Evidence for a Stabilizer Element in the Untranslated Regions of Drosophila Glutathione S-Transferase D1 mRNA*

The neighboring genes gstD1 andgstD21 share 70% sequence identity. gstD1 encodes a 1,1,1-trichloro-2,2-bis-(P-chlorophenyl)ethane dehydrochlorinase; gstD21, a ligandin. Both of their mRNAs are inducible by pentobarbital but otherwise behave very differently. Intact gstD21 mRNA is intrinsically labile, but becomes stabilized when separated from its native untranslated region (UTR). In contrast, whereasgstD1 mRNA is very stable in its entirety, without its native UTRs it becomes even more labile than that ofgstD21. Decay patterns from four chimeric D1-D21 mRNAs, designed to reveal the individual importance of each molecular region to stability, strongly indicate the presence of destabilizing elements in the coding region of gstD1 mRNA. Thus, the UTRs of this molecule must contain a dominant stabilizer element that overrides the destabilizing influence of the coding region and confers overall stability to the entire molecule. The suspected presence of such a stabilizer element in gstD1 mRNA extends a concept from mRNA metabolism in yeast and cultured mammalian cells to include a multicellular organism, Drosophila melanogaster. The complementary presence of destabilizing and stabilizer elements on the same mRNA reveals a regulatory mechanism by which an abundant mRNA can be further induced by a chemical stimulus, or otherwise be returned to normal levels during recovery.

An effective way to regulate gene expression involves controlling mRNA stability (1)(2)(3). The major measure of mRNA stability is its half-life (1,4), which determines the time required for a mRNA to reach a new steady state following a change in transcription rate (e.g. by inducers such as pentobarbital).
We have been using mRNAs of the Drosophila glutathione S-transferase (gst) 1 genes D1 and D21 as reporters to investigate pentobarbital-mediated changes in mRNA stability. The early paradigm for RNA metabolism associates mRNA decay rates largely with the strength of the destabilizing sequences of the molecule (1)(2)(3)5). But the recent discovery of a handful of active stabilizer elements (STE) in certain mammalian and yeast mRNAs (6 -9) has called for a revision to this model.
Although gstD21 mRNA is labile, the coding region of the gene gains stability when separated from its native UTRs. Just the opposite is true for mRNA of the D21 homologue, gstD1. This mRNA is very stable, but the coding region of the molecule alone, without native UTRs, is even more labile than intrinsically unstable gstD21 mRNA. To further investigate the nature and the cause of this instability, we assembled chimeric D1-D21 mRNAs containing various segments of the D1 coding sequence. We observed that these chimeras were also unstable in the same context of heterologous UTRs as the D1 coding sequence. We repeatedly detected putative decay intermediates from the D21 portion, but seldom the D1 segment of these chimeric mRNAs. Such patterns are strong evidence for the presence of cryptic destabilizing cis-acting elements in the coding region of gstD1 mRNA. Our observations also suggest that the stability of already abundant gstD1 mRNA is maintained by a stabilizer element in its UTRs, which overrides any destabilizing elements in the coding region. We speculate that this combination of stabilizer and destabilizing elements helps to regulate gstD1 mRNA levels in response to pentobarbital induction and to generally maintain mRNA stability. As we compare the characteristics of labile gstD21 mRNA with those of gstD1 mRNA, in which a completely different arrangement of cis-acting elements govern RNA metabolism, we note the potential for significant diversity in the regulation of different members of a single multigene family.

EXPERIMENTAL PROCEDURES
Materials-Bacteriological media were purchased from Invitrogen, and chemicals from ICN, Invitrogen, or Sigma. Oligonucleotides were products of either Integrated DNA Technologies, Inc. (Coralville, IA) or Invitrogen. Radioactive nucleotides ([␣-32 P]UTP) were purchased from ICN (Irvine, CA). RPA III kits were purchased from Ambion (Austin, TX). Restriction enzymes were products of New England Biolabs (Beverly, MA) or American Allied Biochemicals (Aurora, CO). T4 polynucleotide kinase and T4 DNA ligase were products of New England Biolabs and Promega (Madison, WI), respectively. Pfu DNA polymerase was purchased from Stratagene (San Diego, CA). SP6 RNA polymerase and the plasmid vector pSP64(A) for in vitro transcription were purchased from Promega. Tobacco acid pyrophosphatase was a product of Epicentre Technologies (Madison, WI). T7 RNA polymerase was a generous gift from Bi-Cheng Wang (University of Georgia, Athens, GA). Escherichia coli DH5␣ competent cells and Pfx DNA polymerase were products of Invitrogen. The plasmid vector pCaSpeR-hs-act for Drosophila transformation was obtained from C. S. Thummel of the University of Utah (10). The ⌬2-3 line {P[ry ϩ ⌬2-3](99B)} (11) expressing transposase and the yw line were obtained from Susan Abmayr and David Gilmour, respectively, both of the Department of Biochemistry and Molecular Biology, The Pennsylvania State University. The E. coli expression plasmids for GST D1 (pGTDm1-KK) and GST D21 (pGTDm21-KK) were previously reported (12). The plasmids for C-terminal FLAG derivatives of D1 (D1-F) and D21 (D21-F), as well as plasmid for the G8S,G9S mutant of D21-F, were unpublished laboratory stocks.
Transgenic Constructs and Nomenclature-Site-directed mutagenesis was carried out according to the QuikChange TM mutagenesis procedure (Stratagene). All clones were sequenced at the Penn State Nucleic Acid Facility prior to microinjection into embryos.
To switch segments of GST D1 and GST D21 at the 66th and 65th residues (Fig. 1), primers D1-F-R66-SmaI-S (GTGGGAGTCCCGCGC-CATCCAGGTG) and D1-F-R66-SmaI-AS (CACCTGGATGGCCCGGG-TCTCCCAG) were used to introduce a SmaI site into pGTDm1-KK by site-directed mutagenesis without changing the amino acid sequence encoded by the template. The SmaI fragment of the resulting plasmid (pGTDm1-SmaI-KK) was cloned into calf intestinal alkaline phosphatase-treated, SmaI-digested pGTDm21-FLAG-G8S,G9S-KK to obtain pBA12-KK. The transgene, which contains the N-terminal 65 amino acids of D21 and C-terminal 150 amino acids of D1F, is called D21-65-FIG. 1. Complete gstD1 mRNA sequence and organization of chimeric D1-D21 genes. Panel A, the 5Ј-UTR and 3Ј-UTR of gstD1 mRNA. An analysis of the genomic sequence (20) revealed the presence of a 627-nucleotide intron between the 4th and 5th nucleotides upstream of the ATG initiation codon. The gene orthologous to gstD1 in the housefly Musca domestica also contains an intron at this same position (21). Panel B, a comparison of GST D1 and GST D21 amino acid sequences (20,22). Arg 66 of GST D1 is marked by an asterisk (*). Phe 171 of GST D1 and Phe 170 of GST D21 are marked by a vertical arrow (s). The extra C-terminal sequence of D21, ARKLAAK (22), is not shown in panel B. Panel C, DNA fragments were cloned into pCaSpeR-hs-act for microinjection, a step toward establishing transgenic lines. All chimeric mRNAs contain the 5Ј-UTR of hsp70 (not shown), CCCCAAC of the 5Ј-UTR of gstD21 (not shown), the coding region of gstD1, gstD21, or chimeric D1-D21 with the FLAG octapeptide (filled squares) at the C terminus, and the 3Ј-UTR of actin5C (not shown). The open rectangle represents the D21 coding sequence (214 amino acids) and the cross-hatched rectangles represent the D1 coding sequence (208 amino acids) (22). ‡, G8S,G9S mutations in GST D21. D1. The opposite swap, resulting in pBA10-KK, was made by inserting the SmaI fragment of pGTDm21-FLAG-KK into calf intestinal alkaline phosphatase-treated, SmaI-digested pGTDm1-FLAG-KK. The transgene, which contains the N-terminal 66 amino acids of D1 and C-terminal 157 amino acids of D21F, is called D1-66-D21. The exchange at the 171st (D1) and 170th (D21) amino acid position was initiated by the introduction of an XhoI site into both pGTDm1-FLAG-KK (D1F-F171L-XhoI-S (GCAGGTGGCCAAACTCGAGATCAGCAAGTAC) and D1F-F171L-XhoI-AS (GTACTTGCTGATCTCGAGTTTGGCCACCTC-G)) and pGTDm21-FLAG-G8S,G9S-KK (D21-F170L/D171E-XhoI-S (GTTCGAAGTTAGTGATCTCGAGTTCAGCAAGTACTCC) and D21-F170L/D171E-XhoI-AS (GGAGTACTTGCTGAACTCGAGCTCACTAA-CTTCGAAC)), which yield pBA1-XhoI-KK and pBA2-XhoI-KK, respectively. The XhoI-SmaI fragment of pBA2-XhoI-KK was cloned into XhoI-SmaI-digested pBA1-XhoI-KK to generate pBA13-KK. This led to the transgene D1-171-D21, which contains 171 amino acids of D1 followed by 52 amino acids of D21-F at the C terminus. Because the gstD21 coding region has a second SmaI site that interferes with cloning, the XhoI-PstI fragment of pBA1-XhoI-KK was cloned into pBA2-XhoI-KK, resulting in pBA15-KK. The corresponding transgene is called D21-170-D1, which contains 170 amino acids of D21 followed by 45 amino acids from the C-terminal of D1-F. Prior to cloning into pCaSpeR-hs-act DNA, BamHI and EcoRI restriction sites were introduced by PCR at the desired ends of the constructs described thus far, using the appropriate pair of starting primers from the set D1-F-5Ј, D1-F-3Ј, D21-F-5Ј, and D21-F-3Ј. A FLAG-less version of pBA1-CaSpeR (i.e. pBA22-CaSpeR) was constructed using the same strategy, but changing the 3Ј-end PCR primer to D1-3Ј (CGGGATCCGTGAATAT-CAGGCTTATTC) from D1-F-3Ј. This transgene is called D1-UTR. Construction of transgene D21L-UTR will be described elsewhere.
All clones in pKK223-3 and pCaSpeR-hs-act vectors were sequenced at the Penn State Nucleic Acid Facility. Results also showed that the four chimeric proteins (D1-66-D21, D21--65-D1, D1-171-D21, and D21-170-D1) were successfully expressed from the pKK223-3-based expression constructs in E. coli. 2 Plasmid DNA was prepared for microinjection using a Concert TM rapid plasmid DNA isolation kit (Invitrogen). Microinjection of embryos was subsequently performed, as previously described (13,14). The newly enclosed G 0 flies were crossed singly to yw to remove any transposase background. Yellow-to red-eyed G 1 progeny with longer body bristles (Sb Ϫ ) were re-crossed with yw. Stable lines were established through sibling crossings of colored-eye G 2 virgin flies. Three separate lines were maintained for each transgene.
Pentobarbital and Heat Shock Treatments-Adult flies (2-3 days old) were distributed into clean milk bottles in approximately equal numbers for 5 h starvation at room temperature (21-23°C) (15). Control flies received a blotting paper strip (3 ϫ 10 cm) saturated with a solution of 5% sucrose; PB-treated flies received a strip soaked in 5% sucrose plus 200 mg/ml PB. The strips were placed in the fly bottles for 2 h at room temperature. Heat shock was administered by incubating flies at 35°C for 1 h in clean bottles containing 5% sucrose paper strips in a Robbins Scientific Co. (Sunnyvale, CA) hybridization oven (model 2000). (An empty milk bottle with a foam plug requires ϳ15 min to reach 35°C from room temperature and takes ϳ6 min to drop to 31°C after removal of the bottle from the 35°C oven.) In addition, heat shock treatments of varying duration were carried out at 35°C for 5-40 min (instead of 1 h) to detect labile transgenic mRNAs. The flies were subsequently snap-frozen in liquid nitrogen and stored at Ϫ70°C until use.
RNA Isolation and RPA Analysis-RNA was isolated from pulverized flies according to a protocol from Ullrich et al. (16). The templates used to prepare our radiolabeled riboprobes, pSP64(A).D21AS, pSP64(A).D1AS, and pSP64(A).RP-49AS, were constructed by RT-PCR amplification. Each plasmid DNA was linearized with an appropriate restriction enzyme then transcribed in vitro using [␣-32 P]UTP. RPA analyses of 40 g of total RNA samples were conducted according to procedures specified by Ambion, manufacturer of the RPAIII kits. In our figures, we call the protection product of endogenous gstD1 mRNA "endo-D1." For transgenes, protection products of expected sizes are labeled as "transgene"; those smaller than the expected sizes are called "decay intermediates" (Int).
Mapping the 5Ј End of gstD1 mRNA-The 5Ј end sequence of the gstD1 mRNA was determined by primer extension using the primer 5Ј-AGCGGCAGGGGGAGGAGCCGGGCA-3Ј and by circular RT-PCR (17,18). Decapping, DNase I treatment, and circularization of RNA were carried out according to a procedure by Couttet et al. (18). 5 g of circularized RNAs was used for reverse transcription using a gstD1specific primer 5Ј-GCGGATCCTTGGCGGTCATGATCACGGAGC-3Ј. The resulting cDNA reaction mixture was boiled for 5 min and then digested with a mixture of RNase A and RNase T1. The treated cDNA was recovered by phenol extraction and ethanol precipitation. One percent of the recovered cDNA was taken for PCR amplification, using 1 unit of Pfx DNA polymerase and the primer pair, 5Ј-GCGGATCCTT-GGCGGTCATGATCACGGAGC-3Ј and 5Ј-GCGAGCTCTCCCGGAT-GGGAGGAGAACTGGGC-3Ј, as instructed by Invitrogen. The PCR product was digested, gel-purified, and then cloned into BamHI-SacIdigested pSP64(A). Two clones were randomly selected for sequencing to determine the 5Ј and 3Ј end sequences of the gstD1 mRNA. Finally, Southern blot analysis was carried out with 5Ј-end-labeled oligo(dT) 18 to determine the length of the poly(A) of each mRNA (19).

RESULTS
Complete Sequence of the gstD1 mRNA-Based on sequencing results obtained for the cRT-PCR clones, the 5Ј-UTR of gstD1 mRNA spans 64 nucleotides, and primer extension yielded multiple bands, marking D1 mRNAs with 5Ј-UTR sequences of 67, 66, 64, 63, 61, and 60 nucleotides in length (data not shown). Two cRT-PCR clones were sequenced. One had a 5Ј-UTR of 63 nucleotides, the other, one of 64 nucleotides. The 3Ј-UTR of gstD1 mRNA is 132-135 nucleotides long, with variation because of uncertain cleavage over a stretch of As in the genomic sequence (20,22). Total RNAs from control or PBtreated flies yielded cRT-PCR products of gstD1 mRNA. Moreover, they did so regardless of decapitation by tobacco acid pyrophosphatase, indicating the presence of uncapped D1 mRNA (Fig. 2). The presence of multiple primer extension products of varying size supports the notion that some gstD1 mRNAs are uncapped and missing a few nucleotides at the 5Ј end.
In contrast, the same preparations of RNAs yielded no cRT-PCR products for gstD21 in the absence of tobacco acid pyrophosphatase when a D21 primer pair was used under the same set of experimental conditions (data not shown). This affirms that a small contingent of stable, uncapped gstD1 mRNA exists among a capped majority population whose molecules have an average of ϳ40 As at the 3Ј ends under both control and PB treatment conditions. The uncapped and shorter than fulllength gstD1 mRNA molecules with short poly(A) tails are probably decay intermediates stabilized by a stabilizer element (Ref. 9 and see "Discussion").
Identification of Cryptic Destabilizing Element(s) in the gstD1 mRNA-Despite a slow transcription rate, gstD1 mRNA is relatively abundant under control conditions. In contrast, gstD21 mRNA, which has a faster transcription rate than gstD1, holds at barely detectable levels in the same environment. gstD1 mRNA must therefore be significantly more stable than gstD21 mRNA under control conditions (15). We observe that the coding sequence alone of gstD21 was stable as part of a chimeric RNA with the 5Ј-UTR of hsp70 and the 3Ј-UTR of act5C (i.e. transgene D21L-UTR, Fig. 3B). Thus, we attribute the instability of gstD21 mRNA to the influence of a cis-acting, destabilizing element in the UTRs. This association reflects the current paradigm that mRNA stability depends largely upon the strength of its destabilizing element (for reviews, see Refs. 1-3 and 5). We were surprised to observe, then, that contrary to the standing model, chimeric D1 mRNA, which contains only the D1 coding sequence, was actually very labile (Fig. 3C) in the same heterologous UTR contexts (i.e. D1-UTR) in which chimeric D21L-UTR mRNA was stable.
To test the effect of nonspecific 3Ј extension on D1-UTR mRNA stability, we added the FLAG TM sequence (GACTA-CAAGGACGACGATGACAAG) at the 3Ј end of the D1 coding region to yield transgenic line D1F-UTR. We used RPA to compare the mRNA expression levels of three chimeric constructs: D1F-UTR (D1F-UTR ϭ D1 coding sequence with FLAG TM tag minus the native UTRs of gstD1 mRNA); D1-UTR, the same sequence minus the FLAG TM ; and D21L-UTR, the D21L sequence minus the native UTRs. Whereas chimeric D21L-UTR mRNA expression was induced to a great extent by 1 h of heat shock at 35°C (Fig. 3B), the same treatment reduced chimeric D1-UTR mRNA to barely detectable levels (Fig.  3C). Also, under the same conditions, control endogenous gstD1 mRNA levels were elevated 1.8 Ϯ 0.4-fold by heat shock (Fig.  3C). Results of shortened heat shock treatments (of incubation lasting 5-40 min at 35°C) showed that this chimeric D1F-UTR mRNA was inducible by heat shock but yielded very labile product (Fig. 3D). Chimeric D1F-UTR mRNA levels increased with the duration of heat shock for up to 40 min, but always remained much lower than those of endogenous gstD1 mRNA (Fig. 3, C and D). Thus, this particular nonspecific 3Ј end extension to the D1 coding sequence did not sufficiently replace the function of the native 3Ј-UTR sequence.
The critical difference between the endogenous gstD1 mRNA and the transgenic D1 mRNAs, from D1-UTR and D1F-UTR, is in the presence or absence of native gstD1 UTRs. We had previously observed that endogenous gstD1 mRNA is stable under both control and PB treatment conditions (15). Given the current paradigm of mRNA stability, we did not anticipate the relative instability of the transgenic D1 mRNAs. These unexpected results strongly suggest that the coding region of gstD1 mRNA contains one or more cryptic destabi- lizing element(s), which exert their influence in the absence of the native UTRs.
Effect of the gstD1 mRNA Destabilizing Element(s) on gstD21 Coding Sequences-Recalling that D1 and D21 coding sequences share 70% sequence identity (Ref. 22 and Fig. 1), and given how the chimeric D21L-UTR mRNA, which contains the D21 coding region, is stable, we set out to localize these suspect destabilizing cis-acting elements in gstD1 mRNA and observe their destabilizing effect on an otherwise stable sequence. We constructed two pairs of chimeric D1-D21 genes, and established corresponding transgenic lines for each (D1-66-D21 and D1-171-D21 for D1-D21 chimeras; D21-65-D1 and D21-170-D1 for D21-D1 chimeras, see "Experimental Procedures" for nomenclature of chimeric genes). Each of the induced chimeric RNAs was analyzed by RPA with antisense D21 and D1 riboprobes. Results are shown in Figs. 4, B-E, and 5, B-E, respectively. The D21 probe detected an RNA band of ϳ450 nt (Fig. 4B, D1-66-D21) from the total RNAs of transgene D1-66-D21; the D1 probe, however, failed to detect anything (Fig. 5B).
These results suggest that the chimeric D1-66-D21 mRNA was unstable, particularly in the 201-nucleotide region of D1 (codons numbers 1-67). A time course analysis of heat shock induction (Fig. 4B, D1-66-D21) revealed that the chimeric mRNA was induced as soon as the inside of the experiment bottle reached 32°C, between 5 and 10 min inside the 35°C oven. This protected band from the D21 segment appeared exclusively for the transgene D1-66-D21 and was not observed in other transgenic or nontransgenic lines. The D1 portion, on the other hand, for which no band showed, probably degraded rapidly (Fig. 5B, D1-66-D21). Because the stable D21 component lies downstream from the D1 region in the chimeric mRNA (D1-66-D21), we know that degradation of the D1 sequence cannot be caused by 3Ј 3 5Ј exonucleases from the poly(A) end (23)(24)(25).
In analyses of other chimeric mRNAs, the full-length D21 probe protected multiple fragments of the chimeric D21-170-D1 RNA (Figs. 4E and 5E). But this same probe protected only one fragment, and which was smaller than expected, in chimeric D21-65-D1 mRNA (Figs. 4C and 5C), and three very small fragments in D1-171-D21 mRNA (Figs. 4D and 5D). Recalling that endogenous gstD21 mRNAs are not induced by heat shock, the protected D21 subfragments must trace to the induced chimeric mRNAs.
The D1 riboprobe clearly protected endogenous gstD1 mRNA but as for the D1 portions of chimeric mRNAs yielded protection products at low to undetectable levels. No bands appeared for the D1-66-D21 and D21-170-D1 constructs (Fig. 5, B and E), and the D1 portions of chimeric mRNAs from D21-65-D1 and D1-171-D21 were detectable but only at very low levels (Fig. 5,  C and D). Our results show these four D1-D21 (D21-D1) chimeric RNAs to be very labile, conceivably because of the presence of destabilizing D1 sequences. The instability of these chimeras was manifest very early into heat shock induction, with some decay intermediates appearing before induced chimeric mRNAs could be detected (Fig. 4, D and E). These decay intermediates were not generated during RNA isolation but, rather, increased along with the duration of heat shock. Meanwhile the reference RP-49 mRNAs remained intact throughout our time course analysis.
Results in Figs. 4 and 5 along with the demonstrated stability of the chimeric D21L-UTR mRNA (Fig. 3B), support the notion that the gstD1 coding sequence contains cryptic destabilizing elements. These cis-acting elements apparently exert their degradative influence in the absence of the native UTRs from the mRNA. A conceivable explanation, then, for how fulllength gstD1 mRNA maintains its stability is that the UTRs contain a dominant STE(D1) that overrides any destabilizing influence from the coding region. This hypothesis, substantially supported by our findings, expands the current paradigm of mRNA stability regulation with this new detail of an additional stabilizer element.
Mapping Putative Decay Intermediates-We set out to identify a decay pattern for the D21 portion of the chimeric D21-D1 mRNAs by mapping the decay intermediates of each molecule from D21-65-D1, D1-171-D21, and D21-170-D1 flies with a nested set of D21 riboprobes. (We passed over the D1 portion because its intermediates were barely detectable.) RPA results are shown in Fig. 6. The decay intermediate from D21-65-D1 (Fig. 6A, Int-Sa) spanned ϳ100 nucleotides of the D21 sequence (numbers 81-181 of the 198 nucleotides from codons 1 to 66). The D1 portion of chimeric D21-65-D1 mRNA was barely detectable and only so very early into heat shock treatment (Fig.  5C). Three pieces of decay intermediates spanning ϳ50, 55, and 60 nucleotides were detected from D1-171-D21 by the D21 probe (Fig. 6A, Int-1, Int-2, and Int-3). As the sum of these lengths exceeds the entire D21 stretch (129 nt, numbers 567- 696 of the complete gstD21(L) mRNA) of the chimeric mRNA; they must partially overlap. Decay of the D21 part of the chimeric D21-170-D1 mRNA probably involves an endonucleolytic cleavage near the SmaI site (number 228 of the gstD21(L) sequence). 2 One major decay intermediate (Int-1 in Fig. 4E and fragments marked by arrows in Fig. 6B) was mapped to the region of 81-547 of the D21 mRNA sequence, another intermediate (Int-2) to the region of 245-460, and a third (Int-3 in Fig.  4E and Int-3-Sa in Fig. 6B) to the region of 81-245 (Fig. 6, B and C). The early appearance of stable decay intermediates from the induced transgene(s) indicates that the half-lives of the intact chimeric D21-D1 mRNAs are shorter than 20 min (Figs. 4 and 5). Given that the half-life of chimeric D21L-UTR mRNA in the same context of UTRs is much longer (Fig. 3B), 2 dramatically shorter half-lives for the D21-D1 chimeras strongly suggests, then, that the D1 coding sequence contains active destabilizing elements. The summary of mapping results (Fig. 6C) suggests that these destabilizing elements most likely are located in the first 67, and the last 37, codons of the D1 coding sequence. Chimeric RNAs containing these sequences are either undetectable (transgenes D1-66-D21 and D21-170-D1) or detectable only at very low levels (transgenes D21-65-D1 and D1-171-D21). These destabilizing elements are suppressed in endogenous gstD1 mRNA, which must contain the proposed dominant stabilizer element STE(D1) in its native UTRs. DISCUSSION mRNA stability is an important regulatory factor in gene expression. The relative stability of mRNA determines its lifespan and thus, its translatability, in the cytoplasm (1)(2)(3)(4)(5). Endogenous gstD1 mRNA is quite stable under both control and PB treatment conditions. A ϳ2-fold PB-induced increase in the transcription rate of gstD1 accordingly resulted in a ϳ2-fold increase in the steady-state level of gstD1 mRNA (15). How, then, does stable gstD1 mRNA return to normal levels after the PB inducer is removed from the flies? The presence of cryptic, cis-acting, destablizing elements in the coding region of gstD1 provides a possible avenue. Just how they exert their influence, however, remains to be elucidated.
In the absence of their native UTRs, D1 portions of the D1-D21 chimeric RNAs were shown not only to be degraded themselves, but also to destabilize segments of the D21 coding sequence that we know is stable in the absence of its native UTRs (Figs. 3 and 4). Mapped decay intermediates from the chimeric D1-D21 mRNAs display patterns that are consistent with the hypothesis that the destabilizing elements are, most likely, located in the N-terminal (codons 1-67) and C-terminal (codons 172-209) regions of the GST D1 coding sequence. It also indicates that endonucleolytic cleavage(s) are probably involved in the decay pathway(s).
The stability of endogenous gstD1 mRNA, therefore, must rely on a dominant STE that overrides these destabilizing elements. The fact that both chimeric D1 mRNAs from transgenes D1-UTR and D1F-UTR, which lack native UTRs, are both very labile is strong evidence that this putative STE(D1) resides in the UTRs of gstD1 mRNA. The presence of STE(D1) may also explain the occurrence of a small fraction of stable, decapped, and shorter than full-length gstD1 mRNAs in control and PB-treated RNA populations (see Fig. 2).
Studies in the yeast Saccharomyces cerevisiae have revealed that decapping triggered poly(A)-shortening leads to 5Ј 3 3Ј exonucleolytic degradation (26,27). This pathway of mRNA decay has also been detected in mammalian cells (18). If this same pathway also persists in D. melanogaster then the presence of stable decapped gstD1 mRNA would indeed be impossible without the function of a stabilizer element. The P-STE stabilizer, found in the coding region of the yeast PGK1 mRNA has been shown to block deadenylation-dependent mRNA decay (9). The short poly(A) (ϳ40 As) of the intact molecules and presence of decapped gstD1 mRNA in the natural population suggest that the putative STE(D1) functions similarly to the yeast P-STE by blocking 5Ј 3 3Ј exonucleolytic degradation.
STEs are a known feature of yeast and mammalian mRNAs (6 -9). Several yeast mRNAs that contain upstream open reading frames in the 5Ј-UTR are degraded through nonsensemediated decay (3,8,28). But for certain genes, such as GCN4 and YAP1, mRNAs are known to harbor a stabilizer element in the 5Ј-UTR just upstream of the main open reading frame. In GCN4 mRNA this STE protects the molecule from rapid decay by interacting with the RNA-binding protein Pub1p, which is required in the nonsense-mediated decay pathway (8). Our findings provide solid evidence that such a STE(D1) works similarly on the mRNA of a multicellular eukaryotic organism. Moreover, this STE would be similarly located to the stabilizer element of the ␣-globin mRNA if it should fall in the 3Ј-UTR of gstD1 mRNA (6,7). There is, however, no pyrimidine-rich segment in gstD1 UTRs as there is in the 3Ј-UTR of ␣-globin mRNA (7).
The Drosophila gstD1 and gstD21 genes are adjacently located but divergently transcribed (20). Although their coding sequences share 70% identity, their products perform very different enzymatic functions. GST D1 is a 1,1,1-trichloro-2,2bis-(P-chlorophenyl)ethane dehydrochlorinase as well as a glutathione S-transferase. GST D21, on the other hand, does not exhibit normal GST activity (12) but may be an important ligand-binding protein (i.e. ligandin). The UTRs of both the gstD1 and gstD21 mRNAs appear to contain cis-acting regulatory element(s), but ones which function quite differently. The native UTRs of the gstD1 mRNA are essential to the stability of the molecule, whereas those of gstD21 mRNA contain one or more element(s) that render the molecule very unstable in the absence of PB. 2 The coding regions of gstD1 and gstD21 also exhibit contrasting behaviors with respect to mRNA stability. In the same context of the hsp70 5Ј-UTR and the actin5C 3Ј-UTR, we observe that, on the one hand, the D21 coding sequence remains very stable (Fig. 3B), but that, on the other, the D1 coding sequence becomes very labile. The stark differences in behavior between gstD1 mRNA and gstD21 mRNA with regard to stability show the potential for diversity in yet another aspect of expression regulation within a multigene family.