Function of the SmpB tail in transfer-messenger RNA translation revealed by a nucleus-encoded form.

Stalled bacterial ribosomes are freed when they switch to the translation of transfer-messenger RNA (tmRNA). This process requires the tmRNA-binding and ribosome-binding cofactor SmpB, a beta-barrel protein with a protruding C-terminal tail of unresolved structure. Some plastid genomes encode tmRNA, but smpB genes have only been reported from bacteria. Here we identify smpB in the nuclear genomes of both a diatom and a red alga encoding a signal for import into the plastid, where mature SmpB could activate tmRNA. Diatom SmpB was active for tmRNA translation with bacterial components in vivo and in vitro, although less so than Escherichia coli SmpB. The tail-truncated diatom SmpB, the hypothetical product of a misspliced mRNA, was inactive in vivo. Tail-truncated E. coli SmpB was likewise inactive for tmRNA translation but was still able to bind ribosomes, and its affinity for tmRNA was only slightly diminished. This work suggests that SmpB is a universal cofactor of tmRNA. It also reveals a tail-dependent role for SmpB in tmRNA translation that supersedes a simple role of linking tmRNA to the ribosome, which the SmpB body alone could provide.

Translating ribosomes can stall, for example, when they arrive at the 3Ј-end of an mRNA lacking an in-frame stop codon (nonstop mRNA) (1). Stalling could sequester ribosomes and produce incomplete polypeptides. In bacteria these problems are ameliorated by tmRNA, 1 a specialized RNA with both tRNA-like and mRNA-like properties. The stalled ribosome can switch mRNA templates, leaving nonstop mRNA and resuming translation on the reading frame in tmRNA, where the ribosome is freed upon release at the tmRNA stop codon. The protein encoded by the nonstop mRNA gains a hydrophobic peptide tag encoded by tmRNA that is a signal directing the proteolysis of the entire tagged protein.
In model bacteria, the protein SmpB is a requisite cofactor of tmRNA (2,3). SmpB binds to the tRNA portion of tmRNA in the elbow region on the D-loop face (4), retaining the ␤-barrel structure found in the free protein (5)(6)(7). A C-terminal tail of ϳ30 amino acids, emerging from the ␤-barrel opposite to the tmRNA-binding face, appears unstructured in both free and tmRNA-bound SmpB.
SmpB has been found to improve aminoacylation of tmRNA and to allow simultaneous tmRNA binding by EF-Tu⅐GTP (4,8). SmpB is further required for interaction of tmRNA with the ribosome (9,10). It can bind the ribosome directly in the absence of tmRNA, and its position in the complex with the tRNA domain of tmRNA suggests that its tail might contact the ribosome in the vicinity of the decoding center (7,11).
Both ssrA (tmRNA) and smpB genes have been found in all fully sequenced bacterial genomes (12). ssrA has also been identified in certain whole plastid genomes (12,13) producing the expected RNA (14), but smpB has not been identified in these same plastid genomes, raising the possibility that tmRNA in plastids might function without its usual cofactor.
In this article we show that for two eukaryotes with a plastidial ssrA gene, the diatom Thalassiosira pseudonana and the red alga Cyanidioschyzon merolae, SmpB is encoded in the nucleus with an apparent signal for import into the plastid. The predicted mature T. pseudonana SmpB is active in tmRNA translation with Escherichia coli components in vivo and in vitro. Its lower activity relative to that of E. coli SmpB is ascribed, through the study of chimeric proteins, to its ␤-barrel domain and not to its C-terminal tail. The latter result does not mean that the tail has no function; indeed, with deletion of just half the tail tmRNA translation was undetectable. The affinity of SmpB for either ribosomes or tmRNA was not significantly affected by tail truncation, indicating that the role of SmpB in tmRNA translation is more complex than providing a simple link between tmRNA and the ribosome.

EXPERIMENTAL PROCEDURES
Construction of the T. pseudonana smpB Gene-A coding sequence was designed for T. pseudonana pre-SmpB residues 52-197 (protein TT0 of Fig. 1; see also Fig. 2), with the codon bias approximating as closely as possible that of E. coli SmpB, and flanked by start and stop codons and NsiI and BamHI restriction sites. The gene was constructed from 16 ϳ45-mer oligonucleotides sharing ϳ15-nucleotide sequence overlaps with neighbors by mutually extending eight oligonucleotide pairs hybridized at their 3Ј-ends with TaqDNA polymerase, followed by three rounds of the following: 1) amplification of one strand of each DNA with TaqDNA polymerase; 2) hybridization at the 3Ј-end to a neighboring partner; 3) mutual extension with TaqDNA polymerase; and 4) amplification of the product by PCR. The final product was subsequently cloned into the pCR2.1 Topo vector (Invitrogen) and sequenced. The remaining 3Ј-codons for the tail of the mature SmpB were added with designed oligonucleotides using PCR as above.
SmpB Expression Plasmids-A vector was prepared for an in vivo SmpB assay by amplifying the E. coli smpB promoter and the 5Јuntranslated region from genomic DNA, adding an NdeI restriction site overlapping the start codon and a BamHI site further downstream during the PCR, and ligating into the low copy plasmid pBBR-MCS2 (15) at its NsiI and KpnI restriction sites. The E. coli smpB coding region amplified from genomic DNA and the T. pseudonana construct of the previous section were used as starting points to prepare, by PCR methods, various chimeric and truncated smpB genes (Fig. 1), which were effectively inserted between the NdeI and BamHI sites of the above vector.
Overproducing clones for N-terminally His-tagged versions of the TTT, EEE, and EE0 proteins ( Fig. 1) were constructed by subcloning amplified smpB gene inserts from the above in vivo assay plasmids into the NcoI and BamHI restriction sites of pET-15b (Novagen). All clones were confirmed by sequencing the inserts.
In Vivo Tagging Assay-An earlier two-plasmid tmRNA assay (16,17) was modified for a three-plasmid SmpB assay. The smpB-ssrA region of E. coli BW25113, corresponding to positions 8618 -9685 of GenBank TM file AE000347.1, was replaced with the kan region amplified from plasmid pKD4 using the phage Red recombinase (18). This mutation was transduced with phage P1 into E. coli X90, and kan was removed by FLP recombinase using the subsequently evicted plasmid pCP20, as confirmed by PCR analysis of the genomic DNA. The resulting ⌬smpB-ssrA strain was triply transformed with a low copy tmRNA plasmid (p22T) encoding the tag peptide AANDENYALDD (which does not induce substantial proteolysis) (1), a high copy plasmid (p79A) encoding an inducible reporter mRNA without an in-frame stop codon, and one of the low copy smpB plasmids described in the previous section. Expression of reporter nonstop mRNA was induced by adding 0.75 mM isopropyl 1-thio-␤-D-galactopyranoside to a mid-log (A 600 of 0.4) culture in Luria-Bertani medium supplemented with antibiotics and shaking at 37°C for 1 h. The His-tagged reporter protein was purified from harvested cells, and the amount from the 2.5-ml culture was analyzed by electrophoresis in a protein gel with Coomassie staining (17).
The PCR product for the transcription of E. coli tmRNA by T7 RNA polymerase was prepared using a 3Ј-oligonucleotide with two 2Ј-methoxy nucleotides at its 5Ј end; such templates greatly reduce the fraction of RNAs with nontemplated 3Ј-extensions (19). tmRNA was gel-purified and folded in translation buffer by heating for 2 min at 80°C and cooling to 20°C over 10 min. Tightly coupled ribosomes were prepared from E. coli strains KW1063 and KW2073, the products of transducing either the ssrA::cat or ⌬smpB-ssrA::kan alleles into the strain CAN20 -12E (20) deficient in the RNases I, II, D, and BN. SmpB was detected in the preparation from the ssrA::cat strain by quantitative Western analysis at a molar ratio (SmpB/ribosome) of 0.005 Ϯ 0.002. All additional factors were purified as six-histidine fusions of E. coli proteins (21).
tmRNA Binding Assay-tmRNA was radiolabeled by randomly incorporating [ 32 P]CMP during in vitro transcription and purified and folded as described above in binding buffer (0.2 M KCl, 50 mM MES-KOH, pH 6.5, 5 mM MgCl 2 , and 0.1% Nonidet P-40). Reactions in 20 l of binding buffer containing 25 pM tmRNA (ϳ10000 cpm), 2 l of SmpB storage buffer, 5 mM ␤-mercaptoethanol, 0.1 mg/ml bovine serum albumin, and varying amounts of SmpB were incubated for 30 min at 22°C and then filtered through 0.45-m nitrocellulose disks prewet with binding buffer. Filters were washed three times with 2 ml of binding buffer and air-dried, and the fraction of tmRNA retained was measured by scintillation counting. Data were fit to the formula B ϭ SF/( where B is the fraction of tmRNA bound, S is the concentration of SmpB, F is the fraction of tmRNA that binds tightly, K d is the dissociation constant for F, and K 2 is the apparent dissociation constant for the remaining tmRNA. The three fitted values for F averaged 0.48 with a S.D. of Ϯ0.03. tmRNA Aminoacylation Assay-Reactions in 70 l of translation buffer containing 50 M [ 14 C]alanine, 1 M tmRNA, 2 l storage buffer, and no SmpB or various forms of E. coli SmpB (1 M) were initiated by the addition of 300 nM alanyl-tRNA synthetase and incubated at 37°C. At various time points 10 l was removed, quenched in 2 ml of ice-cold 5% trichloroacetic acid, incubated on ice for 30 min; alanine incorporation was then measured as described above.
Ribosome Binding Assay-The ⌬smpB-ssrA E. coli strain described above was transformed with the low copy plasmid encoding either the 000, EEE, or EE0 ( Fig. 1) form of SmpB, and grown to an A 600 of 0.41-0.42 in 800 ml of Luria-Bertani medium supplemented with kanamycin. Cells were pelleted, resuspended in 8 ml of lysis buffer (100 mM NH 4 OAc, 25 mM Tris-HCl, pH 7.5, 10.5 mM MgOAc, 2 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride) supplemented with 5 g/ml DNase I, and lysed using a French pressure cell. Lysates were cleared by centrifugation for 45 min at 26,000 rpm in a Beckman TLA-100.4 rotor, and then ribosomes were pelleted by centrifugation for 90 min at 50,000 rpm in the TLA-100.4 rotor and resuspended in 0.4 ml of lysis buffer. The pellet fraction was subjected to centrifugation in an 11-ml 10 -30% linear sucrose gradient in lysis buffer for 90 min at 35,000 rpm in a Beckman SW41 rotor. Gradient fractions (0.5 ml) were concentrated 10-fold and washed with lysis buffer using Microcon YM-10 centrifugal filter devices (Millipore). Samples were subjected to electrophoresis in a protein gel (14% polyacrylamide) followed by semidry electroblotting onto a nitrocellulose sheet. The blot was developed using a rabbit antibody raised against wild-type E. coli SmpB and horseradish peroxidase-coupled goat anti-rabbit antibody.

RESULTS
Nuclear smpB in Two Eukaryotes-Both plastidial ssrA and nuclear smpB genes were identified in highly redundant shotgun sequencing data for the diatom T. pseudonana (www.jgi.doe.gov) by BLAST searching.
The plastidial location of ssrA is indicated by the match that it and its flanking sequence make to a previously determined plastid sequence from other diatoms. tmRNA is encoded in 341 bp (scaffold 118, 10569 -10229) that can be aligned with a few short gaps to the Thalassiosira weissflogii plastidial tmRNA sequence (22) with 85% identity. The flanking sequence continues to match the entire 1.8-kbp ssrA-containing sequence from the T. weissflogii plastid (GenBank TM accession AF049491) with 82% identity, and the whole 68-kpb scaffold matches throughout the plastid genome of the diatom Odontella sinensis (GenBank TM accession Z67753), albeit with rearrangements. The tmRNA encoded in the T. pseudonana plastid would appear to be functional because it contains the key tmRNA features, namely a tRNA-like structure involving the RNA termini and a reading frame encoding a hydrophobic peptide tag.
The nuclear location of smpB is indicated by the two typical Thalassiosira nuclear introns it contains (Fig. 2) (23). This first example of eukaryotic SmpB shows remarkable similarity to bacterial SmpB, in particular to cyanobacterial SmpB (Fig. 3). Like introns identified in T. weissflogii nuclear genes (23), the smpB introns are short with strong matches to consensus features at the splice sites and branch region. The assigned start codon is the first in its reading frame; the only other in-frame AUG codon in the first exon is absent from one of the two T. pseudonana smpB alleles (see below). SmpB is much better conserved from bacteria to plastids than is tmRNA. The best BLAST hit at GenBank TM was to SmpB from the cyanobacterium Trichodesmium erythreum (Fig. 3), in keeping with the cyanobacterial origin of plastids. Identity is 48% over the length of the T. erythreum SmpB. Matching is strongest in the regions that are the best conserved among bacterial SmpBs and where contacts with tmRNA have been observed. One or two extra residues are inserted at three T. pseudonana positions, but these map to the ends of ␤-strands in the SmpB structural models (5,6) where similar insertions occur in various bacterial SmpBs. The major change relative to bacterial SmpB is a 50-residue N-terminal extension. The sequence of this extension matches that of the bipartite signal/transit peptide for the complex diatom plastid, which is removed stepwise during import into the plastid (24). It thus appears that diatoms deliver a very bacteria-like mature SmpB to a compartment containing tmRNA.
T. pseudonana smpB had excellent sequencing coverage, revealing two alleles in the sequenced diploid strain. Querying the original sequencing reads with BLAST showed that 47 of them from independent shotgun clones impinged on the gene. There were 805 nearly invariant positions between the start and stop codons inclusive, sequenced by the BLAST hits 19.8 times on average with a 1.7% error rate (base miscalls and single-nucleotide insertions/deletions). However, variation rose above noise at the five remaining gene positions, each of which showed two alternative bases at approximately equal frequencies. One of the base variations creates a Met/Leu alternative in the putative signal peptide of the pre-protein, but the other four are either silent codon changes or at unimportant intron positions; the mature SmpB sequence is thus unaffected by any of the variations (Fig. 2). These five positions covaried among clones, establishing a pair of alleles with allelic linkage of each pair of neighboring variant positions observed in at least nine clones (Table I). One clone suggested an effective allelic recombination within a very short segment containing an intron/exon junction. Of the remaining clones, 22 could be assigned to one allele and 18 to the other, approximating the even distribution expected for a heterozygous diploid.
We also identified a nuclear smpB gene in new data from the red alga Cyanidioschyzon merolae 10D (25), whose chloroplast genome contains ssrA (26). The best BLAST hit for this second nucleus-encoded SmpB is also from a cyanobacterium; identity is 52% over the length of the Synechococcus elongatus SmpB. The algal protein is predicted to contain an N-terminal signal peptide for its import into the simple plastid (Fig. 3). Like all but 27 of C. merolae nuclear genes, smpB has no introns.
smpB genes found in other eukaryotic genome projects are best ascribed to bacterial contamination because they are un-

FIG. 2. T. pseudonana smpB allele 2.
The nucleotide sequence is indexed at the left, and the protein sequence is indexed at the right with coding and predicted mature sequences in uppercase. The stop codon for the adjacent expressed gene is also in uppercase. Matches to consensus splice and branch sequences for U2/U6spliceosomal introns are underlined, and positions that differ in allele 1 (see Table  I interrupted by introns, in contigs with other bacterial affiliations, and encode proteins that are especially similar to known bacterial SmpBs. Sequences obtained from the nematode Brugia malayi, known to contain a Wolbachia endosymbiont, encode an SmpB that matches that from the Wolbachia endosymbiont of Drosophila melanogaster with 90% identity. One rice genome project (for Oryza sativa ssp. indica) contains at least four smpB sequences encoding proteins with identities of 76% to Sphingomonas elodea SmpB, 53% to Bacterioides thetaiotaomicron SmpB, 74% to Burkholderia fungorum SmpB, and 75% to Rubrivivax gelatinosus SmpB; such bacterial sequence contamination was identified and purposefully removed from another rice genome project. A segment containing adjacent smpB and ssrA genes, affiliated with Gram-positive bacteria, is found in a contig from the Trypanosoma cruzi genome project and not in sequence data from the other kinetoplastids Trypanosoma brucei or Leishmania major.
Diatom SmpB Promotes tmRNA Translation in E. coli Cells-The extensive sequence similarity that eukaryotic SmpB shares with bacterial SmpB, together with its predicted import into plastids that encode tmRNA, strongly suggests that the role of SmpB in plastids is equivalent to its only known role in bacteria, in tmRNA translation. The homology further suggests that the plastid protein might function in bacteria as has been demonstrated for other plastid proteins (27,28). Very high sequence similarity between the tRNA domains of the E. coli and T. pseudonana tmRNAs, with base identity at all the positions where contacts were observed between SmpB and the tRNA domain (7), encouraged a test for activity of T. pseudonana SmpB in E. coli cells.
An in vivo SmpB assay was developed from an earlier tmRNA assay (16,17). An E. coli strain with a chromosomal deletion encompassing the neighboring smpB and ssrA genes was transformed with three plasmids; the first plasmid allowed for induction of a discrete nonstop mRNA, the second produced a tmRNA variant encoding a peptide tag that does not induce proteolysis, and the third expressed the test SmpB under control of the E. coli smpB promoter. In this system, active SmpB allows tmRNA to stably tag the reporter protein encoded by the nonstop mRNA; the 9% increase in reporter protein mass is readily detected in a protein gel. Control cells lacking SmpB produced only the untagged reporter protein, and those expressing E. coli SmpB produced only tagged reporter protein (Fig. 4). The latter control shows increased activity of the variant tmRNA relative to systems in which SmpB is provided from the chromosomal gene. Presumably, smpB expression is   SmpB Tail Function increased in this assay even though the same promoter as the chromosomal gene was used, perhaps due to a mildly elevated gene copy number on a low copy plasmid.
To assay the predicted mature diatom SmpB in this system, its gene was constructed de novo. The codon bias was matched as closely as possible to E. coli smpB to keep synthesis rates equal. The T. pseudonana SmpB strongly promoted tmRNA translation in E. coli, albeit with some reduction of activity relative to E. coli SmpB (Fig. 4; compare lanes TTT and EEE).
Diatom SmpB Promotes tmRNA Translation in an in Vitro E. coli System-SmpB activity was assayed in a recently described in vitro system (9) adapted from an earlier tmRNA assay (29). Ribosomes from ⌬smpB-ssrA E. coli were programmed with poly(U) in the presence of radiolabeled alanine, bulk E. coli tRNA, enzymatically synthesized tmRNA, elongation factors, all required aminoacyl-tRNA synthetases, and the test SmpB. Poly(U) serves as a nonstop mRNA that should not promote the incorporation of alanine into polypeptide when SmpB is omitted but, nonetheless, produces a high background. With each round of tmRNA translation, five equivalents of alanine are incorporated.
Preliminary titrations showed that alanine incorporation reached a plateau at 20 M E. coli SmpB or at 90 M diatom SmpB (data not shown). Plateau activity was lower with diatom SmpB and yet was distinct from the background of the assay (Table II). The same result was obtained with ribosomes that were purified from ssrA Ϫ cells with negligible contamination (0.5% on a molar basis) by endogenous SmpB.
SmpB Tail Is Required for tmRNA Translation-Various chimeras between the diatom and E. coli smpB genes were constructed and assayed in vivo (Fig. 4) to examine whether the reduced activity of the diatom SmpB was due to a particular domain of the protein (the ␤-barrel body or the C-terminal tail). Neither partial nor complete replacement of the T. pseudonana SmpB tail with the E. coli equivalents improved tagging activity (Fig. 4, lanes TTT, TTE, and TEE). Instead, slight drops in activity suggest either that a normal interaction between cognate ␤-barrel and tail domains is diminished in these chimeras or that the chimeras create new negative interactions. Replacement of the E. coli tail with that from T. pseudonana did not diminish activity (Fig. 4, lanes EEE and ETT). Thus, the activity difference between the bacterial and diatom SmpB correlates with the presence of the ␤-barrel domain, not the tail.
One interpretation of the above conclusion is that the Cterminal tail plays no role in tmRNA translation. To examine this possibility, tail truncations were constructed that were guided by SmpB structural studies and the intron structure of the diatom gene. If the second exon were unspliced, SmpB would effectively be truncated by 16 C-terminal residues (Fig.  2), approximately half of its tail. No tagging activity was detected with such a truncation (TT0 in Fig. 4). This observation led us to test the effect of the equivalent truncation in E. coli SmpB, which was likewise completely inactive in tmRNA translation in vivo (EE0 in Fig. 4). These results demonstrate the importance of the distal half of the C-terminal tail of SmpB for tagging.
A mutation arose during cloning that had little effect on tmRNA translation activity despite altering a residue in the proximal tail (Asp-139, underlined in Fig. 3, changed to Asn) that is absolutely conserved in a collection of 113 SmpB sequences (compare TEE* to TEE in Fig. 4).
SmpB Tail Is Not Required for tmRNA Binding or Enhancement of Aminoacylation-The reduced activity of T. pseudonana SmpB and inactivity of tail-truncated SmpB in tmRNA translation could be due to lower affinity for E. coli tmRNA. This possibility was tested using a filter-binding assay to measure the affinity of purified SmpB for tmRNA (Fig. 5A). Approximately half of the tmRNA reproducibly showed high affinity binding to E. coli SmpB, and the other half was trapped only at higher SmpB concentrations, which could be explained by low affinity SmpB binding to a substantial fraction of improperly folded tmRNAs. A substantial misfolded fraction of tmRNA is not surprising given that, in our hands and for others (8), alanylation yields for tmRNA are only half of those for similarly prepared tRNA Ala . The lower affinity binding phase can be explained by the observation that SmpB has an affinity of 20 M for simple double-stranded RNA (30). Fitting the data for the high affinity phase with the assumption of 1:1 binding stoichiometry yielded a dissociation constant of 0.34 Ϯ 0.18 nM for full-length E. coli SmpB. This K d value is much lower than those reported previously (ranging from 20 to 400 nM) from gel retardation and indirect assays that usually have omitted magnesium ions from the binding buffer (2,4,30). The K d was increased Ͻ2-fold for partially or fully tail-truncated E. coli SmpB, showing that the tail contributes little to tmRNA binding and further suggesting that the ␤-barrel portions of the truncated proteins were well folded despite their complete inactivity in tmRNA translation in vivo.
Functional binding of tail-truncated SmpB to tmRNA was confirmed by an aminoacylation assay. It has been shown that, although tmRNA alone is a substrate for alanyl-tRNA synthetase, one equivalent of SmpB increases the rate of alanylation ϳ2-fold (4). We reproduced this effect and found that partially or fully tail-truncated SmpB provided the same enhancement as full-length SmpB (Fig. 5B). This result shows both that the truncated SmpB retains function and that its inactivity in promoting tmRNA translation is not due to a negative effect on aminoacylation.
The diatom SmpB had a substantially higher K d (5.2 Ϯ 2.3 nM), which might either reflect its imperfect folding in these experimental conditions or species-specificity in SmpB⅐tmRNA binding. Its lower affinity for E. coli tmRNA may account for its lower activity in E. coli tmRNA translation.
SmpB Tail Is Not Required for Stability or for tmRNAindependent Ribosome Binding-SmpB has recently been shown to bind to ribosomes even in the absence of tmRNA, and it has been suggested that tmRNA translation depends on this activity (11). Thus, the inability of the tail-truncated SmpB to promote tmRNA translation might be explained if such ribo- FIG. 4. SmpB assay in E. coli. E. coli X90 ⌬smpB-ssrA (p79A, p22T) was additionally transformed with the indicated smpB assay plasmid (see Fig. 1), and the reporter protein was purified and resolved in a Coomassie-stained gel. Bands for tagged and untagged reporter protein are marked and should not be confused with the two unidentified minor bands that appear between them in all samples. Thus no untagged reporter was detected for EEE or ETT, and no tagged reporter was detected for 000, TT0, or EE0. some binding activity were reduced. We investigated this possibility by fractionating lysates of ⌬smpB-ssrA E. coli expressing either no, wild-type, or truncated SmpB (000, EEE, or EE0 of Fig. 1) from the native smpB promoter on a low copy plasmid. Both SmpB forms were found in the pellet with ribosomes after high speed centrifugation and further co-purified with ribosomes in sucrose density gradient centrifugation (Fig. 6). Thus, it does not appear that the defect of truncated SmpB in tmRNA translation is explained by the failure of tmRNA-independent ribosome binding. Fig. 6 also shows that the tail-truncated SmpB accumulates to approximately the same level as full-length SmpB. Thus, the effect of the tail on tmRNA translation is not due to an effect on SmpB stability. DISCUSSION This work shows that eukaryotes with a plastidial tmRNA gene also have the smpB gene. Eukaryotic SmpB is encoded in the nucleus, with a peptide presequence expected both to promote import into the plastid and to be removed during the process. The expected mature eukaryotic SmpB promotes tmRNA translation in bacterial in vivo and in vitro systems. These results suggest not only that plastid tmRNA has the same protein cofactor requirement as in bacteria but also that the tmRNA⅐SmpB complex has the same function in plastids as in bacteria.
The implication that SmpB is a universal cofactor of tmRNA encourages a search for smpB in the nuclear genome of jakobids, whose mitochondria have no recognizable smpB but do produce a truncated tmRNA homolog that retains features important for SmpB binding (1, 31). Among completed eukaryotic genomes, excluding apparent contamination from bacteria, either the ssrA and smpB genes are both found, or neither is found. An apparent lack of smpB could result from a sufficient divergence such that its sequence is no longer recognizable, but this possibility has become less tenable with the finding that two eukaryotes have a readily recognizable smpB. Thus, the correlation of ssrA and smpB genes appears to be genuine, which is in line with the idea that SmpB has little function beyond its role as a tmRNA cofactor.
The SmpB tail emerges from the ␤-barrel opposite to the tmRNA-binding face (7), rationalizing our observation that it contributes little either to tmRNA binding or to aminoacylation by the class II alanyl-tRNA synthetase (Fig. 5). Neither does the tail contribute significantly to the stability of SmpB in vivo (Fig. 6). We therefore look to a role for the tail within the ribosome during tmRNA translation. One specific suggestion has been that the SmpB tail promotes peptidyl transfer to tmRNA by stimulating the decoding center in the small ribosomal subunit as codon-anticodon pairing does for tRNA (7,32). Removal of only 15 residues from its C terminus is sufficient to completely block tmRNA translation in vivo. Conserved sequence features of the tail suggest that it would form an ␣-helix (Fig. 3) that would present several positively charged residues on one face, yet the region appears unstructured in solution and in the crystal complex with the tRNA domain of tmRNA (5)(6)(7). We propose that the tail does adopt a helical structure during tmRNA translation. The positive face of this helix could be stabilized by a negative charge in ribosomal RNA. It is also possible that even though the SmpB tail does not appear to interact with tmRNA outside of the ribosome, it does so inside the ribosome. It will be of great interest to elucidate the course that SmpB takes in the ribosome during tmRNA translation and the encounters that it makes, perhaps by derivitization with Fe(II) for directed hydroxyl radical probing.  Fig. 1) was fitted to a model in which one fraction of tmRNA binds SmpB tightly and the other fraction binds less tightly (see "Experimental Procedures"). B, stimulation of tmRNA aminoacylation.

FIG. 6. Co-purification of tail-truncated SmpB with ribosomes.
A, anti-SmpB Western blot of the cleared lysate (S30) and the supernatant (S100) and pellet (P100) fractions of the S30 after high speed centrifugation from E. coli cells expressing no tmRNA and no (None), full-length, or tail-truncated SmpB. B, sucrose density gradient centrifugation of the P100 fraction for tail-truncated SmpB. Ribosomes in the fractions were traced by absorbance at 260 nm (A 260 ), and SmpB by a Western blot.