| Nucleic Acids Research | Pages |
The 5[prime]-untranslated region of GM-CSF mRNA suppresses translational repression mediated by the 3[prime] adenosine-uridine-rich element and the poly(A) tail
Introduction
Materials And Methods
cDNA constructs
In vitro transcription
In vitro translation
Analysis of translation products
mRNA isolation and northern blotting
Results
In vitro stability of GM-CSF mRNAs
In vitro translation of GM-CSF mRNAs
Removal of the native 5[prime]-UTR unmasks ARE-mediated translational inhibition
Effects of the ARE and poly(A) tail on GM-CSF mRNA translation
Mapping of the 5[prime]-UTR ARE antagonist
A decrease in global translation rate promotes ARE-mediated inhibition of GM-CSF translation in the presence of a full-length 5[prime]-UTR
Discussion
Acknowledgements
References
The 5[prime]-untranslated region of GM-CSF mRNA suppresses translational repression mediated by the 3[prime] adenosine-uridine-rich element and the poly(A) tail
Received June 8, 1999; Accepted July 23, 1999
ABSTRACT Granulocyte-macrophage colony stimulating factor (GM-CSF) mRNA levels are controlled post-transcriptionally by the 3[prime]-untranslated region (UTR) adenosine-uridine-rich element (ARE). In untransformed, resting cells, the ARE targets GM-CSF mRNA for rapid degradation, thereby significantly suppressing protein expression. We used a rabbit reticulocyte lysate (RRL) cell-free system to examine translational regulation of GM-CSF expression. We uncoupled decay rates from rates of translation by programming the RRL with an excess of mRNAs. Capped, full-length, polyadenylated human GM-CSF mRNA (full-length 5[prime]-UTR AUUUA+A90) and an ARE-modified version (full-length 5[prime]-UTR AUGUA+A90) produced identical amounts of protein. When the 5[prime]-UTR was replaced with an irrelevant synthetic leader sequence (syn 5[prime]-UTR), translation of syn 5[prime]-UTR AUUUA+A90 mRNA was suppressed by >20-fold. Mutation of the ARE or removal of the poly(A) tail relieved this inhibition. Thus, in the absence of a native 5[prime]-UTR, the ARE and poly(A) tail act in concert to block GM-CSF mRNA translation. Substitutions of different regions of the native 5[prime]-UTR revealed that the entire sequence was essential in maintaining the highest rates of translation. However, shorter 10-12 nt contiguous 5[prime]-UTR regions supported 50-60% of maximum translation. The 5[prime]-UTR is highly conserved, suggesting similar regulation in multiple species and in these studies was the dominant element regulating GM-CSF mRNA translation, overriding the inhibitory effects of the ARE and the poly(A) tail.
INTRODUCTION
The expression of many cytokines and proto-oncogenes is regulated at the post-transcriptional level by alterations in the rates of mRNA degradation and translation (1-3). These mRNAs often contain 3[prime]-untranslated region (UTR) adenosine-uridine-rich elements (AREs) (4,5) which are typically composed of reiterated AUUU repeats within a U-rich context. When arranged as the nonamer UUAUUUA(U/A)(U/A), AREs function as potent mRNA destabilizers (6,7).
In normal, quiescent lymphocytes, ARE mRNAs are rapidly degraded (4-7). However, entry into the cell cycle in response to phorbol ester, ionophore or mitogenic antibodies transiently and substantially stabilizes ARE mRNAs (8-10). Because the resulting increase in steady-state mRNA levels occurs concurrently with an increase in the activity of several ARE RNA-binding proteins (AUBPs), it has been proposed that ARE mRNA stability may be mediated in trans by these AUBPs (11-13). Therefore, the ARE indirectly influences protein synthesis by modulating mRNA decay and accumulation.
The ARE also has potent effects on translation. Interferon-[beta] (IFN-[beta]), c-fos and granulocyte-macrophage colony stimulating factor (GM-CSF) mRNAs without their respective AREs showed enhanced translation in Xenopus oocytes (14). Recently, several reports of translational regulation by 3[prime]-UTR-binding proteins have been published and lend support to models involving long-range interactions between the 5[prime]- and 3[prime]-ends of these mRNAs (15-18). This hypothesis is further supported by data showing that the 5[prime]-m7GpppG cap and poly(A) tail synergistically stimulated translation (19), perhaps secondary to an association between poly(A)-binding protein (PABP) and eIF4F (20). The longer the poly(A) tail, the greater its effectiveness as a translational enhancer (21-23). Finally, it has been proposed that the uridine-rich ARE and the poly(A) tail could also interact, either through direct base pairing or indirectly via associations between trans factors. Recently, Elav-like proteins were shown to bind both the ARE and the poly(A) tail of their cognate mRNAs (24).
We have used a rabbit reticulocyte lysate (RRL) cell-free system to examine which cis elements control GM-CSF mRNA translation. By evaluating the translation of mRNAs differing in their 5[prime]-UTRs, ARE and poly(A) tails, we have begun to elucidate the key elements responsible for the regulation of GM-CSF translation. Our data, in contrast to earlier reports in Xenopus oocytes (14), show that full-length GM-CSF mRNAs are well translated despite the presence of an ARE. However, in the absence of the 5[prime]-UTR, the ARE in cooperation with the poly(A) tail becomes a potent translational repressor. The 5[prime]-UTR is evolutionarily conserved and folds into a predicted stem-loop structure. Substitution and deletion analyses revealed that while the complete 5[prime]-UTR sequence wholely suppressed ARE and poly(A) tail-mediated inhibition of translation, partial but significant suppression was also achieved by 10-12 nt contiguous subdomains.
MATERIALS AND METHODS
cDNA constructs
Plasmids containing the wild-type and mutant human GM-CSF cDNAs have been described previously (25,26). Briefly, plasmids pT7syn5[prime]GM AUUUA+T90 and pT7syn5[prime]GM AUGUA+T90 contain GM-CSF cDNAs downstream of a T7 promoter and immediately upstream of a 90 nt poly(dT) tract, followed by a unique HindIII site. These plasmids contain the complete coding region and 3[prime]-UTR of GM-CSF, but lack the native 5[prime]-UTR, instead having a synthetic 28 nt leader made up of p[alpha]19 plasmid sequence (Gibco BRL, Gaithersburg, MD). In this report, we have referred to mRNAs from these templates as syn 5[prime]-UTR AUUUA+A90 and syn 5[prime]-UTR AUGUA+A90.
GM-CSF-encoding plasmids containing full-length 5[prime]-UTRs were cloned from the syn versions above by removing the synthetic 5[prime]-UTR via a SacI/PstI double digest and ligating in complementary oligonucleotides corresponding to the full-length, native 5[prime]-UTR. These plasmids are referred to as pT7full-length 5[prime]GM AUUUA+T90 and pT7full-length 5[prime]GM AUGUA+T90. The wild-type and mutant GM-CSF mRNAs produced from these constructs are referred to as full-length 5[prime]-UTR AUUUA+A90 and full-length 5[prime]-UTR AUGUA+A90, respectively. Constructs with substituted 5[prime]-UTRs were produced in an analogous fashion to full-length cDNAs by ligating in complementary oligonucleotides into SacI/PstI double digested GM-CSF cDNA backbones. Substitutions of the 5[prime]-UTR were made between nucleotides -1 and -12, -12 and -22 and -22 and -32 with respect to the start codon. DNA templates for transcription, lacking the 90 nt poly(dT) tract were produced by PCR, using T7 sequence for forward priming and sequence corresponding to the GM-CSF 3[prime]-end for reverse priming. All mRNAs synthesized from these plasmids lacked poly(A) tails. All GM-CSF constructs were verified by manual sequencing (Sequenase 2.0 kit; Amersham, Arlington Heights, IL) from both directions, using T3-, T7- and GM-CSF-specific primers (data not shown).
In vitro transcription
Capped mRNAs were produced using the T7 mMessage mMachine kit (Ambion, Austin, TX) as described by the manufacturer. Uncapped mRNAs were produced using the T7 Megascript kit (Ambion). RNAs were purified by phenol:chloroform extraction followed by isopropanol-ammonium acetate precipitation. Concentrations were assessed spectrophotometrically and the size and integrity of all transcripts were verified by electrophoresis on denaturing formaldehyde-agarose gels.
In vitro translation
Translation reactions were generally programmed with 200 ng of GM-CSF mRNA/50 µl reaction. Flexi RRL (Promega, Madison, WI) was reconstituted with amino acids, 20 µCi L-[35S]methionine (New England Nuclear, Boston, MA) and KCl as recommended by the manufacturer. Translation reactions were incubated at 30°C for 0-2 h. Experiments were performed in triplicate, using different batches of reagents and in vitro transcribed RNAs. Data are shown as means SD of three to five determinations.
Analysis of translation products
Aliquots were taken at various time points and snap frozen at -80°C. Samples were then thawed and treated with 0.1 mg/ml RNase A at 37°C for 15 min to digest residual mRNA and charged tRNAs. SDS-PAGE loading buffer was added and the samples were electrophoresed by 15% SDS-PAGE with low range molecular weight protein markers (Bio-Rad, Hercules, CA). Gels were dried under vacuum, visualized, and quantitated by phosphorimaging (Molecular Dynamics, Sunnyvale, CA).
mRNA isolation and northern blotting
At the indicated time points, 1 µl aliquots were removed from each translation reaction, diluted 10-fold in diethylpyrocarbonate-treated water and mixed with 990 µl of TriReagent (Molecular Research Center, Cincinnati, OH). Samples were then stored at -80°C until the time course was completed. Total RNA was then quantitatively isolated as per the manufacturer's standard protocol and analyzed by northern blotting as described previously (26). Signals were quantitated by phosphorimaging.
RESULTS
In vitro stability of GM-CSF mRNAs
The RRL was initially programmed with capped, full-length 5[prime]-UTR AUUUA+A90 and capped, full-length 5[prime]-UTR AUGUA+A90 GM-CSF mRNAs. Reactions were incubated under standard translation conditions (see Materials and Methods) and aliquots were taken at the indicated time points for RNA and protein analysis. Northern blotting revealed that the AUUUA+A90 mRNA (Fig. 1, top) decayed somewhat more rapidly (t = 40 min) than the AUGUA+A90 mRNA (t > 60 min; Fig. 1, middle). Although there was no detectable signal at 60 min for the AUUUA+A90 mRNA, it must be noted that this image was produced after blots were exposed for only 1 h to a phosphorimager screen. Longer, overnight exposure showed a substantial amount of full-length mRNA persisting at 60 min (data not shown). Further, the amount of sample used for northern analysis at each time point represented only 2% of the entire translation reaction. These results were highly reproducible and in agreement with decay measurements made in intact cells (25,27). Removal of the poly(A) tail stabilized the AUUUA-containing mRNA (Fig. 1, bottom) but did not alter the stability of the AUGUA mRNA (data not shown).
Figure 1. Decay of GM-CSF mRNAs with full-length 5[prime]-UTRs. Rabbit reticulocyte lysates were programmed with the indicated full-length GM-CSF mRNAs (see Materials and Methods). At the indicated times, aliquots were taken and total RNA was isolated and analyzed by northern blotting with 32P-radiolabeled GM-CSF cDNA probes.
In vitro translation of GM-CSF mRNAs
Aliquots of RRL programmed with full-length GM-CSF mRNAs were analyzed for protein production by SDS-PAGE. As shown in Figure 2A and B, full-length 5[prime]-UTR AUUUA+A90 and full-length 5[prime]-UTR AUGUA+A90 mRNAs were translated with identical efficiencies and produced protein of the expected size (~18.5 kDa). Translation was therefore not influenced by differences in the decay rates of these mRNAs (see Fig. 1). Further, despite mRNA stabilization, removal of the poly(A) tail did not increase translation from full-length AUUUA mRNA (Fig. 2A and B). The uncoupling of translation from decay in this system is most likely due to the translation of only a fraction of the exogenously added mRNAs. In addition, these data showed that the 3[prime]-UTR ARE, in the context of full-length GM-CSF mRNA, did not inhibit translation, contrary to previous observations with ARE-containing GM-CSF, IFN-[beta] and c-fos mRNAs in Xenopus oocytes (14,22). We therefore began to look for regions within GM-CSF mRNA that could account for this discrepancy.
Figure 2. Influence of the the ARE and poly(A) tail on translation of full-length 5[prime]-UTR GM-CSF mRNAs. Full-length 5[prime]-UTR GM-CSF mRNAs were translated in an RRL system containing 35S-radiolabeled methionine. (A) At the indicated times, 2 µl of each reaction was analyzed by SDS-PAGE for the synthesis of 35S-labeled GM-CSF. (B) Quantitation of GM-CSF production by phosphorimaging. Data shown are means SD of five determinations.
Removal of the native 5[prime]-UTR unmasks ARE-mediated translational inhibition
The IFN-[beta], c-fos and GM-CSF mRNAs injected into oocytes (14) lacked full-length, wild-type 5[prime]-UTRs. We hypothesized that our inability to observe ARE-mediated translational suppression in RRL might reflect differences in the 5[prime]-UTR. We therefore transcribed capped, polyadenylated, wild-type and mutant GM-CSF mRNAs from plasmids which had the native 5[prime]-UTR replaced with a synthetic 20 nt leader sequence (syn 5[prime]-UTR AUUUA+A90 and syn 5[prime]-UTR AUGUA+A90; see Materials and Methods). These mRNAs thus closely resembled those used previously by Kruys et al. (14). Decay of syn 5[prime]-UTR AUUUA+A90 mRNA (Fig. 3, top) was very similar to that of full-length 5[prime]-UTR AUUUA+A90 mRNA (Fig. 2, top). However, 20-fold more GM-CSF was synthesized from full-length mRNA than the version with the synthetic 5[prime]-UTR (Fig. 4A and B, compare with Fig. 2). Thus in the absence of a full-length native 5[prime]-UTR, translation of polyadenylated, AUUUA-containing mRNA was significantly suppressed. Taken together, our data supports the presence of a dominant translational regulatory element in the native 5[prime]-UTR of GM-CSF mRNA.
Figure 3. Decay of GM-CSF mRNAs with synthetic 5[prime]-UTRs. Rabbit reticulocyte lysates were programmed with the indicated synthetic GM-CSF mRNAs (see Materials and Methods). At the indicated times, aliquots were taken and total RNA was isolated and analyzed by northern blotting with 32P-radiolabeled GM-CSF cDNA probes.
Figure 4. Influence of the ARE and poly(A) tail on the translation of syn 5[prime]-UTR GM-CSF RNAs. Synthetic 5[prime]-UTR GM-CSF mRNAs were translated in an RRL system containing 35S-radiolabeled methionine. (A) At the indicated times, 2 µl of each reaction was analyzed by SDS-PAGE for the synthesis of 35S-labeled GM-CSF. (B) Quantitation of GM-CSF production by phosphorimaging. Data shown are means SD of five determinations.
Effects of the ARE and poly(A) tail on GM-CSF mRNA translation
Several groups have suggested that repression of cytokine mRNA translation was a consequence of U-A base pairing between the 3[prime] ARE, AUUU repeats and the poly(A) tail (19,20,23). In addition, there is growing evidence implicating 3[prime]- and 5[prime]-UTR interactions in the regulation of translation (19,20,28). To further examine these hypotheses, we programmed RRLs with syn 5[prime]-UTR mRNAs either without the ARE (syn 5[prime]-UTR AUGUA+A90) or lacking a poly(A) tail (syn 5[prime]-UTR AUUUA-A90). Both mRNAs were stable over the time course of the experiment (Fig. 3, middle and bottom, respectively). Mutation of the ARE (syn 5[prime]-UTR AUGUA+A90) caused a 10-fold increase in translation, while removal of the poly(A) tail (syn 5[prime]-UTR AUUUA-A90) produced an even more dramatic 20-fold increase (Fig. 4A and B). Thus, in the absence of a full-length 5[prime]-UTR, suppression of translation by the 3[prime] ARE likely required its interaction with the poly(A) tail.
Mapping of the 5[prime]-UTR ARE antagonist
As the full-length 5[prime]-UTR overcame ARE and poly(A) tail-mediated translational inhibition, we hypothesized that it contained a regulatory element. The 5[prime]-UTR of human GM-CSF mRNA is a relatively short 32 nt sequence that is highly conserved across species (Fig. 5). We therefore modified the composition and length of the 5[prime]-UTR of GM-CSF and assessed translation in the context of the ARE and poly(A) tail. When the native 5[prime]-UTR sequence was replaced by a synthetic 28 nt sequence, translation was inhibited 20-fold (Figs 2, 4 and 6). Since the lengths of the native and synthetic sequences were similar, we inferred that the 5[prime]-UTR sequence was more important than its length. A truncated 5[prime]-UTR containing the nucleotides -1 to -12 upstream from the start codon directed the synthesis of 10-fold more GM-CSF than the synthetic 28 nt sequence (Fig. 6). Increasing the length of this truncated 5[prime]-UTR to 32 nt, with a synthetic sequence, did not further enhance translation (Fig. 6), thereby reaffirming the importance of sequence specificity rather than length. mRNA containing nucleotides -1 to -22 of the wild-type sequence in the 5[prime]-UTR showed marginally increased translation, but was still 35% less than with the entire 5[prime]-UTR. Substitutions of 5[prime]-UTR sequences between nucleotides -33 and -22 or between -22 and -12 resulted in an ~50% inhibition of translation, compared to full-length control (Fig. 6). These data therefore indicate that while multiple portions of the native 5[prime]-UTR can substantially overcome ARE and poly(A) tail-mediated suppression of translation, the entire 5[prime]-UTR is needed to completely attenuate inhibition. The dominant translational regulatory element is therefore the entire 5[prime]-UTR.
Figure 5. Evolutionary conservation and proposed secondary structure of the GM-CSF 5[prime]-UTR. (A) Comparison of the 5[prime]-UTR of the GM-CSF sequences from humans, mice and pigs. Both the murine and porcine 5[prime]-UTRs share 85.7% homology to the human sequence. (B) Computer predicted modeling (47) of the human 5[prime]-UTR, with a calculated thermodynamic stability of -13.9 kcal/mol.
Figure 6. The entire 5[prime]-UTR is the minimal essential element needed for optimal translation. GM-CSF production was measured in the RRL system from mRNAs containing an ARE and a poly(A) tail but with modified 5[prime]-UTRs as shown on the left side of the figure. The numbering starts at -1, just upstream of the start codon. Open bars represent native 5[prime]-UTR sequences and shaded bars represent synthetic sequences. GM-CSF production is presented as a percentage of that synthesiszed by mRNA with a full-length 5[prime]-UTR.
A decrease in global translation rate promotes ARE-mediated inhibition of GM-CSF translation in the presence of a full-length 5[prime]-UTR
The sequence, length and secondary structure of the 5[prime]-UTR can specifically regulate translational initiation by controlling the rate of cap-dependent ribosomal scanning (29). We therefore hypothesized that productive scanning of GM-CSF mRNA initiated by the native, full-length 5[prime]-UTR could suppress the 3[prime] ARE as a translational inhibitor by promoting continuous ribosomal movement in a 5[prime]->3[prime] direction. However, under suboptimal translation (initiation/elongation) conditions, decreased ribosomal transit could unmask the ARE, permitting it to function as a translational repressor even in the presence of a full-length 5[prime]-UTR.
To test this, we programmed RRLs with either full-length 5[prime]-UTR AUUUA+A90 or full-length 5[prime]-UTR AUGUA+A90 GM-CSF mRNAs. Translation was allowed to proceed at either 30 (optimal), 22 or 15°C (sub-optimal). At 30°C, translation of both mRNAs was efficient and equivalent (Fig. 7, top). Translation of both mRNAs was suppressed at 22 (40% of optimal) and 15°C (20% of optimal). The blots shown here (Fig. 7, middle and lower), however, reflect longer exposures. Translation of ARE-containing mRNA was more significantly inhibited than that of ARE-negative mRNA (Fig. 7, middle and lower). At 15°C, full-length 5[prime]-UTR AUUUA+A90 mRNA failed to synthesize any detectable GM-CSF. Thus, global inhibition of translation caused by lowering the temperature unmasked the ARE as a translational repressor, despite the presence of a full-length 5[prime]-UTR.
Figure 7. The ARE suppresses translation from mRNA containing the full-length 5[prime]-UTR as reaction temperature is lowered. RRLs were programmed with GM-CSF mRNAs containing a full-length 5[prime]-UTR with either an intact or mutated ARE. Translation reactions were carried out at the indicated temperatures.
DISCUSSION
GM-CSF is a prototypical member of a family of post-transcriptionally regulated genes encoding several cytokines and proto-oncogenes (4,5). Conserved AREs in the 3[prime]-UTRs of their mRNAs target them for rapid degradation in resting cells (4-10). Transfection of human wild-type (hGM-AUUUA) and mutant (hGM-AUGUA) GM-CSF mRNAs into normal human lymphocytes demonstrated that deletion of the ARE increased mRNA half-life from 9 to 35 min and increased protein production 20-fold (25). However, since mRNA decay was concurrent with translation, the individual contributions of each process towards GM-CSF production was impossible to assess. In another study, when GM-CSF mRNAs were transfected into Xenopus oocytes the ARE was shown to function as a translational repressor (14). The GM-CSF mRNAs used in these studies, however, did not contain wild-type 5[prime]-UTRs (14, 25).
A number of recent studies have proposed that translation could be regulated through long-range interactions between the 5[prime]- and 3[prime]-ends of mRNAs (15-20). We have therefore re-examined the regulation of GM-CSF mRNA translation in the context of a full-length 5[prime]-UTR, utilizing an RRL cell-free translation system. Since the RRL contained specific ribonucleases that targeted ARE-containing mRNAs for rapid degradation (30), we programmed RRLs with an excess of mRNAs, so despite turnover, mRNAs did not become limiting to translation.
We have identified, for the first time, a key regulatory element that comprised the entire 5[prime]-UTR of GM-CSF mRNA which suppressed ARE and poly(A) tail-mediated inhibition of translation in the RRL. Our observations raised the question of how a regulatory element at the 5[prime]-end could suppress a distal second element at the 3[prime]-end of the same mRNA. In the RRL system, the ARE was able to funtion as a translational repressor when either: (i) the native 5[prime]-UTR was replaced by a synthetic leader sequence of similar length; (ii) global translation rate was diminished by lowering the temperature of the reaction. In either instance, decreased rates of translation initiation and (or) elongation unmasked the ARE. We presume that the underlying mechanism reflects ribosomal slowing along the mRNA. Under these circumstances, the ARE also loses its destabilizing capability (31). Thus, a high rate of translation as promoted by the full-length GM-CSF 5[prime]-UTR would support the ARE function as a destabilizing element while suppressing its potential as a translational inhibitor. The ARE-mediated decay of v-/c-fos recombinants was shown to be modulated by a splicing event that removed an optional exon located entirely in the 5[prime]-UTR (32). The full-length 5[prime]-UTR sequence was essential for rapid v-/c-fos mRNA decay and for the decay of chimeric mRNA containing a GM-CSF 3[prime]-UTR ARE (32).
Substitution and deletion mapping of different regions of the 5[prime]-UTR revealed that any changes in sequence or reduction of length substantially reduced translation. Thus the entire 5[prime]-UTR sequence was the minimal essential element needed for maintaining the highest rates of translation. This relatively short 32 nt leader has a high degree of sequence and length homology across species, consistent with a regulatory function. A putative stem-loop structure involving almost the entire sequence, had a [Delta]G of folding of -13.9 kcal/mol. The 5[prime]-UTRs of several other cytokine mRNAs, including TNF[alpha], IFN-[gamma], IL-1[alpha], IL-1[beta], IL-2 and IL-3, are similarly conserved among species. Although these ARE-containing mRNAs do not exhibit 5[prime]-UTR homology to each other, computer modeling showed stable stem-loop structures immediately 5[prime] of the start codons (results of computer modelling not shown; 33,34). Thus it is possible that many cytokines use highly conserved 5[prime]-UTRs as translational control elements.
Clusters of AUUU in the 3[prime]-UTR of TNF-[alpha] mRNA have been implicated as translational attenuators in unstimulated cells (35). The role of the 5[prime]-UTR in regulating translation of this mRNA has not been examined. In PBMC stimulated with recombinant C5a, IL-1[beta] mRNA was up-regulated, but with little or no expression of protein (36). The precise translational regulatory element(s) was not mapped. For a number of other cytokines/growth factors and proto-oncogenes, long 5[prime]-UTRs attenuate translation either through domains of strong secondary structure ([Delta]G = -40 to -80 kcal/mol) or through upstream open reading frames (37). The human GM-CSF 5[prime]-UTR contained neither of these features. However, a minimal 5[prime]-UTR secondary structure such as that predicted for GM-CSF has been associated with efficient translation of a number of different mRNAs (38,39). In transformed cells, mRNA isoforms containing truncated 5[prime]-UTRs with weak secondary structures permitted dysregulated expression of interleukin 15 (40) and transforming growth factor [beta]3 (41). Only a single form of human GM-CSF mRNA has been described and, based on our data, its 5[prime]-UTR should permit constitutive translation. In mouse, transcription from two alternative promoters produced GM-CSF mRNAs that differed only in the lengths of their 5[prime]-UTRs (42). The shorter mRNA contained a 5[prime]-UTR with high sequence and length homology to the human 5[prime]-UTR (Fig. 5), while the longer isoform contained an upstream AUG codon (42) that is generally inhibitory to translation (37). Currently, it is not known whether GM-CSF production is regulated in a cell- or tissue-specific manner through the use of alternatively processed 5[prime]-UTRs.
Expression of some growth factors are translationally modulated through the binding of cellular proteins to specific sequences in the 5[prime]-UTRs of their mRNAs. The 5[prime]-UTR of transforming growth factor [beta]1 contains a strong stem-loop structure that suppresses translation in adenocarcinomas, but permits translation in pheochromocytomas (43). This portion of the 5[prime]-UTR bound different cytoplasmic proteins in the two cell lines, likely accounting for the differences in translation. The insulin-like growth factor II (IGF-II) gene transcribed several mRNAs with different 5[prime]-UTRs that were expressed in a tissue-specific manner (44). Recently, a family of IGF-II mRNA-binding proteins (IMPs) were identified that suppressed translation during late embryonic development, through their specific association with the 5[prime]-UTR (45). One possible mechansim governing GM-CSF expression could therefore involve the masking of the 5[prime]-UTR element by a specific binding protein. When bound by protein, the 5[prime]-UTR would be unavailable to overcome translational repression by the 3[prime] ARE, effectively silencing GM-CSF mRNA translation.
In this study, the ARE failed to suppress GM-CSF mRNA translation in the absence of a poly(A) tail. A similar observation was made with IFN[beta] mRNA in an RRL system (22). An interaction between the ARE and the poly(A) tail could either involve direct base pairing or could be mediated by an mRNA-binding protein with affinities for both the ARE and the poly(A) sequence. Recently, a family of RNA-binding proteins (Elav-like proteins) was shown to have precisely this binding activity (46), suggesting a potential role for these proteins in ARE-mediated suppression of translation. Together, our data suggest that GM-CSF mRNA embodies opposing 5[prime] and 3[prime] regulatory elements that control its translation. In the RRL system, the 5[prime]-UTR element is predominant, driving high rates of translation. However, in intact resting cells, the regulatory balance between the 5[prime]- and 3[prime]-UTR elements reflects greater mRNA decay, silencing GM-CSF production.
ACKNOWLEDGEMENTS
We are grateful for the assistance and creative suggestions of members of the laboratory. This work was supported by Molecular Biosciences Training Grant T32 GM07215 (to J.A.J.) and National Institutes of Health Grant SCOR-Asthma P50HL56396 (to J.S.M.).
REFERENCES
*To whom correspondence should be addressed. Tel: +1 608 263 6043; Fax: +1 608 263 1568; Email: js.malter{at}uwmsg.hosp.wisc.edu The authors wish it to be known that, in their opinion, the first three authors should be regarded joint First Authors
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. Grosset, R. Boniface, P. Duchez, A. Solanilla, B. Cosson, and J. Ripoche In Vivo Studies of Translational Repression Mediated by the Granulocyte-Macrophage Colony-stimulating Factor AU-rich Element J. Biol. Chem., April 2, 2004; 279(14): 13354 - 13362. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Houzet, D. Morello, P. Defrance, P. Mercier, G. Huez, and V. Kruys Regulated control by granulocyte-macrophage colony-stimulating factor AU-rich element during mouse embryogenesis Blood, September 1, 2001; 98(5): 1281 - 1288. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Hasday, D. Bannerman, S. Sakarya, A. S. Cross, I. S. Singh, D. Howard, B.-E. Drysdale, and S. E. Goldblum Exposure to febrile temperature modifies endothelial cell response to tumor necrosis factor-{alpha} J Appl Physiol, January 1, 2001; 90(1): 90 - 98. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||