Portal de Periódicos da CAPES

The 5′-Untranslated Region of the FMR1 Message Facilitates Translation by Internal Ribosome Entry

2001; Elsevier BV; Volume: 276; Issue: 41 Linguagem: Inglês

10.1074/jbc.m105584200

1083-351X

Pei-Wen Chiang, Lauren E. Carpenter, Paul J. Hagerman,

RNA regulation and disease

Fragile X syndrome, the leading heritable form of mental impairment, is generally caused by large expansions of a CGG repeat in the promoter region of the FMR1 gene followed by transcriptional silencing. However, there is growing evidence that translation of the FMR1 message is also impaired, presumably because of the expanded CGG element in the 5′-untranslated region (5′-UTR) of the FMR1 message. To study the properties of the FMR1 5′-UTR, deletions were generated within a normal 5′-UTR with 16 CGG repeats for both monocistronic and dicistronic (luciferase) reporter constructs. Transient transfection experiments revealed a ∼20-nucleotide region upstream of the CGG repeat element that functions as an internal ribosome entry site (IRES). The normal CGG element itself does not appear to influence the efficiency of IRES-mediated stimulation of downstream reporter activity (∼18-fold over controls). Additional controls indicate that the enhanced activity of the downstream reporter is not due to readthrough from the upstream cistron, nor is it due to translation of cryptic monocistronic transcripts. The role of the FMR1 IRES element is not known at present; however, by analogy to other IRES-containing mRNAs expressed in neurons, the FMR1IRES element may help to promote translation in dendrites. Fragile X syndrome, the leading heritable form of mental impairment, is generally caused by large expansions of a CGG repeat in the promoter region of the FMR1 gene followed by transcriptional silencing. However, there is growing evidence that translation of the FMR1 message is also impaired, presumably because of the expanded CGG element in the 5′-untranslated region (5′-UTR) of the FMR1 message. To study the properties of the FMR1 5′-UTR, deletions were generated within a normal 5′-UTR with 16 CGG repeats for both monocistronic and dicistronic (luciferase) reporter constructs. Transient transfection experiments revealed a ∼20-nucleotide region upstream of the CGG repeat element that functions as an internal ribosome entry site (IRES). The normal CGG element itself does not appear to influence the efficiency of IRES-mediated stimulation of downstream reporter activity (∼18-fold over controls). Additional controls indicate that the enhanced activity of the downstream reporter is not due to readthrough from the upstream cistron, nor is it due to translation of cryptic monocistronic transcripts. The role of the FMR1 IRES element is not known at present; however, by analogy to other IRES-containing mRNAs expressed in neurons, the FMR1IRES element may help to promote translation in dendrites. 5′-untranslated region fragile X mental retardation 1 protein internal ribosome entry site firefly luciferase Renilla luciferase cytomegalovirus Fragile X syndrome (Mendelian Inheritance in Man number 309550) is the most common heritable form of mental impairment (1Hagerman R.J. Neurodevelopmental Disorders: Diagnosis and Treatment. Oxford University Press, New York1999: 61-132Google Scholar). Clinical involvement generally arises as a consequence of large expansions of a CGG trinucleotide repeat within the 5′-untranslated region (5′-UTR)1 of the fragile X mental retardation 1 (FMR1) gene (2Oberle I. Rousseau F. Heitz D. Kretz C. Devys D. Hanauer A. Boue J. Bertheas M.F. Mandel J.L. Science. 1991; 252: 1097-1102Crossref PubMed Scopus (1309) Google Scholar, 3Pieretti M. Zhang F.P. Fu Y.H. Warren S.T. Oostra B.A. Caskey C.T. Nelson D.L. Cell. 1991; 66: 817-822Abstract Full Text PDF PubMed Scopus (1235) Google Scholar, 4Verkerk A.J. Pieretti M. Sutcliffe J.S. Fu Y.H. Kuhl D.P. Pizzuti A. Reiner O. Richards S. Victoria M.F. Zhang F.P. Eussen B.E. van Ommen G.-J.B. Blonden L.A.I. Riggins G.J. Chastain J.L. et al.Cell. 1991; 65: 905-914Abstract Full Text PDF PubMed Scopus (2912) Google Scholar, 5Sutcliffe J.S. Nelson D.L. Zhang F. Pieretti M. Caskey C.T. Saxe D. Warren S.T. Hum. Mol. Genet. 1992; 1: 397-400Crossref PubMed Scopus (570) Google Scholar, 6Yu S. Mulley J. Loesch D. Turner G. Donnelly A. Gedeon A. Hillen D. Kremer E. Lynch M. Pritchard M. Sutherland G.R. Richards R.I. Am. J. Hum. Genet. 1992; 50: 968-980PubMed Google Scholar, 7Devys D. Lutz Y. Rouyer N. Bellocq J.P. Mandel J.L. Nat. Genet. 1993; 4: 335-340Crossref PubMed Scopus (633) Google Scholar). For expansions exceeding ∼200 repeats, the CG-rich promoter region and repeat element are normally hypermethylated, leading to transcriptional silencing. In the absence of message, no FMR1 protein (FMRP) is produced, resulting in fragile X syndrome (3Pieretti M. Zhang F.P. Fu Y.H. Warren S.T. Oostra B.A. Caskey C.T. Nelson D.L. Cell. 1991; 66: 817-822Abstract Full Text PDF PubMed Scopus (1235) Google Scholar). Despite its success in explaining much of the relationship between genotype and the cognitive component of the fragile X phenotype, the transcriptional silencing model cannot account for more recent observations in which FMRP levels are reduced in the premutation (55–200 repeats) and full mutation (>200 repeats) ranges for those individuals whose FMR1 mRNA levels are normal or even elevated (8Tassone F. Hagerman R.J. Loesch D.Z. Lachiewicz A. Taylor A.K. Hagerman P.J. Am. J. Med. Genet. 2000; 94: 232-236Crossref PubMed Scopus (126) Google Scholar, 9Tassone F. Hagerman R.J. Taylor A.K. Gane L.W. Godfrey T.E. Hagerman P.J. Am. J. Hum. Genet. 2000; 66: 6-15Abstract Full Text Full Text PDF PubMed Scopus (678) Google Scholar, 10Tassone F. Hagerman R.J. Chamberlain W.D. Hagerman P.J. Am. J. Med. Genet. 2000; 97: 195-203Crossref PubMed Scopus (170) Google Scholar, 11Tassone F. Hagerman R.J. Taylor A.K. Hagerman P.J. J. Med. Genet. 2001; 38: 453-456Crossref PubMed Scopus (64) Google Scholar). These results support an earlier observation of Fenget al. (12Feng Y. Zhang F. Lokey L.K. Chastain J.L. Lakkis L. Eberhart D. Warren S.T. Science. 1995; 268: 731-734Crossref PubMed Scopus (260) Google Scholar) that translation is impaired in the full mutation range and also suggest that translation is compromised even in the premutation range (8Tassone F. Hagerman R.J. Loesch D.Z. Lachiewicz A. Taylor A.K. Hagerman P.J. Am. J. Med. Genet. 2000; 94: 232-236Crossref PubMed Scopus (126) Google Scholar, 10Tassone F. Hagerman R.J. Chamberlain W.D. Hagerman P.J. Am. J. Med. Genet. 2000; 97: 195-203Crossref PubMed Scopus (170) Google Scholar). The GC-rich 5′-UTR of the FMR1 mRNA has the potential for forming a substantial secondary structure. Thus, it is not surprising that an expanded CGG repeat element might partially block cap-dependent translation initiation and scanning (13Kozak M. Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 2850-2854Crossref PubMed Scopus (500) Google Scholar, 14Kozak M. Mol. Cell. Biol. 1989; 9: 5134-5142Crossref PubMed Scopus (490) Google Scholar, 15Kozak M. J. Cell Biol. 1991; 115: 887-903Crossref PubMed Scopus (1451) Google Scholar). However, there are two potential problems with a simple scanning mechanism for the translation of FMR1 mRNA. First, forFMR1 alleles spanning the premutation range there are only modest reductions (<2–3-fold) in protein levels (8Tassone F. Hagerman R.J. Loesch D.Z. Lachiewicz A. Taylor A.K. Hagerman P.J. Am. J. Med. Genet. 2000; 94: 232-236Crossref PubMed Scopus (126) Google Scholar), and although those reductions may be attenuated by (partially) compensatory increases in mRNA levels, the free energy (ΔG0) required to disrupt the CGG element is predicted to increase from roughly 123 to 469 kcal/mol over that range (25Zuker M. Mathews D.H. Turner D. Barciszewski J. Clark B.C.F. RNA Biochemistry and Biotechnology. NATO ASI Series, Kluwer Academic Publishers Group, Dordrecht, The Netherlands1999: 11-43Crossref Google Scholar, 26Mathews D.H. Sabina J. Zuker M. Turner D.H. J. Mol. Biol. 1999; 288: 911-940Crossref PubMed Scopus (3211) Google Scholar). 2ΔG0 values were estimated for folding the GCC repeat element (RNA) by using Zuker's mfold server, version 3.1 (bioinfo.math.rpi.edu).2ΔG0 values were estimated for folding the GCC repeat element (RNA) by using Zuker's mfold server, version 3.1 (bioinfo.math.rpi.edu). Second, there are no reported trends in FMRP levels in the normal range (∼5–50 repeats), despite a substantial increase over that range (from 3.6 to 123 kcal/mol) in the predicted ΔGo for disruption of secondary structure within the CGG repeat. Thus, it appears that formation of a structured CGG repeat element does not affect a rate-limiting step (e.g. scanning) in the initiation of translation of the FMR1 mRNA. The absence of a strong dependence of translation efficiency on CGG repeat length may reflect the actions of transacting factors that could prevent the formation of the secondary structures predicted by the folding algorithms. However, it is also possible that initiation is mediated, at least in part, by an internal ribosome entry site (IRES) (16Pelletier J. Sonenberg N. Nature. 1988; 334: 320-325Crossref PubMed Scopus (1392) Google Scholar, 17Jang S.K. Krausslich H.G. Nicklin M.J. Duke G.M. Palmenberg A.C. Wimmer E. J. Virol. 1988; 62: 2636-2643Crossref PubMed Google Scholar, 18Chen C.Y. Sarnow P. Science. 1995; 268: 415-417Crossref PubMed Scopus (502) Google Scholar, 19Sachs A.B. Cell. 2000; 101: 243-245Abstract Full Text Full Text PDF PubMed Scopus (93) Google Scholar). IRES elements were first identified as elements used by picornaviruses to avoid viral-induced suppression of host 5′ cap-dependent translation. IRES elements bypass the translational requirement for 5′ cap-dependent scanning by promoting translation via internal ribosome entry. IRES elements have also been identified in the 5′-UTR regions of several key cellular regulatory proteins, including platelet-derived growth factor (20Bernstein J. Sella O. Le S.Y. Elroy-Stein O. J. Biol. Chem. 1997; 272: 9356-9362Abstract Full Text Full Text PDF PubMed Scopus (153) Google Scholar), c-Myc protooncogene (21Nanbru C. Lafon I. Audigier S. Gensac M.C. Vagner S. Huez G. Prats A.C. J. Biol. Chem. 1997; 272: 32061-32066Abstract Full Text Full Text PDF PubMed Scopus (210) Google Scholar), apoptotic protease-activating factor (22Coldwell M.J. Mitchell S.A. Stoneley M. MacFarlane M. Willis A.E. Oncogene. 2000; 19: 899-905Crossref PubMed Scopus (171) Google Scholar), and ornithine decarboxylase (23Pyronnet S. Pradayrol L. Sonenberg N. Mol. Cell. 2000; 5: 607-616Abstract Full Text Full Text PDF PubMed Scopus (279) Google Scholar). To determine whether IRES-mediated initiation could contribute to the translation of the FMR1 mRNA, we utilized dicistronic mRNAs in which a normal FMR1 5′-UTR element possessing 16 CGG repeats was placed in the intercistronic region between two reporter cistrons. IRES elements have been shown to increase translation of a downstream reporter cistron when placed in the intercistronic region of a dicistronic construct (24Jackson R.J. Kaminski A. RNA (NY). 1995; 1: 985-1000PubMed Google Scholar). The presence of the full-length 5′-UTR increased the translation of the second cistron by nearly 20-fold, thus establishing the presence of an IRES in the 5′-UTR of the FMR1 mRNA. Plasmid pGL3-Basic (Promega, Madison, WI) was digested with SacI andNcoI. The SacI site of the linearized plasmid was ligated to a SacI/BlpI adapter (5′-pCTAATACGACTCACTATAGGCCTCAGTCAGGCC /5′-pTGAGGCCTGACTGAGGCCTATAGTGAGTCGTATTAGAGCT; overhangs underlined; T7 promoter sequence in italics), and theNcoI site was ligated to an NheI/NcoI adapter (5′-pCTAGCAGGGCTGAAGAGAA/5′p-CATGTTCTCTTCAGCCCTG). The resulting (linear) plasmid with adapter ends was then ligated to aBlpI/NheI fragment of the 5′-UTR ofFMR1, which had been released from plasmid pE5.1 (4Verkerk A.J. Pieretti M. Sutcliffe J.S. Fu Y.H. Kuhl D.P. Pizzuti A. Reiner O. Richards S. Victoria M.F. Zhang F.P. Eussen B.E. van Ommen G.-J.B. Blonden L.A.I. Riggins G.J. Chastain J.L. et al.Cell. 1991; 65: 905-914Abstract Full Text PDF PubMed Scopus (2912) Google Scholar) byBlpI/NheI double digestion. The 5′-UTR of pE5.1 contains a trinucleotide repeat element of the form, (CGG)16 (nucleotides −117 to −70) (Fig.1). The BlpI/NheI fragment represents nucleotides −238 to −21 of the FMR15′-UTR (5′ nucleotides of the restriction sites; numbering relative to +1 of the initiation codon). The −20 to −2 region of the 5′-UTR is restored by the NheI/NcoI adapter, which differs from the native FMR1 sequence only at the −1 position (G to C change). The SacI/BlpI adapter restores the remaining portion of the native 5′-UTR (through nucleotide −252), with the 5′ G of the native FMR1 mRNA equivalent to the 5′ G of transcripts generated from the T7 promoter cassette. The resulting plasmid, pGL3-FMR-FL (FL, firefly luciferase), was used as a starting point for the construction of the 5′-UTR deletion plasmids. The various 5′-UTR deletion constructs, pGL3-Δ(i/j)-FL (i and j are the terminal nucleotides of the deletions, relative to the start codon), were generated from pGL3-FMR-FL by replacement of theBlpI-NheI segment with synthetic adapters encompassing different portions of the 5′-UTR. The deletion Δ(−222/−62) creates an NruI site (TCGCGA) across the deletion junction. The deletion Δ(−145/−65) was generated by insertion of an adapter into the NruI site, creating aStuI site (AGGCCT) across the junction. The deletion Δ(−117/−70) was created by insertion of an adapter into theStuI site; this deletion exactly removes the CGG repeat element. The remaining deletions (Δ(−168/−65), Δ(−188/−65), Δ(−198/−65), and Δ(−208/−65)) were created from Δ(−222/−62) by insertion of the corresponding adapters into the NruI site. The largest deletion, Δ(−222/−22), contains a trinucleotide bridge (5′-CGA/5′-TCG) of heterologous sequence between nucleotides −223 and −21 of the FMR1 5′-UTR. Sequencing of this construct revealed a C to G change at position −232. An XbaI linker (5′-CTAGTGGTACCTGCT/5′CTAGAGCAGGTACCA) containing a KpnI site was inserted into the XbaI site of plasmid pRL-CMV (Promega); the XbaI site lies immediately 3′ of the TAA stop codon of the Renilla luciferase (RL) gene. The resulting plasmid (pRL-CMV-K) retains a single XbaI site downstream of the KpnI site. For each of the pGL3-Δ(i/j)-FL plasmids, the region containing the coding portion of the firefly luciferase gene and the FMR1 5′-UTR was isolated byKpnI/XbaI double digestion. Those fragments were then inserted between the KpnI and XbaI sites of pRL-CMV-K to yield the dicistronic expression plasmids, pRL-Δ(i/j)-FL. For each plasmid of this set, the intergenic region possesses an additional 36 nucleotides between the TAA termination codon and the 5′ G residue of the native FMR1 mRNA. The dicistronic control plasmid, pRL-C-FL, was generated by inserting theKpnI-XbaI fragment from pGL3-basic into pRL-CMV-K, yielding a 94-nucleotide intercistronic region that consists mainly of the multiple cloning site of pGL3-basic. An inverted repeat element was introduced into pRL-CMV-K by inserting the synthetic fragment, 5′-pCTAGCTGAACTGGGAGTGGACACCTG/5′-pCTAGCAGGTGTCCACTCCCAGTTCAG into the NheI site 10 nucleotides upstream of the start codon of the Renilla luciferase gene. Twenty-one base pairs of the insert (italics) are complementary to an upstream sequence block, such that the transcript forms a hairpin with a 21-base pair stem and 50-nucleotide loop. The predicted ΔG0of stabilization for this hairpin is −47.9 kcal/mol (mfold server) (25Zuker M. Mathews D.H. Turner D. Barciszewski J. Clark B.C.F. RNA Biochemistry and Biotechnology. NATO ASI Series, Kluwer Academic Publishers Group, Dordrecht, The Netherlands1999: 11-43Crossref Google Scholar, 26Mathews D.H. Sabina J. Zuker M. Turner D.H. J. Mol. Biol. 1999; 288: 911-940Crossref PubMed Scopus (3211) Google Scholar). The resulting plasmid is designated pHPRL-CMV-K. TheKpnI-XbaI fragment from pGL3-FMR-FL, containing the full-length 5′-UTR of FMR1 and the firefly luciferase gene, was then inserted into the KpnI/XbaI double-digest of pHPRL-CMV-K to create pHPRL-FMR-FL. Two plasmids harboring sequence elements that span the region of potential IRES activity were constructed by inserting syntheticBlpI-NheI fragments into the corresponding location in pRL-FMR-FL. The plasmids are designated pRL-I[−222(−231)/−190(−21)]-FL and pRL-I[−222(−231)/−146(−21)]-FL, where numbers without parentheses are the positions of the 5′ and 3′ nucleotides, respectively, for the element being tested (original numbering). The numbers in parentheses represent the new locations of those nucleotides (original numbering). Thus, the G residue at position −222 is relocated to a position immediately 3′ to the Blp I site, and the G residues at positions −190 or −146 are relocated to position −21 (5′ nucleotide of the NheI site). Fragments released from pGL3-Δ(i/j)-FL by digestion with KpnI andXbaI (firefly luciferase coding region plus deletions of theFMR1 5′-UTR) were ligated to KpnI/PstI adapters (5′-GCTGCAGGTAC/5′-CTGCAGCGTAC; nestedPstI site in italics). After digestion with PstI, the resulting (PstI-XbaI) fragments were inserted into pSP64 Poly(A) vector (Promega) between the PstI andXbaI sites to form a series of in vitroexpression plasmids, designated pSP6-Δ(i/j)-FL. For in vitro transcription reactions, these plasmids were linearized withEcoRI, which cuts just downstream of the A30element, and were transcribed using SP6 RNA polymerase (Life Technologies, Inc.) according to the manufacturer's protocols. A control plasmid, pSP6-RL, was generated by inserting theNheI-XbaI fragment from pRL-CMV into theXbaI site of pSP6-poly(A). SK-N-MC cells (neuroepithelial origin; ATCC) were maintained in Eagle's minimal essential medium supplemented with 2 mm l-glutamine and Earle's balanced salt solution. The medium was adjusted to final concentrations of 1.5 g/liter sodium bicarbonate, 0.1 mm non-essential amino acids, 1.0 mm sodium pyruvate, and 10% fetal calf serum. Lymphoblastoid cell lines GM (GM4025; ATCC) and AG (AG09391; NIA Cell Repository) were grown in RPMI 1640 supplemented with 10% fetal calf serum and glutamine. The GM line was derived from a fragile X male with a large, full mutation expansion; the cells produce no detectable FMR1 mRNA or FMRP. The AG line was derived from a normal female. Cells were grown in 24-well plates at 37 °C in an atmosphere of 5% CO2. The cells were 90% confluent at the time of transfection. For lymphoblastoid cell lines, 4 × 105 cells were used for each transfection. Transfection reactions used 3 μl of LipofectAMINE2000 (Life Technologies, Inc.) mixed with 1 μg of plasmid DNA or RNA in Opti-MEM media (Life Technologies, Inc.). Twenty-four hours post-transfection, cells were lysed with passive lysis buffer (Promega), and dual luciferase assays were performed following the manufacturer's protocol. Twenty microliters of cell lysate was mixed with 100 μl of Luciferase Assay Reagent II (Promega). The firefly luciferase activity of each mixture was measured for 10 s, following a 2-s delay, using an Analytical Luminescence Lab Monolight 3010 luminometer with dual injectors. One hundred μl of Stop and Glo reagent (Promega) was added to quench the firefly luciferase activity, and the activity of the Renillaluciferase was measured for 10 s, following a 2-s delay. All readings were greater than 100-fold above background, and all experiments were performed at least in triplicate. Total RNA was extracted from cells using Trizol as recommended by the manufacturer (Life Technologies, Inc.). Purified (total) RNA was quantified spectrophotometrically, and 25 μg of total RNA was run on formaldehyde-containing agarose gels. Digoxygenin-labeled probes were generated using the DIG High Prime DNA labeling kit (Roche Molecular Biochemicals). Hybridization and detection were carried out using the DIG Detection Kit II (Roche Molecular Biochemicals) as recommended by the manufacturer. To evaluate the potential of theFMR1 5′-UTR to initiate translation via internal ribosome entry, dicistronic reporter plasmids were constructed in which the 5′-UTR of FMR1 was placed in the intercistronic region between upstream RL and downstream FL coding sequences (Fig.2). The dicistronic region was placed under the control of the CMV promoter. For the initial set of experiments, dicistronic reporter plasmids possessing the entire 5′-UTR of FMR1 (pRL-FMR-FL), a substantial deletion thereof (pRL-Δ(−222/−22)-FL), or a control sequence (pRL-C-FL) were used to transfect SK-N-MC cells. Twenty-four hours after transfection, cells were harvested and lysed, and the ratios of FL to RL luminescence were measured. The results of these measurements are presented in Fig. 2. In this initial set of experiments, the luminescence ratio (FL/RL) was found to be ∼18-fold higher in cells transfected with pRL-FMR-FL compared with cells that were transfected with either deletion or control plasmids. This observation suggests that translation of the downstream (FL) reporter is initiated by internal ribosome entry within the FMR1 5′-UTR. To address the possibility that the elevated levels of FL fluorescence were due to translation from monocistronic (FL) mRNA species, a Northern blot analysis was performed using a DNA fragment from the FL coding region as a probe, and equivalent amounts of total RNA were isolated from SK-N-MC cells that had been transfected with each of the three reporter plasmids (Fig. 3). Each of the transformants expressed predominantly a single RNA species of the expected size (∼4.2 kilobases). The sizes of the deletion and control RNAs are slightly reduced relative to the dicistronic mRNA with the full UTR, as expected for their shorter intercistronic regions. Finally, the reporter mRNAs were present at comparable levels relative to whole cell RNA levels, thus ruling out altered transcriptional efficiency or stability as explanations for increased downstream reporter activity with the full 5′-UTR. In a second control experiment, we examined whether expression of the FL cistron was the result of readthrough from the RL reporter by introducing a stable hairpin (see "Experimental Procedures") upstream of the RL coding sequence (i.e. in the 5′-UTR of the RL reporter). After addition of the hairpin element, the activity of the RL reporter was strongly decreased (Fig.4), with reductions comparable with those observed by Kozak (13Kozak M. Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 2850-2854Crossref PubMed Scopus (500) Google Scholar) for a −50 kcal/mol stem-loop structure. However, the activity of the FL reporter was not reduced; in fact, its activity actually increased. These data support the conclusion that translation of the second cistron is not due to readthrough from the upstream cistron but occurs as a consequence of IRES activity within the FMR1 5′-UTR. The IRES activity of the FMR1 5′-UTR was also tested using the pRL constructs in two lymphoblastoid cell lines, AG (normalFMR1 allele) and GM (full mutation allele producing noFMR1 mRNA or FMRP). 3F. Tassone and R. J. Hagerman, unpublished data. In both of these non-neural cell lines, the relative luminescence ratios (FL/RL) were elevated for dicistronic mRNAs with the full-length FMR15′-UTR. The level of enhancement in AG cells (fold enhancement: 16.9 ± 1.3, relative to deletion or control plasmids) was nearly identical to the degree of enhancement found with the SK-N-MC line. For GM, the elevation was not as great (fold enhancement: 11.2 ± 0.8). We do not yet know whether the moderate reduction in GM cells is due to dominant negative effects within the non-expressing cell line (e.g. because of the absence of FMRP); this issue is currently under investigation. To further define the sequence elements that facilitate internal ribosome entry, expression plasmids, pRL-Δ(i/j)-FL, were generated with nested deletions within theFMR1 5′-UTR (i and j are the 5′- and 3′-nucleotides of the deleted region). Each plasmid was used to transfect SK-N-MC cells, and the resultant transformants were tested for relative FL activity (Fig.5). Deletions of nucleotides −198 through −65, which include the CGG repeat element, appeared to have little effect on IRES activity. Deletion of an additional 10 nucleotides in the 5′ direction, through −208, produced a slight drop in translation of the second cistron. Deletion through −222 resulted in a dramatic drop in translation of FL, approaching background (control) levels. These observations point to sequence elements between nucleotides −223 and −209 as being important for IRES activity within the FMR1 5′-UTR. Portions of the 5′-UTR spanning this region, when introduced between the Blp I andNheI sites, were capable of partially restoring IRES activity. Thus, the region between −223 and −209 appears to act directly as a site (or a portion thereof) for internal ribosome entry. It is not surprising that inserted sequences (−222 through −145 and −222 through −189) fail to restore full IRES activity because the length of the intercistronic region and the secondary and/or tertiary structure effects are likely to participate in internal ribosome entry. Analysis of the 5′-UTR of FMR1 using RNA folding algorithms (mfold) (25Zuker M. Mathews D.H. Turner D. Barciszewski J. Clark B.C.F. RNA Biochemistry and Biotechnology. NATO ASI Series, Kluwer Academic Publishers Group, Dordrecht, The Netherlands1999: 11-43Crossref Google Scholar,26Mathews D.H. Sabina J. Zuker M. Turner D.H. J. Mol. Biol. 1999; 288: 911-940Crossref PubMed Scopus (3211) Google Scholar) suggests that 5′-UTRs with CGG expansions in the normal range have the potential to form very stable secondary structures. For example, folding of the entire 5′-UTR with a twenty-repeat CGG element yields ΔGo values in the −150 kcal/mol range for the most stable structures. Secondary structures predicted by such algorithms are hypothetical in the absence of direct experimental evidence for their formation; however, if formed in vivo, such structures would likely be inhibitory to scanning (13Kozak M. Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 2850-2854Crossref PubMed Scopus (500) Google Scholar, 14Kozak M. Mol. Cell. Biol. 1989; 9: 5134-5142Crossref PubMed Scopus (490) Google Scholar, 15Kozak M. J. Cell Biol. 1991; 115: 887-903Crossref PubMed Scopus (1451) Google Scholar). To determine whether the FMR1 5′-UTR is inhibitory to translation in an upstream context, the order of the two reporter genes was reversed with the FMR1 5′-UTR at the 5′ end of the dicistronic transcript preceded by a 5′cap and followed by FL and RL coding regions, in that order. In this arrangement, the RL cistron is translated at low levels and acts as a baseline control. Deletion of major portions of the FMR1 5′-UTR, including the CGG element, increased translation of the lead (FL) cistron by only 1.8 (± 0.25)-fold. More extensive deletions, including the putative IRES region, had little additional effect. Thus, at least for a 5′-UTR with a CGG repeat element in the normal range, the full-length 5′-UTR did not display strong inhibition or promotion of translation when preceded by a 5′-cap. In a second approach, monocistronic constructs were generated by placing the full-length or deleted 5′-UTRs upstream of the FL cistron in the pSP64 Poly(A) vector. Monocistronic mRNAs possessing both a 5′-cap and an A30 tail were then produced in vitro using SP6 RNA polymerase. An internal control (RL) mRNA, with both a cap and poly(A) tail, was also produced. The monocistronic RL and FL mRNAs were then cotransfected into SK-N-MC cells. Again, little effect was observed upon deleting major portions of the 5′-UTR; specifically, neither the deletion of the CGG repeat nor the deletion of the IRES element had any significant influence on the level of translation of the FL cistron (Fig. 6). Thus, a normal CGG repeat element (16 repeats) has only a modest influence on the efficiency of translation of a heterologous reporter. In addition, these results suggest that the IRES element makes only a marginal contribution (less than 10%) to the efficiency of translation of the monocistronic reporter under the experimental conditions used for these investigations. An IRES is generally defined operationally as a sequence element that can facilitate the initiation of translation of a downstream cistron in a heterologous, dicistronic mRNA. (24Jackson R.J. Kaminski A. RNA (NY). 1995; 1: 985-1000PubMed Google Scholar, 27Johannes G. Sarnow P. RNA (NY). 1998; 4: 1500-1513Crossref PubMed Scopus (221) Google Scholar) By this measure, the 5′-UTR of the FMR1 mRNA possesses significant IRES activity, with translation of the downstream (FL) reporter cistron enhanced by nearly 20-fold when the FMR15′-UTR constitutes the intercistronic region. The IRES activity is enhanced when translation of the upstream (RL) reporter cistron is blocked by the addition of a hairpin element in the upstream UTR. This observation militates against a simple readthrough mechanism for translation of the second cistron. When compared with reported levels of IRES-mediated stimulation for other candidate IRESs, the FMR1 IRES activity would appear to be of moderate strength. However, there are several caveats associated with such comparisons, including intrinsic differences in the cell type used for transfection and the nature of the reporter constructs and controls. In addition, Kozak (28Kozak M. Gene (Amst.). 1999; 234: 187-208Crossref PubMed Scopus (1128) Google Scholar) has raised the possibility that some of the translational activity attributed to IRES-mediated initiation may, in fact, be due to translation of a monocistronic message comprising the downstream reporter and part or all of the intercistronic region. Such monocistronic species may be produced either by cryptic downstream transcriptional start sites (29Akiri G. Nahari D. Finkelstein Y. Le S.Y. Elroy-Stein O. Levi B.Z. Oncogene. 1998; 17: 227-236Crossref PubMed Scopus (222) Google Scholar) or by cleavage of the dicistronic mRNA through various mechanisms, including splicing. The current study has addressed the issue of monocistronic species in two ways. First, we have found that for total RNA isolates from cells harboring dicistronic mRNAs, the (fractional) amount of mRNA with the full-length (FMR1) 5′-UTR is essentially equal to the corresponding mRNA with the Δ(−222/−22) deletion, arguing against reduced stability of the dicistronic construct with the full-length (FMR1) intercistronic region. Furthermore, using Northern analysis (Fig. 3), no monocistronic RNA species are detected by densitometry at the 1–2% level using probes specific to the downstream (FL) coding region. This latter observation argues against UTR-dependent splicing events or cryptic downstream promoters that could potentially favor the full-length FMR1UTR. A slightly smaller RNA species was detected at the 1–2% level in all three lanes. However, this species could not account for the enhanced translation observed for the dicistronic construct with the full-length FMR1 UTR sequence, because the monocistronic form would have to initiate translation with ∼18-fold greater efficiency than the species in the other two lanes. In fact, as noted above, the full-length FMR1 5′-UTR initiates with lower efficiency than the deletion or control species. Kozak (28Kozak M. Gene (Amst.). 1999; 234: 187-208Crossref PubMed Scopus (1128) Google Scholar) raised another possible source of adventitious IRES activity, one involving potential misinterpretation of the "hairpin test" due to unexpected downstream splicing events or a second promoter. Again, the absence of detectable monocistronic species supports the interpretation of the hairpin test in the current investigation, namely, that translation of the second cistron is initiated in the intercistronic region. Sequences between −223 and −209 (Fig. 5) appear to contribute most of the observed IRES activity of the FMR1 5′-UTR in the dicistronic context. This sequence is located mainly within a 21-nucleotide, pyrimidine-rich region (81% Y versus 43% Y overall for the 5′-UTR) spanning nucleotides −232 through −212. This region includes two UUUC sequences and a CUUC sequence, each separated by one or two purines. Thus, with respect to the presence of UUUC elements, the −232 through −212 region of the FMR1 5′-UTR has characteristics of IRES elements found in some other viral and non-viral 5′-UTRs (23Pyronnet S. Pradayrol L. Sonenberg N. Mol. Cell. 2000; 5: 607-616Abstract Full Text Full Text PDF PubMed Scopus (279) Google Scholar, 30Nicholson R. Pelletier J. Le S.Y. Sonenberg N. J. Virol. 1991; 65: 5886-5894Crossref PubMed Google Scholar, 31Pilipenko E.V. Gmyl A.P. Maslova S.V. Svitkin Y.V. Sinyakov A.N. Agol V.I. Cell. 1992; 68: 119-131Abstract Full Text PDF PubMed Scopus (222) Google Scholar). After removal of part of this IRES region (through −222), leaving the upstream UUUC sequence, translation was reduced to baseline levels. Other features of viral IRESs, including their relative proximity to the start codon, are not preserved in the FMR1 5′-UTR. It is unclear at present what biological role the FMR1 IRES normally plays in regulating the expression of FMRP. It is possible that the IRES is involved with the regulation of translation within dendrites in response to synaptic activity (32Weiler I.J. Irwin S.A. Klintsova A.Y. Spencer C.M. Brazelton A.D. Miyashiro K. Comery T.A. Patel B. Eberwine J. Greenough W.T. Proc. Natl. Acad. Sci. U. S. A. 1997; 94: 5395-5400Crossref PubMed Scopus (544) Google Scholar, 33Weiler I.J. Greenough W.T. Am. J. Med. Genet. 1999; 83: 248-252Crossref PubMed Scopus (131) Google Scholar), perhaps through the action of IRES-transacting factors (34Pilipenko E.V. Pestova T.V. Kolupaeva V.G. Khitrina E.V. Poperechnaya A.N. Agol V.I. Hellen C.U. Genes Dev. 2000; 14: 2028-2045PubMed Google Scholar) or pyrimidine tract-binding protein (35Kim Y.K. Hahm B. Jang S.K. J. Mol. Biol. 2000; 304: 119-133Crossref PubMed Scopus (64) Google Scholar, 36Gosert R. Chang K.H. Rijnbrand R. Yi M. Sangar D.V. Lemon S.M. Mol. Cell. Biol. 2000; 20: 1583-1595Crossref PubMed Scopus (111) Google Scholar, 37Hunt S.L. Jackson R.J. RNA (NY). 1999; 5: 344-359Crossref PubMed Scopus (130) Google Scholar, 38Hellen C.U. Wimmer E. Curr. Top. Microbiol. Immunol. 1995; 203: 31-63PubMed Google Scholar). It is also possible that the IRES acts only at specific stages of development or only in certain regions, for example, to promote translation in dendrites (39Pinkstaff J.K. Chappell S.A. Mauro V.P. Edelman G.M. Krushel L.A. Proc. Natl. Acad. Sci. U. S. A. 2001; 98: 2770-2775Crossref PubMed Scopus (147) Google Scholar). In this regard, we do not yet know whether the comparable levels of IRES activity observed in SK-N-MC, lymphoblastoid, and HeLa (data not shown) reflect the absence of a highly cell type-specific function for the FMR1 IRES, or whether the lack of cell type specificity is simply the result of the cell lines chosen for the current experiments. Clearly, more work is needed in this area. One inference can be drawn from the maintenance of IRES activity in the GM cell line, namely that the absence of FMRP does not strongly influence IRES activity. More generally, the absence of FMRP does not decrease the activity of reporters (e.g. FL) whose translation is controlled by the FMR1 5′-UTR. Thus, IRES function does not appear to be coupled to FMRP-mediated mechanisms of translation. If the purpose of the IRES is to avoid problems of secondary structure impeding translation, it is difficult to rationalize why the IRES element would be located upstream of the CGG repeat because the latter is presumed (although not demonstrated) to block translation via scanning when the repeat length is expanded into the premutation or full mutation ranges (8Tassone F. Hagerman R.J. Loesch D.Z. Lachiewicz A. Taylor A.K. Hagerman P.J. Am. J. Med. Genet. 2000; 94: 232-236Crossref PubMed Scopus (126) Google Scholar, 12Feng Y. Zhang F. Lokey L.K. Chastain J.L. Lakkis L. Eberhart D. Warren S.T. Science. 1995; 268: 731-734Crossref PubMed Scopus (260) Google Scholar). Deletion of the normal range CGG repeat resulted in only a marginal increase in the level of translation of the FL reporter in the monocistronic construct, despite the expectation that the FMR1 5′-UTR (current study) should have extensive secondary structure. Interestingly, addition of a hairpin element with a much smaller predicted free energy (ΔGo) of stabilization did block translation. Taken together, these results suggest that the postulated 5′-UTR structure may not be forming as proposed, an important caveat with the use of folding algorithms. If the role of the IRES element is to promote translation in dendrites, rather than to avoid the CGG repeat, the upstream location of the IRES would not be a problem for normal IRES activity. Future studies will explore IRES function in the premutation range to further characterize the affect of CGG repeat length. It is clear that additional investigation is required to elucidate the roles played by the CGG repeat element and the IRES element in regulating the translation of the FMR1 mRNA. We have not addressed the possibility of interplay between the 5′ and 3′UTRs; the latter does play important regulatory roles, and interaction between 5′ and 3′UTRs via the poly(A)-binding protein and eIF4G is important for initiation (19Sachs A.B. Cell. 2000; 101: 243-245Abstract Full Text Full Text PDF PubMed Scopus (93) Google Scholar). Such interactions may underlie the apparent increase in the activity of the downstream reporter (FL) when translation of the upstream reporter (RL) is blocked with a hairpin element, perhaps due to the absence of competition with the 5′-cap site for interaction with the 3′UTR. Finally, although the biological significance of the observed IRES activity has not been defined, the IRES element represents a potentially important therapeutic target for fragile X. It has been demonstrated (11Tassone F. Hagerman R.J. Taylor A.K. Hagerman P.J. J. Med. Genet. 2001; 38: 453-456Crossref PubMed Scopus (64) Google Scholar) that a significant fraction of males with fragile X syndrome (i.e. with little or no FMRP) nevertheless continues to produce significant amounts of FMR1 mRNA. Thus, for these individuals it may not be necessary to target the gene itself—it is already active. Rather, one could target the mRNA directly in an effort to enhance translation of an existing message. The IRES element represents one such potential target. We thank Dr. Leslie Krushel for fruitful discussions regarding his unpublished IRES-related work.