A B12-responsive Internal Ribosome Entry Site (IRES) Element in Human Methionine Synthase*

Regulation of homocysteine, a sulfur-containing amino acid that is a risk factor for cardiovascular diseases, is poorly understood. Methionine synthase (MS) is a key enzyme that clears intracellular homocysteine, and its activity is induced by its cofactor, vitamin B12, at a translational level. In this study, we demonstrate that translation of MS, which has a long and highly structured 5′-untranslated region, is initiated from an internal ribosome entry site (IRES), which is modulated by B12. The minimal IRES element spans 71 bases immediately upstream of the initiation codon. Electrophoretic mobility shift analysis reveals the presence of a B12 -dependent protein-RNA complex and suggests the possibility that B12-dependent increase of IRES efficiency is mediated via a protein. Modulation of the IRES-dependent translation of an essential gene by the cofactor of the encoded enzyme represents a novel example of a gene-nutrient interaction.

Regulation of homocysteine, a sulfur-containing amino acid that is a risk factor for cardiovascular diseases, is poorly understood. Methionine synthase (MS) is a key enzyme that clears intracellular homocysteine, and its activity is induced by its cofactor, vitamin B 12 , at a translational level. In this study, we demonstrate that translation of MS, which has a long and highly structured 5-untranslated region, is initiated from an internal ribosome entry site (IRES), which is modulated by B 12 . The minimal IRES element spans 71 bases immediately upstream of the initiation codon. Electrophoretic mobility shift analysis reveals the presence of a B 12 -dependent protein-RNA complex and suggests the possibility that B 12dependent increase of IRES efficiency is mediated via a protein.

Modulation of the IRES-dependent translation of an essential gene by the cofactor of the encoded enzyme represents a novel example of a gene-nutrient interaction.
The important interplay between genes and nutrients modulates a number of physiological and pathophysiological processes (1). The influence of nutrients on DNA stability, repair, methylation, and gene expression is well studied (2). Elevated levels of homocysteine, a metabolic product of an essential nutrient, methionine, is correlated with an increased risk for cardiovascular diseases (3), neural tube defects (4), and Alzheimer's disease (5). Intracellular clearance of homocysteine is controlled by the activity of three enzymes that are found at a vitaminrich metabolic junction. These include the B 6 -dependent cystathionine ␤-synthase, B 12 -and folate-dependent methionine synthase (MS) 2 and betaine homocysteine methyl transferase. Of these, only MS is ubiquitous and remethylates homocysteine to methionine. Mutations in MS result in hereditary hyperhomocysteinemia (6,7). The rich B-vitamin dependence of homocysteine metabolism has stimulated studies on the benefits of multivitamin treatment in lowering plasma homocysteine (8 -11).
The activity of MS in cells cultured in normal medium is enhanced by supplementation with vitamin B 12 , an observation that was first reported over 30 years ago (12). This regulation is exerted at the translational level and involves a B 12 responsive element located at the 3Ј-end of the MS 5Ј-UTR (13). B 12 supplementation shifts the equilibrium of MS mRNA from the translationally inactive ribonucleoprotein pool to the active polysomal pool and thereby increases the level of MS.
The MS 5Ј-UTR is 394 bases in length and is predicted to be highly structured, which raises questions about how it can be efficiently translated. Translation initiation of most eukaryotic genes commences with recruitment of the initiation complex to the 5Ј-m 7 G cap structure of the mRNA. The initiation complex then scans the leader region until the initiator AUG codon is located. Leader-burdened mRNAs in eukaryotes can employ alternative mechanisms for translation initiation, i.e. ribosomal shunting or an internal ribosome entry site (IRES) element (14). Whereas there is little information on the mechanism of ribosome shunting, IRES elements are capable of recruiting ribosomal subunits in the vicinity of the initiation codon, thus bypassing the canonical caprecognition and scanning route. Cellular IRES activities can be modulated in response to amino acid availability, hypothermia, or during apoptosis and development and afford a route for translation initiation under conditions where global cap-dependent initiation is inhibited (14 -16).
In this study, we have investigated whether the human MS mRNA is translated through an internal initiation mechanism. We demonstrate that the MS mRNA harbors an IRES element in its leader region, and its boundaries overlap with those of the previously described B 12 response element. The IRES element is located immediately upstream of the initiator AUG, and its activity is modulated by vitamin B 12 . B 12 does not appear to bind directly to the MS 5Ј-UTR and supports a model in which the cofactor regulates the IRES element via an auxiliary protein.

EXPERIMENTAL PROCEDURES
Materials-Eagle's MEM (minimum essential medium), OHCbl, AdoCbl, MeCbl, and CNCbl were purchased from Sigma. Fetal bovine serum was from HyClone. Cell lines were purchased from American Type Culture Collection. Radiolabeled [␥-32 P]ATP (5000 mCi/mmol) was purchased from Amersham Biosciences. HeLa S100 cell extracts were purchased from Paragon.
Cell Culture Conditions-Cells were grown in Eagle's MEM supplemented with 10% fetal bovine serum and incubated at 37°C, 5% CO 2 . B 12 derived from fetal bovine serum is present at a concentration of ϳ125 pM in this medium. For B 12 induction studies, the cells were grown to 60 -80% confluency, and fresh medium supplemented with 5 mg liter Ϫ1 OHCbl (3.6 M final concentration) was added. For reporter studies (luciferase and CAT), cells were grown in 6-well 35-mm plates, harvested, and lysed according to the manufacturer's protocols (Promega).
Reporter Constructs-Plasmids pSVCAT/ICS/LUC, pSVCAT/BIP/ LUC, and pSVhpCAT/BIP/LUC were kindly provided by Dr. Maria Hatzoglou (Case Western Reserve University) and were initially developed in Dr. Peter Sarnow's laboratory (Stanford University). The pSV-CAT/ICS/LUC vector contains 400 nucleotides of antisense antennapedia cDNA of Drosophila melanogaster (17) in the intercistronic (ICS) region and was used as a negative control for IRES-mediated translation. * This work was supported by Grant DK64959 from the National Institutes of Health and by a predoctoral fellowship from the Heartland Affiliate of the American Heart Association (to S. O.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The online version of this article (available at http://www.jbc.org) contains supplemental Table S1. 1 To whom correspondence should be addressed. The plasmid pSVCAT/BIP/LUC contains the IRES of the immunoglobulin-binding protein (17) and was used as a positive control. The plasmid pSVhpCAT/BIP/LUC contains a stable hairpin inserted in front of the first cistron. The MS 5Ј-UTR (394 bases) was PCR-amplified and cloned into the SalI/NcoI sites of the pSVCAT/ICS/LUC plasmid by replacing the ICS sequence to give pSVCAT/MS1-394/LUC. Using a similar procedure, the vector pSVhpCAT/MS1-394/LUC was obtained. Deletion constructs containing the last 340, 270, 220, 140, and 71 bases, of the MS 5Ј-UTR were generated by PCR and subcloned into the SalI/NcoI sites of the bicistronic vector (with or without the hairpin) as described above for the full-length construct. The deletion constructs containing the last 67, 64, 61, 55, 52, 47, 42, 35, and 26 bases, respectively, of the MS 5Ј-UTR were generated by PCR and subcloned into the SalI/BstEII sites. Plasmid MS-pGL3-basic was constructed by cloning the MS 5Ј-UTR into the HindIII/NcoI sites of the pGL3-basic promoter-less vector (Promega). Mutants were generated using the QuikChange site-directed mutagenesis kit (Stratagene). The primers employed in this study are described in supplemental Table S1.
Transient Transfection and Reporter Assays-Transfections were performed using the Lipofectamine reagent (Invitrogen) for 293 cells or GeneJammer (Stratagene) for COS-1, HepG2, and NIH3T3 cells. Briefly, 1.5 g of plasmid DNA was mixed with 10 l of transfection reagent according to the manufacturer's specifications, and the mixture was added to 6-well plates. When needed, B 12 was added to half of the plates 24 h after transfection. At the end of the incubation, reporter gene activity was determined according to the vendor's protocol (Promega).
Bicistronic RNA Northern Analysis-Analysis was performed as previously described (13). A 32 P-labeled probe consisting of a DNA fragment encompassing 650 bases at the 5Ј-end of the luciferase gene was employed to detect the bicistronic message.
In Vitro Transcription-DNA primers containing a T7 promoter sequence were designed to amplify various fragments of the MS 5Ј-UTR. The PCR products were used as templates for in vitro transcription using the RiboMax kit (Promega) according to the vendor's protocol. The transcripts were resolved on a 10% polyacrylamide gel, isolated and 5Ј 32 P-labeled as previously described (18).
In-line Probing of RNA Constructs-Labeled RNA fragments were subjected to in-line probing as previously described (18). Briefly, ϳ1 nM 5Ј 32 P-labeled RNA was incubated for ϳ 40 h at 25°C in 20 mM MgCl 2 , 50 mM Tris-HCl (pH 8.3), and 100 mM KCl in the presence or absence of ligand (i.e. 200 M B 12 derivative). The reaction mixtures were resolved on a 6 -15% polyacrylamide gel (depending on the length of the RNA) and analyzed using a PhosphorImager.
REMSAs-For RNA electrophoretic mobility shift assays, DNA templates were generated by PCR using the following primers: Forward: TAATACGACTCACTATAGGGAGCCAACGGCAGGCGTCAAA and Reverse: GTTGTCGAGTCTCCTCCTT. The forward primer incorporated the T7 promoter sequence. RNA probes were synthesized by in vitro transcription with T7 polymerase (Maxiscript kit, Ambion) in the presence of [␣-32 P]UTP. 15,000 cpm of labeled RNA was incubated with 75 g of HeLa S100 cell extract (Paragon) in a total volume of 15 l at room temperature for 30 min, followed by the addition of 1 l of RNase T 1 (1 unit/l) and incubated for an additional 10 min at room temperature. Heparin was added to a final concentration of 5 mg/ml. When indicated, different forms of cobalamin were added at a final concentration of 200 M or cell extracts were pretreated with proteinase K (100 g/ml) before mixing with the RNA. The ribonucleoprotein complexes were electrophoresed on a 4% native acrylamide gel and visualized by autoradiography.
UV Cross-linking Experiments-RNA-protein complexes for UV cross-linking were prepared as described above. Before adding RNase T 1 , the samples were transferred into a 96-well dish and irradiated on ice with a 254-nm UV light source at 400,000 J/cm 2 . RNA-protein complexes were then resolved by 12% SDS-PAGE and visualized by autoradiography.
Statistical Analysis-Each experiment was repeated at least three times. Statistical analysis was performed using one-way analysis of variance (Microcal Origin software), and results were considered significant if the p value was Ͻ0.05.

RESULTS
The MS 5Ј-UTR Contains an IRES Element-To determine whether the MS 5Ј-UTR is able to initiate translation internally, a series of bicistronic vectors were employed as shown in Fig. 1A. In these constructs, the first cistron (encoding CAT) is translated via a canonical cap-dependent mechanism. The second cistron (encoding LUC) is translated efficiently only if translation initiation occurs in the intercistronic region. The following vectors were tested: CAT/ICS/LUC as a negative control, CAT/BIP/LUC as a positive control (15), and CAT/MS 5Ј-UTR/LUC containing the MS 5Ј-UTR instead of the BIP sequence. To rule out the possibility that the observed IRES activity does not result from ribosomal read-through, vectors having stable hairpins preceding the first cistron were employed (Fig. 1A).
Translation of the first but not the second cistron (Fig. 1B, upper panel) is strongly inhibited by the presence of the hairpin in the CAT/MS 5Ј-UTR/LUC construct and in the positive control containing the BIP-IRES (Fig. 1B, lower panel). These results provide evidence that translation of the second cistron occurs by initiation in the intercistronic region and are consistent with the presence of an IRES element in the MS 5Ј-UTR.
As an additional control, the LUC/CAT activity was measured in 293 cells treated with rapamycin, which inhibits cap-dependent translation by promoting dephosphorylation and activation of 4E-BP1, a repressor of the cap-binding protein, 4E (19). A ϳ10-fold increase in LUC/CAT activity was seen in the presence of rapamycin and resulted from diminished CAT activity whereas the LUC activity was unaffected (data not shown).
A potential source of error in interpreting data for internal initiation of translation is the possible presence of a cryptic promoter in the test sequence, which would result in a monocistronic luciferase mRNA in addition to the bicistronic message. Cap-dependent translation of the monocistronic RNA would confound the results. We thus tested a luciferase reporter construct in which the SV40 promoter was replaced by the MS 5Ј-UTR. Only background levels of luciferase activity were detected demonstrating that the MS 5Ј-UTR does not exhibit promoter activity (Fig. 1C).
A second source of error that needs to be considered is the possible presence of a splice site in the MS 5Ј-UTR. In this case, even if a bicistronic message is produced initially, splicing could generate a smaller fragment containing the second cistron. However, Northern analysis revealed the presence of only a single message that is long enough to accommodate both cistrons, ruling out the presence of a cryptic splice site in the MS 5Ј-UTR (Fig. 1D).
Because optimal IRES-dependent initiation requires trans-acting factors that have a tissue-specific distribution (20), we have examined the efficiency of MS-and BIP-IRES activities in different cell lines (Fig. 1E). Efficiency of the IRES varied within a 2-5-fold range compared with the ICS-containing negative control. The pattern observed with the BIP-IRES was comparable to that reported previously in the same cell lines (20). It is interesting to note that the highest MS-IRES activity was observed in the 293 kidney cell line, in which MS activity is also reported to be highest (21).
Deletion Mapping of the IRES Element-To map the minimal sequence that is required for IRES activity, we generated a series of nested deletions from the 5Ј-end of the MS 5Ј-UTR and measured IRES activity in the resulting bicistronic constructs. Fig. 2 shows that the deletions modulate IRES activity, probably because of the inhibitory or stimulatory influences of different regions in the MS 5Ј-UTR. For instance, the first 54 bases of the MS 5Ј-UTR are inhibitory, and their deletion increases IRES activity. As the deletions progress toward the 3Ј-end, stimulatory regions are eliminated, as indicated by a decrease in IRES activity. However, the last 71 bases of the MS 5Ј-UTR retain ϳ80% of the IRES activity versus the full-length MS 5Ј-UTR.
To confirm that the 71-mer indeed harbors IRES activity, we have tested a plasmid containing the 71 bases inserted in the intercistronic region and a hairpin preceding the first cistron. When the hairpin is present, translation of the first cistron is inhibited, whereas translation of the second cistron is enhanced, consistent with retention of IRES activity in this sequence (data not shown).

MS IRES Is Modulated by B 12 -
The boundaries of the MS-IRES element and of the B 12 responsive element reporter earlier (13) overlap, and raises the obvious question as to whether or not the IRES activity is modulated by B 12 . A B 12 -dependent increase (ϳ60%) in the translational efficiency of the bicistronic vector was observed, and the effect was clearly specific to the second cistron, which is under control of the MS-IRES element (Fig. 3). The fold increase in translation in the reporter construct is lower than the B 12 effect on the endogenous gene (ϳ3.8-fold in COS-1 cells), which could result from placement of the MS 5Ј-UTR in an artificial context in the bicistronic vector. A similar observation has been reported for the X-linked inhibitor of apoptosis protein, which is elevated 3.5-fold upon radiation treatment but shows a 50% increase in reporter activity when its IRES element is inserted in a bicistronic vector (16,22). B 12 had no effect on the LUC/CAT ratio in constructs containing the BIP IRES or the ICS sequence between the two reporter genes (not shown). It should be noted that the control cells are cultured in medium containing ϳ125 pM B 12 as described under "Experimental Procedures." Thus, the effect of B 12 supplementation, which activates MS (13), is correlated with IRES-dependent translation of MS in this study. (24) predicts the presence of a hairpin (between bases Ϫ9 and Ϫ44) with a long stem (Fig. 4A). Interestingly, this particular hairpin is retained in secondary structure models for all the deletion constructs described in Fig. 2. To test whether this hairpin exists in vivo, we probed the RNA secondary structure using the in-line probing method (18,25), which relies on degradation of RNA at room temperature and analysis of the resulting degradation pattern (Fig. 4B). Many of the regions that are predicted to be double-stranded in the stem (e.g. bases Ϫ36 to Ϫ44) are protected from degradation. In contrast, loops are more accessible to cleavage and generate strong banding patterns in the gel (e.g. bases Ϫ34, Ϫ26 and Ϫ25). We also used this method to determine whether B 12 binds directly to the MS 5Ј-UTR. This analysis is based on the premise that binding of a ligand to RNA can cause changes to the secondary/tertiary structure and thus influences the pattern of degradation (18). Fig. 4B shows the degradation pattern of a transcript spanning the last 71 bases of the MS 5Ј-UTR in the presence or absence of B 12 . None of the B 12 derivatives visibly altered the degradation pattern, which suggests but does not establish that B 12 does not bind directly to the MS 5Ј-UTR. We similarly found no evidence for modulation of degradation by B 12 of the full-length MS leader or of the last 140 bases of the leader (data not shown).

Secondary Structure Probing of MS-IRES-Computer modeling of the 71-mer using Mfold
The degradation pattern is in generally good agreement with the predicted long hairpin secondary structure in the model for the 71-mer (Fig. 4). The stronger degradation bands are seen in the predicted loops whereas the protected regions correspond to bases in the stem. These results provide evidence for the in vivo existence of the predicted hairpin structure in the MS IRES-element.
Fine Mapping of the Minimal IRES Element-The 71-mer was employed further for fine mapping analysis to probe the essential features of a minimal IRES element. To this end, 3-5 bases were deleted consecutively, starting at the 5Ј-end of the 71-mer. The secondary structure model predicts that the bases Ϫ9 to Ϫ44 form a long hairpin (Fig.  4A) and its role in IRES activity is confirmed by the bicistronic reporter assays (see loss of activity in deletion constructs Ϫ42, Ϫ35, and Ϫ26 in Fig. 5). The integrity of the bicistronic message was established by Northern analysis (Fig. 5B). The equal intensity of mRNA in these sam- ples is consistent with the reporter assays reflecting changes at the translational level.
The deletion analysis revealed that the sequence from Ϫ47 to Ϫ43 (CACGU) is important for IRES activity. In an Mfold analysis, both the last 47 and the last 42 bases of MS 5Ј-UTR form a stable hairpin with the only difference being that the main stem is 9-bp long in the 47-base hairpin and 7-bp long in the 42-base hairpin. To determine whether the primary sequence or the length of the stem was important, we mutated residues Ϫ44 and Ϫ43 and introduced complementary changes at Ϫ9 and Ϫ10 so that the 9-bp stem length was retained. The IRES activity of this mutant was 100% of the activity of the wild-type IRES (not shown). In contrast, mutating residues at the base of the stem so that pairing between residues Ϫ44 and Ϫ9 and Ϫ43 and Ϫ10 were disrupted resulted in ϳ50% of wild-type IRES activity (not shown). This is consistent with the length of the hairpin stem being important rather than the specific sequence at the positions that were tested. However, these results do not exclude the possibility that the mutations disrupted a tertiary interaction that modulates IRES activity. Because of the small magnitude of the B 12 effect in the bicistronic construct (Fig. 3), fine mapping of the B 12 effect was not pursued in this study.
A B 12 -responsive Protein Binds to MS 5Ј-UTR-Whereas the structure probing data do not rule out direct B 12 binding to the MS 5Ј-UTR (Fig. 4), they suggest that the B 12 effect may be expressed via a trans factor. To test this hypothesis, we performed electrophoretic mobility shift assays to detect proteins that bind to the MS 5Ј-UTR. A labeled RNA probe encompassing the last 140 nucleotides of the MS 5Ј-UTR, which showed the highest B 12 -sensitivity in a reporter gene assay (13), was incubated with cytosolic S100 extracts from HeLa cells in the presence or absence of different forms of B 12 and analyzed on native PAGE. Of the four complexes that were detected, three bind to the RNA independently of B 12 (Fig. 6A). The fourth complex (with the highest molecular weight) exhibits B 12 specificity, binding RNA in the order of decreasing strength in the presence of OHCblϾMeCblϷCNCbl. Binding was not observed in the presence of AdoCbl or in the absence of cobalamins. Preincubation of the cell extract with proteinase K prior to electrophoresis abolished complex formation confirming that binding of proteins to the RNA probe retarded its mobility (Fig. 6B). UV-crosslinking of the ribonucleoprotein complexes and separation on denaturing acrylamide gels provided an estimate of ϳ100 kDa for the molecular mass of the B 12 -responsive protein-RNA probe complex (data not shown).

DISCUSSION
Our studies reveal a novel mechanism of translational regulation of human MS that is IRES-dependent and is modulated by its cofactor, B 12 . The MS 5Ј-UTR is 394 bases long in contrast to the 5Ј-UTRs of most cellular mRNAs that are between 25 and 70 bases in length (26). The presence of extensive secondary structure (with a predicted ⌬G of Ͼ Ϫ50 kcal/mol) in the 5Ј-leader sequence is generally believed to inhibit ribosome scanning (26). Secondary structure predictions for the fulllength MS 5Ј-UTR, estimate a ⌬G of Ϫ175 kcal/mol for melting the multiple hairpins. Structure probing experiments of the full-length MS 5Ј-UTR confirm that it is indeed highly structured raising the possibility that IRES-dependent initiation could be important for this leader-burdened mRNA.
Several cellular IRES elements are activated under conditions of stress (viz. amino acid starvation, irradiation, and apoptosis) during which general cap-dependent translation is inhibited (14 -16, 27). IRESdependent translation initiation confers an advantage under these conditions by bypassing general translation arrest and allowing expression of proteins essential for adaptation/survival. MS is an essential gene as revealed by the embryonic lethality of MS-null mice (28). IRES-dependent translation of MS could have evolved to maintain production of this enzyme under different environmental conditions. B 12 is a relatively rare, albeit essential, vitamin with limited distribution and is absent in the plant kingdom. We have speculated previously that translational up-regulation of MS by B 12 may represent an evolutionary adaptation to the presence of this nutrient (13), which leads to rapid synthesis and sequestration of the vitamin by the only known B 12 -utilizing enzyme in the cytoplasm, MS (29,30). Modulation of IRES activity by B 12 as uncov- ered by this study is novel and affords regulation of MS translation initiation by its cofactor.
Recently, a novel mechanism for gene regulation in prokaryotes by vitamins and metabolites has been described, i.e. via riboswitches. The latter afford control of transcription termination and/or translation through direct binding of small ligand molecules to RNA, thereby circumventing the need for proteins (31). Ligand binding can either stabilize or melt RNA secondary structures with regulatory consequences (18,(32)(33)(34). A B 12 riboswitch is present in the 5Ј-UTR of the Escherichia coli and Salmonella typhimurium btuB transporter (18). B 12 binding inhibits ribosome recruitment thus diminishing production of the transporter when the cofactor is abundant. A phylogenetic analysis reveals that a B 12 riboswitch consensus sequence is wide spread in Gram-positive and Gram-negative organisms (35). Although most riboswitches have been described in prokaryotes, recent reports reveal their presence in eukaryotes as well (36). The negative results of the structure probing analyses do not exclude the possibility that B 12 binds directly to a responsive element in the MS 5Ј-UTR (Fig. 4). However, in combination with the observation that the MS 5Ј-UTR sequence is retarded by B 12 only in the presence of cell extract (Fig. 6), they support mediation of the B 12 effect by a protein factor. IRES activities are modulated by a number of proteins named ITAFs (IRES trans-activating factors) such as the hnRNP I/polypyrimidine tract-binding protein (37), the La autoantigen (38) and the ribonucleoproteins C1 and C2 (39). Variation in the MS-IRES potency in different cell lines (Fig. 1E) suggests the involvement of tissue-specific expression of ITAFs interacting with the MS-IRES. Interestingly, factorless ribosome assembly on the IRES of cricket paralysis virus has been reported (40), and it was suggested that distinct pseudoknot-like structures are important for the correct positioning of the ribosome on the mRNA, bypassing the need for initiation factors.
Based on the current study, a model for regulation of MS translation is proposed (Fig. 7). According to this model, the MS IRES is translated more efficiently in the presence of B 12 presumably via its interaction with an ITAF. It is potentially significant that the putative B 12 -dependent ITAF is insensitive toward AdoCbl, a cofactor derivative that is synthesized in the mitochondrion. In contrast, MS, which is cytoplasmic, requires MeCbl as a cofactor and can synthesize it in situ from CNor OHCbl (41), to which the putative ITAF responds. Alkylcobalamins, viz. MeCbl and AdoCbl, are photosensitive, and it is likely that the form of the vitamin that is delivered to the cytoplasm is predominantly in the OHCbl or cob(II)alamin state.
Another example of B 12 -dependent gene regulation involves the B 12binding protein, CarA, which is a transcriptional regulator in Myxococcus xanthus (42,43). CarA is a repressor of the carB operon encoding carotenoid biosynthetic functions and has a DNA binding domain that is directly fused to a B 12 binding domain. Binding of B 12 to CarA is proposed to release the repressor from the promoter sequence. B 12 has also been reported to inhibit the hepatitis C virus IRES-driven translation (44) by stalling the 80 S ribosomal complex on the IRES (23). The activity for each construct is reported after subtraction of the background ICS activity (negative control) and compared with the Ϫ394 MS 5Ј-UTR construct, which was set at 100. B, Northern analysis demonstrates integrity of and equal expression of the bicistronic mRNA in the deletion constructs. COS-1 cells were transfected with bicistronic plasmid constructs Ϫ71, Ϫ67, Ϫ52, Ϫ47, and Ϫ42. 24 h following transfection, half of the cells were treated with B 12 . Cells were harvested 24 h later, and RNA was isolated and subjected to Northern analysis using a luciferase probe as described under "Experimental Procedures" that detected a single ϳ3.3-kb transcript. Equal loading was established by monitoring 18 S RNA levels in each lane (not shown). FIGURE 6. RNA electrophoretic mobility shift analysis of RNA-protein complexes with the MS 5-UTR. A, a B 12 -responsive complex and three B 12 -independent complexes form on the MS IRES element. Gel mobility shift assays were performed with a 32 P-labeled probe encompassing the last 140 bases of the MS 5Ј-UTR and S100 extracts from HeLa cells as described under "Experimental Procedures." B, the mobility shifted complexes disappear upon treatment with proteinase K (100 g/ml (ϩ) and 1 mg/ml (ϩϩ) confirming the presence of protein-RNA complexes. Ado, -AdoCbl; CN, -CNCbl; Me, -MeCbl; OH, -OHCbl; *, artifacts derived from cassette used for autoradiography. The data are representative of three independent experiments.
Our studies reveal the presence of three other protein complexes that bind to the MS 5Ј-UTR independently of B 12 (Fig. 6). These putative ITAFs may be involved in modulating IRES activity. The identities of these proteins as well as of the B 12 responsive protein are currently under investigation in our laboratory.
In summary, we have shown that MS, which is essential for survival, is translated via an IRES mechanism and that a protein factor appears to be involved in mediating the B 12 -dependent response of the IRES element. Modulation of IRES-dependent translation of MS by B 12 represents a novel mode of gene regulation by a nutritional factor, which is also the cofactor for the encoded enzyme.