoradiography, and the corresponding gel slices were excised, rehydrated in 50 l of Cleveland buffer (50) and cut into small pieces. The pieces were then loaded on a 15% SDS-polyacrylamide gel, and the proteins were allowed to enter the gel by electrophoresis. 1, 10, or 100 ng of V8 protease in 50 l of Cleveland buffer (50) were then loaded into the wells, and the gel was run until the samples reached the interface between the stacking and the running gel. At this point, the run was stopped for 30 min to allow the V8 protease digestion to occur. The gel was then run, dried, and exposed to x-ray film as described above.
Genomic DNA Laddering-Cells (0.5-2 ϫ 10 6 ) were stimulated for various times in 24-well plates, washed once with ice-cold phosphatebuffered saline, and lysed in 30 l of 50 mM Tris-HCl (pH 8), 10 mM EDTA, 0.5% laurylsarcosinate (Life Technologies, Inc.), 0.5 mg/ml proteinase K and incubated for 2-10 h at 50°C. 10 l of 0.5 mg/ml RNase A was then added, and the digestion was performed for 2 h at 50°C. 5 l of loading buffer (10 mM EDTA, 0.25 mM bromphenol blue, 40% sucrose) was added to the sample that was then heated to 65°C for 15 min and run on a 1.8% agarose gel containing 0.5 g/ml of ethidium bromide. The DNA was visualized by photographing the gel on a UV light box.
DNA and RNA Degradation Assay-Cellular DNA or RNA was labeled by pulsing 10 6 cells with 5 Ci of [ 3 H]thymidine or [ 3 H]uridine in 1 ml of complete medium for 2 h at 37°C. After three washes, 5 ϫ 10 4 labeled cells were distributed in triplicate in a 96-well microtiter plates in the presence of various stimuli. After 8 h, cell-associated high molecular weight DNA or RNA was harvested onto fiberglass filters and quantitated by liquid scintillation counting as described previously (24). A reduction in the amount of radioactivity present in high molecular weight DNA or RNA after culture with a given stimulus was taken as evidence of DNA or RNA degradation, respectively.

IL-2 mRNA Expression Is Post-transcriptionally Regulated by TCR-generated Signals in the A.AE7.2 T Cell Hybridoma-
IL-2 mRNA was detectable in anti-CD3-stimulated T hybridoma cells after 2 h, was maximally expressed after 4 h, and then declined thereafter ( Fig. 2A). Because growth factor mRNAs are often regulated by RNA stability, we tested whether an increased rate of mRNA degradation may explain the rapid disappearance of IL-2 mRNA that occurs after peak expression. IL-2 mRNA decay rates were measured by terminating transcription with cyclosporin A (22) after 3-4 h of activation with anti-CD3 mAb, the time of peak mRNA accumulation, or after 6 -8 h, during the decline phase. The amount of IL-2 mRNA remaining at various later times was then measured by Northern blot analysis. As shown in Fig. 2B, the half-life of IL-2 mRNA present after 3-4 h of TCR stimulation was significantly longer than the half-life of the IL-2 mRNA present after 6 -8 h of TCR stimulation (60.5 Ϯ 2.5 and 45.2 Ϯ 4.7 min, respectively, mean Ϯ S.D., p Ͻ 0.005, based on a two-sample t test). The fact that IL-2 mRNA became increasingly unstable as the total amount of IL-2 mRNA decreased indicated that RNA destabilizing factors were induced late in the response.
Identification of AU-specific Binding Proteins in the Cyto-plasm of Cells Stimulated with anti-CD3 mAb-In an attempt to define the molecular basis for the drop in the steady-state levels of IL-2 mRNA, we searched for proteins that bind to IL-2 mRNA and might be involved in RNA destabilization. IL-2 mRNA contains four AUUUA sequences in the 3Ј-UTR that have been suggested to function as binding sites for RNAbinding proteins involved in mRNA turnover (4). Extracts from resting or activated cells were incubated with a 32 P-labeled riboprobe consisting of 278 bases from IL-2 mRNA that contains the four AUUUA sequences clustered together in the 3Ј-UTR (HindIII-SpeI, Fig. 1). The RNA-protein complexes were detected in SDS-gels following UV cross-linking and RNase A digestion. RNase A was used at a concentration that would be expected to completely degrade all portions of the probe that were not protected by the protein, minimizing the contribution of the bound RNA to the molecular weight of the complex. Cytosolic extracts from unstimulated T cells reproducibly contained proteins with molecular masses of 90, 65, 40 -50, and 34 kDa that could be cross-linked to this RNA (Fig. 3A). Digestion of the reaction mixtures with proteinase K eliminated the labeled complexes, confirming that these factors are at least partially composed of protein (data not shown). Cytosolic extracts from T cells stimulated for 6, 8, or 10 h with anti-CD3 mAb reproducibly showed no change in the 65-kDa protein, a decrease in the 90-kDa protein, and an increase in the 34-and 40 -50-kDa proteins when compared with extracts from resting T cells. The extracts from stimulated T cells also contained novel proteins of 70 and 30 kDa. An identical pattern was obtained with another independently derived T cell hybridoma The appearance of the 70-kDa complex following anti-CD3 mAb stimulation often correlated with a decrease in the 90-kDa complex (Fig. 3A). This suggested that the protein present in the 70-kDa complex could be derived from the 90-kDa complex. Partial V8 protease digestion of the 70-and 90-kDa RNAprotein complexes (produced by UV cross-linking using a 32 Plabeled IL-2 RNA and cytosolic extracts from resting or activated T cells) generated overlapping sets of peptide fragments (Fig. 4). Thus, the proteins in the two complexes are highly related to each other at least in their RNA binding domains. Similar V8 protease experiments were performed on gel slices corresponding to AU-A and AU-B. As shown in Fig. 4, these two proteins showed distinct protease patterns as reported previously (9) and do not appear to be related to the 70-and 90-kDa proteins.
The RNA-binding Factors Require at Least Three AUUUA Elements in the IL-2 Probe and Bind an RNA Probe Derived from the c-myc cDNA-To further characterize the fine specificity of these RNA-binding proteins, competition experiments were performed by including unlabeled RNAs in the binding reactions. Cytosolic extracts from activated cells were incu-bated with a 32 P-labeled IL-2 riboprobe containing the entire AUUUA-rich region (IL-2 SpeI in Fig. 1) in the absence or in the presence of a 10, 100, or 500ϫ molar excess of unlabeled IL-2 RNAs containing different numbers of AUUUA sequences. As shown in Fig. 5, A and B, only the unlabeled IL-2 SpeI RNA with four AUUUA elements and an unlabeled 39-base RNA oligonucleotide containing the four AUUUA elements and only 5 additional bases at each end (IL-2 RNA oligo ARE in Fig. 1) were able to inhibit the binding of AU-A and AU-B proteins to the 32 P-labeled IL-2 SpeI riboprobe. The RNAs with no (IL-2 RsaI in Fig. 1) or two (IL-2 SspI in Fig. 1) AUUUA elements did not compete even at the highest concentration tested. These results are consistent with the previous observation that AU-A  Fig. 1; 2 ϫ 10 5 cpm) and processed as described in Fig. 3. Gel slices corresponding to AU-A, AU-B, the 90-and the 70-kDa complexes were excised from the gel, rehydrated, and reloaded on a 15% polyacrylamide SDS gel. Different concentrations (0 -1 g) of V8 protease were then loaded in each lane, and the digestion was performed in the gel at room temperature for 30 min. The cleavage products were then detected by autoradiography. and AU-B bind to synthetic RNAs with at least three tandem repeats of the AUUUA sequence (12). These proteins were also detected in UV cross-linking experiments using the 32 P endlabeled RNA oligonucleotide as a probe (data not shown), providing further evidence that they bind specifically to the AUUUA-rich sequence.
The 90-and 70-kDa complexes had a different specificity. The cross-linking of these proteins to the 32 P-labeled 278-base IL-2 SpeI riboprobe (see Fig. 1) that contains four AUUUA elements was inhibited by the unlabeled IL-2 SpeI RNA (Fig.  5A) and by the 39-base RNA oligonucleotide that also contains the 4 AUUUA elements (Fig. 5B). Although these proteins could be cross-linked directly to the 32 P-labeled 39-base IL-2 RNA oligonucleotide (data not shown), suggesting that they can bind to the minimal ARE sequence, they were not competed as efficiently as AU-A and AU-B by the cold oligonucleotide (Fig. 5B). In addition, unlabeled RNAs containing two (IL-2 SspI) or no (IL-2 RsaI) AUUUA elements inhibited the crosslinking of the 70-and 90-kDa proteins to the 32 P-labeled IL-2 SpeI RNA (Fig. 5A), indicating that these proteins also bind to a sequence outside the ARE. The 70-and 90-kDa proteins were not degenerate RNA-binding proteins, however, in that their binding to the 32 P-labeled IL-2 SpeI RNA was not inhibited by an unlabeled c-myc RNA that was an efficient inhibitor of the binding of both AU-A and AU-B (Fig. 5B). These results suggest that AU-A and AU-B proteins bind only to the ARE but do not discriminate between IL-2 and c-myc mRNAs, whereas the 90-and 70-kDa bind specifically to IL-2 mRNA but appear to bind the ARE and another site.
[ 32 P]UTP-labeled riboprobes corresponding to various portions of the 3Ј-end of the IL-2 mRNA were used to identify the non-ARE p90/p70 binding site. As shown in Fig. 6, and as summarized in Fig. 1, the 90-and 70-kDa proteins, but not AU-A, -B, or -C, could be cross-linked to a 56-nucleotide segment spanning nucleotides 508 -564 at the 3Ј-end of the IL-2 coding region. In contrast, AU-A, -B, and -C, but not p90 and p70, bound to a c-myc RNA. All of the RNA-binding proteins bound to an IL-2 probe spanning nucleotides 508 -660 that contains the ARE, and p90 and p70 could be cross-linked to this RNA better than to the RNA spanning nucleotides 508 -564 that lacks the ARE. Together, our results show that p90 and p70 bind independently to two different elements in the IL-2 mRNA, one encompassing nucleotides 508 -564, and the other encompassing nucleotides 589 -627 that contains the ARE. In contrast, AU-A, -B, and -C bind only to the IL-2 ARE containing greater than 2 AUUUA repeats.
The Appearance of the TCR-induced RNA-binding Factors Requires Calcineurin and PKC Activation and Is Induced in the Absence of RNA/Protein Synthesis-TCR-mediated signaling FIG. 5. Specificity of the AU-binding factors. Five g of cytosolic extracts derived from unstimulated cells (med) or anti-CD3 (␣CD3) mAb-stimulated cells were incubated with 1.5 fmol of [ 32 P]UTPlabeled IL-2 riboprobe (IL-2 SpeI-4 AUUUA) in the absence (0) or in the presence of a 10-fold (10), 100-fold (100), or 500-fold (500) molar excess of the following unlabeled competitors. A, transcripts derived from the IL-2 sequence containing four AUUUA (SpeI), two AUUUA (SspI), no AUUUA (RsaI); B, an IL-2 RNA oligonucleotide restricted to the ARE sequence (IL-2) and transcripts derived from the 3Ј-UTR of the c-myc cDNA (myc). Following UV cross-linking, the RNA-protein complexes were digested with RNase A, separated on SDS-polyacrylamide gel electrophoresis and detected by autoradiography.
FIG. 6. The 90-and 70-kDa proteins bind to a second site in the IL-2 coding region. A, RNA probes corresponding to the HindIII-TaqI (TaqI), the TaqI-DdeI (DdeI), the DdeI-RsaI (RsaI), and the DdeI-SpeI (SpeI) sequences of IL-2 coding region were transcribed in the presence of [ 32 P]UTP as indicated under "Experimental Procedures." B, 5 fmol of each RNA was analyzed by UV cross-linking experiments as described in Fig. 3 for the ability to form specific complex with RNA binding proteins that were detected by autoradiography.
involves protein kinase C activation and increased intracellular calcium, which in turn activates the calcium-dependent phosphatase calcineurin (25). Cyclosporin A is known to bind cyclophilin and to inhibit the activity of calcineurin, thus preventing the signal transduction cascade and downstream events such as IL-2 production (26,27). As shown in Fig. 7A, the induction of all of the RNA-binding proteins by anti-CD3 mAb-stimulation was blocked by cyclosporin A. A similar inhibition was observed when the protein kinase C-inhibitor, staurosporin was used (Fig. 8B). Moreover, when phorbol myristate acetate, a protein kinase C activator, and a calcium ionophore were used together, but not separately, to stimulate the cells, the same RNA-binding proteins induced by anti-CD3 mAb were detected in the cytosol (data not shown). These results suggest that both protein kinase C activation and calcium-dependent events are necessary and sufficient to activate these cytosolic RNA-binding factors.
Correlation between Induction of the RNA-binding Proteins and Apoptosis-To better understand the signals required to induce these proteins, we determined whether they were newly synthesized in response to TCR signaling. The TCR-stimulated appearance of the inducible cytosolic RNA-binding proteins occurred in the presence of 10 g/ml of the protein synthesis inhibitor cycloheximide (Fig. 7A). Moreover, when the cells were treated with 10 g/ml of cycloheximide in the absence of TCR signaling, a set of cytosolic RNA-binding proteins indistinguishable from the set observed after TCR-mediated activation (Fig. 7B) was induced. The RNA synthesis inhibitors, actinomycin D and DRB, as well as emetine, another protein synthesis inhibitor, also induced the cytosolic RNA-binding proteins ( Fig. 7B and data not shown). Therefore, the same set of RNA-binding proteins appeared in the cytoplasm of T cells either when RNA or protein synthesis was blocked or when TCR signaling occurred.
Previous work by others demonstrated that TCR stimulation results in activation-induced cell death of T cell hybridomas (17). Apoptosis depends on transcription and translation of new gene products as shown by the finding that at low doses, inhibitors of these processes prevent TCR-induced DNA fragmentation (19). However, at high concentrations, the inhibitors themselves were shown to induce apoptotic cell death in T cell blasts (19,28,29). Based on these results and those shown in Fig. 7, A and B, it was possible that the RNA-binding proteins were induced under any condition where apoptosis was induced, e.g. TCR signaling or exposure to high doses of protein synthesis inhibitors. To test this, we cultured a T cell hybridoma with anti-CD3 mAb, cycloheximide, or actinomycin D at doses that induced the cytosolic RNA-binding proteins and then analyzed genomic DNA for apoptosis-associated, oligonucleosomal cleavage. As shown in Fig. 7D, and as previously reported, TCR signaling resulted in DNA fragmentation that was inhibited by cyclosporin A (Ref. 30 and Fig. 7D). Cycloheximide and actinomycin D, at 10 and 5 g/ml, respectively, induced DNA fragmentation by themselves, suggesting that apoptosis was induced under these conditions. Therefore, cyclosporin A is capable of blocking the TCR-induced DNA fragmentation and the appearance of the cytosolic RNA-binding proteins, whereas TCR-mediated signaling alone and inhibitors of transcription and translation induce both genomic DNA fragmentation and the appearance of the RNA-binding proteins. Moreover, the accumulation of the most inducible RNAbinding factors, AU-B and p70, appeared to be specific for cells that were dying by apoptosis because they were not detected in the cytosol of cells undergoing necrotic cell death following antibody plus complement or azide treatment (Fig. 7C).
To further test the idea that the appearance of RNA-binding proteins correlated with apoptosis induction, we reproduced the previous finding (31) that glucocorticoids and TCR signaling each induce apoptosis in T cell hybridomas but that the combination does not (Fig. 8A). The inducible cytosolic proteins that could be cross-linked to IL-2 RNA (AU-B and p70) were detected only in cells undergoing apoptosis (cells treated with FIG. 7. The appearance of the RNA-binding proteins does not require new RNA and protein synthesis and correlates with apoptotic cell death. Cells were treated with medium alone (med); anti-CD3 mAb in the absence (␣CD3) or presence of 0.2 g/ml of cyclosporin A (␣CD3ϩcsa); or 10 g/ml cycloheximide (␣CD3ϩcx) for 6 h (A). In panel B, the cells were treated with medium alone (med); immobilized anti-CD3 mAb (␣CD3); cyclosporin A 0.2 g/ml (csa); 10 g/ml of cycloheximide (cx); or 5 g/ml of actinomycin D (actD) for 6 h. In panel C, the cells were left untreated (med) or treated with 0.5% azide (azide) or with anti-CD4 mAb and rabbit complement (CK) for 30 min at 37°C. The inducible RNA-binding proteins were detected as described in Fig. 3. Three fmol of 32 P-labeled IL-2 RNA probe were used in the binding reactions. The molecular masses of the standards are expressed in kDa. In panel D, genomic DNA was isolated from cells (5 ϫ 10 5 ) treated for 6 h with medium alone (med); immobilized anti-CD3 mAb in the absence (aCD3) or in the presence of 0.2 g/ml cyclosporin A (␣CD3ϩcsa) or 10 g/ml cycloheximide (␣CD3ϩcx); 10 g/ml cycloheximide (cx) alone, or 5 g/ml actinomycin D (actD) alone. The cells were lysed in 30 l of 50 mM Tris-HCl (pH 8), 10 mM EDTA, 0.5% laurylsarcosinate (Life Technologies, Inc.), 0.5 mg/ml proteinase K and incubated for 10 h at 50°C. The mixtures were then digested with RNase A and loaded on a 1.8% agarose gel containing 0.5 g/ml of ethidium bromide. anti-CD3 mAb or dexamethasone, but not both) (Fig. 8B). Unlike TCR-mediated signals, dexamethasone treatment does not induce IL-2 mRNA expression (data not shown), thus disassociating the appearance of the RNA-binding proteins and lymphokine production. These results suggest that the inducible RNA-binding proteins probably bind to AU-rich RNAs in addition to IL-2 RNA.
We also tested whether the up-regulation of the RNA-binding proteins correlated with apoptosis in nontransformed T cells. Treatment of A.E7, a CD4 ϩ nontransformed T cell clone, with dexamethasone resulted in both DNA fragmentation (Fig.  8C) and the appearance of the cytosolic RNA-binding proteins (Fig. 8D). Moreover, addition of the T cell growth factor IL-2 to the T cell cultures containing dexamethasone inhibited both apoptosis induction (Fig. 8C) and the accumulation of the cytosolic RNA-binding proteins (Fig. 8D).
Taken together, our results show that the cytosolic appear-ance of the RNA-binding factors and the development of an apoptotic phenotype were correlated in two different T cell types in response to a variety of stimuli. RNA Degradation and Apoptosis Are Temporally Related-All of the evidence indicated that the cytosolic RNA-binding proteins were present at times when the cells were undergoing apoptosis. If these proteins are involved in RNA degradation, then the rate of RNA turnover should be increased as apoptosis is induced. T cells were labeled with [ 3 H]uridine and then cultured in medium alone or in the presence of immobilized anti-CD3 mAb for 8 h. The amount of labeled RNA remaining was then measured as an indication of RNA degradation. RNA degradation was induced in T cells cultured with anti-CD3 mAb, and this was inhibited by cyclosporin A (Fig. 9A). Similarly, RNA degradation was induced by treatment of the cells with 5 g/ml of actinomycin D (Fig. 9A), a dose that induced maximal apoptosis, but not 0.5 g/ml, a dose that blocked transcription completely but induced minimal apoptosis (data not shown). Finally, as shown in Fig. 9B, the same doses of actinomycin D that induced DNA fragmentation, elicited RNA degradation, suggesting that the development of an apoptotic phenotype is associated with a faster RNA turnover. Therefore, RNA degradation correlated well with the induction of apoptosis and the appearance of cytosolic RNA-binding proteins.  A, B). A.E7 T cells were stimulated with medium alone (med) or with 1 M dexamethasone (dex) for 24 h in the absence (Ϫ) or in the presence (ϩ) of 10 units/ml recombinant IL-2 (C, D). In panels A and C, the cells were lysed and genomic DNA was analyzed in an ethidium bromide-agarose gel as described in Fig. 7. In panels B and D, the cytosolic extracts were recovered, and the RNA-binding proteins were analyzed as described in where the RNA and/or DNA degradation was considered to be zero. DISCUSSION TCR engagement elicits IL-2 production and then apoptotic cell death in T cell hybridomas. IL-2 mRNA is expressed shortly after TCR-stimulation, and its kinetic is shaped by effects on the rate of transcription of the IL-2 gene and by effects on stability of the IL-2 mRNA. Engagement of the TCR by anti-CD3 mAb quickly and transiently induces the expression of IL-2 mRNA that peaks around 4 h and rapidly disappears thereafter (Fig. 2A). Our results show that the IL-2 mRNA molecules present during the decline phase (i.e. after 4 h) are less stable than those present during the accumulation phase (i.e. before 4 h) (Fig. 2B), suggesting that the decline in IL-2 mRNA after its peak accumulation may be explained by cessation of transcription, as previously suggested (32), and by decreased stability of the remaining mRNA. The observation that AU-rich sequences have been shown to confer instability on otherwise stable mRNAs (2), and the fact that all of the inducible RNA-binding proteins appear in the cytosol during the period when IL-2 mRNA becomes less stable (compare Figs. 2 and 3A), suggests but does not prove that these proteins play a role in targeting mRNA to rapid degradation.
Several of the proteins we describe here in a murine system are similar to proteins previously characterized in human T cells. Bohjanen et al. (8,9) described three human proteins that bound to the ARE of the granulocyte-macrophage colony-stimulating factor 3Ј-UTR that they called AU-A, AU-B, and AU-C. Based on UV cross-linking and RNase T1 digestion experiments, these investigators reported molecular masses of 43, 40, and 50 -60 kDa for AU-A, AU-B, and AU-C, respectively. However, because RNase T1 (8,9) cuts RNA infrequently and leaves a relatively large fragment of RNA associated with the proteins, they calculated that the true molecular masses of AU-A, AU-B, and AU-C were 34, 30, and 43 kDa, respectively. These molecular masses correspond well with 34-, 30-, and 40 -50-kDa RNA-binding proteins that we observed in extracts from murine T cells. AU-A and the 34-kDa murine protein are both present predominantly in the nucleus, although some of the protein is also present in the cytoplasm. Following TCR signaling, in both cases, a fraction of the nuclear protein translocates to the cytoplasm. Both AU-A and the 34-kDa murine protein bind to the c-myc and IL-2 3Ј-UTRs, and in the latter case they do so by recognizing at least three AUUUA sequences. AU-B and the 30-kDa murine protein are both present only in the cytoplasm and only after TCR signaling and, like AU-A, both bind to the IL-2 3Ј-UTR by recognizing at least three AUUUA sequences (Fig. 5). AU-C and the 40 -50-kDa set of murine proteins, are a heterogeneous group of proteins that are present in the cytoplasm constitutively but increase after TCR signaling (Fig. 3). The specificity of AU-C is less well understood, although our competition results indicate that it has a relatively low affinity for the IL-2 3Ј-UTR. The simplest interpretation of these results is that the 34-, 30-, and 40 -50-kDa proteins described here are the murine homologues of AU-A, AU-B, and AU-C. It should be noted, however, that we did observe a difference in the specificity of the 30-kDa murine protein from the one reported for human AU-B. Human AU-B was shown to not bind to the c-myc 3Ј-UTR (8) whereas the 30-kDa murine protein does. This discrepancy could be due to a subtle difference in the RNA-binding sites of the human and the murine forms of the protein or to differences in the binding conditions used in the two studies.
We also identified two novel, related proteins of 90 and 70 kDa that could be cross-linked to IL-2 mRNA. The 90-kDa protein is present in the cytosol of unstimulated cells, whereas the 70-kDa protein is detectable only after TCR signaling (Fig.  3, A and B). The findings that the appearance of the 70-kDa protein correlates with the disappearance of the 90-kDa protein (Fig. 3), and that the two proteins have overlapping V8 protease maps (Fig. 4) and RNA cross-linking specificities (Fig. 5) raise the possibility that the 70-kDa protein is derived from the 90-kDa protein. If this is the case, then the conversion of the 90-kDa protein to the 70-kDa protein must not require new RNA or protein synthesis because it occurs in the presence of inhibitors of these processes (Fig. 7B). It is possible that activation induces a post-translational conversion of the 90-kDa form to the 70-kDa from via phosphorylation, glycosylation, or proteolytic cleavage.
The simplest explanation for our UV cross-linking results is that the 90-and 70-kDa proteins bind to the AUUUA minimal element in the 3Ј-UTR and to a second site in the IL-2 coding region (Fig. 6). V8 protease mapping experiments indicate that the domain on the 90-kDa protein that binds to the non-AUUUA sequence is the same as the one that binds to the ARE. 2 The capacity of this protein to bind more than one site on the IL-2 mRNA could allow it to form multimers, a property that could be important for its function. Cooperation between two different instability regions in the same mRNA has been previously described for the c-fos mRNA. As in the case of p90 and p70, one binding site is located in the c-fos coding region, and the other in the ARE present in the 3Ј-UTR. These elements appear to function by initiating the deadenylation of the mRNA as an early step in c-fos mRNA decay (33,34).
AU-A, AU-B, AU-C, and the 70-kDa protein were coordinately regulated. All appeared and disappeared at roughly the same times after the initiation of TCR signaling (Fig. 3) and all were dependent on the activity of calcineurin and protein kinase C (Fig. 8B). In addition, the induction of all four proteins by TCR signaling could occur in the presence of RNA or protein synthesis inhibitors (Fig. 7). This observation suggests that these proteins are already present in resting T cells but are unable to bind RNA until a TCR-dependent signal confers on them the binding activity. The idea that TCR-mediated signals could be involved in controlling RNA processing is supported by a recent finding that demonstrates a physical association between the signal transduction molecules p95vav (35) and c-Src (36) and the hnRNP-K.
Conversion to the RNA-binding forms of these proteins in the absence of new protein synthesis could be achieved by posttranslational modification of the proteins, translocation of the proteins from a storage compartment to the cytoplasm, already described for the heterogeneous nuclear ribonucleoprotein A (37), or by turnover of a labile inhibitor of the binding activity. Post-translational modifications of RNA-binding proteins have been described for the hnRNP A, B, and C proteins. The two modifications described for these proteins are phosphorylation of serine and threonine residues and methylation of arginine residues (for review, see Ref. 38). Phosphorylation and a redox switch have also been suggested for the activation of the 36-kDa protein expressed in Jurkat cells (39). Of particular relevance to the current study is the finding that the 70-kDa protein component of the U1 small nuclear ribonuclear protein is proteolytically cleaved in apoptotic cells (40). Moreover, posttranslational changes in subcellular compartmentalization have already been proposed for the regulation of endonuclease activity during apoptosis induction (41,42). DNase I, a calcium, magnesium-dependent endonuclease of thymocytes and lymphocytes that is capable of cleaving DNA to oligonucleosomesize fragments, is thought to be bound in an inactive state to G-actin in the rough endoplasmic reticulum (42,43). It has been proposed that calcium-dependent changes in the rough endoplasmic reticulum cause DNase I to dissociate from Gactin and enter the nucleus where it could cause DNA fragmentation. It is thus possible that a labile inhibitor sequesters or inhibits the RNA-binding proteins until apoptosis is induced.
Dexamethasone also induced the appearance of the cytosolic RNA-binding proteins (Fig. 8). This result raises the interesting possibility that the induction of specific RNA-binding proteins by dexamethasone is responsible for the glucocorticoidmediated destabilization of target mRNAs described in lymphocytes (10) and fibroblasts (44). The observation that the increased turnover of interferon-␤ in dexamethasone-treated cells is mediated by the AU-rich sequence found in the 3Ј-UTR (44) is consistent with this idea.
All of the agents that caused the cytosolic appearance of the RNA-binding proteins also induced apoptosis (Figs. 7 and 8). This correlation held for stimuli as diverse as glucocorticoids, anti-CD3 mAb, and high dose RNA or protein synthesis inhibitors, that are each likely to induce apoptosis via different signaling pathways. Recently it has been reported that autocrine stimulation through the Fas receptor is essential for TCR-activation induced T cell death (45,46,47) and that Fas probably signals via the ceramide pathway. Interestingly, cyclosporin A inhibits the induced expression of Fas ligand (45), and glucocorticoids inhibit the expression of Fas (47), perhaps explaining how these agents could block TCR-induced apoptosis. In contrast, oxidants, DNA-damage, and glucocorticoids probably trigger apoptosis by different signaling pathways. However, these diverse signals are thought to converge on a common death pathway that involves the activation of a protease called interleukin-1␤ converting enzyme, which is antagonized by the Bcl-2 family of survival proteins (for review, see Ref. 48). If the RNA-binding proteins described here are involved in apoptosis, then they must be activated at a point after the diverse cytotoxic signals have converged. The observation that apoptosis induced by TCR activation, like other routes to cell death, depends on the activity of a serine or cysteine protease (49) suggests that the induction of interleukin-1␤ converting enzyme or proteases like it is one such point. It is conceivable that activation of interleukin-1␤ converting enzyme-like proteases is responsible for the conversion of p90 to p70 that we observed in apoptotic T cells. As reported above, this would not be unprecedented because it has been reported that a specific proteolytic cleavage of the 70-kDa protein component of the U1 small nuclear ribonuclear protein is induced in apoptotic cells (40). Once activated, the RNA-binding proteins could facilitate the induction of apoptosis, perhaps by facilitating the turnover of AU-rich mRNAs that encode proteins that antagonize death, such as growth factors or oncogenic proteins. Alternatively, the RNA-binding proteins could play a more general role in RNA degradation to ensure that all new transcripts are eliminated in dying cells.