Candida albicans - NCBI

18 downloads 13 Views 2MB Size Report
genetic code which invalidate the theory that the code is frozen and universal. The unexpected finding that some organisms evolved alternativegenetic codes ...

The EMBO Journal vol. 15 no. 18 pp.5060-5068, 1996

Transfer RNA structural change is a key element in the reassignment of the CUG codon in Candida albicans Manuel A.S.Santos1, Victoria M.Perreau and Mick F.Tuite Research School of Biosciences, University of Kent, Canterbury, Kent CT2 7NJ, UK

'Corresponding author

The human pathogenic yeast Candida albicans and a number of other Candida species translate the standard leucine CUG codon as serine. This is the latest addition to an increasing number of alterations to the standard genetic code which invalidate the theory that the code is frozen and universal. The unexpected finding that some organisms evolved alternative genetic codes raises two important questions: how have these alternative codes evolved and what evolutionary advantages could they create to allow for their selection? To address these questions in the context of serine CUG translation in C.albicans, we have searched for unique structural features in seryl-tRNACAG, which translates the leucine CUG codon as serine, and attempted to reconstruct the early stages of this genetic code switch in the closely related yeast species Saccharomyces cerevisiae. We show that a purine at position 33 (G33) in the C.albicans Ser-tRNACAG anticodon loop, which replaces a conserved pyrimidine found in all other tRNAs, is a key structural element in the reassignment of the CUG codon from leucine to serine in that it decreases the decoding efficiency of the tRNA, thereby allowing cells to survive low level serine CUG translation. Expression of this tRNA in S.cerevisiae induces the stress response which allows cells to acquire thermotolerance. We argue that acquisition of thermotolerance may represent a positive selection for this genetic code change by allowing yeasts to adapt to sudden changes in environmental conditions and therefore colonize new ecological niches. Keywords: Candida albicans/evolution/genetic code/ serine CUG decoding/tRNA structure and function

Therefore, alternative genetic codes must have evolved from the standard genetic code, yet code changes would introduce amino acid substitutions in a large number of, if not all, proteins encoded by any genome and should have been lethal. This raises the questions of how and why alternative genetic codes have evolved. To date, with the exception of molecular phylogeny studies which unexpectedly show that reassignment of the UAA and UAG stop codons to glutamine have arisen as independent events in several ciliates (Tourancheau et al., 1995), very little experimental evidence has been obtained that allows us to develop a model for the evolution of alternative genetic codes. However, the establishment of alternative genetic codes has been subjected to theoretical consideration, from which two distinct theories have emerged: (i) the 'Codon Capture Theory', which postulates that alterations to the genetic code can only evolve via a neutral mechanism to preclude introduction of mutations into proteins (Osawa et al., 1992); (ii) the 'Ambiguous Intermediate Theory', which postulates that genetic code changes are relatively fast processes driven by selection (Schultz and Yarus, 1994, 1996). The Codon Capture Theory was derived from the observation that variations in genome GC content influence codon usage (Osawa et al., 1992) and could 'force' rarely used codons to disappear from the entire pool of mRNAs, as is illustrated by the non-assignment of the CGG codon in Mycoplasma capricolum, whose genome GC content is only 25% (Durasevic et al., 1994). Such an evolutionary pathway appears to have operated in Mycoplasma and Spiroplasma and in non-plant mitochondrial translation systems, where release factors have lost affinity for UGA allowing for its reassignment as a tryptophan codon (Jukes and Osawa, 1993). In these cases, all UGA stop codons are assumed to have mutated to UAA or UAG and were subsequently reintroduced by back mutation of tryptophan UGG codons and simultaneously captured by Trp-tRNA isoacceptors with novel decoding properties that allowed for decoding of both UGG and UGA codons (Osawa et

Introduction The human pathogenic yeast Candida albicans and a number of other Candida species translate the standard leucine CUG codon as serine (Kawaguchi et al., 1989; Ohama et al., 1993; Ueda et al., 1994; Santos and Tuite, 1995; Sugiyama et al., 1995). This genetic code alteration is the most recent addition to an increasing number of alternative genetic codes uncovered in prokaryotic, eukaryotic and non-plant mitochondrial translation systems (reviewed by Osawa et al., 1992). This finding is unexpected, since the biological evidence obtained to date suggests that all forms of life evolved from a common ancestor.

5060

al., 1992).

An alternative theory to explain the existence of genetic code changes has recently been proposed by Schultz and Yarus (1994, 1996). This theory was developed from an initial observation that 10 out of the 18 known codon reassignment events involved single nucleotide changes at either the first or third codon position and naturally occurring tRNAs are able to misread the reassigned codons (reviewed by Schultz and Yarus, 1994, 1996). This 'Ambiguous Intermediate Theory', postulates that tRNA isoacceptors which are able to misread near cognate codons can gradually take them over from their cognate tRNAs or release factors in a process driven by selection, i.e. in any codon reassignment event, an intermediary step is required in which a codon codes either for two different © Oxford University Press

Serine CUG decoding in C.albicans

amino acids or for an amino acid and translation termination. The theory assumes that codon ambiguity is not lethal and therefore codons do not have to disappear from the genome during their reassignment. If this theory is confirmed experimentally, then one is left with the intriguing question of what kind of evolutionary advantage arises from a genetic code switch that allows for its selection? In order to better understand how and why a number of Candida species have evolved an alternative assignment for the CUG codon, we have attempted to reconstruct the early stages of this genetic code change in S.cerevisiae. The data conclusively show that S.cerevisiae is able to tolerate serine CUG mistranslation and that it does so by inducing the stress response and efficiently degrading mistranslated proteins. This is the first experimental evidence for evolution of a genetic code change driven by selection using a novel mechanism that does not require that codons disappear from the entire pool of mRNAs.

A C C G C U G U G C

G A U A C G U CG CC A U D DGA CGUCC GA GmD GC GAGG T ac4C C G G M5C GG G C G

Experimental approach Genetic code changes are mediated by novel tRNAs that either (i) have mutated anticodons but retain their identities or (ii) have changed their identities without changing their anticodons. In either case, studying the decoding and/or aminoacylation properties of these tRNAs is bound to elucidate how alternative genetic codes have evolved. The Ser-tRNACAG responsible for translation of the CUG codon as serine in Candida species is the only tRNA known to contain guanosine at position 33 in the anticodon loop (Steinberg et al., 1993). This critical position is usually occupied by a pyrimidine (U in general) in all known tRNAs (Steinberg et al., 1993). The presence of G33 suggests a mechanistic link between the unusual structure of Ser-tRNACAG and the genetic code switch. We reasoned that by investigating whether or not the decoding properties of Ser-tRNACAG are modulated by G33 it would be possible to establish a functional link between the novel structure of Ser-tRNACAG and development of the genetic code switch. We speculated that if the CUG codon did not disappear prior to its reassignment in Candida, expression of Ser-tRNACAG in a closely related yeast species which translates the CUG codon as leucine (i.e. S.cerevisiae) would allow evaluation of the negative impact that serine CUG decoding had during the evolution of the code switch. Saccharomyces cerevisiae and C.albicans have genomes of similar GC content, 40 and 36% G + C respectively, and similar codon usage (Lloyd and Sharp, 1992). More importantly, the CUN codon family are rarely used in the mRNAs of both species (Lloyd and Sharp, 1992), making S.cerevisiae an ideal host for these studies.

Expression of Calbicans Ser-tRNACAG in S.cerevisiae To determine whether or not it would be possible to express C.albicans Ser-tRNAcAG in S.cerevisiae, an expression system based on S.cerevisiae single copy vector pRS315 (Sikorski and Hieter, 1989) was constructed. A 250 bp genomic DNA fragment of Calbicans containing the SertRNACAG gene and its associated polymerase III terminator

G C

mG

U

_ A, C, U, G Results

m1A

GA U

G C A v

CA

v

mIG G

Fig. 1. Cloverleaf structure of Calbicans Ser-tRNACAG. The arrow indicates the location of nt 33 in the secondary structure. A/CIU indicate the mutations introduced in the tRNA.

was cloned into pRS315 to give plasmid pUKC701-G33. pUKC701-G33 transformants of S. cerevisiae were obtained in selective medium; these transformants grew significantly slower than control pRS315 transformants (data not shown), suggesting that Ser-tRNAcAG could be expressed and was functional in S.cerevisiae. Variants of the SertRNACAG gene containing either C, A or T at position 33 (Figure 1) were synthesized as described in Materials and methods and cloned into pRS315. The resulting plasmids (pUKC71X-N33) were then transformed into S.cerevisiae. Transformants were obtained for GIAIC33 but not for pUKC716-T33. Five independent attempts were made to transform S.cerevisiae with pUKC716-T33, but no clone was obtained, despite the fact that in all cases control transformations carried out with the pRS315 vector alone or with one or the other of the pUKC71X-N33 vectors gave on average 104 transformants/tg DNA. These data indicate that the U33 mutant tRNA is lethal in S.cerevisiae even when expressed from a single copy vector. That expression of canonical U33 Ser-tRNAcAG is apparently not compatible with viability suggests that G33 in SertRNACAG decreases the ability of Ser-tRNAcAG to compete with endogenous CUG-decoding Leu-tRNA and its expression can thus be tolerated by S.cerevisiae. Determination of growth rates for each class of transformants and control cells showed that expression of the wild-type (G33) and A33 and C33 mutant tRNAs induced relative growth decreases of 47.9, 38.6 and 5.1% respectively (Figure 2). While the strong growth inhibition induced by the wild-type and A33 mutant tRNAs was expected, the relatively low level of inhibition induced by the C33 mutant was surprising, both because expression of the U33 tRNA mutant was lethal and previous work had shown that U/C33 tRNAs from S.cerevisiae and Escherichia coli have similar ribosomal binding constants and therefore decoding efficiencies (Curran and Yarus, 1986; Dix et al., 1986). With the exception of the wild-

5061

M.A.S.Santos, V.M.Perreau and M.FTuite 10 -.

100

0 0

90 80

62

~70

0

1

L.)

60 50 40 30

0.1

20

10 0

0

0.01 0

10

20

30

40

50

Time (hr) Fig. 2. Growth profiles of S.cerevisiae transformants of C.albicans Ser-tRNACAG. Cultures were grown in 100 ml flasks at 30°C. The relative decrease in growth rates for S.cerevisiae transformed with pRS315 (02), pUKC718-C33 (0), pUKC717-A33 (U) and pUKC715G33 (A) shown in the box were determined from the specific growth rates calculated for each transformant. The values represent the percentage decrease in relation to S.cerevisiae transformed with the pRS315 vector alone.

type (G33) tRNA, the N33 mutants did not induce a significant decrease in the final density of the cultures (OD600 5.0), indicating that viability is minimally affected by serine CUG decoding.

Mutations at position 33 in Ser-tRNACAG affect CUG decoding efficiency in vivo To quantify CUG decoding efficiencies of each of the SertRNACAG variants and confirm that the growth inhibition observed was caused by serine CUG decoding and not by any other secondary effect, we exploited a ,B-galactosidase thermolability assay we have developed to quantify misreading in S.cerevisiae (M.A.S.Santos, I.Stansfield and M.F.Tuite, in preparation). Sense codon misreading affects the thermostability of 3-galactosidase and the rates of thermoinactivation give a relative measure of amino acid misincorporation (Branscomb and Galas, 1975). Since the E.coli f-galactosidase gene has 54 CUG codons, misreading of these codons as serine would reduce Pgalactosidase thermostability and its thermoinactivation rate could be used to measure serine CUG mistranslation in vivo. For this, S.cerevisiae was co-transformed with pGL-C 1, which encodes a fusion between glutathione S-transferase and ,-galactosidase genes (GST-f-gal) and one or other of the pUKC7 1X-N33 plasmids. f3-Galactosidase thermoinactivation profiles were then determined directly in S.cerevisiae cells as described in Materials and methods. 1-Galactosidase synthesized in the presence of the A33 mutant tRNA displayed a faster rate of inactivation than f-galactosidase synthesized in the presence of either the wild-type G33 or the C33 mutant tRNA (Figure 3), suggesting that the A33 tRNACAG mutant is the most efficient decoder of the three tRNAs examined. These data are in agreement with previous results which showed that N33 mutations affect decoding efficiency by increasing the binding constant of tRNAs to the ribosomal A site (Dix et al., 1986). Considering that expression of the SertRNACAG U33 mutant is lethal and that previous data showed that U33 and C33 mutants of other tRNAs have 5062

2.5

5.0

7.5

10.0

12.5

15

Time (min)

Fig. 3. Mutations at position 33 in Ser-tRNACAG affect decoding efficiency of the CUG codon. Decoding efficiency of the CUG codon by Ser-tRNACAG was determined by measuring the effect of serine misincorporation at CUG codons on 3-galactosidase thermostability. For this, f-galactosidase thermoinactivation profiles were determined by measuring f-galactosidase activity directly in S.cerevisiae cotransformed with pGL-Cl and either pRS315 (E), pUKC718-C33 (0), pUKC717-A33 (U) or pUKC715-G33 (A). Cells were first incubated at 47°C to denature mistranslated f-galactosidase and the activity of the fraction of P-galactosidase that remained functional after each time point at 47°C was determined by incubation for 2 min at 37°C in the presence of ONPG. The percentage of ,B-galactosidase activity for each time point was determined in relation to total activity determined from cells that were not incubated at 47°C.

similar decoding efficiencies (Curran and Yarus, 1986), the poor decoding efficiency of the C33 mutant is unexpected, but correlates well with the small inhibitory effect on growth we observed (Figure 2). The A33 mutant tRNA was a more efficient decoder than the G33 wild-type tRNA (Figure 3), which was unexpected given the observation that S.cerevisiae transformants expressing the wild-type tRNA (G33) grew slower than those expressing the A33 mutant tRNA (Figure 2). To exclude the possibility that this result was not simply due to plasmid instability, several independent transformants were tested. Although some variability in ,3galactosidase thermoinactivation profiles was detected between transformants, the inactivation trend always remained the same, i.e. expression of the A33 mutant tRNA induced a faster rate of ,-galactosidase thermoinactivation than expression of the G33 mutant (data not shown). The negative impact of expression of Ser-tRNACAG in S.cerevisiae was also evaluated by measuring the relative decrease in total ,B-galactosidase activity in transformed cells (Figure 4A). The A33/G33 transformants displayed a dramatic reduction in total P-galactosidase activity, 5.9 and 5.7 U respectively, in comparison with the pUKC7 18C33 and pRS315 transformants, which had 13 and 16.5 U respectively (Figure 4A). To determine the cause of such a dramatic reduction in 3-galactosidase activity, total protein extracts were prepared from the different transformants and the encoded GST-p-galactosidase was quantified both by Western blotting (Figure 4C) and gel scanning densitometry. Protein extracts prepared from S.cerevisiae transformed with either the wild-type (G33) or A33 mutant tRNAs had decreases in 3-galactosidase levels of 67.5 and 59.2% respectively, relative to the control extract prepared from cells transformed with the pRS3 15 vector (Figure 4B and C). No significant reduction

Serine CUG decoding in C.albicans

A.IfI.-k

1B

-A

1) -4-

.

"Now

'Room

1..

%\

I

()

+

_ ;

_

315

\

4

--

1

Fig. 4. Effect of N,3 mutations of Ser-tRNACAG on -galactosidase S.cer-eiisiae. (A) The variation in total f-gyalactosidase activity at 37°C for S.cerev7isiae co-transformed with pGL-C I and either pRS315. pUKC718-C33, pUKC717-A33 or pUKC715-G33 is shown. (B) Amount of GST-p-galactosidase present in total protein extracts from S.cereviisiae co-transformed with pGL-C1 and either pRS315 (lane 2) pUKC717-A33 (lane 3). pUKC718-C3, (lane 43 or pUKC715-G3 (lane 5). Lane M shows protein molecular weight markers and lane 1 shows purified GST-3-galactosidase. The arrow indicates the position of GST-3-galactosidase. A 10c%c SDS-PAGE stained with commiassie blue is shown. (C) Western blot of the SDSPAGE eel shown in (B). Lanes 1-5 correspond to samples loaded in lanes 1-5 in (B). turnover in

in GST-P-galactosidase level was observed in extracts prepared from cells expressing the C33 mutant tRNA (Figure 4B and C). The lower levels of GST-f3-galactosidase present in the wild-type and pUKC717-A33 extracts (Figure 4B and C) correlate well with the low P3-galactosidase activity detected in those cells, i.e. 65.5 and 64.2% decrease in enzymatic activity respectively, indicating that the primary cause of the reduction in total enzyme activity was most likely degradation of mistranslated 3-galacto-

sidase.

Candida albicans Ser-tRNACAG is not efficiently charged by S.cerevisiae seryl-tRNA synthetase (SerRS) in vivo tRNA decoding efficiency, as measured by f-galactosidase thermal inactivation, represents a combination of the ribosomal binding affinity of the tRNA and its stability, expression level and aminoacylation kinetics. In order to determine if any of these parameters was affected by the N33 mutations, Northern blot analysis was performed tRNAs extracted from the S.cerevisiae transformants fractionated by denaturing and acidic PAGE. The A33/G33 Ser-tRNA were expressed at similar levels, but there was a 98% reduction in the relative levels of the C33 mutant tRNA compared with the A33 mutant tRNA (Figure 5A). To ensure that the apparent decrease in expression levels of the C33 mutant tRNA was not simply due to loading or blotting artefacts, the membranes were reprobed and washed at low stringency in order to detect S.cerevisiae Ser-tRNAIGA, which is 80% identical to Calbicans SertRNACAG (Santos et al., 1993) and hybridizes with the Calbicans [32P]tDNACAG probe at low stringency. Similar levels of cross-hybridizing S.cer-evisie Ser-tRNAIGA were detected in all lanes (Figure SB), indicating that the lower levels of the C33 mutant were either due to decreased transcription by polymerase III or due to an enhanced

on

1

2

>

>

3

4

-

i;.

t

t;

{

Fig. 5. Northern blot analysis of Cailbicans Ser-tRNACAG expressed in S. cerevisice. Total tRNAs extracted from S. cerevisiae were fractionated on a 12.5%, polyacrylamide-8 M urea gel and blotted onto Hybond N membranes. Detection of Ser-tRNACAG was carried out using a [-V-P]tRNACAG probe labelled by PCR. Membranes were washed at either high stringency (A) or low stringency (B). Lanes 1-5 coITespond to tRNAs extracted from S.cerevisicae transformed with either pRS315. pUKC717-A33. pUKC718-C33. pUKC715-G33 and from Calbicains respectively. The small arrow indicates hybridization of the Calbicains [32PItDNACAG probe to Ser-tRNACAG and the bigger arrow indicates hybridization to S.cerevisiace Ser-tRNAIGA. which only hybridizes with the Cailbicans tRNA probe at low stringency.

of turnover of the tRNA. This finding explains the observation that unlike the U33 mutant tRNA, the C33 mutant tRNA is not lethal to S.cerevisiae and suggests a major difference in polymerase III transcription and/or stability between the two mutant tRNAs. Since the N33 base is outside the known polymerase III promoter elements present in tRNAs (Geiduschek and Tocchini-Valentini, 1988), the low levels of the C33 tRNA are most likely due to an enhanced rate of turnover of this mutant tRNA and not decreased efficiency of polymerase III transcription. To test whether the N33 mutant tRNAs were fully charged in viVo in S.cerevisiae, total tRNA samples were extracted under acidic conditions and fractionated in an acidic urea PAGE system to separate charged from uncharged tRNAs (Varshney et al., 1991). Charged and uncharged tRNAs were detected by Northern blot analysis as before. Wild-type Ser-tRNACAG was not fully charged in xix'o, since two bands corresponding to uncharged and charged tRNAs were visible (Figure 6A). Quantification of total radioactivity present in each band using a Phosphorlmager showed that levels of charged and uncharged G33 Ser-tRNACAG were 57.1 and 42.9% respectively (data not shown). A similar result was obtained for the A33 mutant tRNA (Figure 6B). In contrast, the C33 mutant was fully charged (Figure 6C), suggesting that this mutation does not act as a negative determinant for serine charging of the tRNA by the S.cereivisiae SerRS. To ensure that the aminoacylation data was not an artefact of deacylation during extraction or due to aberrant migration of the tRNAs, the Northern blots were reprobed and washed at lower stringency to detect S.cereiisiae SertRNAIGA. This tRNA was fully charged in all transformants (Figure 6D). rate

Expression of C.albicans Ser-tRNACAG in S.cerevisiae induces the stress response The data presented above clearly show that wild-type Ser-tRNACAG was functional in S.cerevisiae and that Scerev,isiace can tolerate low levels of serine CUG translation. If one assumes that the negative impact observed in S.cerev isiae mimics what happened in the CaIdida 5063

M.A.S.Santos, V.M.Perreau and M.FTuite

._ -b 41...

Fig. 7. Quantification of thermotolerance in S.cerevisiae expressing N33 mutants of Calbicans Ser-tRNACAG. (A) The percentage of survival on YEPD plates of S.cerevisiae pRS315 (D), pUKC718-C33 (0), pUKC717-A33 (U) and pUKC715-G33 (A) transformants

Fig. 6. Aminoacylation of the N33 mutants of Calbicans SertRNACAG expressed in S.cerevisiae. Total tRNAs were extracted under acidic conditions from S.cerevisiae transformed with N33 mutants of Ser-tRNACAG. (A-C) Separation of charged (+aa) from deacylated (-aa) Ser-tRNACAG. (D) Separation of charged and deacylated S.cerevisiae Ser-tRNAIGA, which hybridizes with the tDNASerCAG probe at low stringency. Aliquots of 10 tg total tRNA were loaded per lane for tRNAs extracted from pRS315, pUKC717-A33 and pUKC715G33 transformants and 40 .tg/lane were loaded for the pUKC716-C33 mutant. Overnight exposures are shown in (A), (B) and (D) and a 48 h exposure is shown in (C). The symbol - indicates tRNAs extracted from S.cerevisiae transformed with pRS315. Autoradiographs of high stringency washed membranes are shown except in (D), which shows a low stringency washed blot.

ancestor in which the genetic code change first occurred, then these data confirm our prediction that the CUG codon need not have disappeared from Candida genes prior to its reassignment, i.e. the reassignment of the CUG codon might have been driven by selection, as predicted by the Ambiguous Intermediate Theory (Schultz and Yarus, 1994, 1996). If this was so, then what type of selective pressure could have driven this dramatic event? In an attempt to answer this key question, we studied the physiological alterations in S.cerevisiae induced by expression of SertRNACAG in the hope of identifying a positive selection mechanism for the genetic code switch. Antibiotic-induced mistranslation in S. cerevisiae induces the stress response and cells exposed to the drug acquire thermotolerance (Grant et al., 1989). These observations prompted us to investigate if the stress response was also induced in S.cerevisiae transformed with the pUKC7 1 X-N33 plasmids by determining the levels of HsplO4 and Hsp7O (two key indicator proteins of the stress response in S.cerevisiae; Craig, 1992) and quantitatively measuring the acquisition of thermotolerance in these cells. The synthesis of Hsp7O and Hsp 104 was significantly increased in the transformants compared with the control cells (Figure 7B and C) and such cells were significantly more thermotolerant than control cells transformed with pRS315 alone (Figure 7A). Remarkably, cells expressing the A33 mutant tRNA showed a 28-fold increase in tolerance after 25 min at 50°C, while cells expressing the wild-type (G33) tRNA and the C33 mutant tRNA showed

5064

incubated for different periods of time at 50°C. (B and C) Overexpression of HsplO4 and Hsp7O in S.cerevisiae transformed with N33 mutants of Ser-tRNACAG respectively. Lanes 1-4, total S.cerevisiae proteins extracted from pRS315, pUKC717-A33, pUKC718-C33 and pUKC715-G33 transformants respectively and probed with either an anti-HsplO4 antibody (B) or an anti-Hsp7O

antibody (C).

6- and 7-fold increases in tolerance in relation to control cells transformed with the pRS3 15 plasmid. No correlation between HsplO4/Hsp7O overexpression levels and thermotolerance levels was found, indicating that thermotolerance is not primarily due to induction of these two Hsps alone, although they must play a contributory role. This suggests that other Hsps are overexpressed and therefore it will be interesting to investigate how many other Hsps are induced

by misreading.

Discussion Reassignment of the standard leucine CUG codon to serine in several Candida species occurred -150 million years ago (Pesole et al., 1995) and therefore this genetic code change, like all others, is a relatively recent event which evolved from the standard genetic code. Considering the difference in biochemical properties between leucine (hydrophobic) and serine (polar), one would expect that, unless a molecular mechanism that precludes replacement of leucines by serines at CUG sites is in place, reassignment of the CUG codon should be detrimental to protein structure and hence function, as predicted by the Frozen Accident (Crick, 1968) and the Codon Capture Theories (Osawa et al., 1992). However, data presented here unequivocally show that expression of Calbicans Ser-tRNAcAG in a closely related yeast species (S.cerevisiae) was compatible with viability and that S.cerevisiae minimizes the deleterious effect of serine CUG translation by inducing the stress response and efficiently degrading mistranslated proteins. This raises the possibility that reassignment of the CUG codon in Candida evolved via a molecular mechanism that did not require elimination of CUG codons prior to their reassignment, as predicted by the Ambiguous Intermediate Theory (Schultz and Yarus, 1994, 1996). Rather, a mechanism which lowered the efficiency with

Serine CUG decoding in Calbicans

which the CUG codon was translated could provide an alternative evolutionary pathway for this genetic code change. Such a mechanism must involve a single mutation (U33-G33) in one of the most conserved nucleotides in tRNA and this mutation would therefore have played a key role in the development of the genetic code switch in Calbicans and other Canldida species That the reassignment of the CUG codon in Canididca species could have evolved from inefficient misreading of the leucine CUG codon as serine is further supported by studies on missense suppression carried out in E. coli (reviewed by Murgola, 1985). Among all missense suppressor tRNAs isolated from E.coli, slapH (Thorbjarnardottir et al. 1985) is of particular importance for understanding the evolution of the genetic code switch in Canldida. because it is a minor serine tRNA derived from wild-type Ser-tRNACAG by a single mutation in the anticodon which creates Ser-tRNACAA. As in the case of serine CUG decoding in S.cerevisiae by Calbicans Ser-tRNACAG, decoding of the leucine UUG codon as serine in E.coli by supH also has a negative impact on growth rate, but more importantly in the context of selection-driven genetic code changes, supH suppresses ilvD and lac mutations. which cause auxotrophy for isoleucine/valine and lactose respectively (Eggertsson. 1968). indicating that under certain physiological conditions it is not only beneficial but critical for survival, i.e. under specific growth conditions ambiguous decoding of the UUG codon allows cells that encode supH to survive, while ilvD and Itc cells that do not encode slupH are eliminated due to their inability to overcome the block in amino acid synthesis or lactose transport pathways. That E.coli glycine missense suppressor tRNAs could be isolated (Murgola, 1985) on the basis of their ability to suppress mutations in trpA and that S.cereiisiae missense suppressors are known to suppress his2-1. trp5-18, trp5-67 and ilvi -83 mutations (reviewed by Sherman. 1982) broadens the spectrum of mutations which can be reverted and indicates that ambiguous decoding of certain codons can provide selective advantage to cells that carry missense suppressors.

G33 as a key structural element in the reassignment of the CUG codon The tRNA phosphate backbone changes direction at position 33 in the anticodon loop with a sharp U-turn (Ladner et al., 1975; Woo et at.. 1980) and conservation of U33 in more than 99% of tRNAs so far sequenced suggests that it must be functionally important. U-turns are highly conserved and ubiquitous RNA structural motifs and are found not only in the anticodon loop of tRNAs, but also in the TVC loop and in other RNAs such as hammerhead ribozymes (Jucker and Pardi, 1995). In the anticodon loop,

the U-turn exposes the anticodon bases to the solvent region in a stacked fashion, allowing for the codonanticodon interaction (Ladner et al., 1975). Thus mutation of U33 to any other base must affect tRNA function. In litro experiments have shown that replacement of U33 with A or G dramatically decreases decoding efficiency of both yeast Phe-tRNA, by increasing the rate of aminoacyltRNA rejection from the ribosomal A site (Dix et al.. 1986), and the efficiency of UAG suppression by a mutant of yeast Try-tRNA~.GwA (Bare et al.. 1983). Similar results were obtained in vivo in E.cali using the glutamine

suppressor tRNA si7. which is able to translate the UAG stop codon (Curran and Yarus, 1987). Data obtained with C.albicants Ser-tRNACAG are consistent with these early studies, since they show that mutations at position 33 in the anticodon loop of Ser-tRNACAG affect its decoding efficiency, as measured by the f3-galactosidase thermoinactivation mistranslation assay. This is also reflected in the growth inhibition observed for the N33 mutants. However. our data also show that decreased decoding efficiencies are not simply the result of altered ribosomal binding constants, but may also be due to altered aminoacylation kinetics and/or tRNA stability, which are affected at different levels by different mutations. Of central importance to the development of a model to explain how the CUG codon was reassigned from leucine to serine is the fact that expression of the U33 mutant in S.cerevisiae is lethal, i.e. if our experimental model reliably mimics the early stages of the evolution of this genetic code change. then the data clearly show that reassignment of the CUG codon could not have occurred with a canonical tRNA with the conserved U at position 33, because it would have been lethal. This is further confirmed by the converse experiment in which canonical Leu-tRNA isoacceptors which translate the CUG codon as leucine could not be expressed in C.albicans (Leuker and Ernst, 1994). Thus, efficient mistranslation of the CUG codon is lethal to both S.cer-evisiace and C.albicans. This could be interpreted as providing strong evidence for a neutral mechanism for the evolution of this genetic code switch involving elimination of CUG codons prior to reassignment. However, if this genetic code change evolved through a neutral mechanism which did not introduce mutations into proteins, then there is no obvious reason for the unusual structure of Ser-tRNACAG and instead a canonical tRNA with U33 would be expected to decode the CUG codon as serine. tRNAs with unusual structures are known to decode the standard leucine CUN codon family as threonine in S. cerevisiae and Toruidlopsis glabra1ta mitochondrial translation systems (Osawa et al.. 1990). These mitochondrial tRNAs also have unconventional anticodon loops with 8 or 9 nucleotides (nt) instead of the canonical 7 nt (Li and Tzagoloff, 1979) and tRNAs with 8-9 nt anticodon loops are known to frameshift in E.coli (Tuohy et al., 1992). Therefore. novel tRNA decoding properties arising from alterations in tRNA structure may be key to the evolution of alternative genetic codes. We have previously shown that C.albicans Ser-tRNACAG is able not only to decode the standard leucine CUG codon as serine, but might also be able to translate stop codons (Santos et al., 1993). Therefore. it will be interesting to investigate if G33 is required for other such non-standard translation events and this may explain why G33 was preserved during evolution of Ser-tRNACAG.

Mistranslation of the CUG codon as serine induces the stress response Perhaps one of the most intriguing aspects of CUG mistranslation in S.cerevisiace is induction of thermotolerance with little or no effect on cell viability. That microorganisms are able to tolerate significant levels of mistranslation is well documented (Edelmann and Gallant, 1977: Ellis and Gallant. 1982). For example. E.coli can

5065

M.A.S.Santos, V.M.Perreau and M.FTuite

tolerate 6- to 10-fold increases in the basal translational error rate (4 X 1 04 errors/codon) in the presence of 2 gg/ ml streptomycin, a concentration which causes only 15% inhibition of growth (Edelmann and Gallant, 1977). That the stress response in yeast can be induced not only by environmental factors (Craig, 1992) but also by mistranslation (Grant et al., 1989) is well established. Therefore, it is likely that the sensor(s) which detects protein misfolding, aggregation and/or unfolding upon exposure to a higher than optimal growth temperature, or other environmental factors, may also detect misfolding or aggregation of proteins caused by mistranslation. Interestingly, induction of the stress response by exposure to high temperature or mistranslation seems to have similar effects in protecting S.cerevisiae from otherwise lethal temperatures (Grant et al., 1989). Indeed S.cerevisiae cells grown at 25°C but pre-incubated at 37°C induce synthesis of a number of different Hsps and become much more resistant to sudden exposure to high temperatures (45-50°C) than do control cells which are not pre-incubated at 37°C (Craig, 1992). This exactly parallels what happens in S.cerevisiae upon expression of Calbicans Ser-tRNACAG. In both cases, S.cerevisiae overexpresses Hsp7O, which, among other functions, protects the mRNA splicing machinery from inactivation (Yost and Lindquist, 1991), and HsplO4, which helps cells to recover from damage induced by sudden severe stresses (Sanchez and Lindquist, 1990). However, when S.cerevisiae is exposed to supra-optimal growth temperatures it not only overexpresses those two Hsps, but a number of other Hsps (Craig, 1992). It will be interesting to monitor how many of these Hsps are also overexpressed or induced in S.cerevisiae in the presence of Ser-tRNACAG. In the context of reassignment of the CUG codon from leucine to serine in Candida species, the protective effect that arises from induction of the stress response most likely played a critical role in minimizing the deleterious effect of serine CUG mistranslation. Acquisition of thermotolerance could have allowed colonization of new ecological niches which exist at higher than ambient temperatures. Furthermore, acquisition of thermotolerance by S.cerevisiae expressing Ser-tRNACAG also raises the question of knowing if all Candida species which translate CUG as serine (of which there are at least 10 identified to date) have a constitutive stress response. Intriguingly, the optimal growth temperature for Calbicans is 37°C and not 25-30°C, as is the case for most other yeasts, and a range of Hsps appear to be constitutively expressed in C.albicans (Maresca and Carratu, 1992). C.albicans is an opportunistic pathogen whose pathogenicity appears not to be caused by a toxin, but rather by its ability to colonize mammalian tissues (Cutler, 1991), suggesting that the heat-protective effect observed after induction of the stress response could have been important during the development of Candida pathogenesis by allowing Candida to survive the increase in temperature at the onset of infection, as appears to be the case in other fungi, such as Histoplasma capsulatum (Caruso et al., 1987).

Evolution of genetic code changes If one considers that codons do not have to disappear prior to their reassignment then the Ambiguous Intermediate Theory (Schultz and Yarus, 1994, 1996) provides a good

5066

model to rationalize almost all the alternative genetic code changes found to date. The data we present here add weight to this theory, since we have proven that expression of Calbicans Ser-tRNACAG in S.cerevisiae is not only compatible with viability, but has little or neutral effect on cell viability. However, serine CUG translation in Candida and threonine CUN translation in S.cerevisiae and Tglabrata mitochondria (Ho and Kan, 1987; Osawa et al., 1990) do not involve capture of codons by nearcognate anticodons. Instead, at least in the case of C.albicans, an alternative pathway based on low decoding efficiency of the CUG codon was probably the critical factor in the development of this genetic code change. That Ser-tRNACAG is not fully charged in S.cerevisiae would suggest that the Candida SerRS might also have played an important role in the evolution of this codon reassignment. Interestingly, in yeast mitochondria ThrtRNAUAG (which decodes the CUN codon family) is aminoacylated by a unique ThrRS which is unable to charge canonical Thr-tRNA (Pape et al., 1985). Both SertRNACAG and Thr-tRNAAUG have non-canonical anticodons (Li and Tzagoloff, 1979; Santos et al., 1993) and most likely non-standard decoding properties (Santos et al., 1990), indicating that these unusual anticodons could also have played important roles in the reassignment of those codons. Furthermore, the requirement for additional non-standard decoding properties mediated by G33 could explain why G33 has been preserved during evolution of C.albicans Ser-tRNACAG. Overall, our data suggests that different genetic code changes evolved via different molecular pathways and that no single unifying mechanism is responsible for the evolution of them all, most likely because evolution of different genetic code changes was driven by different selective pressures.

Materials and methods Chemicals were purchased from Sigma, BDH or Fisons. DNA restriction and modification enzymes were purchased from Boehringer Mannheim, Promega or Gibco BRL. All media were purchased from Difco. Oligonucleotides were from Oswell DNA Service (UK). DNA purification and extraction kits were from Qiagen. Secondary antibodies for Western blotting were from Dako (Denmark).

Strains and growth conditions Escherichia coli strain JM109 (recAl SupE44 endAl hsdRJ7 gyrA96 relAI thi A(Lac-proAB) F'[traD36 proAB+ lacI lacZ AM15]) was used as a host for all DNA manipulations. Saccharomyces cerevisiae AS24 (leu2-3, leu2-112, his3-11, his3-15, ura3-251, ura3-373, trpl) was grown at 30°C in YEPD (2% glucose, 1% yeast extract, 1% peptone) or minimal medium lacking the required amino acid (0.67% yeast nitrogen base without amino acids, 2% glucose, 2% agar and 100 tg/ml of each of the required amino acids).

Plasmid transformation Transformation of E.coli was carried out by electroporation using a BioRad Gene Pulser according to the manufacturer's instructions. Transformation of S.cerevisiae AS24 was carried out using the lithium acetate method (Gietz and Woods, 1994). Plasmids For expression of Calbicans Ser-tRNACAG in S.cerevisiae, a SpeI-RsaI genomic DNA fragment containing the Ser-tRNAcAG gene (250 bp), previously isolated from a genomic library prepared from Calbicans strain 2005E (Santos and Tuite, 1995), was cloned into the single copy vector pRS315 at the SpeI/SmaI sites (Sikorski and Hieter, 1989). The resulting plasmid was named pUKC701-G33. Mutants of Ser-tRNACAG were synthesized using an oligonucleotide with degeneracy at position

Serine CUG decoding in 33. This oligonucleotide was used as a template for PCR amplification with Taq DNA polymerase. DNA was amplified in a GeneAmp PCR system 2400 from Perkin Elmer using the following PCR protocol: 65°C. 30 s annealing, 72°C, 30 s extension and 94°C, 30 s denaturation. Primers for PCR were tRNACAG5'/BanmHI (5'-AATTGGATCCGATA CGATGGCCGAGTGG-3') and tRNACAG3'polII/HindIII (5'-AAACCAAGCTTGAAAAAAAATAACGACACGAGCAGGGTTC-3'). BatnHI and HioidIll sites were incorporated in the primers at the 5'- and 3'-ends respectively to allow for cloning into the pRS315 vector at the BalmlHI and HioidIll sites. The 3' primer also contained the natural C.albicanls polymerase III terminator sequence, A8TA,. to allow' for correct transcription termination. Amplified DNA was digested with BaniHI and HiidIlll and cloned into pRS315 and the resulting DNA was then transformed into E.coli JM109. Individual clones were then isolated and the plasmids purified and sequenced to screen for N33 mutants. A/G/C/T33 tRNA genes were obtained. The resulting clones were named pUKC715-G33, pUKC716-T33, pUKC7 17-A33 and pUKC7 18-C33. f-Galactosidase was expressed in S.cerevisiae under the control of the GPD promoter as a fusion with glutathione S-transferase (GST-fgal) from the multicopy vector pGL-C 1.

/3-Galactosidase thermoinactivation f-Galactosidase thermal inactivation was also monitored directly in S.cerev'isiae cells. For this, 200 or 500 p1 exponentially growing S.cerevisiae (OD600 1-1.5) transformed with pGL-Cl/pRS315, pGL-C1/ pUKC7 18-C33, pGL-C l/pUKC7 17-A33 or pGL-C l/pUKC7 1 5-G33 were harvested, washed and then resuspended in 800 VI Z buffer (60 mM Na,HPO4. 40 mM NaH1PO4.2H0, 10 mM KCI. 1 mM MgSO47H0. 50 mM 2-mercaptoethanol. pH 7.0). 20 pi 0.1%57 SDS, 50 mM 2-

mercaptoethanol and 50 p1 chloroform. Cells suspensions were vortexed for 30 s and incubated in triplicate at 47°C in a water bath for different periods of time. Permeabilized cells were put on ice for 30 min. then transferred to 37°C for 5 min and 200 p1 4 mg/ml o-nitrophenyl-f-Dgalactopyranoside (ONPG) were added to each tube. Reactions were allowed to proceed for 2 min and were stopped by addition of 400 p1 1 M Na1CO3. f-Galactosidase activity was determined by monitoring synthesis of o-nitrophenol at 420 nm (Branscomb and Galas. 1975: Finkelstein and Strausberg. 1983).

Northern blot analysis For total tRNA extractions, 200 ml cultures of S.cerevisiae grown overniaht in minimal medium were harvested when the OD600 reached 1.2-1.5 and cell pellets were washed with ice-cold sterile H,O. Pellets were frozen at -70°C overnight and then resuspended in 10 ml lysis buffer (0.3 M sodium citrate, pH 4.5. 10 mM EDTA). I vol phenol equilibrated with 0.1 M sodium citrate, pH 4.5. and baked glass beads were added and the cell suspension was vrigorously shaken for I h at 4°C (Durasevic et al., 1994). The aqueous phase containing RNAs was separated from the phenolic phase by centrifugation at 3000 g for 15 min at 4°C and then transferred to a new Falcon tube and re-extracted with fresh phenol for 10 min. Extracted tRNAs were collected from the aqueous phase and precipitated overnight at -20°C with 2 vol ethanol. tRNAs were harvested by centrifugation at 3000 g for 30 min at 4°C, resuspended in 10 ml 0.1 M sodium acetate, pH 4.5. and applied to a 20 ml DEAE-cellulose (Whatmann) column equilibrated with the same buffer. tRNAs were eluted with 0.1 M sodium acetate containing I M sodium chloride, precipitated wvith ethanol, resuspended in 10 mM sodium acetate, pH 5.0, and stored at -70°C. For deacylation, tRNAs were resuspended in I M Tris-HCl. pH 8.0, 10 mM EDTA and incubated at 37°C for 1 h (Santos et a/., 1993). Deacylated tRNAs were precipitated with ethanol and 0.3 M sodium acetate, pH 4.5, resuspended in 10 mM sodium acetate, pH 5.0, and stored at -70°C. tRNAs were fractionated at room temperature in 15%7c acrylamide-8 M urea gels or at 4°C in 7.5% acrylamide-8 M urea gels (30 cm long. 0.5 mm thick). buffered with TBE. pH 8.0. or 0.1 M sodium acetate. pH 5.0, respectively. The 15%7 gels w\ere run at 35 mA until the xylene cyanol dye reached the bottom and the 7.5% acidic gels were run at 300 V until the bromophenol blue dye was 5 cm from the bottom (Varshney et al., 1991). Fractionated tRNAs were transferred to nitrocellulose membranes (Hybond N: Amersham International), according to described methods (Varshney et al.. 1991). Membranes were pre-hybridized for 8 h in a Techne Hybridization oven at 45°C in 50% v/v formamide. SX SSC, 1% SDS, 0.04% Ficoll. 0.04% polyvinylpyrrolidone and 250 pg/ ml sheared salmon sperm DNA (Heitzler et al.. 1992). Hybridization was performed overnight in the above buffer using a Ser-tRNAcAG probe labelled with [u.-32P]dATP by PCR using standard protocols (Innis et al.. 1990) except that the amount of dATP was reduced from 100 to

C.albicans

50 mM and 5 nmol (30 pCi) 6000 Ci/mmol [ux-32P]dATP was added to the reaction mixture. In order to minimize the backaround level of free radioactivity, 50 PCR cycles were performed to decrease to a minimum the amount of non-incorporated [ux-32P]dATP. Membranes were washed at low stringency in Ix SSC. 0.1%7 SDS at 50°C or at high stringency in 0.lx SSC. 0.1% SDS at 65°C for I h.

Monitoring of thermotolerance

SDS-PAGE was performed according to standard protocols (Laemmli, 1970). Western blotting onto nitrocellulose was perform according to standard protocols (Towbin et al., 1979: Harlow and Lane, 1988). Western blots were probed with either an anti-HsplO4 polyclonal rabbit antibody used at 1:5000 dilution or a polyclonal rat anti-Hsp7O antibody used at 1:1000 dilution. Both antibodies were obtained from Prof. Susan Lindquist. The anti-Hsp7O antibody is a broadly cross-reacting antibody. Bound antibody was detected using the Amersham ECL system according to the manufacturer's instructions. Quantitation of thermotolerance was carried out as described (Grant et al., 1989) with the following alterations. Saccharomnyces cerevisiae transformed with the pRS315 vector alone or carrying N33 mutants were growin at 25°C to exponential phase (OD600 0.8) in minimal medium lacking leucine. Cells were diluted in fresh minimal medium to a density of 10 000 cells/ml and were incubated in a water bath at 500C. At 5 min intervals. 0.1 ml aliquots of cells (1000 cells) were removed and plated in triplicate onto YEPD/agar. Viable counts were determined after 3 days for cells transformed with pRS315 and pUKC718-C33 and after 5 days for cells transformed with pUKC715-G33 and pUKC717-A33. Percentage survival was calculated by comparing the results obtained with those from samples grown in parallel at 30°C prior to incubation at 500C.

Acknowledgements We thank Dr Gerard Keith and Prof. Guy Dirheimer from the IBMCCNRS in Strasbourg and Prof. Michael Yarus from the University of Colorado for useful discussions. Plasmid pGL-Cl was a kind gift of Norma Wills from the HHMI, Univeristy of Utah (Salt Lake City, UT). Anti-Hsp7O and anti-HspIO4 antibodies were a kind gift of Prof. Susan Lindquist from the HHMI, University of Chicago (Chicago, MI). This work was supported by a Wellcome Trust grant (M/91/2902) (M.F.T.), a Career Development Fellowship (045548/Z/95/Z/JRS/SH) (M.A.S.S.) and a BBSRC Ph.D. studentship (V.P.).

References BranscombE.W. and Galas.D.J. (1975) Progressive decrease

synthesis accuracy induced by streptomycin Nature. 254, 161-163.

in

in

protein

Eschler-ichia coli.

Uhlenbeck,O.C. (1983) Uridine33 in yeast tRNA not essential for amber suppression. Nature. 305. 554-556. Caruso,M., Sacco,M., Medoff.G. and Maresca.B. (1987) Heat shock 70 gene is differentially expressed in Histoplasoiia cajpsulatumn strains with different levels of thermotolerance and pathogenicity. Mol. Micmy-biol., 1. 151-158. Craig,E.A. (1992) The heat-shock response of Saccha o-ny ces cerevisiae. In Jones,E.W., Pringle,J.R. and Broach,J.R. (eds), The Molecu/lar and Cellil/ar- Biology of the Yeast Sacchlar-oorrces. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 501-537. Crick.F.H. (1968) The origin of the genetic code. J. Mol. Biol., 38, 367-379. Curran.J.F. and Yarus,M. (1986) Base substitutions in the tRNA anticodon arm do not degrade the accuracy of reading frame maintenance. Proc. Ncatl Acad. Sci. USA. 83, 6538-6542. Curran.J.F. and Yarus,M. (1987) Reading frame selection and transfer RNA anticodon loop stacking. Scienice, 238. 1545-1550. albicanis. Annloi. Cutler,J.E. (1991) Putative virulence factors of Rei: Micr-obiol.. 45, 187-218. Dix.D.B., Wittenberg.W.L., Uhlenbeck.O.C. and Thompson,R.C. (1986) Effect of replacing uridine 33 in yeast tRNAPhe on the reaction with ribosomes. J. Biol. Cliemii.. 261, 10112-10118. of DurasevicI.W.. Nalaskowska.M. and SolID. (1994) Coexpression eukarvotic tRNASer and yeast seryl-tRNA synthetase leads to functional amber suppression in Esclier-ichlia coli. J. Bacteriol.. 176, 232-239.

Bare,L., Bruce.A.G., Gesteland,R. and

Canldida

5067

M.A.S.Santos, V.M.Perreau and M.F.Tuite

Edelmann,P. and Gallant.J. (1977) Mistranslation in E.coli. Cell, 10, 131-137.

Eggertsson,G. (1968) Suppressors causing temperature sensitivity of growth in Escher-ichiai coli. Genetics, 60, 269-280. Ellis,N. and Gallant.J. (1982) An estimate of the global error frequency in translation. Mol. Geti. GenZet.. 188, 169-172. Finkelstein,D.B. and Strausberg,S. (1983) Heat shock-regulated production of Escher ichia coli f-galactosidase in Saccharonyzces cerevisiae. Mol. Cell. Biol., 3, 1625-1633.

Geiduschek.E.P. and Tocchini-Valentini,G.P. (1988) Transcription by RNA polymerase III. Annuiii. Rev. Biochein., 57, 873-914. Gietz,R.D. and Woods,R.A. (1994) High efficiency transformation with lithium acetate. In Johnston,J.R. (ed.). Moleclulac- Genietic-s o° Yeast: A Practical App-roach. IRL Press at Oxford University Press. Oxford, UK, pp. 121-134. Grant,C.M., Firoozan,M. and Tuite,M.F. (1989) Mistranslation induces the heat-shock response in the yeast Saccharoinyces cerevisiae. Mol. Microbiol., 3, 215-220. Harlow,E. and Lane,D. (eds), (1988) Anitibodies: A Laborator-y Malnu1al. Cold Spring Harbor Laboratory Press. Cold Spring Harbor, NY. Heitzler,J., Drouard,L.M.. Dirheimer,G. and Keith,G. (1992) Use of a dot blot hybridization method for identification of pure tRNA species on different membranes. Biochibni. Biophv.s. Actca, 1129, 273-277. Ho,Y. and Kan,Y.W. (1987) In vivo aminoacylation of human and Xenoplus suppressor tRNAs constructed by site-specific mutagenesis. Proc. Ncatl Acadi. Sci. USA, 84, 2185-2188. Innis,M.A., Gelfand,D.H., Sninsky.J.J. and White.T.J. (1990) In Innis,M.A., Gelfand,D.H., Sninsky,J.J. and White,T.J. (eds), PCR pr-otocols: A GCui(le to Metlhods aniid Applications. Academic Press, San Diego, CA. Jucker,F.M. and Pardi,A. (1995) GNRA tetraloops make a U-turn. RNA, 1, 219-222. Jukes,T.H. and Osawa,S. (1993) Evolutionary changes in the genetic code. Coinp. Biochein. Physiol., 106B, 489-494. Kawaguchi,Y., Honda.H., Morimura.T.J. and Iwasaki,S. (1989) The codon CUG is read as serine in an asporogenic yeast Candida cv/indracea. Natur-e, 341, 164-166. Ladner,J.E., Jack,A., Robertus,J.D., Brown,R.S., Rhodes,D., Clark,B.F.C. and Klug,A. (1975) Structure of yeast phenylalanine transfer RNA at 2.5 A resolution. Pr-oc. Natl Acad. Sci. USA. 72, 4414-4418. Laemmli,U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. NMatre. 227, 680-685. Leuker,C.E. and Ernst,J.F. (1994) Toxicity of a heterologous leucyltRNA (anticodon CAG) in the pathogen Candida albicans: in vivo evidence for non-standard decoding of CUG codons. Mol. Ceii. Ce,vet., 245, 212-217. Li,M. and Tzagoloff,A. (1979) Assembly of the mitochondrial membrane system: sequence of yeast mitochondrial valine and an unusual threonine tRNA gene. Cell, 18, 47-53. Lloyd,A.T. and Sharp,P.M. (1992) Evolution of codon usage patterns: the extent and nature of divergence between Cacndidlc albicans and Saccharoninvces cer-evisiae. Nucleic Acids Res., 20, 5289-5295. Maresca,B. and Calratu,L. (1992) The biology of the heat shock response in parasites. Parcasitol. Today, 8, 260-266. Murgola,E. (1985) tRNA suppression, and the code. Annui. Rev. Geniet., 19, 57-80. OhamaT., Suzuki,T.. Mori,M., Osawa,S., Ueda,T., Watanabe.K. and Nakase,T. (1993) Non-universal decoding of the leucine codon CUG in several Cam/didati species. Nucleic Acidis Res., 21, 4039-4045. Osawa,S., Collins.D.. Ohama,T., Jukes,T.H. and Watanabe,K. (1990) Evolution of the mitochondrial genetic code. III. Reassignment of CUN codons from leucine to threonine during evolution of yeast mitochondria. J. Mol. Eva!., 30, 322-328. Osawa,S., Jukes,T.H., Watanabe,K. and Muto,A. (1992) Recent evidence for evolution of the genetic code. Micr-obial. Rev.. 5, 229-264. Pape,L.K., Koerner,T.J. and Tzagoloff,A. (1985) Characterization of a yeast nuclear gene (MSTI) coding for the mitochondrial threonyltRNA synthetase. J. Biol. Chele., 260. 15362-15370. Pesole,G., Lotti,M., Alberghina,L. and Saccone,C. (1995) Evolutionary origin of non-universal CUGS,r codon in some Canldidal species as inferred from molecular phylogeny. Genietics, 141, 903-907. Sanchez,Y. and Lindquist,S.L. (1990) HSPI04 required for induced thermotolerance. Science, 248, 1112-1115. Santos,M.A.S. and Tuite.M.F. (1995) The CUG codon is decoded in vivo as serine and not leucine in Ciic/idaiCI albicans. Nucleic Aci(s Res.. 23, 1481-1486. Santos,M.A.S., Colthurst,D.R., Wills,N., McLaughlin,C.S. and

5068

Tuite,M.F. (1990) Efficient translation of the UAG termination codon in Candida species. Clure Cen7et., 17, 487-491. Santos,M.A.S., Keith.G. and Tuite.M.F. (1993) Non-standard translational events in Camc/licda albicanis mediated by an unusual seryltRNA with a 5'-CAG-3' (leucine) anticodon. EMBO J., 12, 607-616. Schultz,D.W. and Yarus,M. (1994) Transfer RNA mutation and the malleability of the genetic code. J. Mol. Evol., 235, 1377-1380. Schultz,D.W. and Yarus,M. (1996) On malleability in the genetic code. J. Mol. Evol., 42. 597-601. Sherman,F. (1982) Suppression in the yeast Saccharom-yvces cerevi.aiae. In Strathern,J., Jones,E. and Broach,J. (eds), The Molecutlar- Biology of the Yeast Saccharoinvces: Metabolismn1 andcl Genie Expression. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp.463-486. Sikorski.R.S. and Hieter,P. (1989) A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Sacc/haroinwces cerev.isiae. Cenetics, 122. 19-27. Steinberg,S., Misch,A. and Sprinzl,M. (1993) Compilation of tRNA sequences and sequences of tRNA genes. Nuicleic Acids Res., 21, 3011-3015. Sugiyama,H., Ohkuma,M.. Masuda,Y.. Park.S.M., Ohta,A. and Takagi,M. (1995) In vivo evidence for non-universal usage of the codon CUG in Canldidai inaltosa. Yeast, 11, 43-52. Thorbjarnardottir,S., Uemura,H., Dingermann,T., Rafnar,T., Thorsteinnsdottir,S., Soll,D. and Eggertsson,G. (1985) Escherichia coli supH suppressor: temperature-sensitive missense suppression caused by an anticodon change in tRNASC',. J. Bacte-iol., 161, 207-211. Tourancheau.A.B., Tsao,N., Klobutcher,L.A., Pearlman,R.E. and Adoutte,A. (1995) Genetic code deviations in the ciliates: evidence for multiple and independent events. EMBO J., 14, 3262-3267. Towbin,H., Staehelin.T. and Gordon,J. (1979) Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. P,oc. Natl Acadbl. Sci. USA, 76, 4350-4354. Tuohy,T.M.F., Thompson,S., Gesteland.R.F. and Atkins,J.F. (1992) Seven, eight and nine-membered anticodon loop mutants of tRNAA(, which cause + 1 frameshifting. J. Mol. Biol., 228, 1042-1054. Ueda.T., Suzuki,T., Yokogawa,T._ Nishikawa,K. and Watanabe,K. (1994) Unique structure of new serine tRNAs responsible for decoding leucine codon CUG in various Cam/ida species and their putative ancestral tRNA genes. Biochinfie, 76, 1217-1222. Varshney,U., Lee,C.P. and RajBhandary,U.L. (1991) Direct analysis of aminoacylation levels of tRNAs in vOivml. J. Biol. Clveni., 266, 24712-24718. Woo,N.H., Roe,B.A. and Rich,A. (1980) Three-dimensional structure of Eschverichia coli initiator tRNAMetf. Nature, 24, 346-351. Yost,H.J. and Lindquist,S. (1991) Heat shock proteins affect RNA processing during the heat shock response of Sacc/haroinyces cerevi.siae. Mol. Cell. Biol., 11, 1062-1068.

Received on7 Apr-il 30, 1996; revised oni Jlun7e 12, 1996