Expression of the Gene for Mitoribosomal Protein S12 Is Controlled in Human Cells at the Levels of Transcription, RNA Splicing, and Translation*

The human gene RPMS12 encodes a protein similar to bacterial ribosomal protein S12 and is proposed to represent the human mitochondrial orthologue. RPMS12reporter gene expression in cultured human cells supports the idea that the gene product is mitochondrial and is localized to the inner membrane. Human cells contain at least four structurally distinctRPMS12 mRNAs that differ in their 5′-untranslated region (5′-UTR) as a result of alternate splicing and of 5′ end heterogeneity. All of them encode the same polypeptide. The full 5′-UTR contains two types of sequence element implicated elsewhere in translational regulation as follows: a short upstream open reading frame and an oligopyrimidine tract similar to that found at the 5′ end of mRNAs encoding other growth-regulated proteins, including those of cytosolic ribosomes. The fully spliced (short) mRNA is the predominant form in all cell types studied and is translationally down-regulated in cultured cells in response to serum starvation, even though it lacks both of the putative translational regulatory elements. By contrast, other splice variants containing one or both of these elements are not translationally regulated by growth status but are translated poorly in both growing and non-growing cells. Reporter analysis identified a 26-nucleotide tract of the 5′-UTR of the short mRNA that is essential for translational down-regulation in growth-inhibited cells. Such experiments also confirmed that the 5′-UTR of the longer mRNA variants contains negative regulatory elements for translation. Tissue representation of RPMS12 mRNA is highly variable, following a typical mitochondrial pattern, but the relative levels of the different splice variants are similar in different tissues. These findings indicate a complex, multilevel regulation of RPMS12 gene expression in response to signals mediating growth, tissue specialization, and probably metabolic needs.

The human gene RPMS12 encodes a protein similar to bacterial ribosomal protein S12 and is proposed to represent the human mitochondrial orthologue. RPMS12 reporter gene expression in cultured human cells supports the idea that the gene product is mitochondrial and is localized to the inner membrane. Human cells contain at least four structurally distinct RPMS12 mRNAs that differ in their 5-untranslated region (5-UTR) as a result of alternate splicing and of 5 end heterogeneity. All of them encode the same polypeptide. The full 5-UTR contains two types of sequence element implicated elsewhere in translational regulation as follows: a short upstream open reading frame and an oligopyrimidine tract similar to that found at the 5 end of mRNAs encoding other growth-regulated proteins, including those of cytosolic ribosomes. The fully spliced (short) mRNA is the predominant form in all cell types studied and is translationally down-regulated in cultured cells in response to serum starvation, even though it lacks both of the putative translational regulatory elements. By contrast, other splice variants containing one or both of these elements are not translationally regulated by growth status but are translated poorly in both growing and non-growing cells. Reporter analysis identified a 26-nucleotide tract of the 5-UTR of the short mRNA that is essential for translational down-regulation in growth-inhibited cells. Such experiments also confirmed that the 5-UTR of the longer mRNA variants contains negative regulatory elements for translation. Tissue representation of RPMS12 mRNA is highly variable, following a typical mitochondrial pattern, but the relative levels of the different splice variants are similar in different tissues. These findings indicate a complex, multilevel regulation of RPMS12 gene expression in response to signals mediating growth, tissue specialization, and probably metabolic needs.
In mammals, mitochondrial DNA encodes 13 polypeptide subunits of the mitochondrial respiratory chain and oxidative phosphorylation system (1). Expression of these genes depends upon a dedicated apparatus of transcription, RNA processing, and translation, mainly encoded by nuclear genes. Typically these exhibit clear eubacterial affinities compared with their counterparts that specify cytosolic or nuclear isologues.
Genes encoding components of the apparatus of mitochondrial gene expression are regulated in a variety of ways. For example, the mRNA for the mitochondrial (mt) 1 transcription factor-A is present at different levels in different tissues (2), and its expression is in turn regulated by transcription factors of the nuclear respiratory factor group (3). Despite examples such as this, plus some knowledge obtained from studies in yeast (4), rather little is known of how genes for the apparatus of mitochondrial gene expression are regulated in mammalian cells. In principle this may have both developmental and physiological aspects and would be expected to involve regulation at different levels in the pathway of gene expression.
We have set out to study a gene encoding a key mitochondrial ribosomal protein, the homologue of Escherichia coli ribosomal protein S12 (rps12). Tentative identification of the single copy gene designated RPMS12 (5,6) as encoding the mitochondrial isologue of rps12 in humans was based on phylogenetic analysis of the encoded polypeptide, which groups it with homologues found in eubacteria and plant/protistan organelles, but correspondingly distant from the identified cytosolic homologue in yeast (Rps28p). Moreover, compared with the Rps12 proteins of eubacteria, or those encoded by organelle DNAs, the human gene product shows an N-terminal extension exhibiting features similar to those of mitochondrial targeting peptides. However, formal proof that the RPMS12 gene product is mitochondrially localized is lacking. We therefore set out initially, using a reporter gene approach, to verify that RPMS12 encodes a mitochondrial protein.
Ribosomal protein genes are regulated in unusual ways, perhaps not surprisingly, given their central importance in biosynthesis. In eubacteria and archaea, operons containing ribosomal protein genes are typically autoregulated by one of the proteins they encode (7). The 11 gene S10 operon of E. coli, for example, is regulated by ribosomal protein L4 binding to the S10 leader and simultaneously repressing translation and also promoting transcriptional attenuation (8). In yeast, ribosomal protein expression does not appear to be controlled by such a translational feedback mechanism (9), although a distinct type of post-transcriptional regulation in which Rps14p binds to its own pre-mRNA has been documented (10).
In vertebrate cells, cytosolic ribosomal protein mRNAs are translationally regulated in response to growth status (Ref. 4 and references therein). Such regulation has been demonstrated in response to a number of physiological stimuli, such as serum starvation (11)(12)(13), growth factor or mitogen stimulation (14 -16), dexamethasone treatment (17), and contact inhibition (18). In all these cases the fraction of ribosomal protein mRNA associated with polysomes varies according to cellular need, a higher proportion being loaded on polysomes (translationally active state) in rapidly growing cells as opposed to resting cells, in which most of it is stored in the subpolysomal or mRNP fraction (translationally inactive state, Ref. 19). In all cases so far analyzed, a 5Ј-UTR containing a terminal oligopyrimidine (TOP) tract plays a critical role in the translational control of such genes (20,21). This sequence element is believed to act via specific interactions with one or more regulatory proteins (13,(22)(23)(24).
Little is known of how or even in what context the synthesis of mitochondrial ribosomal proteins is regulated. It is probable that, like their cytosolic counterparts, mitochondrial ribosomal protein genes are regulated in respect to cellular growth, although other types of regulation are also to be expected, for example in response to cellular bioenergetic state, the availability of different kinds of substrate and cell differentiation. Tissues such as heart, skeletal muscle, or pancreas, which are highly dependent on mitochondrial ATP synthesis or in which rapid response of the bioenergy-generating system is needed, may be predicted to exhibit specific modes of regulation of the mitochondrial translational apparatus. The mRNA for one mitochondrial ribosomal protein (MRPL12) is known to be regulated in mouse cells by growth induction but at the transcriptional rather than the translational level (25,26).
Sequence analysis of a full-length or nearly full-length cDNA derived from the human RPMS12 gene (5) revealed a long (Ͼ300 nt) 5Ј-UTR that contains two features strongly suggestive of translational regulation. A short, upstream open reading frame (uORF), potentially encoding the pentapeptide MRACG, is located near the middle of the 5Ј-UTR. Such uORFs have been demonstrated to play a role in translational regulation of many genes in both fungi and vertebrates (27)(28)(29)(30)(31)(32)(33). They are believed to act by facilitating futile ribosome initiation in competition with the real AUG start site and also appear to regulate mRNA stability (30,34). In some cases the short, encoded peptide also interferes directly with ribosome function (35). The 5Ј-UTR of the human RPMS12 mRNA furthermore contains a 16-nt oligopyrimidine (oligo(Y)) tract located about 40 nt upstream of the AUG start codon. Although its placement differs from that seen in "classical" TOP mRNAs, the presence of this sequence element and of the uORF prompted us to investigate the translational behavior of the mRNA.
In this paper we provide evidence, via a reporter gene approach, supporting the idea that the RPMS12 gene does indeed encode a mitochondrially targeted protein. Transcript analysis via a combination of cyberscreening, Northern blots, and RT-PCR assays reveals a complex set of transcripts generated by alternate splicing within the 5Ј-UTR as well as 5Ј end heterogeneity. Surprisingly, the fully spliced mRNA, which completely lacks the putative elements of translational control (oligo(Y) tract and uORF), is translationally regulated in cultured cells according to growth status, whereas the other mRNA variants are not. The tissue representation of all splice variants follows a typical "mitochondrial" pattern, but one transcript found in cultured cells is absent from solid tissues. Our findings indicate a complex multilevel regulation of the expression of a mitochondrial ribosomal protein gene in response to various physiological and developmental stimuli.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK293-EBNA cells were maintained in Dulbecco's modified Eagle's medium, HyClone), 10% fetal calf serum, supplemented with 50 g/ml uridine, 1 mM glutamine, and 1 mM sodium pyruvate plus 100 units/ml penicillin and 100 g/ml streptomycin. HeLa, HEK293, and Xenopus kidney B3.2 cultured cells were grown in Dulbecco's modified Eagle's medium (Sigma), 10% fetal calf serum, containing 50 g/ml gentamicin and 2 mM glutamine. To induce a "downshift" to serum starvation conditions, cells were rinsed twice with PBS and detached with a limited amount of trypsin. After resuspension in PBS to dilute the trypsin, cells were centrifuged at 2000 ϫ g max for 5 min at 4°C, resuspended in serum-free medium, and incubated at 37°C for a further 4 h.
DNA Transfection of Cultured Cells-Cells were grown to 80% confluence on 100-mm plates and transfected in 6 ml of Opti-MEM serumfree medium (HyClone) with 10 g of plasmid DNA plus 40 l of LipofectAMINE reagent (Life Technologies, Inc.), following the manufacturer's recommendations. After 5 h of incubation 6 ml of Dulbecco's modified Eagle's medium (HyClone) containing 20% fetal calf serum was added. For serum starvation conditions no serum was present in this added medium. Cells were harvested 24 h after the start of transfection. Smaller scale transfections on 35-mm plates used 1 g of plasmid DNA plus 10 l of LipofectAMINE.
Subcellular Fractionation-Cytoplasmic extracts for Western analysis were prepared from transfected HEK293-EBNA cells grown on 35-mm plates, collected by pipetting up and down in 500 l of PBS, and centrifugation at 12,000 ϫ g max for 5 min. The cell pellet was vortexed and lysed in 50 l of PBS containing 1.5% (w/v) lauryl maltoside and 2.5 mM phenylmethylsulfonyl fluoride at 4°C for 30 min, and then centrifuged at 16,000 ϫ g max for 5 min at 4°C. Ten l of the supernatant was used immediately for SDS-PAGE. Finer subcellular fractionation of transfected cells was carried out using a standard procedure for mitochondrial isolation (Ref. 36, adapted from Ref. 37). Essentially, transfected cells from each 100-mm plate were washed once with PBS, dislodged from the plate by pipetting up and down in 1.3 ml of ice-cold PBS, and centrifuged at 300 ϫ g max for 10 min at 4°C. The cell pellet was resuspended by gentle pipetting in 10 volumes of ice-cold 0.133 M NaCl, 5 mM KCl, 0.7 mM Na 2 HPO 4 , 25 mM Tris-HCl, pH 7.5, and centrifuged again at 300 ϫ g max for 10 min at 4°C. The pellet was resuspended by pipetting up and down in 500 l of ice-cold 10 mM NaCl, 1.5 mM CaCl 2 , 10 mM Tris-HCl, pH 7.5, kept on ice for 15 min, and then homogenized in a glass homogenizer with 18 -25 strokes of a tight fitting pestle. Disruption of the cells was monitored by microscopy. An equal volume of ice-cold 0.68 M sucrose, 2 mM EDTA, 20 mM Tris-HCl, pH 7.5, was added, and nuclei and cell debris were pelleted by two sequential centrifugations at 1,200 ϫ g max for 10 min at 4°C. The combined pellets ("nuclear fraction") were washed once with 1 ml of ice-cold PBS, resuspended in a final total volume of 300 l of ice-cold PBS, and repeatedly passed through a syringe needle to shear chromatin. Mitochondria from the post-nuclear supernatants of four 100-mm plates of transfected cells were recovered by centrifugation at 16,000 ϫ g max for 30 min at 4°C. The supernatant from this step was saved as the cytosol fraction. Mitochondria were washed once with 1 ml of ice-cold PBS and finally resuspended by gentle pipetting in 160 l of ice-cold PBS. From this suspension 20 l was saved as the mitochondrial fraction, the remainder being further processed to yield various submitochondrial fractions, essentially as described in Ref. 38. Briefly, the mitochondrial suspension was adjusted to 350 l with ice-cold PBS, and an equal volume of ice-cold digitonin solution (4 mg/ml in PBS) was added. The suspension was kept on ice for 5 min, after which 700 l of ice-cold PBS was added, and mitoplasts were recovered by centrifugation at 16,000 ϫ g max for 10 min at 4°C. The supernatant was ultracentrifuged at 144,000 ϫ g max for 50 min at 5°C to separate intermembrane space (supernatant) and outer membrane (pellet) fractions. The outer membrane fraction was resuspended in 20 l of ice-cold PBS. Mitoplasts were resuspended in 140 l of ice-cold PBS, and 20 l was saved as the mitoplast fraction. The remainder was sonicated in a total volume of 300 l on ice, 30 times for 2 s with 30-s intervals, using a Vibracell High Intensity Ultrasonic Processor (Sonics & Materials, Inc.), fitted with the manufacturer's micro-tip. The suspension was then ultracentrifuged at 144,000 ϫ g max for 50 min at 5°C to separate matrix (supernatant) and inner membrane (pellet) fractions. The matrix frac-tion was concentrated by trichloroacetic acid precipitation and resuspended in 10 l of 2ϫ SDS-PAGE sample buffer (39). 2 M Tris was added until the yellow color changed to blue, and the volume was adjusted to 20 l with PBS. The inner membrane fraction was gently resuspended in 100 l of ice-cold PBS. Fractions were stored on ice and generally used for SDS-PAGE the same day (10 l from each of the above fractions). Samples were stored as pellets (nuclei, mitochondria, mitoplasts, outer and inner membrane) or in suspension (cytosol, matrix) at Ϫ80°C. Protease resistance of imported mitochondrial proteins was verified by incubating mitochondrial suspensions in minimal volumes of PBS for 10 min at room temperature either without any further additions, with 50 g/ml trypsin (Fluka), or with 0.05% lauryl maltoside plus 50 g/ml trypsin. For preparing polysomes, the procedure for cell lysis, sucrose gradient sedimentation, and analysis of the polysome/ subpolysome distribution of mRNAs was essentially that described by Meyuhas et al. (19). Cells were directly lysed on the plate with 300 l of lysis buffer (10 mM NaCl, 10 mM MgCl 2 , 10 mM Tris-HCl, pH 7.5, 1% Triton X-1000, 1% sodium deoxycholate, 36 units/ml RNase inhibitor (Amersham Pharmacia Biotech), 1 mM dithiothreitol) and transferred to an Eppendorf tube. After 5 min incubation on ice with occasional vortexing, the lysate was centrifuged at 6,000 ϫ g max for 8 min at 4°C. The supernatant was frozen in liquid nitrogen and stored at Ϫ70°C for later analysis or immediately sedimented in a 5-70% (w/v) sucrose gradient in a buffer containing 100 mM NaCl, 10 mM MgCl 2 , 30 mM Tris-HCl, pH 7.5. Fractions, collected while monitoring the optical density at 254 nm, were ethanol-precipitated overnight at Ϫ20°C.
Polyacrylamide Gel Electrophoresis and Western Blotting-SDS-PAGE used 12% polyacrylamide (Laemmli) gels run under standard conditions (39). Samples were heated at 95°C for 5 min in SDS-PAGE sample buffer prior to loading. Wet blotting to Hybond TM -C extra nitrocellulose membrane (Amersham Pharmacia Biotech) was carried out at 100 V for 1 h at 4°C (40). Blots were blocked 1 h at room temperature in TBS-T (0.1% Tween) containing 0.5% freeze-dried fat-free milk powder, washed several times with TBS-T, and reacted with primary antibody in TBS-T overnight at room temperature. Primary antibodies and dilutions used were mouse anti-Myc monoclonal 9E10 (Roche Molecular Biochemicals), 1:15,000 dilution of a 5 mg/ml stock, rabbit anti-glutamate dehydrogenase (kind gift of Dr. R. N. Lightowlers), 1:10,000, anti-complex IV (41), 1:10,000, mouse anti-cytochrome oxidase subunit II monoclonal antibody (kind gift of Dr. R. A. Capaldi), 1:10,000. Blots were washed 2 times for 10 s, 1 time for 10 min, and 2 times for 5 min in TBS-T and then incubated for 1 h at room temperature with peroxidase-conjugated goat secondary antibody (anti-mouse IgG (Bio-Rad), 1:10,000, or anti-rabbit IgG (Vector Laboratories, Inc.)) in TBS-T containing 0.5% normal goat serum (Vector Laboratories, Inc.). Blots were re-washed as above and finally with PBS. 5 ml of luminol solution (0.25 mg/ml sodium luminol (Sigma), 0.009% H 2 O 2 , 0.1 M Tris-Cl, pH 6.8) and 50 l of enhancer solution (1.1 mg/ml para-hydroxycoumaric acid (Sigma) in Me 2 SO) were mixed and incubated on each blot for 1 min. Film (Kodak Biomax TM ML) was exposed from 15 s to 45 min, as necessary.
Plasmid DNA Constructs-The RPMS12 coding region was amplified from a previously characterized full-length cDNA clone (5) using the Expand High Fidelity PCR System (Roche Molecular Biochemicals) plus chimeric primers S12-59 (CGGGATCCCGCACAGGGACGGCCC-AGGTGGC) and S12-36 (CCAAGCTTGGCTTCTTCTGCACGTGGCC-ACA). The PCR products were digested with restriction enzymes BamHI and HindIII and ligated into the pcDNA3.1(Ϫ)/Myc-His B vector DNA (Invitrogen) cut with the same enzymes, to generate the RPMS12-Myc/S plasmid, encoding the RPMS12-Myc fusion protein. A similar strategy was employed for cloning of the full mRNA sequence, i.e. including the entire 5Ј-UTR, to generate plasmid RPMS12-Myc/B, except that the 5Ј PCR primer used was S12-58 (CGGGATCCCGCCG-CGACCTCACCTTTAGGTC). All constructions were verified by complete DNA sequencing on both strands, using dye-terminator chemistry on the Perkin-Elmer ABI 310 Genetic Analyzer, with kit reagents supplied by the manufacturer.
In Vitro Translation-1 g of circular plasmid DNA of construct RPMS12-S/pcDNA3.1(Ϫ)/Myc-His B was used as template in the TNT T7 Quick Coupled Transcription/Translation System (Promega), in the presence of [ 35 S]methionine (Amersham Pharmacia Biotech, 1000 Ci/ mmol), according to the manufacturer's instructions. Five l of the in vitro translation reaction was adjusted to a final volume of 15 l by the addition of PBS, plus lauryl maltoside to 1.5% (w/v) and phenylmethylsulfonyl fluoride to 2.5 mM, and then centrifuged at 16,000 ϫ g max for 1.5 min at 4°C. The mitochondrial pellet from one plate of RPMS12-Myc-transfected cells was lysed in 25 l of PBS containing 1.5% (w/v) lauryl maltoside and 2.5 mM phenylmethylsulfonyl fluoride at 4°C for 30 min and then centrifuged at 16,000 ϫ g max for 1.5 min at 4°C. 150 l of each supernatant (in vitro translate and mitochondrial lysate) were used immediately for SDS-PAGE, in adjacent tracks of a 12% polyacrylamide gel. The gel was blotted to Hybond TM -C extra nitrocellulose membrane (Amersham Pharmacia Biotech) as above, autoradiographed to detect the 35 S-labeled translation product, and then processed for immunodetection of the Myc-tagged polypeptide.
Oligonucleotides and PCR-Sequences of oligonucleotides used to create PCR probes for Northern blots and for RT-PCR analysis of RPMS12 mRNAs were as shown in the figures and legends. Those used for rpL4 transcript analysis were rpL4-A (CATCGTATTGAGGAGTTC-C), representing nucleotide pair (np) 439 -458 of the rpL4 cDNA (42), and rpL4-B (TGGTGCTCGAAGGGCTCT), corresponding with np 927-909, for ␤-actin transcript analysis ␤-ACT-A (CGCTCGTCGTCG-ACAACG) and ␤-ACT-B (AGGTCTCAAACATGATCT), which amplify a 360-bp PCR product (43). For the control rpL22 transcript used in RT-PCR assays, the primers were rpL22 (CGTGGGCACGTCAGTCAC) (np 57-75, Ref. 44) and the KS vector primer (Promega), which amplify a 600-bp PCR product encompassing the cloned cDNA and a small portion of the pBSK plasmid (Promega), thus avoiding the amplification of the endogenous rpL22 mRNA. Probes for Northern analysis of RPMS12 mRNAs were derived by PCR amplification of a full-length RPMS12 cDNA clone and were gel-purified using the QIAquick gel extraction kit (Qiagen) before radiolabeling. PCR was carried out on HeLa cell genomic DNA (200 ng) to check the specificity of the various primer pairs designed for transcript analysis. Products were purified using either the QIAquick PCR purification or QIAquick gel extraction kit (Qiagen) and were sequenced by means of the Amplicycle sequencing kit (Perkin-Elmer). For RT-PCR total RNA samples were reversetranscribed into cDNA by the random hexanucleotides technique (39)  To perform quantitative RT-PCR analysis on RNA extracted from polysome gradient fractions, each sample was reverse-transcribed into cDNA together with 15 pg of an in vitro transcribed RNA for rpL22, included as an internal control to confirm that the amount of product was not influenced by experimental variations in the reactions. For each PCR, different cycles and template amounts were tested, in order to avoid conditions of saturation.
Extraction and Analysis of RNA-Total RNA was extracted from gradient fraction pellets by the proteinase K method (19). For Northern blot analysis RNA was fractionated on formaldehyde-agarose gels and transferred to GeneScreen Plus membrane (NEN Life Science Products). Blotting and Northern hybridization were carried out essentially according to the manufacturer's instructions. Northern hybridization to human cell line and tissue RNAs used filters purchased from CLON-TECH. Radioactive probes were prepared by the random priming technique (39) using as templates the inserts of plasmids containing cDNAs for human RPMS12 (5) and rpL4 (42), plus chicken ␤-actin (45) or PCR products for specific regions of the human RPMS12 mRNA 5Ј-UTR (see legends to Figs. 4 -6). Probes were synthesized in the presence of [␣-32 P]dCTP (Amersham Pharmacia Biotech, 3000 Ci/mmol). Standard hybridization conditions were used (35), with final washes generally at 55°C in 0.1ϫ SSC, 0.1% SDS. Re-washing at higher temperatures up to 65°C gave indistinguishable results.
Sequence Analysis-Analysis of data base sequences used the GCG package (46) (UK HGMP Resource Center, Cambridge) and on-line facilities of the NCBI.

RPMS12 Encodes a Mitochondrially Targeted Polypeptide-
The mitochondrial localization of the RPMS12 gene product was investigated by means of a reporter fusion to an epitope from the human c-MYC protein, for which a monoclonal antibody is available. Transient expression of the RPMS12-Myc fusion peptide was monitored by Western blotting, 24 h following lipofection into cultured HEK293-EBNA cells (Fig. 1). In unfractionated cytoplasmic extracts solubilized by lauryl maltoside treatment, a prominent, Myc antibody-reactive polypeptide of approximately 21 kDa was detected in cells transformed with the RPMS12-Myc construct (Fig. 1a) but not in cell extracts from cells transfected with Myc fusion constructs for TUFM or LacZ or from mock-transfected cells. Subcellular fractionation revealed that whereas the LacZ-Myc control polypeptide was mainly in the cytosol, the RPMS12-Myc polypeptide was localized to mitochondria (Fig. 1b). RPMS12-Myc was resistant to trypsin digestion in intact mitochondria (Fig. 1f), but lauryl maltoside solubilization rendered it trypsin-digestible. Further subfractionation confirmed that it was present in mitoplasts and exclusively in the inner membrane fraction after sonication (Fig. 1, c-e), colocalizing with a subunit of the respiratory chain (cytochrome oxidase subunit II), whereas glutamate dehydrogenase partitioned mainly to the matrix fraction. Careful alignment revealed that the electrophoretic mobility of the mitochondrially localized product on SDS-PAGE gels was less than that of the in vitro translated RPMS12-Myc polypeptide (Fig. 1g), indicating post-translational proteolytic processing upon mitochondrial import. A faint band corresponding to the unprocessed precursor is just visible in mitochondrial extracts.
RPMS12 mRNA Is Alternately Spliced within the 5Ј-UTR-Cyberscreening of dbEST followed by alignment into cDNA contigs revealed the existence of three distinct classes of transcript of the RPMS12 gene, with alternate splicing in the 5Ј-UTR (Table I and Fig. 2). The isoforms differ in respect to the two previously noted sequence elements suggestive of translational regulation, namely the uORF and oligopyrimidine tract, as illustrated in Fig. 3. The long isoform a remains unspliced within the 5Ј-UTR, whereas isoform b is spliced to remove 101 nt including the oligopyrimidine tract, which could function as part of the splice-acceptor recognition sequence (Fig. 2). The splice sites respect the conventional GT . . . AG rule. The uORF remains intact in isoform b mRNA, with the splice donor site located 11 nt beyond the uORF stop codon. All 15 cDNAs representing isoforms a and b that extended over the coding region splice site were correctly spliced at that position, indicating that these isoforms do not represent unprocessed nu-clear RNA. Isoform c is spliced in a different way, removing a 274-nt segment of 5Ј-UTR containing both the uORF and oligopyrimidine tracts. The splice acceptor is the same as for isoform b, and the coding region intron was again spliced out in every case.
Ten cDNAs (one representing isoform a, two representing isoform b, and seven representing isoform c) commence at approximately the same position, indicating a probable common transcriptional initiation region for all three isoforms. This is also consistent with the results of Northern analysis b Shows numerous discrepancies from the other sequences, especially after np 250. We have independently resequenced this cDNA (Image Clone 247801, GenBank TM Y11681, Ref. 5) finding all these discrepancies to be sequencing errors. Another GenBank TM entry, AA806508, also shows many discrepancies that are probably errors. c These cDNAs extend to a putative upstream transcriptional start site, as shown by the dashed line in Fig. 3a. (see below). Three of these ESTs with the isoform c splice commence at np 17 or 18 of the sequence shown in Fig. 2, which may represent a predominant 5Ј end of this isoform. Two other cDNAs spliced as isoform c, (one from testis and the other from pooled organs) were found to extend at least a further 200 nt upstream and may represent an additional mRNA isoform generated by transcription from an upstream start site. This putative variant of isoform c is denoted by a dashed line in Fig.  3a. Isoform d, as proposed in Fig. 3a on the basis of Northern blot data (see below), was not convincingly detected in dbEST. A small number of additional, RPMS12-related sequences in dbEST probably represent aberrantly spliced RNAs or cloning artifacts.
Cyberscreening of dbEST for the 3Ј region of RPMS12 mRNA revealed no significant heterogeneity. Allowing for a low frequency of sequencing errors, all 36 cDNA sequences that extended beyond the stop codon, plus 7 others located exclusively in the 3Ј-UTR, formed a single contig, terminating at or very close to the previously inferred 3Ј end of the RPMS12 mRNA and just downstream of a conventional poly(A) addition signal. No consensus 3Ј ends were found upstream of this position.
Northern analysis of RNA from various cell lines was carried out using probes for successively more inclusive portions of the RPMS12 5Ј-UTR. These were designed to detect the three splice variants revealed by cyberscreening, in the order of increasing abundance (based on preliminary experiments). Blots were initially probed at high stringency for a region of the isoform a splice variant (Fig. 3b) that is absent from isoforms b and c. This revealed transcripts of two distinct size classes, estimated at 1 and 1.25 kb, whose relative abundance varied between the cell types tested. The shorter transcripts are proposed to represent the 5Ј-truncated isoform d, cyberscreening having revealed no evidence for 3Ј heterogeneity, nor any reasonable match to the consensus poly(A) addition signal located elsewhere in the 3Ј-UTR. Even if the length of the poly(A) tail in these shorter transcripts is much reduced, their overall size precludes that they contain the full 5Ј-UTR of isoform a. 5Ј-Truncated transcripts containing the oligopyrimidine tract unique to isoform a are evident in dbEST (e.g. GenBank TM entry AA257081), although they do not constitute a coherent class indicating a specific 5Ј end. Our best guess is that they are heterogeneous, as indicated by the dashed line in Fig. 3a. It is unclear whether these transcripts indicate a downstream transcriptional start region or else 5Ј truncation in vivo. The shorter transcripts were prominent in A549 lung carcinoma cells but almost undetectable in the K-562 leukemia cell line and were also absent from solid tissues (see below).
After stripping and reprobing for a region contained within both isoforms a and b, but not c, an additional, prominent transcript of intermediate size was detected (Fig. 3, b and c), estimated at 1.15 kb. Both transcript classes detected by the earlier probe were recognized only weakly by the second probe, indicating that the intermediate sized transcripts, which must represent isoform b, are considerably more abundant in all cell lines tested than those of isoforms a or d. Blots were stripped and reprobed again for the full-length RPMS12 mRNA, detecting all classes of transcript. In this case, a prominent additional species was detected in the 1.0-kb size range, i.e. migrating faster than the isoform b transcripts, which must represent the fully spliced isoform c not detected by other probes. This transcript appears to be the major isoform in all cells tested. Larger transcripts, which would be derived from a far upstream start such as tentatively inferred from cyberscreening, were present only at very low abundance. The ratio of isoforms (a ϩ d), b and c appears to be similar in all cell lines studied.
RPMS12 mRNAs Are Tissue Differentially Expressed-The isoforms of RPMS12 mRNA were represented in the different cell lines tested in very variable relative amounts. In order to investigate their expression in vivo, the same probes were hybridized sequentially to Northern blots of RNA from human tissues, as documented in Fig. 4. The pattern of relative abundance between tissues was similar for all three of the 5Ј-UTR splice variants, although the unspliced, 5Ј-truncated isoform d, seen prominently in some cultured cells, was not detected. The unspliced isoform a was detected in heart (Fig. 4b), but only weakly in most other tissues, and in many cases a larger transcript of approximately 2 kb was also detected by the isoform a-specific probe, possibly representing unspliced nuclear RNA from which the coding region intron had also not been removed. The isoform b splice variant, which retains the uORF, was more highly represented in all tissues but was especially prominent in heart, skeletal muscle, and kidney (Fig.  4, b and c). A similar pattern of hybridization was seen using the full-length probe (Fig. 4c), which detects also (and mainly) the shorter, fully spliced isoform c. Some minor differences can be discerned, for example isoform c was detected more strongly in peripheral blood lymphocytes (PBLs) than in colonic mucosa or thymus, and at about the same level as in testis, whereas isoform b was more prominent in colonic mucosa, thymus, and testis than in PBLs (Fig. 4c). In general, isoform b appeared to show more pronounced differences between tissues than either of isoforms a or c. Higher molecular weight transcripts that could correspond with the use of a far upstream transcriptional start site were not detected, except in PBLs, where a (ϳ2.5 kb) species of unknown origin was detected by the isoform b probe.
The Major RPMS12 mRNA Splice Variant Is Translationally Regulated-In order to determine whether RPMS12 mRNAs were translationally regulated in a manner akin to the TOP mRNAs, sucrose density gradient centrifugation was used to separate post-mitochondrial supernatants from cell lysates into polysomal and subpolysomal (mRNP) fractions. Such gradients were prepared from HeLa and HEK293 cells cultured under standard conditions and under serum starvation. RNA was extracted from successive fractions across the gradients and probed for RPSM12 by Northern hybridization. The blots were then reprobed for two control mRNAs, ␤-actin, which is efficiently translated in both growing and non-growing cells, and rpL4, a cytosolic ribosomal protein encoded by a typical, FIG. 3. Northern blots of RNA from human cell lines. a, the various probes used are indicated by double-headed arrows and were derived as follows: probe 1 was a PCR product synthesized from the cloned, full-length RPMS12 cDNA (5) using primer pair b3/b5 (see Fig. 4); probe 2 was a PCR product synthesized from the same template, using primer pair bg3/bg5 (see Fig. 4); probe 3 was a PCR product synthesized from the same template using vector-specific primers to generate the full-length cDNA insert. The coding region, uORF, and oligo(Y) sequences are denoted, respectively, as black, white, and shaded boxes. Isoforms a, b and c are as inferred from cyberscreening (Table I and Fig. 4), and isoform d is inferred from the blots shown here. Dashed lines indicate the uncertain 5Ј ends of isoform d and also of some transcripts with the splice pattern of isoform c that appear to extend far upstream. b, the blot, supplied by CLONTECH, contained RNA from the following cell lines: promyelocytic leukemia HL-60, HeLa cell S3, chronic myelogenous leukemia K-562, lymphoblastic leukemia MOLT-4, Burkitt's lymphoma Raji, colorectal adenocarcinoma SW480, lung carcinoma A549, and melanoma G361. c, the lower panel shows precisely aligned tracks of K-562 cell RNA (plus one of A549 cell RNA) from the three autoradiographs, to indicate the relative positions of the various transcripts detected. Approximate transcript sizes were extrapolated from marker positions as supplied by the blot manufacturer. The blot was probed, stripped, and re-probed in the order probe 1, probe 2, and then probe 3, with effective stripping verified autoradiographically between each hybridization.
FIG. 4. Northern blots of RNA from human tissues. a, the probes used and transcripts detected are depicted as in Fig. 5, except that isoform d was not convincingly detected in tissue RNAs. Two different sets of RNAs were tested using two different filters, as shown in b and c. The blots were probed, stripped, and re-probed in the order probe 1, probe 2, and then probe 3, with effective stripping verified autoradiographically between each hybridization. b, the left-hand panels show hybridization to probes 1 and 2 only. Probe 3 (not shown) gave a similar pattern of relative signals between tissues on this blot as probe 2 but with higher background. c, the right-hand panels show hybridization to probes 2 and 3 only. Probe 1 (not shown) gave only very low signals to these RNAs, even at the longest exposures (72 h, as in the left-hand panel). The exposure times for probe 2 are comparable (16 h) in the two blots shown alongside one another, with a longer exposure (48 h) in c for the right-hand blot, to show the relative signals more clearly and to reveal the higher molecular weight transcript visible in PBLs (peripheral blood lymphocytes). translationally regulated TOP mRNA. As seen in Fig. 5b, RPMS12 mRNA behaved qualitatively like a typical TOP mRNA in HeLa cells, being mainly polysomal in growing cells but mainly non-polysomal in serum-starved cells. The effect was less dramatic than for rpL4 mRNA, mainly due to the fact that even in growing cells most of the RPMS12 mRNA seemed to be translated on rather small polysomes. Similar results were obtained for HEK293 cells, and also, via inter-specific cross-hybridization, for Xenopus B3.2 cells (data not shown).
In order to determine which splice variants of RPMS12 mRNA were translationally regulated in this fashion, a quantitative RT-PCR approach was adopted, using primer pairs capable of generating specific products diagnostic for each splice variant, plus primers for ␤-actin and rpsL4 as controls. Direct sequencing was used to confirm the identity of the major RT-PCR products synthesized from total RNA using all four primer pairs. Primer pairs R1/R5 and R2/R5 generated, respectively, 86-and 72-bp products corresponding to the fully spliced isoform c. Primer pair R3/R5 generated a major product of 74 bp, representing isoform b, whereas primer pair R4/R5 gave a product of 76 bp containing the oligo(Y) tract found only in isoforms a and d. PCR on genomic DNA using these primer pairs (not shown) yielded products identical in sequence to cDNA clones for isoform a (Fig. 2), confirming that the latter is colinear with the genome.
The RT-PCR products generated by primer pairs R1/R5 and R2/R5, representing the fully spliced mRNA isoform c, showed a similar pattern of distribution between polysomal and mRNP fractions as did the total RPMS12 mRNA on Northern blots (Fig. 5b). Although quantitatively less dramatic than for rpL4, isoform c was thus found to be translationally regulated, being mainly polysomal in growing cells, but predominantly nonpolysomal in resting (serum-starved) cells. This is not unexpected, since isoform c was the mRNA variant predominantly detected on Northern blots (Fig. 3). By contrast, the RT-PCR product generated by primer pair R3/R5, representing mRNA isoform b, showed identical distributions in growing and serum-starved cells, being found in both cases associated with monoribosomes and short polysomes. It behaved similarly to the ␤-actin control, found mainly on large polysomes, which showed no translational response to serum starvation. The RT-PCR product from primer pair R4/R5, representing isoforms a and d, was distributed approximately equally between the polysome and mRNP fractions in both growing and nongrowing cells, indicating that these isoforms were also not translationally regulated. Interestingly, the distribution of this RNA in the gradient was wider than for isoform b, encompassing both the postribosomal and larger polysomal fractions.
Thus, the only RPMS12 mRNA variant that was found clearly to be regulated translationally in response to growth status is the one in which both of the putative elements originally hypothesized to be involved in translational control are removed by splicing. Furthermore, the translational properties of each of the 3 splice variants appear to be distinct.
Sequences in the RPMS12 mRNA 5Ј-UTR Mediate Its Translational Regulation-In order to localize the signals responsible for translational regulation of RPMS12 mRNAs, we compared the effects of serum deprivation on the transient expression of two RPMS12-Myc fusion constructs in HEK293-EBNA cells. The first construct was the one referred to above (RPMS12-Myc/S), which includes a 155-nt stretch of vectorderived RNA at its 5Ј end, fused directly to the RPMS12 coding sequence preceded by just 24 nt of untranslated sequence (Fig.  6a). The second construct includes the same vector sequence, but followed by essentially the entire 5Ј-UTR of RPMS12, commencing in the region of the major 5Ј end found in the short (fully spliced) mRNA isoform c and including the splice donor and acceptor sites between which are located the uORF and oligopyrimidine tract. The latter construct (RPMS12-Myc/B) is predicted to give rise to a major spliced transcript that has the 155-nt vector sequence, followed by the 5Ј-UTR of the "short" RPMS12 mRNA isoform c. RT-PCR analysis (Fig. 6b) confirmed that the transiently expressed constructs gave rise to the predicted transcripts. Western analysis (Fig. 6c) showed that both constructs were efficiently expressed in the presence of serum. However, under serum starvation conditions, a much lower amount of product was expressed reproducibly from construct B, whose transcript is spliced in the isoform c-specific mode, whereas the expression of construct S or the LacZ-Myc control was only slightly affected by serum starvation. In other words, despite the presence of an additional 155 nt of vector sequence, the fully spliced RNA was regulated by serum in a   FIG. 5. Northern blot and RT-PCR analysis of RPMS12 mRNAs in polysome gradient fractions from growing and serum-starved HeLa cells. a, schematic map of 5Ј-UTR of RPMS12, showing the four variants detected previously by cyberscreening and Northern analysis, and the five primers used for RT-PCR. Reactions used a common 3Ј primer R5, with each of four 5Ј primers R1 through R4, as shown in Fig. 4. b, Northern blot strips of RNA extracted from gradient fractions, as indicated by A 254 traces, probed successively for ␤-actin, rpL4, and RPMS12 (full-length probe). c, RT-PCR reactions, using various RPMS12 primer pairs, alongside reference reactions using primers for rpL4 and ␤-actin. Also shown is a quantitation control (15 pg of an in vitro synthesized RNA for rpL22, see "Experimental Procedures"). The agarose gel photographs are aligned with the A 254 traces in b above. similar manner to the endogenous transcript, whereas the isoform completely lacking the natural 5Ј-UTR signals was efficiently expressed but not regulated. This identifies a 26-nt region, present in the spliced construct B mRNA but absent from the construct S mRNA (see Fig. 6a), that is required for growth regulation of translation. The region is located immediately upstream of the splice site. Furthermore, the fact that construct S was efficiently expressed under both conditions tested is consistent with the earlier inference that the 5Ј-UTR of the longer isoforms, containing the uORF and oligopyrimidine tract, is a negative element for translation. DISCUSSION These findings support the identification of the RPMS12 gene product as a mitochondrial ribosomal protein, reveal unexpected complexity in the regulation of its expression, and identify specific regions of the 5Ј-UTR involved in translational control. We now address the implications of these findings.
Mitochondrial and Submitochondrial Localization of the RPMS12 Gene Product-The targeting to mitochondria of the RPMS12-Myc reporter protein strongly supports the previous assignment of RPMS12 as encoding the mitochondrial isologue of E. coli ribosomal protein S12. The protein is a member of a well characterized and conserved family of ribosomal proteins (5) and in most plants is even mitochondrially encoded (47)(48)(49). Comparable experiments with an RPMS12-GFP reporter fusion 2 also showed colocalization to mitochondria. The assignment is further supported by the fact that the mitochondrially localized fusion protein was inaccessible to external protease, i.e. had been imported into the mitochondria, appears to have been proteolytically processed, and was localized to the inner mitochondrial membrane. In yeast, the inner membrane is the site of productive synthesis of the hydrophobic, mtDNA-encoded mitochondrial proteins that contribute to the respiratory chain (50). It is therefore logical that an epitope-tagged mitoribosomal protein will be localized there.
Multilevel Control of RPMS12 Gene Expression-The above findings lead to the conclusion that RPMS12 is regulated at the levels of transcription, RNA processing, and translation. The generation of mRNA isoforms with different patterns of translational behavior would seem to be the major outcome of the alternate synthetic pathways.
Each of the isoforms of RPMS12 mRNA is clearly represented in highly tissue-variable amounts. We were unable to perform a meaningful loading control hybridization with the commercially supplied Northern blot membranes, since the final probe could not be completely stripped. However, the claim of the manufacturer that the lanes are evenly loaded with 2 g of poly(A) RNA and quality controlled to check RNA integrity means that minor variations in loading cannot account for the order of magnitude differences in signal seen between the highest and lowest expressing tissues. The pattern of relative abundance between tissues represents a typical pattern for a gene involved in mitochondrial respiratory function, with prominent expression in tissues highly dependent on oxidative metabolism such as heart, skeletal muscle, kidney, and to a lesser extent brain, liver, testis, and pancreas. The gene encodes a conserved ribosomal protein indispensable for translation; hence, its expression is expected to follow a similar profile to that of the mtDNA-encoded mRNAs that are translated by mitoribosomes. The fact that the various mRNA isoforms show similar patterns of tissue distribution as one another suggests that differential RNA processing is not of major importance in generating these different tissue levels of expres-2 Z. H. Shah, unpublished data. The two constructs share 21 nt of sequence located immediately upstream of the start codon. Construct B contains an additional 26 nt of RPMS12 sequence from the other side of the splice junction unique to isoform c, in the place of which construct S has 3 nt of intron-derived sequence but not the full splice acceptor. b, RT-PCR analysis of transgene-specific transcripts cells transiently transfected with RPMS12-Myc constructs B and S. PCR primers were the T7 promoter primer (vector-specific) and primer R5 (see Fig. 2). The only construct B-derived transcript detected is the fully spliced isoform (194-bp product, whose structure was confirmed by direct sequencing). Construct S gives a 170-bp product that is colinear with the DNA. c, Western analysis of RPMS12-Myc fusion protein expression in mitochondria from cells mock-transfected (vector only) or transfected with constructs B or S and then grown in the presence or absence of serum. Equal amounts of protein are loaded on each lane. Shown alongside is the cytosolic expression of a LacZ-Myc control in transfected cells grown with or without serum. sion. Instead the gene is most likely transcribed at different rates in different tissues, although a contribution from RNA stability cannot be ruled out.
Transcription may also be regulated in another way, via the selection of alternate start sites. Tentative evidence for a far upstream start active in at least some tissues was obtained by cyberscreening of dbEST and is also suggested by the detection in PBLs of a larger transcript carrying the b isoform splice pattern. More convincingly, the 5Ј-truncated isoform d detected in cultured cells suggests strongly the use of a downstream initiation site. Although 5Ј truncation by an exonucleolytic activity in vivo cannot be ruled out, the fact that the isoform d transcripts detected on Northern blots constitute a discrete size class argues strongly that they derive instead from the use of a separate initiation site, which would be located just upstream of the oligopyrimidine tract, based on the transcript size. The physiological significance of alternate 5Ј termini in RPMS12 mRNAs remains unknown. It may relate to translational regulation, as discussed further below. The expression of a minor but constitutively translated form of RPMS12 mRNA in the stem cell compartment which cultured cells represent makes obvious sense.
Alternate splicing in a 5Ј-UTR is relatively uncommon. It has been reported, for example, in the bovine gene encoding connexin-32 (51) and the human genes encoding reduced folate carrier (52) and thrombopoietin (53). In principle, 5Ј splice heterogeneity inferred from cyberscreening could be due to the presence in the data base of sequences derived from partially processed nuclear RNA. In the case of RPMS12 this is highly unlikely. Essentially all of the RPMS12 cDNA sequences deposited in dbEST represent transcripts from which the coding region intron has been correctly spliced out, yet a clear majority of them are unspliced or partially spliced within the 5Ј-UTR. Moreover, if the oligo(Y) tract is essential for recognition of the splice acceptor site just upstream of the RPMS12 start codon, then isoform b could not be efficiently processed further to isoform c. In addition, unspliced or partially spliced 5Ј-UTR variants were found at least partly in the polysomal fraction in cultured cells. Alternate splicing must therefore give rise to several different forms of translatable mRNA. Only in those human tissues showing low expression (e.g. placenta, prostate, and ovary) was a significant fraction of the 5Ј-UTR-unspliced RPMS12 RNA of a size indicating that it might also be unspliced in the coding region. These very low abundance transcripts may be nuclear but are a very minor fraction in tissues such as heart or skeletal muscle, showing prominent overall expression.
The physiological significance of alternate splicing within the RPMS12 5Ј-UTR remains unknown. In this study we have demonstrated that the major RPMS12 mRNA (isoform c) is subject to a translational control mechanism that results in a higher fraction of mRNA being loaded onto polysomes in proliferating cells compared with resting cells. However, the fact that only one of the four RPMS12 isoforms expressed in cultured cells appeared to be thus regulated suggests that the other splice variants may be subject to different kinds of control at the translational level, such as described for other genes (54 -56). The four isoforms differ by virtue of the presence of putative elements involved in such regulation. Isoforms b and d contain, respectively, the uORF and the oligo(Y) tract alone, whereas isoform c contains neither, and isoform a contains both. These putative elements could regulate translation in another context than growth control.
Translational Regulation of Mitoribosome Biogenesis-Like cytosolic ribosomes, mitochondrial ribosomes are an essential component of the cellular machinery whose biosynthesis needs to be stimulated in growing cells. The observed translational regulation of RPMS12 therefore makes physiological sense. An important difference is the fact that the growth-regulated mRNA lacks a 5Ј-terminal oligopyrimidine tract. The regulation must therefore employ a mechanism somewhat distinct from that of TOP mRNAs, although it could share features with it. Our reporter assay identified a 26-nt region located at the extreme 5Ј end of isoform c mRNA that is essential for translational down-regulation under conditions of growth inhibition. It is not especially pyrimidine-rich (10 nt are purines). In the reporter construct we used, it was located 155 nt away from the 5Ј end, implying that its extreme 5Ј location in the natural mRNA is also not critical for function. Importantly, the longer isoforms a and b, which are differently spliced and not growthregulated, also contain this 26-nt tract. Therefore, the signal mediating growth-related translational down-regulation must logically include more than just this 26 nt, i.e. must extend over the unique splice site present in the isoform c mRNA. Alternatively, its function may be over-ridden by other 5Ј-UTR signals present in the longer isoforms.
TOP mRNAs are believed to be controlled via specific interactions between the TOP tract and regulatory proteins mediating mRNA recruitment to ribosomes (23,24). The exact properties of these proteins and the machinery with which they interact have not yet been characterized. It seems logical to postulate that the regulatory sequence unique to isoform c interacts specifically with a negative regulatory protein that is only present in an active state in growth-arrested cells. The same protein may independently interact with TOP mRNAs. Alternatively, different proteins may bind to RPMS12 and TOP mRNAs, independently influencing their ability to interact with a common component of the translational recruitment machinery. The fact that the translational regulation of RPMS12 mRNA is less dramatic than for some TOP mRNAs with an "optimal" oligo(Y) tract at the 5Ј end recalls the behavior of some mRNAs bearing poor pyrimidine tracts, whose translational regulation is less evident and/or dependent on the cellular context (57,58).
The synthetic reporter construct RPMS12-Myc/S, which lacks both the uORF and oligo(Y) motifs, was expressed as efficiently as the LacZ-Myc control. This supports the inference that the uORF and/or oligo(Y) tracts are negative regulatory elements for translation, at least in cultured cells. One possibility is that the negative effect of the uORF is disabled where mitochondrial function must be rapidly enhanced, for example in response to bioenergetic needs or developmental signals. Our data suggest that isoform b is likely to be the mRNA variant responsive to such signals. It is significantly more abundant than isoform a in all tissues but especially in oxidative tissues showing high overall expression, such as heart and skeletal muscle. Moreover, the profile of its polysome distribution in cultured cells, where it is found mainly associated with monoribosomes and small polysomes, suggests that it is indeed susceptible to futile initiation at the uORF, with rather few ribosomes traversing the genuine RPMS12 coding sequence. By contrast, the mRNAs containing the oligo(Y) tract (i.e. isoforms a and d) are distributed quite differently, being more prominent in both the overtly post-ribosomal fraction and in larger polysomes than isoform b. This suggests that the isoform a and d mRNAs are regulated in a quite different fashion, being only inefficiently recruited into polysomes, but once there are less subject to futile initiation at the uORF. Our findings prompt a more thorough investigation of the effects on translation in different cellular contexts of the various 5Ј-UTR elements.
The presence of multiple isoforms of RPMS12 mRNA in all tissues potentially allows each cell type to respond to a variety of intra-and intercellular signals, to enhance the biosynthesis of mitoribosomes according to cellular needs. The complex pattern of RPMS12 mRNAs seen in humans is not shared with the mouse, however, where all cDNA sequences from the orthologous gene represented in dbEST form a single contig, with a relatively short 5Ј-UTR (approximately 90 nt), no uORF, and only weak evidence for 5Ј heterogeneity. Regulation of mitoribosome biosynthesis in the mouse might therefore employ different mechanisms than in humans. By contrast, the presence of a uORF may be a common feature in human mitoribosomal protein mRNAs, as it has been found in the one other example studied in detail, MRPL12 (25). Further analysis of the regulatory roles of the different 5Ј-UTR elements of RPMS12 mRNA will enable the components of the regulatory machinery to be identified. Further clues will also doubtless emerge from studies of other mRNAs encoding mitochondrial ribosomal proteins and other components of the mitochondrial translational apparatus, both in humans and in other vertebrates.