Inefficient Translation of T7 Late mRNA by Bacillus subtilis Ribosomes SPECIES-SPECIFIC

subtilis 30 S subunits inefficiently recognize initiation sites in mRNAs from Gram-hegative bacteria, but they are able to efficiently recognize initiation sites in mRNA derived from Gram-positive bacteria. identification of of the in vitro products subtilis ribosomes correct translation initiation sites in late T7 they Competition not B. subtilis ribosomal of B. subtilis concluded that strong Shine- sequences may be necessary in B. subtilis translation initiation sites; however, additional determinants of initiation which differ from those found in the translation initiation sites of mRNAs

includes an initiation codon, a Shine-Dalgarno sequence, and an appropriate spacing ("window") between these two (1-3, 14,15). The Shine-Dalgarno sequence consists of a polypurine stretch of variable length which is located 5' to the initiation codon and is capable of base pairing to the 3' end of the 16 S rRNA (16).
McLaughlin et al. (13) have suggested that the Shine-Dalgarno complementarity required by Bacillus subtilis and other Gram-positive ribosomes is significantly greater than that required by E. coli and other Gram-negative ribosomes, and is required for species-specific translation. The sequence information for over 40 Gram-positive derived translation initiation sites that has appeared since its formulation supports this hypothesis since all of the sites contain "strong" Shine-Dalgarno sequences (17). The "strength" of the Shine-Dalgarno sequence was estimated by determining the free energy of formation of the most stable double helical complex The known E. coli translation initiation sites (19) include a number which have strong Shine-Dalgarno sequences with calculated free energies of binding that overlap those of B. subtilis (17). The E. coli phage T7 contains an unusually large number of these strong ribosome binding sites (20). To determine whether a ribosome binding site composed of a strong Shine-Dalgarno sequence with an appropriately placed initiation codon is a sufficient determinant for translation by B. subtilis ribosomes, we tested mRNA prepared from T7 DNA for activity with an in vitro B. subtilis translation system. We decided to test the late region of T7 for this purpose because of its complete characterization and abundance of strong ribosome binding sites and the ease of identification of protein products using mutants and because there are few polar effects in T7 gene expression (20). We find that B. subtilis ribosomes do translate authentic T7 late proteins, although at a markedly reduced level compared to E. coli ribosomes. There is some correlation between the strength of the Shine-Dalgarno sequence and the relative expression of the protein by B. subtilis ribosomes; however, there must be other features of a Gram-positive translation initiation site which are important for efficient translation which remain to be elucidated.

EXPERIMENTAL PROCEDURES AND RESULTS~
The late region of T7 (i.e the region transcribed by the T7 RNA polymerase) contains several genes that are preceded by strong Shine-Dalgarno sequences with4ree energies of binding close to the average value of -16.7 kcal/mol characteristic of Shine-Dalgarno sequences derived from Gram-positive oganisms. Fig. 1 shows the response to E. coli and B. subtilis ribosomes to the addition of T7 late mRNA and to $29 mRNA. While B. subtilis ribosomes translate the 429 mRNA, they are practically inactive on the T7 late mRNA, despite the relatively strong binding energies of these sequences. The relatively poor translation of $29 mRNA by E. coli ribosomes would seem to be in contrast to the earlier reports of efficient translation of this mRNA by E. coli ribosomes (11). There are two factors which are responsible for this apparent change in the translatability of $29 mRNA. First, we have used [35S] methionine rather than [3H]lysine to increase the sensitivity of our fluorograms, and B. subtilis, but not E. coli, ribosomes predominantly synthesize the gene 6 product (also referred to as the 13.9-kDa protein) (11,21 The relative amounts of each protein made by E. coli and B. subtilis ribosomes were quantitated by scanning fluorograms with a densitometer (Table I). B. subtilis ribosomes show greater relative translation of proteins that have strong Shine-Dalgarno interactions (note 3.5, 11, and 14) and less relative translation of those which have weaker Shine-Dalgarno interactions, as compared to E. coli ribosomes. The exception to this is seen with protein 9, which does not have a strong Shine-Dalgarno interaction but is translated by B. subtilis ribosomes at an increased relative amount as compared to E. coli ribosomes. These results indicate that while B. subtilis ribosomes do not efficiently translate T7 mRNA they may prefer initiation sites that have strong Shine-Dalgarno sequences. DISCUSSION T7 late mRNA is inefficiently translated by B. subtilis ribosomes despite the strong Shine-Dalgarno sequences in many of these mRNAs. Thus, although strong Shine-Dalgarno sequences appear to be necessary for translation of mRNA by B. subtilis ribosomes (23), such sequences are not sufficient to allow the efficient translation of E. coli phage mRNA by a system containing B. subtilis ribosomes. The products of the low levels of translation include the same proteins made by E. coli ribosomes, so B. subtilis ribosomes appear to recognize the correct initiation sites on T7 late mRNA; however, B. subtilis ribosomes prefer different initiation sites than E. coli ribosomes. The translation of T7 late mRNA by B. subtilis ribosomes could be inhibited at any stage, but the ability of B. subtilis ribosomes to translate $29 mRNA following preexposure to T7 late mRNA ( Fig. 4) suggests that B. subtilis ribosomes inefficiently bind to initiation sites on T7 mRNA. Poor translation from what is otherwise a good initiation site is analogous to some recent examples of translational regulation (23-26). It is possible that there is a factor which inhibits the translation of heterologous mRNA by B. subtilis ribosomes. Unlike gene 32 protein (24), or the ribosomal proteins (25) which inhibit translation of specific mRNAs, such a factor would have to be quite nonspecific. In addition it would have to block translation for B. subtilis ribosomes but not E. coli ribosomes, since the components of the translation systems are interchangeable (11,12).
The inducible resistance genes derived from Gram-positive organisms appear to use a different mechanism for translational regulation (26). The mRNAs are capable of at least two mutually exclusive conformations. In the repressed state the Shine-Dalgarno is sequestered in a stem structure and would Expression of identified proteins is given as the ratio of each protein relative to the total ,incorporation of the reaction, based on densitometry of fluorograms of sodium dodecyl sulfate gels. Also indicated are the binding energies (AG) in kcal/mol for the interaction of the translation initiation site with the 3' end of the 16 S rRNAs of B. subtilis and E. coli Our preferred interpretation of our observations is that E. coli and B. subtilis translation initiation sites share some common elements but differ at least with respect to specific features c-f these elements. An analysis of 42 translation initiation sites derived from Gram-positive organisms supports this view (17). The common elements include the Shine-Dalgarno sequence and its spacing to an initiation codon. The Gram-positive derived Shine-Dalgarno sequences are "stronger" than the average E. coli sequence, but the spacing to the initiation codon is similar. The center of the Shine-Dalgarno sequence is usually found about 10 bases upstream of the initiation codon. Although an initiation codon is an invariable part of a translation initiation site, B. subtilis and other Gram-positive organisms use non-AUG initiation codons more frequently than E. coli (29% to 9%) (17, 19). E. coli ribosomes appear to utilize additional sequence preferences within the ribosome binding site (31, 32); for example, these sites tend to be A-U rich. The Gram-positive derived initiation sites accentuate this characteristic. A collection of B. subtilis and other Gram-positive derived mRNAs contain 42% A residues for a 50-base region around the initiation site (33).
In summary, B. subtilis translation initiation sites differ from E. coli sites, both with respect to the strength of the Shine-Dalgarno and possibly to sequence preferences within the translation initiation site. A strong Shine-Dalgarno sequence is not a sufficient signal for efficient translation of a mRNA by B. subtilis ribosomes, but is probably a necessary one. Additional features of a strong B. subtilis translation initiation site remain to be elucidated.

Mol. Biol. 1 3 5 , 627-638
Continued on next page.  3-29, 4-20, 5-28, 6-147, 7-213, 8-11, 9-17, 10-13, 11-37, 12-3, 13-149, 14-140, 15-31, 16-9, 17-290, 18-182, 19-10 The amber mutants showed reversion on the nonpermissive host of less than 0.01% that On the permissive host except for 10-13. 10-13 had a reversion to wild type of 1%. which is not uncommon since gene 10 is the major coat protein of the T7 phage particle (F. W. Studier, personal communication). are identified in Fig. 3. The proteins which are most easily identified Some of the products of the translation of Tl by 8. subtilis ribosomes include 1.3, 2.5, 9, 10. 11, and 14. In the cases of 3.5 and 5.5, the protein band at the expected location is missing, but there appear to be other changes in the pattern of translation products as well. There are two prominent protein bands from the B. subtilis in vitro translations which have not been identified (found between 3.5 and 5.5 in Fig. 31. When equal amounts of acid-precipitable counts are compared, B. subtilis and E. ribosomes prefer to translate different products (see Fig. 3,

and i].
Since the B. subtilis ribosomes do translate several of the T7 proteins, they It is not clear why B. subtilis ribosomes translate Tl mRNA poorly. recognize the Correct initiation sites for these proteins. This inability to efficiently translate Could be due to poor initiation of translation 1l.e. weak binding to the initiation regionl. Alternatively, initiation complex formation could proceed normally followed by a block at a later point in the bind to Tl mRNA in Unproductive complexes, these complexes might prevent the translation as Suggested by Sharrock & (45). If 8. subtilis ribosomes ribosomes from translating a mRNA presented at a later time. Such a competilate mRNA and then adding 629 mRNA as shown in Fig. 4 . T7 late mRNA lor the tion experiment was performed by peeincubating E. &tb ribosomes with Tl transcription reaction) does not inhibit translation of 629 since E. subtilis poration. E.. S -ribosomes are not inactivated by preexposure to Tl ribosomes with 629 sRNA or with 629 plus T7 late mRNA give the same incorlate mRNA, since addition of 629 mRNA at 5 rnin results in immediate incorporation at a rate similar to the initial rate using 629 mRNA alone. The absolute level of incorporation for this competition assay is lower than for 629 mRNA alone, but this is probably a consequence of the loss of activity these ribosomes display after 15 mi" 133). Thus, the relatively inefficient binding to Tl late mRNA l n nonproductive complexes.