CA Repeats in the 3′-Untranslated Region of bcl-2 mRNA Mediate Constitutive Decay of bcl-2 mRNA*♦

An AU-rich element (ARE) in the 3′-untranslated region (UTR) of bcl-2 mRNA has previously been shown to be responsible for destabilizing bcl-2 mRNA during apoptosis through increasing AUF1 binding. In the present study, we investigated the effect of the region upstream of the ARE on bcl-2 mRNA stability using serial deletion constructs of the 3′-UTR of bcl-2. Deletion of 30 nucleotides mostly consisting of the CA repeats, located upstream of the ARE, resulted in the stabilization of bcl-2 mRNA abundance, in the absence or presence of the ARE. The specificity of the CA repeats in terms of destabilizing bcl-2 mRNA was proven by the substituting the CA repeats with other alternative repeats of purine/pyriminine, but this had no effect on the stability of bcl-2 mRNA. CA repeats alone, however, failed to confer instability to bcl-2 or gfp reporter mRNAs, indicating a requirement for additional sequences in the upstream region of the 3′-UTR. Serial deletion and replacement of a part of the region upstream of the CA repeats revealed that the entire 131-nucleotide upstream region is an essential prerequisite for the CA repeat-dependent destabilization of bcl-2 mRNA. Unlike the ARE, CA repeat-mediated degradation of bcl-2 mRNA was not accelerated upon apoptotic stimulus. Moreover, the upstream sequences and CA repeats are conserved among mammals. Collectively, CA repeats contribute to the constitutive decay of bcl-2 mRNA in the steady states, thereby maintaining appropriate bcl-2 levels in mammalian cells.

Apoptosis is a tightly controlled cellular suicide program that is critical for the successful development of multicellular organisms, the maintenance of normal tissue homeostasis, and removal of damaged cells (1). The protooncogene bcl-2, originally isolated from the chromosomal breakpoint of a t (14,18)bearing B cell lymphoma, serves as an important repressor of apoptosis in a variety of cell types (2,3). In line with its significant role in altering susceptibility to apoptosis, investigations of the mechanisms by which bcl-2 expression is modu-lated may prove crucial for identifying therapeutic strategies for cancer and some neurodegenerative diseases and for defining the role of bcl-2 in the development of multiple tissues (4).
Recent studies have indicated that bcl-2 is regulated at both the transcriptional and posttranscriptional levels. A number of negative transcriptional regulatory sites have been described in the bcl-2 promoter region (5,6), and several transcription factors, including cAMP response element binding protein, A-Myb and WT1, are known to be involved in the positive regulation of bcl-2 transcription (7)(8)(9). In addition to the promoter region, some sequences within the coding region, such as estrogen response elements, have also been demonstrated to mediate the transcriptional modulation of bcl-2, as was shown in breast cancer cell lines (10). The posttranscriptional modification of bcl-2 includes the phosphorylation of Bcl-2 at the putative mitogen-activated protein kinase sites, which confers resistance against ubiquitination-and proteasome-dependent degradation (11,12), and caspase-dependent cleavage, which results in the loss of anti-apoptotic activity (13,14). Another mechanism of posttranscriptional regulation of bcl-2 expression is based on mRNA stability. Recent reports have described that a conserved AU-rich element (ARE) 1 is present in the 3Ј-untranslated region (UTR) of bcl-2 mRNA (15) and that interaction of ARE with a number of ARE-binding proteins, including AUF1, is associated with bcl-2 mRNA decay during apoptosis (16,17). In addition, transfection of a synthetic ribozyme targeting bcl-2 ARE was shown to successfully downregulate bcl-2 mRNA expression accompanied by an increase in cell death in lymphoma cells (18), thus supporting the central role of ARE in the regulation of bcl-2 gene expression.
A recent study using a cell-free in vitro degradation assay revealed that a 396-nt segment of the bcl-2 3Ј-UTR including the ARE was degraded faster than the ARE motif itself and that the 3Ј-UTR lacking the ARE was still degraded faster than control RNA (19). These findings suggest that, in addition to the ARE, other regulatory sites might exist in the 3Ј-UTR, upstream or downstream of the ARE. Here, we investigated the effect of the region upstream of the ARE on bcl-2 mRNA stability and identified a new putative bcl-2 mRNA destabilization determinant that contains CA repeats, mediating the constitutive decay of bcl-2 mRNA.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-The African green monkey kidney cells (COS7) were obtained from the American Type Culture Collection and grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum (Hyclone) and antibiotic solutions (100 units/ml penicillin and 100 g/ml streptomycin, Invitrogen) under a 5% CO 2 humidified atmosphere at 37°C. COS7 cells (1 ϫ 10 5 ) were transfected in 6-well plates with 1 g of plasmids using FuGENE 6 reagent (Roche Applied Science) according to the manufacturer's instructions. COS7 cells were harvested at 48 h after transfection for Northern and Western blotting. To evaluate mRNA degratation rates, further transcription was inhibited by adding actinomycin D (Act D) at 10 g/ml (Sigma) 24 h after transfection, and then further incubating cells, before total RNA was extracted. To determine the degradation rate of bcl-2 mRNA under apoptotic condition, 1 mM H 2 O 2 was added to COS7 cells 1 h prior to treating Act D.
Plasmid Construction-A 1114-nt segment for the BU4 plasmid including the coding region (1459 -2178) and 394 nt of the 3Ј-UTR was first obtained by PCR amplification using cDNA from HL60 cells as a template and corresponding primers flanked by an EcoRI restriction site (see Fig. 1A; all nucleotide positions were based on the sequence of accession number M13994 in GenBank TM ). The PCR products were purified from a 1% agarose gel and inserted into the TA cloning site of pGEM-T easy vector (Promega); sequences were confirmed by automated DNA sequencing. The EcoRI fragment was then subcloned in the EcoRI site of pCR3.1, named BU4. Other bcl-2 cDNA constructs containing various lengths of the 3Ј-UTR were also produced by PCR using BU4 as a template for B, BU1, BU2, and BU3 and using BU3 for BU2-1, BU2-2, BU2-3, and BU2-4 with corresponding primers. For plasmids attached by other alternative repeats of purine and pyrimidine (BU2-2 AT , BU2-2 CG , BU2-2 GT and BU2-2 CA ), reverse primers were designed to include five respective repeats at the last part, and BU2-2 was used as a template.
To prepare a chimeric cDNA, which contains a coding region of bcl-2 and the 3Ј-UTR of unrelated gene followed by CA repeats (BU3 bis ), two separate PCR amplifications were performed as follows. The first part confining the bcl-2 coding region was synthesized with 5Ј-GAATTCAT-GGCGCACGCTGGGAGAACGGGG-3Ј (forward, primer1) and 5Ј-GCT-AGCTCACTTGTGGCTCAGATAGGCACC-3Ј (reverse) flanked at the 5Ј-end by EcoRI and NheI restriction sites, respectively. The second part composed of 125 nt (1742-1866) in the 3Ј-UTR of the bis gene (GenBank TM accession number AF127139) and 30 nt of the CArepeated region (CAR, 2310 -2339) were PCR-amplified with forward primer 5Ј-GCTAGCCCTCTGCCCTGTAAAAATCAGACT-3Ј and reverse primer 5Ј-GAATTCTGTGTGTGTGTGTGTCTGTCTGTGTGTGT-GCACCCACGTTACTGC-3Ј flanked at the 5Ј-end by NheI and EcoRI, respectively, using human bis gene (20) as a template. Each PCR product was independently cloned into pGEM-T easy vector and excised at the EcoRI and NheI sites for the first part and at the NheI and EcoRI sites for the second part. Two fragments were then ligated together into pCR3.1 at the EcoRI site to produce BU3 bis construct in which 131 nt of the 3Ј-UTR of bcl-2 was replaced with unrelated sequences including an NheI restriction site and part of the bis 3Ј-UTR. Two chimeric constructs of GFP, G2 and G3, were also prepared by ligating the two PCR products at the NheI and EcoRI sites as described above.
CAR was deleted from the 3Ј-UTR of BU4 plasmid (BU4⌬ CAR ) by a single step ligation of two the PCR products derived from BU4, upstream and down stream sequences of CAR, into the EcoRI sites of pCR3.1. Primer1 and 5Ј-CTCGAGGATGTTTATATGTGTGTTATTTT-T-3Ј containing a XhoI site (underlined) were used for the region upstream of CAR and 5Ј-CTCGAGCAATTAACAGTCTTCAGGCAAAA-C-3Ј with a XhoI site (underlined) and 5Ј-GAATTCGGTGATCCGGCC-AACAAC-3Ј with an EcoRI site (underlined) for the region downstream of CAR. Expression plasmids for mouse bcl-2 were produced by PCR using cDNA from NIH3T3 cells and the primers corresponding to the various length of the 3Ј-UTR.
Northern Blot Analysis-Total RNA was extracted from cells using the RNAzol B reagent (Tel-Test Inc.), separated by electrophoresis through denaturing 2.2 M formaldehyde, 1% agarose gel, and then transferred onto positively charged nylon membrane by capillary blotting. Before blotting, 28 and 18 S ribosomal RNAs were visualized by ethidium bromide staining to ensure the integrity of the isolated RNA and to quantify RNA loading. RNA was UV cross-linked to the membrane, which was prehybridized for 1 h at 55°C in high SDS hybridization buffer (5ϫ SSC, 7% SDS, 50% formamide deionized, 0.1% Nlauroylsarcosine, 2% milk, and 50 mM sodium phosphate, pH 7.0), and then hybridized with digoxigenin (DIG)-labeled specified DNA probes, prepared by PCR labeling, in the same buffer at 55°C overnight. The following primers were used to incorporate DIG-11 dUTP (Roche Applied Science) for different DNA probes; 5Ј-GGCAAGCTTGTGAAC-TGGGGGAGGATTGTG-3Ј and 5Ј-GGCAAGCTTGAGCAGAGTCTTCA-GAGACAG-3Ј for bcl-2; 5Ј-TGAAGTTCATCTGCACCACC-3 and 5Ј-AC-GAACTCCAGCAGGACCA-3 for GFP, and 5Ј-ACCACAGTCCATGCC-ATCAC-3Ј and 5Ј-TCCACCACCCTGTTGCTGTA-3Ј for GAPDH. The membrane was washed sequentially in 2ϫ SSC, 0.1% SDS for 15 min at room temperature twice and in 0.1ϫ SSC, 0.1% SDS for 15 min at 68°C twice. Specific hybridization was then immunodetected with alkaline phosphatase-conjugated anti-DIG antibody (1:10,000, Roche Applied Science) followed by chemiluminescent substrate CDP-star (Roche Applied Science).
To determine the rate of bcl-2 mRNA decay, total RNA was isolated from cells at multiple time points after treatment with Act D and then subjected to Northern blot analysis as described above. The membrane was stripped and rehybridized with GAPDH probe. The bcl-2 mRNA signal was quantified at each time point by densitometry using Scion image software (Scion Corp.), normalized to GAPDH, and plotted as a percentage of the initial value against time.
In Vitro Transcription-To examine the effect of the presence of the CA repeats in the 3Ј-UTR on the transcription of bcl-2, the transcripts of BU2 or BU3 constructs were synthesized in vitro using T7 RNA polymerase (21). BU2 or BU3 was first subcloned into the EcoRI site of pBluescript SKϪ in sense orientation with respect to the T7 promoter. In vitro transcripts were then synthesized using 0.5 g of DNA linearized at the SpeI site with 20 units of T7 polymerase (Roche Applied Science) in transcription reaction buffer containing 20 units of RNase inhibitor and a 0.05 mM concentration each of ATP, CTP, GTP, and UTP for 2 h at 37°C according to the manufacturer's instruction. After treatment of DNase I (Promega, 1 unit/reaction), the purity as well as the amount of RNA transcripts were determined by using 1% formaldehyde-agarose gel followed by ethidium bromide staining.

RESULTS
The CA Repeats in the 3Ј-UTR Contribute to the Destabilization of bcl-2 mRNA-To investigate the contribution of the sequence of the 3Ј-UTR upstream of the ARE of bcl-2 to the stability of bcl-2 mRNA, we prepared a series of 3Ј-UTR deletion mutants of bcl-2 between its stop codon and the ARE by PCR, as shown in Fig. 1A. Each cDNA fragment-containing coding region and the indicated portion of the 3Ј-UTR was inserted into the pCR3.1 vector. The resulting constructs were transiently transfected into COS7 cells, and the level of bcl-2 expression from each construct was analyzed by Northern and Western blotting. Fig. 1B shows that no significant differences of bcl-2 expression were found in cells transfected with B, BU1, or BU2. However, the bcl-2 mRNA level was lower in cells transfected with the BU3 cDNA than in cells transfected with B, BU1, or BU2 but was higher than in cells transfected with BU4 cDNA, which includes the ARE previously reported to be an apoptosis-dependent destabilizing element for bcl-2 mRNA (15). Similar results were obtained in Western blotting assays of Bcl-2 protein expression levels (Fig. 1B). Since all constructs used in this experiment were cloned into the same pCR3.1 vector, which has a cytomegalovirus promoter, the difference in the bcl-2 mRNA level between BU2 and BU3 constructs was likely to be due to the differences in mRNA stability rather than differences in the transcription rate. Supporting our presumption, in vitro transcription experiments using T7 polymerase and BU2 or BU3 as templates revealed no significant difference in transcript level from each template (Fig. 1C).
We then compared the mRNA degradation rates of the transcripts from these two constructs by Act D chase experiments. Fig. 1D shows that the mRNA level of BU3 was reduced to 29% of the control at 5 h after adding Act D, while the mRNA level of BU2 was maintained at 56% of the control level, indicating that the BU3 mRNA was degraded faster than the BU2 mRNA. Therefore, these results suggest that the part of the 3Ј-UTR region present in BU3 but deleted in BU2 has the potential to affect bcl-2 mRNA stability rather than the transcription processes.
To identify in detail the sequences with the mRNA destabilizing potential, the last 50 nt of the 3Ј-UTR sequences of the BU3 cDNA were serially deleted in 10-nt increments, and the expression levels of each construct were then compared by Northern blotting. As shown in Fig. 2B, transfection of the BU2-1 or BU2-2 construct, which, respectively, has 10 or 20 bases more than the BU2 construct, did not cause any decrease in the bcl-2 mRNA level compared with BU2 cDNA. However, the mRNA level gradually decreased in the cells transfected with BU2-3, BU2-4, or BU3, which had serial additions of 10 nt, up to 30 nt. Interestingly, the sequences of these 30 nt (2311-2340), which dramatically reduced the bcl-2 mRNA level, are mostly composed of CA repeats (ϩ2 GA), indicating Northern blot analysis was performed using 3 g of total RNA using DIG-labeled bcl-2 probe. Ethidium bromide-stained 28 and 18 S ribosomal RNAs were used as loading controls. Western blot analysis was performed in parallel with Northern blotting. One g of the total protein was separated by 12.5% SDS-PAGE, transferred to a membrane, and probed with an anti-Bcl-2 monoclonal antibody followed by a ␤-actin antibody. C, in vitro transcription of the coding region and subsequent region of the 3Ј-UTR of BU2 and BU3 was performed as described under "Experimental Procedure." RNA was visualized by staining with ethidium bromide. D, comparison of bcl-2 mRNA degradation mediated by 3Ј-UTR of BU2 and BU3 constructs. mRNA decay rates were determined by Act D chase experiments and Northern blotting. 24 h after transfection, 10 g/ml Act D was added to COS7 cells, which were harvested at the indicated times for hybridization with the bcl-2 probe, and subsequently with a GAPDH probe as a loading control (left panel). bcl-2 mRNA remaining at each time point was plotted against time as a percentage of its initial value, after being normalized to GAPDH (right graph). Signal intensity at time 0 was defined as 100%. Results are shown as means Ϯ of S.D. of three independent experiments. that bcl-2 mRNA stability decreased in proportion to the length of the CA repeats. We designated this region as CAR. It is notable that the BU2-3 construct, which has only five CA repeats, significantly reduced the bcl-2 mRNA level. When the five repeats of CA in the BU2-3 cDNA were replaced with five repeats of AT, CG, or GT, as shown in Fig. 2C, the mRNA levels derived from these three constructs were similar to the mRNA level of the BU2-2 construct without the CA repeats, excluding the possibility that a simple repeat of purine and pyrimidine rather than sequence-specific CA repeats affect bcl-2 mRNA stability.
Sequences Upstream of the 3Ј-UTR Are Required for CA Repeat-dependent Destabilization of bcl-2 mRNA-To establish whether the presence of CA repeats alone is sufficient to regulate bcl-2 mRNA stability, we made two chimeric constructs, termed BU3 bis and B CAR , respectively, as indicated in Fig. 3A. Transfection with the B CAR construct, in which 131 nt of the bcl-2 3Ј-UTR upstream of CA repeats were deleted, resulted in a marked stabilization of bcl-2 mRNA despite the presence of the CA repeats, indicating an additional requirement of the proximal sequences of 3Ј-UTR for the CA repeat-dependent destabilization of bcl-2 mRNA. Replacement of the proximal 131 nt of the bcl-2 3Ј-UTR with the unrelated sequences of 131 nt of the bis 3Ј-UTR also failed to destabilize bcl-2 mRNA as efficiently as the 3Ј-UTR of bcl-2, even though a slight destabilization was observed. Therefore, it is not the simple distance between the stop codon and the CA repeats but the specific sequences in the 131 nt of the 3Ј-UTR that seem to confer the destabilizing potential of the CA repeats.
The destabilizing activity of the proximal region of the 3Ј-UTR of bcl-2 was confirmed using a GFP reporter gene. The 3Ј-end of the open reading frame of the GFP cDNA was linked to the 131 nt of the 3Ј-UTR of bcl-2 and subsequently to 30 nt of CA repeats, or directly to the CA repeats only, as described under "Experimental Procedures." Compared with the mRNA level expressed from the GFP coding region only, the insertion of 131 nt of the 3Ј-UTR together with CA repeats into the GFP coding region resulted in a comparable decrease in the GFP mRNA level, whereas insertion of the CA repeats only caused no decrease (Fig. 3B). The incorporation of 131 nt of the 3Ј-UTR of the bis gene, instead of the 3Ј-UTR of bcl-2, had no effect on gfp mRNA stability as in bcl-2 mRNA, verifying the potential destabilizing activity of the 3Ј-UTR of bcl-2, which demands both the proximal 3Ј-UTR of bcl-2 and subsequent CA repeats.
To determine the essential region affecting the CA repeatdependent bcl-2 mRNA destabilization, we attached 30 nt of CA repeats to 42 or 82 nt of the 3Ј-UTR of bcl-2. The bcl-2 mRNA expressed from these two chimeric constructs showed significantly high levels, which corresponded to the mRNA levels from the construct in which all 131 nt were deleted (Fig.  4A). These results suggest that the last part of the 131 nt, lacking in these two constructs but present in the BU3 wild type, might be important for the CA repeat-dependent destabilization of bcl-2 mRNA.
Subsequently, the last part of the proximal 131 nt of the 3Ј-UTR, as well as the first and middle part, were substituted with unrelated sequences. As shown in Fig. 4B, the replacement of each part increased bcl-2 mRNA stability to a level similar to that of construct B, which contains only the coding region (Fig. 4B). These results, taken together, indicated that the overall mRNA structure provided by the linear sequences of the complete 131 nt of the 3Ј-UTR might be a prerequisite for the destabilizing ability of the subsequent CA repeats.
CA Repeat-dependent bcl-2 mRNA Destabilization Is Not Affected by Apoptosis-It has been reported previously that AREmediated decay in bcl-2 mRNA is enhanced during apoptosis and that this is accompanied by increased binding of several AUBFs (15)(16)(17). To determine whether CA repeat-mediated bcl-2 mRNA degradation also participates in the down-modulation of Bcl-2 during apoptosis, we investigated whether the degradation rate of bcl-2 mRNAs with the 3Ј-UTR including the CA repeats but lacking ARE is affected by apoptotic stimulation. Act D chase experiments up to 5 h following transient transfection for 24 h revealed that the degradation of bcl-2 mRNA with both CA repeats and ARE (BU4) was significantly accelerated by H 2 O 2 treatment (Fig. 5, B and central panel of   FIG. 2. The bcl-2 mRNA stability is regulated by CA repeats in the 3-UTR. A, schematic representation of the 3Ј-UTR of several bcl-2 constructs. The coding region is not shown for convenience. The upper panel represents deletion mutants of 3Ј-UTR in which 10 nt were serially deleted from the 3Ј-end of the BU3 construct. The lower panel shows that five repeats of CA of BU2-3 were substituted with other purine-pyrimidine repeated sequences. B and C, Northern and Western blottings of bcl-2 mRNA and protein, respectively, from deletion constructs, BU2, BU2-1, BU2-2, BU2-3, BU2-4, and BU3 (B) and from BU2-2, BU2-2 AT , BU2-2 CG , BU2-2 GT , and BU2-2 CA (C). Transfection and detection were performed as described in the legend to Fig. 1.   FIG. 3. CA repeats are required but not sufficient for destabilization of bcl-2 mRNA. Schematic diagrams of bcl-2 constructs (A) and GFP constructs (B) are shown in the left panel and the mRNA expressions of these constructs in the right panel. The coding region of the bcl-2 and GFP genes are presented as dashed and dotted boxes, respectively. The BU3 bis and G3 constructs include the 3Ј-UTR of the bis gene (*) and the CAR of the bcl-2 gene, and G2 includes 131 nt of the 3Ј-UTR of the bcl-2 gene (**), which was attached to the coding region of GFP. In B CAR and G4 constructs, 30 nt of CA repeats was directly inserted into the 3Ј-end of the coding regions of bcl-2 and GFP, respectively. Transfection and Northern blotting were carried out as described in the legend to Fig. 1. C), which is consistent with the earlier observation (15). However, the bcl-2 mRNA with the CA repeats but lacking the ARE (BU3) was degraded at similar rates in the absence or presence of H 2 O 2 to 38 or 33%, respectively, after treatment with Act D for 5 h, indicating that apoptotic conditions did not influence bcl-2 mRNA decay mediated by CA repeats (Fig. 5, B and left  panel of C). We also examined the decay of bcl-2 mRNA with the ARE but not the CA repeats (BU4⌬ CAR ). Deletion of the CA repeats led to marked stabilization of bcl-2 mRNA levels in the steady state, despite the presence of the ARE, but greatly enhanced the degradation of bcl-2 mRNA following H 2 O 2 treatment, as was revealed by a reduction in the remaining RNA from 107% to 35% after 5 h of Act D treatment (Fig. 5, B and  right panel of C). These results, taken together, indicated that the degradation of bcl-2 mRNA under apoptotic conditions was mainly mediated through the ARE component and that the CA repeats in the 3Ј-UTR may contribute to the constitutive decay of bcl-2 mRNA.
Effect of CA Repeats on Mouse bcl-2 mRNA Stability-Whereas the ARE in the 3Ј-UTR of the bcl-2 gene has been reported to be highly conserved in the human, mouse, chicken, and nematode (15), the CA repeats are confined to the 3Ј-UTRs of mammalian bcl-2 genes. CA repeats in mouse bcl-2, located 115 nt from the stop codon, include 23 repeats of CA without G, and the sequences upstream of the CA repeats share 79% of homology with those of human bcl-2 (Fig. 6A). To investigate whether CA repeats present in the 3Ј-UTR of the bcl-2 genes of other species also exert destabilizing activity on its mRNA, we compared the mouse bcl-2 mRNA levels from three different constructs. The mouse bcl-2 gene with only coding sequences expressed a higher level of bcl-2 mRNA compared with the mouse bcl-2 mRNA with 161 nt of the 3Ј-UTR including the CA repeats. In addition, a 46-nt deletion of the CA repeats resulted in the marked stabilization of mbcl-2 mRNA (Fig. 6B), as seen in human bcl-2 mRNA. Therefore, the mRNA destabilizing ability of CA repeats in the 3Ј-UTR of bcl-2 appears to be preserved among mammals to maintain bcl-2 mRNA levels in various cellular environments. DISCUSSION ARE-mediated mRNA decay has recently been implicated in the regulation of bcl-2 mRNA stability, which is activated by an apoptotic program (15)(16)(17). In this study we found that bcl-2 mRNA decay is also modulated in part by the CA repeats in the 3Ј-UTR, which are located about 131 nt from the stop codon and upstream of the previously characterized ARE. The destabilizing potential of the CA repeats was confirmed by deleting the CA repeats from the 3Ј-UTR of bcl-2, which resulted in a marked increase in the level of bcl-2 mRNA despite the presence of the ARE sequence (Fig. 5). These results suggest that the CA repeat is a novel determinant of bcl-2 mRNA decay, which is in line with the previous finding that an ARE-deficient 3Ј-UTR of bcl-2 was still degraded faster than control 3Ј-UTR in vitro, suggesting the presence of an additional bcl-2 mRNA decay-regulatory element in the 3Ј-UTR (19). Therefore, the destabilizing activity of the 3Ј-UTR of bcl-2 including the ARE demonstrated in earlier studies, which primarily focused on the ARE as a representative bcl-2 mRNA destabilizing determinant, seems to be attributable not entirely to the ARE but substantially to CA repeats as well.
CA repeats are the most common dinucleotide polymorphism found in the human genome and thus are routinely used as genetic markers of allelic variants in various genes (22)(23)(24). Recently, a number of reports have described a correlation between the reduced expression levels of several genes such as IFN-␥ and HSD11B2 and either the presence of repeats or an increase in the length of the repeats in the intron, suggesting a regulatory function of CA repeats on the processing of pre-mRNAs (25)(26)(27). Furthermore, CA repeats in intron 13 of the human endothelial nitric-oxide synthase (eNOS) gene seem to specify the cleavage site of the pre-mRNA of eNOS for splicing or degradation, depending on the presence of heterogeneous nuclear ribonucleoprotein L (hnRNPL) (28,29). However, little is known about the functional importance of CA repeats in the exonic context. Here we describe for the first time that CA repeats in the 3Ј-UTR can target bcl-2 mRNA for selective degradation, extending the role of CA repeats as an important element for modulating the stability of mRNA as well as the stability of pre-mRNA.
The mechanism by which CA repeats in an exon exert a negative effect on bcl-2 mRNA abundance has not been clarified. However, exonic CA repeats for mRNA destabilization appear to involve a different pathway from that of intronic CA repeats for specifying the cleavage sites of the pre-RNA of eNOS, given that the CA repeats did not function as a splicing enhancer when moved in an exonic context (28). Furthermore, CA repeats were shown to be sufficient for the cleavage of eNOS pre-mRNA (29), whereas CA repeats were found to be essential but not sufficient to destabilize bcl-2 mRNA, as shown by our results indicating that the incorporation of the CA repeats directly into the 3Ј-end of coding sequences of bcl-2 or GFP mRNA did not affect mRNA stability (Fig. 3, A and B). On the other hand, levels of GFP reporter gene mRNA decreased after introduction of the proximal 131 nt and subsequent CA repeats of the 3Ј-UTR of bcl-2, whereas the incorporation of the same length of the 3Ј-UTR of the bis gene with the CA repeats had no effect on GFP or bcl-2 mRNA stability. Finally, our deletion and replacement experiments indicate that the entire 131 nt region preceding the CA repeats is apparently a requisite for the destabilizing activity of the CA repeats (Fig. 4, A  and B).
In addition to the requirement of the upstream sequences of the 3Ј-UTR, CA repeat-mediated decay of bcl-2 mRNA was found to exhibit features that differed from ARE-mediated processes in several respects. The ARE in the 3Ј-UTR of bcl-2 has been known to mediate bcl-2 mRNA decay during apoptosis, either by increasing binding of destabilizing factors such as AUF1 (16,17) or by reducing interaction with stabilizing factors such as nucleolin (30,31). Our results show that the decay rate of the bcl-2 mRNA from the construct including the CA repeats but not the ARE was not significantly changed upon apoptosis induction, whereas bcl-2 mRNA from the construct including the ARE and CA repeats was degraded faster under the same conditions (Fig. 5). Moreover, the transcript from the bcl-2 DNA including the ARE but devoid of the CA repeats showed increased stability in nonapoptotic conditions but accelerated degradation in apoptotic conditions. Therefore, CA repeat-mediated decay of bcl-2 appears to contribute to maintaining appropriate Bcl-2 levels in the steady states, while the ARE-dependent destabilizing pathway may react to apoptotic stimuli to permit a rapid down-regulation of the Bcl-2 level, FIG. 6. The nucleotide sequences and destabilization potential of an upstream portion of the 3-UTR and CA repeats of the bcl-2 gene are conserved in mammals. A, the nucleotide sequences of the region upstream of the CA repeats in the 3Ј-UTR of bcl-2 gene were compared for human, rat, and mouse (GenBank TM accession numbers on the left). Alignment was performed starting from the first nucleotide of the stop codon up to the last nucleotide of the CA repeats. The shared nucleotides among these three species are in bold and underlined. Multiple alignment comparison was performed using the ClustalW engine (www.justbio.com/aligner/). B, three constructs of mouse bcl-2, prepared as shown in the left panel, were transfected into COS7 cells, and bcl-2 mRNA levels were analyzed (right panel).
FIG. 5. CA repeat-dependent decay of bcl-2 mRNA is not enhanced by apoptosis. COS7 cells were transiently transfected with three bcl-2 constructs including CAR (BU3), ARE only (BU4⌬ CAR ), or both (BU4) in the 3Ј-UTR as shown in A, and bcl-2 mRNA stability was analyzed in the presence of Act D. To induce apoptosis, H 2 O 2 (1 mM) was added to cells 1 h prior to adding Act D (10 g/ml). Total RNA was extracted at 0, 1, 3, or 5 h after Act D treatment and analyzed by Northern blot for bcl-2 and subsequently for GAPDH as shown in B. C, bcl-2 mRNA signals derived from the three constructs (BU3, BU4, and BU4⌬ CAR ) under normal and apoptotic condition were normalized with GAPDH mRNA signal from each lane in B after being quantified by densitometry using Scion image software (Scion Corp.). All data are represented as means Ϯ S.D. from three independent experiments. bcl-2 mRNA signal without Act D was defined as 100% and Act Dtreated bcl-2 mRNA levels were calculated as percentages of decay at the indicated times. thereby leading to cell death. However, more diverse apoptotic stimuli should be applied to confirm the relevance of these two cis-elements in Bcl-2 down-regulation in apoptosis. Interestingly, the presence of Bcl-2 protein was previously proposed as an essential requirement for the activation of ARE-dependent degradation programs (19). We also found that the destabilizing effect induced by the incorporation of the upstream sequences and CA repeats of the 3Ј-UTR of bcl-2 on GFP mRNA was less prominent than that of bcl-2 mRNA (Fig. 3), raising the possibility that CA repeat-mediated decay also entails the presence of Bcl-2 protein. However, the coexpression of Bcl-2 did not influence the level of GFP mRNA with or without the 3Ј-UTR of the bcl-2 gene in our study (data not shown). The amount of Bcl-2 protein was therefore not likely to be a critical prerequisite for the CA repeat-mediated decay of bcl-2 mRNA, but another cis-element in the coding region of bcl-2 might participate in the maintenance of the steady-state level of bcl-2 mRNA. Another difference is that the presence of CA repeats in the 3Ј-UTR of bcl-2 is confined to mammals, including humans, rats, and mice, whereas ARE is widely preserved, even being present in the 3Ј-UTR of nematode bcl-2 (15). Moreover, the sequences 5Ј-upstream of the CA repeats also reveal a high level of homology in mammalian bcl-2 (Fig. 6). Thus, CA repeat-mediated decay systems might have developed during evolutionary division into mammals to tune the bcl-2 level more finely. The presence of diverse modes of regulating the Bcl-2 level, as shown in our report and in previous reports (15)(16)(17), suggests that different mechanisms are involved in the modulation of the Bcl-2 level in response to different physiological and pathological conditions.
Most of the mechanisms that regulate mRNA stability involve specific interactions between structural determinants on mRNA, cis-acting elements and proteins that bind the determinants, trans-acting proteins, which modulate the susceptibility of mRNA to degradation (32). Cis-acting elements could be an actual target site for ribonuclease, or they might regulate ribonuclease attack elsewhere in the mRNA by binding with either stabilizing or destabilizing factors. It has been shown previously that CA repeats or CA-rich sequences have the potential to bind proteins such as heterogeneous nuclear ribonucleoprotein (hnRNP) in the intron of eNOS pre-mRNA or in the 3Ј-UTR of vascular endothelial growth factor (VEGF) mRNA (28,33). Therefore, defining whether a protein binds to the upstream sequences or CA repeats in the 3Ј-UTR of bcl-2 and subsequently affects the bcl-2 mRNA stability will be a subject of further investigation to get a complete understanding of the process of destabilizing bcl-2 mRNA.
In conclusion, we describe a novel pathway of constitutive decay of bcl-2 mRNA that involves CA repeats and their upstream sequences in the 3Ј-UTR, extending the functional significance of CA repeats from an intronic to an exonic context. Further investigation on the molecular mechanisms by which the CA repeats exert their destabilizing activity, including RNA/protein interactions, may contribute to the development of new strategies for reducing Bcl-2 levels in pathological conditions.