NATURE. FEBRUARY 4. 1967 UGA: A Third Nonsense Triplet in the Genetic Code by S. BRENNER L. BARNETT E. R. KATZ F. H. C. CRICK M.R.C, Laboratory of Molecular Biology, ‘Hills Road, Cambridge Mosr of tho sixty-four triplets of the genetic code’ have been allocated to one or other of the. twenty amino-acids. The two known nonsense triplets (UAA, ochre and UAG, amber) are believed to signal the termination of the poly- peptide chain. The only other triplet so far unallocated is UGA, for which binding experiments give uncertain or negative results. . ; In this article we show that UGA is “unacceptable” in our system (Escherichia colt infected with bacteriophage 74) and present suggestive evidence that it is nonsense; that is, that it does not stand for any amino-acid. Theore- tical arguments make it likely that there is no transfer RNA (tRNA) to recognize it.. The reason for this spparent absence of function is not yet known. Neither is it known whether UGA is nonsense in other organiams. Evidence that UGA may be nonsense in also been presented by Garen ef al.". They investigated the reversion of amber and ochre mutants in the alkaline - phosphatase gene of Z. coli. Amber mutants (UAG) reverted, as expected, to seven different amino-acids including tryptophan which is.coded by UGG. Ochre mutants (UAA) reverted to six of these amino-acids, but not to tryptophan. This negative result makes it unlikely that UGA stands for tryptophan (see. also Sarabhai and Brenner*) and. suggests that it might be a nonsense codon. Mutant X666 contains UGA. Much of our genetic work |. has been concerned with the left-hand ond of the B cistron © of. the rII region of bacteriophage T4. We have made extensive and detailed investigations of this region which are being reported elsewhere’. The mutant X655 occurs in the middle of this region. In brief our proof that X655 contains the triplet UGA consista in converting it to an ochre (UAA), using mutagens the behaviour of which is already known. Fe X655 was induced from wild type by 2-aminopurine, and identical mutants are also found after treatment of . wild type phage with hydroxylamine. This shows that it differs from an acceptable triplet by a G-C to A-T base pair change in the DNA. It is.not suppressed by any amber or ochre suppreesor (Table 1) and is therefore neither UAG nor UAA. The reversion ‘properties of X655 are shown in Table 2.. It is strongly induced to revert to r+ by 2-aminopurine, as is expected, but there isno induction to r+ by hydroxylamine. Thus-the triplet in the DNA either contains no G-C pairs or, if it. does contain one, it is connected to another unacceptable. triplet by a G-C to A-T transition, 2-000. Ue Pe ae motes The triplet is in fact connected to UAA by a transition, because X655 can be converted to an ochre and this change is induced by 2-aminopurine (Table 2). The nature of the E. coli has . Two base triplets of the genetic code are known not to represent any amino-acid. It now appears that, in Escherichia. coli, the UGA triplet of the bases uracil, guanine and adenine does not code for an amino-acid and Is therefore also a ‘‘nonsense triplet”. transition is more precisely specified by the finding that the conversion to an ochre is induced by hydroxylamine and that the ochre triplet produced does not require any replication for expression. Using a. previous argument‘ this result. suggests that the change arises from a GA change in the messenger RNA. Because X655 is not an amber, this proves that it contains the triplet UGA. To confirm that an amber at the site of X655 would be sup- pressed by amber suppressors the X655 ochre has been converted to an amber by mutation and its properties tested (Table 1). Table 1. SUPPRESSION PROPERTIES OF Y655 AND ITS DERIVATIVES : Amber suppressors Ochre suppressors Mutant Triplet sup ous autre certs ew$ sug sud £655 UGA 0 Qo o oOo .-.9 Oo .9 X655 ochre UAA 0 0 9 0 + + + X6565 amber UAG 0 + + +. + + + Phage stocks were plated on the following strains’: su-, CA244; sut CA266; suiy,CA180; sufpr, CA265; sug. CA165; suf ,CA167; and sus.” 248, : ’ . : Table 2. REVERSION OF Y655 Reversion index (in units of 10%) : lamin droxylamine Hydro’ e Spontaneou. 2-Aminopurine 7 direct . aiter growth tor? — 4 a2 6 - ee to ochre 4° =. 1,080: 88 X655 was treated with 2-aminopurine and hydroxylamine as previously ' deacribed**, Total phage was assayed on E. coli B and r* revertants on CA244 (su-). Ochre revertants were selected on CA248 (sup) and distinguished. pal revertants by picking and stabbing about 300 plaques into CA2é8 an A244. - o . oo a wee 7 - Other occurrences of UGA. In three cases we have been able to produce the triplet UGA by selected phase shifts in our region. When (+ —) phase shifts are made over the first part of the B cistron, the two phase shift mutants frequently do not suppress each other. We have shown‘ that these barriers to mutual suppreasion are due to the generation of unacceptable triplets in the shifted frame. _One of these barriers, b,, has been identified as an amber and two others, b, and b,, as ochree. Three barriors, 3,; b,.and b,, have now been identified as'UGA by their base-analogue induced reversion to ochres. In each caso the identification has been checked by converting the ochre to an amber at the same site. eS tophan is represented by the single codon UGG. It would therefore be-expected to mutate by transitions to both. UAG (amber) and UGA, and thus in such cases amber and UGA mutants should occur in cloee pairs. ’ The amber mutant, HB74, which maps close to X655, 450 is an examplo of this. Genetic crosses between it, 665, and the ochre and amber derived from X655, show that HB74 maps identically to the amber derived from X655, as expected (Table 3), . Table 3. RECOMBINATION BETWEEN VARIOUS. MUTANTS 655 655 X655 —_—soochre amber HBT4 Triplet X655 0 _ UGA 655 ochre 0 0 UAA 655 amber + 0 0 : UAG AB74 + 0 0 oO UAG The phages were crossed [n E. coli B and the complexes irradiated with ultra-violet light to-stimulate recombination (see ref. 4). In the Table, 0 means that r* recombinants were not significantly above the reversion rate, which was between 2 and 9x 10-?: in those experiments qhere positive resulta were obtained (+), the frequency was between 2 and 6 x 10~5, So far we have found the expected pairs ‘consisting of UGA and an amber in two other cases. In the A cistron, &@ mutant X665* is found with the amber mutant N97, and in the B cistron, N65 is paired with the amber mutant - X237. Both N97 and X237 are likely to have arisen from UGG (tryptophan) which is confirmed by the finding that they respond only poorly to the amber suppressor suj; which inserts glutamine’. Both X665 and 65 have been converted into ochre mutants, showing that they contain the triplet UGA. These ochres have also been converted to ambere at the same site. Mapping investigations, analogous to those in Table 3, aro consistent with these allocations. Unacceptability of UGA. There is very good evidence that the amino-acid sequence coded by the firat part of the B cistron is not critical for the function of the gene‘. It can be replaced by varying lengths of the A ciatron using deletions that join the two genes. Moreover, an extensive (— +) frame shift can be made without notice- able effect on the function. Of the fifteen known base- anslogue mutants in the region, thirteen are either ochres or ambers; one, HD263, is temperature sensitive and X655 is UGA. The extreme bias towards amber and ochre chain-terminating mutants confirms the dispens- ability of the region‘. These results make it unlikely that the unacceptability of UGA in X655 and the three barriers results from the insertion of an amino-acid, and strongly suggest that it is nonsense. _ In addition, the UGA mutant X665 in the A cistron has been combined with the deletion r1589 and has been found to remove the B activity of this phage. This is the test for. nonsense originally used by Benzer and Champe‘. In all these cases, however, it could be argued that UGA might code cysteine, especially as the two known triplets for cysteine are UGU and UGC. If the B protein already contained a cysteine essential for its function the effect of UGA elsewhere might be to produce an S-S bridge between the cysteine inserted by UGA and the (hypo- thetical) essential one, and thus inactivate the protein. Nevertheless we regard this as unlikely for two reasons, one genetic and one chemical. The genetic evidence concerns the anozaalous minutes produced by certain (+ +) combinations in the B cistron‘, In some regions of the first part of the B cistron combina- tions of two (+) phase shift mutants are able to grow to some extent on the restrictive host, E. coli K12. The plaques produced are minute, however, showing that the wild type Phenotype is very far from boing completely restored. A detailed analysis of one set of these combins- tions showed that minutes are obtained only from pairs of (+) mutants which straddlo barrior &,. The presence of the barrier is obligatory because, if it is removed by mutation, the (+ +) doubles are unablo to grow at all-on &. coli K12,. The minutes are cloarly duo to a phase error of one sort or anothor and tho phase crror is dependent on the barrier b, which we now know to be UGA. This result shows that UGA cannot be associated with any * This is not a misprint for Y655. “place of the codon no ‘RNA NATURE, FEBRUARY 4, 1967 , normal amino-acid reading and conclusion that it is nonsense. The chemical reason-for UGA not coding for cysteine comes from the work of Khorana e¢ al.?. They have shown that poly (UGA), when used as a messenger in a cell-free system derived from EH. colt induces the production of poly methionine (corresponding to AUG) and also poly aspartic acid (corresponding to GAU). No other amino- acid appears to be incorporated. In particular, no poly’ cysteine was found. For various reasons this evidence is not completely decisive, but it at least makes it unlikely that UGA is cysteine. Function of ‘UGA. It might be thought that the se- quence containing UGA was nonsense because it was tho signal for the beginning or ending of a gene (or operon). In other words, that it produced its effect during the synthesis of the messenger RNA on the DNA template of the gene. This explanation is highly unlikely because the effects of UGA depend on it being read in phase. The phenotypic effect of X656 can be removed when the mutant is placed in a (— +) shifted frame‘, and the barriers bs, bs and 6, are of course produced by phase shifts. That is, the base sequence UGA actually occurs at these places in the wild type messenger RNA but in such a way that it is out of phase when the message is read correctly. Because we have no reason to suspect that RNA poly- merase synthesizes: messenger RNA in groups of three bases at @ time these results imply that the phenotypic effects of UGA must occur during protein synthesis. It thus seems unlikely that UGA codes for any amino- acid, and in particular it does not ap to code for either cysteine (UGU and UGC) or tryptophan ;(UGG). The wobble thoory of codon-anticodon interaction de- veloped by one of us* makes the prediction that because of a wobble in the recognition mechanism at the third molecule can recognize XYA alone without at the same time recognizing either Pointe strongly to the ,XYG or both XYU and XYC. Such theoretical argu- ments cannot be considered conclusive, but they certainly suggest that UGA is..triplet for which no ¢RNA oxists. . For this reason we think it unlikely that UGA produces the efficient termination of the polypeptide chain, but more direct evidence will be noeded to establish this point. Conclusion. We have thus established that in the phage- infected cell UGA ‘is certainly “ table’’ in the rII cistrons, although it remains to be seen whether this is true for other species. We have produced reasons why it is unlikely to eode for any amino-acid. We are confident that there must be weighty reasons if even a single triplet is not used in the genetic code, because otherwise natural selection would have certainly allocated it to an amino- acid. At the moment we are inclined to believe that UGA may be necessary as. “‘space” to separate genes in a polycistronic . It is possible to make a plausible theory for Z. colé along these lines, but we prefer to leave the discussion of this until we have more experimental evidence to support it. This we are at present attempting to obtain. : : / We thank Drs. A. Garen, H. G. Khorana and A. Sarabhai for interesting discussions and for showing us tkeir papers in advance of publication. One of the authors (E. R. K.) is. @ holder of s- United States Churchill Foundation scholarship. Reoelved December 22,1066 - —— . _ ' For the structure of the genetic code and the evidence for nonsense triplets seo the papers in the Cold Spring Harbor Symposium XXXI on “The Genetic Code”, 1966 (in the prem), . mt * Welgert, M. G., Lanka, E., and Garen, A; J. Mol. Biol. 28, 301 (1967). . *Sarabhal, A., and Brenner,.8., [n preparation, ‘ Barnett, L., Brenner, S., Crick, F. H. C., Shulman, R. G., and Watts-Tobin, R.J., Phil, Trane. Roy. Soe.(in the press), ‘ we * Brenner, S., Stretton, A. O. W., and Kaplan, 3.: Nature, 206, 094 (1065). * Benzer, 8.,and Champe. 8. P., Proe. U.S. Nat. dead. Sei., 48, 1114 (1962). - * Morgan, A. R., Wells, R..D., and Khorana, H. G., Proc. U.S. Nat. dead. Sei. (in the press). * 0 0 oa . * Crick, F. H.C., J. Mol. Biol., 19, 548 (1966), * Brenner, 8..and Beckwith, J. R.. J. Mol. Biol.,18, 620 (1965),