Additive, cooperative and anti-cooperative effects

The EMBO Joumal vol.12 no.7 pp.2949-2957, 1993

Additive, cooperative and anti-cooperative effects between identity nucleotides of a tRNA

Joern Putz, Joseph D.Puglisi1, Catherine Florentz and Richard Gieg62 Unite 'Structure des Macromolecules Biologiques et Mecanismes de Reconnaissance', Institut de Biologie Moleculaire et Cellulaire du Centre National de la Recherche Scientifique, 15 rue Rend Descartes, F-67084 Strasbourg Cedex, France lPresent address: Massachusetts Institute of Technology, Department of Chemistry, 77 Massachusetts Avenue, Cambridge, MA 02139-4307, USA 2Corresponding author Communicated by A.Fersht

We have invetigated the functional relationship between nucleotides in yeast tRNAAsP that are important for aspartylation by yeast aspartyl-tRNA synthetase. Transcripts of tRNAAsP with two or more mutations at identity positions G73, G34, U35, C36 and base pair G10-U25 have been prepared and the steady-state kinetics of their aspartylation were measured. Multiple mutations affect the catalytic activities of the synthetase mainly at the level of the catalytic constant, k,.t. Kinetic data were expressed as free energy variation at transition state of these multiple mutants and comparison of experimental values with those calculated from results on single mutants defined three types of relationships between the identity nucleotides of this tRNA. Nucleotides located far apart in the three-dimensional structure of the tRNA act cooperatively whereas nucleotides of the anticodon triplet act either additively or anti-cooperatively. These results are related to the specific interactions of functional groups on identity nucleotides with amino acids in the protein as revealed by the crystal structure of the tRNAASP/aspartyl-tRNA synthetase complex. These relationships between identity nucleotides may play an important role in the biological function of tRNAs. Key words: aminoacylation/aspartyl-tRNA synthetase/ functional effects of mutations/in vitro tRNA transcripts/ tRNAAsP identity

Introduction The exact aminoacylation of tRNAs by their cognate aminoacyl-tRNA synthetases (aaRSs) is essential to the accuracy of gene expression and is controlled by positive and negative discrimination mechanisms. Positive discrimination leads to aminoacylation of a tRNA by its productive interaction with the cognate synthetase and is determined by a limited number of nucleotides (reviewed by Schimmel, 1989; Normanly and Abelson, 1989; Lapointe and Giege, 1991; Schulman, 1991; Sol1, 1991; Pallanck and Schulman, 1992; Giege et al., 1993). Negative discrimination mechanisms, involving structural features and modified nucleotides, prevent a tRNA from being recognized and aminoacylated by non-cognate synthetases (Muramatsu

et al., 1988; Perret et al., 1990a). Although progress has been made in determining the nucleotides functionally important for specific aminoacylation for various tRNAs, little is known about the molecular mechanisms by which specificity is achieved. Identity nucleotides are either scattered throughout the three-dimensional structure of tRNA or concentrated within the anticodon loop and/or the acceptor stem. It is believed that they usually act by directly contacting critical amino acids on the synthetase, but also by tuning the optimal presentation of these contact nucleotides to the synthetase. Recent high-resolution X-ray structures of the complex between either tRNAG0n and GlnRS (Rould et al., 1989) or tRNAAsP and AspRS (Ruff et al., 1991; Cavarelli et al., 1993) have confirmed that almost all identity nucleotides are in close interaction with the synthetase. Chemical footprinting performed on complexes between synthetase and either wild-type tRNAAsP transcripts or transcripts mutated at the level of identity nucleotides, confirm this view (Rudinger et al., 1992). More precisely, chemical groups on identity nucleotides contact synthetases (Rould et al., 1991; Cavarelli et al., 1993) and mediate the specificity and efficiency of aminoacylation (Musier-Forsyth and Schimmel, 1992). The effect of mutations at a given identity position suggests the functional role of these nucleotides (reviewed in Giege et al., 1993). For tRNA aminoacylation systems studied to date, mutations at identity positions mainly affect the catalytic rate constant (km,a), which can be decreased as much as five orders of magnitude (e.g. Schulman and Pelka, 1988). In general, the Michaelis constant (Km.) is much less affected and exhibits increases of only up to 60-fold, in agreement with the view that specificity of tRNA aminoacylation is governed more by kinetic effects than by the affinity of tRNA for synthetases (Ebel et al., 1973). In a few cases, mutation at identity positions affects predominantly Km (e.g. Yarus et al., 1977; Puitz et al., 1991). Despite these advances, it remains unclear how specificity of aminoacylation is achieved by the ensemble of the identity elements of a given tRNA. The strong effect of mutations of anticodon nucleotides on the catalytic rate of aminoacylation suggests a long range interaction between the anticodon and active site. However, the pathways within the tRNA and aminoacyltRNA synthetase structures by which the chemical information is transmitted from the identity elements to the catalytic sites of the synthetases are unknown. This paper examines whether identity nucleotides in a tRNA contribute to aminoacylation specificity in a coupled or an independent manner. We have chosen the yeast aspartic acid system as a model, since the identity set has been established (Piitz et al., 1991; Figure 1) and since considerable structural data are available. The crystal structures of free (Moras et al., 1980; Westhof et al., 1985) and synthetase-complexed tRNAAsP (Ruff et al., 1991) are known and contacts of tRNAAsP with aspartyl-tRNA

2949

J.Putz et al.

aspartylation activity are analysed. The energetic relationships between identity nucleotides may be important for the biological function of tRNAs and synthetases.

Results

DA

A IPC

G

G-C C-G

30G-U40

C-G

C

GmlG

U

35 Fig. 1. Cloverleaf structure of yeast tRNAASP (Gangloff et al., 1971). 'The shadowed nucleotides are those defined as being responsible for specific aspartylation according to in vitro studies on tRNA transcripts devoid of modified nucleotides (Putz et al., 1991).

synthetase (AspRS) have been defined by solution footprinting experiments (Romby et al., 1985; Garcia et al., 1990; Garcia and Giege, 1992; Rudinger et al., 1992) in addition to X-ray crystallography (Cavarelli et al., 1993). To estimate the contribution of each identity element to specific tRNAAsP aminoacylation, tRNA variants with multiple substitutions of identity nucleotides were constructed and tested for their residual aspartylation activity. Kinetic data of multiple mutants were compared with values predicted from individual mutants. Kinetic specificity constants (kcatIKm) can be related to the free energy change AGt occurring during transition state binding of substrates to enzymes (Fersht, 1985). Thus, comparison of kctlKm values for wild-type and variant systems yields the free energy changes (AAGt) of the variant relative to the wildtype system (e.g. Ackers and Smith, 1985; Wells, 1990). Such quantitative comparisons have been widely used to measure effects of single or double mutations on the activity of variant synthetases (e.g. Wilkinson et al., 1983; Carter et al., 1984) or of single mutations of tRNAs (reviewed in Schulman, 1991; Giege et al., 1993). The free energy change for a multiply mutated tRNA can be calculated from the sum of free energy changes for each single mutant. These calculated free energies can be compared with the experimental values in order to determine if the mutated positions contribute independently or interactively to stabilization of the transition state for aminoacylation. Our results indicate differential effects of mutations of yeast tRNAASP on the aspartylation reaction catalysed by AspRS and show three different types of communication between the five identity determinants. Delocalized identity nucleotides show cooperativity whereas multiple substitutions of the anticodon nucleotides exhibit either additive or anticooperative effects. Correlations between the loss of tRNA/synthetase contacts in mutants and the decrease in 2950

Design and transcription of tRNAASP variants with multiple substitutions Figure 1 displays the sequence of yeast tRNAAsP with the identity nucleotides highlighted. Experiments reported here were carried out on molecules obtained in vitro by transcription with T7 RNA polymerase. Thirteen tRNAAsP genes were constructed using synthetic DNA oligonucleotides and transcribed to generate the tRNA variants displayed in Figure 2. Six variants contain delocalized pairwise substitutions of identity nucleotides from either the anticodon triplet and the discriminator residue in the acceptor stem (mutant 1: G35/A73; mutant 2: C34/A73; mutant 3: G35/U73; mutant 4: C34/U73) (Figure 2A), either D-stem and acceptor stem (mutant 8: C25/U73), either anticodon and D-stem (mutant 9: C25/A34) (Figure 2B). Three double mutants contain pairwise substitutions of anticodon nucleotides (mutant 10: A35/A36; mutant 11: A34/U36; mutant 12: A34/A36) (Figure 2C). Two triple mutants are changed at two anticodon positions in conjunction with the discriminator base (mutant 5: A341U361U73; mutant 6: A351A361A73). In one triple mutant (mutant 13) the aspartic acid anticodon GUC is replaced with the CAU anticodon (methionine). A last mutant contains a quadruple substitution in the anticodon triplet and discriminator base (mutant 7: C34/A35/U36/A73) (Figure 2E). These variants are only a limited number of the possible multiple substitutions of identity nucleotides, but they correspond to a representative- set of combinations; they are expected to exhibit either small, moderate or strong effects on aspartylation activity.

Aspartylation activity of multiple tRNAASP variants The aspartylation of tRNAASP variants decreases with the number of mutated identity positions. Figure 3 compares the aspartylation kinetics of wild-type tRNAAsP transcript with typical kinetics of three variants with either one, two or three substitutions at identity positions. No aspartylation was detected for all triple or quadruple variants except for the CAU triple mutant of the anticodon, which retains a significant aspartic acid acceptance (see below). The Michaelis -Menten parameters kcat and Km of the variants are summarized in Table I together with the relative kinetic specificity constants (kcatIKm)rei = (kcat1Km)tt (kcat/Km)wild-type. A more intuitive number is also included, namely the loss in aminoacylation efficiency, L; this is the inverse of (kcat/Km)re1, i.e. (kcatlKm)wild-typeI(kcatlKm)mutsnt. For simplicity, mutants are classified in terms of combined domains of the tRNA. As for single tRNAASP mutants (Piutz et al., 1991), multiple mutations mainly affect the kinetic rate constant kcat which is decreased by factors up to 2.2 x 104-fold (G35/A73), while the Km is only increased 15- to 150-fold (for G35/A73 and A35/A36). As a consequence the relative specificity constants are drastically decreased and the loss in aminoacylation efficiency of mutants can be up to 8.5 x 105-fold lower than the aminoacylation of the wild-type transcript. For triple and quadruple mutants (except the CAU anticodon mutant) the -

Relationship between tRNA identity nucleotides

(C)

°

G 0-0 0-0 0-0 0-0 0-0

0-0 o-o

0

0

g 0000 (7 0 oooO00U

00000 000

00

I000

0-000 !^

0

0-0 0-0 0-04:'

0-0 0 35

Gt, C y

4-'

~~~~~~

~

00000000

)-0 0 0 000 9~0 0 000

ro ,

000,1 00 000 -000o -0

UO0

~-0

0

0 oOooo00U0. 0? 1-1

0

00

0

0C O

2)0

I60i 000

0-

00-

t

-0

(D)

A*A@

0

(5 c0

(E)

G-

AU

CA L:-

C-AU .....X. .U..e Fig. 2. Cloverleaf structures of multiple mutants of yeast tRNAAsP at the level of identity nucleotides. Mutants have been arranged in five categories (A-E) in order to highlight structural relationships between variants: (A) double mutants involving one nucleotide of the anticodon and the discriminator base; (B) double mutants involving nucleotide 25 and either nucleotide 73 or nucleotide 34; (C) double mutations in the anticodon triplet; (D) triple mutants; (E) quadruple mutant between anticodon nucleotides and discriminator base.

aminoacylation levels were undetectable even under reaction conditions favouring mischarging of tRNAs (Giege et al., 1972). Moreover, assays in the presence of high amounts of substrate and/or enzyme, as was done elsewhere for direct determination of kratlKm (Schulnan and Pelka, 1989), failed as well. Thus, no estimation of the specificity constant could be made for these mutants. Theoretical versus experimental kinetic characteristics of tRNAASP variants with multiple mutations Before drawing conclusions about the relationship between the identity nucleotides in tRNAAsP let us recall the underlying theoretical background. Following Fersht (1985), and assuming Michaelis-Menten reaction kinetics, the relative kinetic specificity constant (kcat/Km)rel of an aminoacylation reaction involving a mutant tRNAAsP molecule as compared with the reaction with wild-type tRNA, is related to the difference in free energy change, AAGt, occurring during transition state binding of tRNA to AspRS in reactions with wild-type and mutant tRNAs, according to equation 1: (1) AAGt = -RT ln (kcatIKm)rej in which R is the gas constant, T the absolute temperature and (krat/Km.)rej defined as above. For a multiple tRNAAsP mutant (mm), the AAGt, during transition state binding to AspRS will be the sum of the free energy changes of the

10080.

20.z

co 20

0

5

10

Time (min) Fig. 3. Time-course of aspartylation of a selected set of identity nucleotide mutants of yeast tRNAASP. Filled squares, wild-type

transcript; filled diamonds, single mutant U36; filled triangles, double mutant C25/U73; filled circles, triple mutant A34/U36/U73Aminoacylation conditions were as described in Materials and methods. It should be noted that the extremely low charging of the double mutant is above background. It levels at a low plateau value because deacylation rates become more important than acylation rates. For more explanation of aminoacylation plateaus see Bonnet and Ebel (1972) and Dietrich et al. (1976). 2951

J.Putz et a/. Table I. Kinetic parameters for aspartylation of yeast tRNAASP variant transcripts with yeast AspRS kcat (x 10-3/s)

Km

kcatfKm (rel)

(IM)

(X 10-3)

La (x-fold)

tRNAASP Anticodon/acceptor stem:

520

0.05

1000

1

G35/A73

0.024 0.044 1.56 0.15 n.m. n.m. n.m.

0.74 3.6 1.0 1.4 n.m. n.m. n.m.

0.0031 0.0012 0.15 0.01 n.m. n.m. n.m.

320 000 830 000 6700 100 000 n.m. n.m. n.m.

3.83

1.5

0.25

4000

1.37

4.5

0.029

35 000

3.82 5.30 1.72 3.7

7.5 5.25 4.8 1.2

0.049 0.097 0.035 0.3

20 000 10 000 29 000 3300

tRNAASP variant wt

1 2 3 4 5 6 7

A35/A36/A73 C34/A35/U36/A73

8

C225U73

C34/A73

G35/U73 C34/U73 A34/U36/U73

D-stem/acceptor stem:

D-stem/anticodon:

C25/A34

9 10 11 12 13

Anticodon/anticodon: A35/A36 A34/U36 A34/A36

C34/A35/U36

aAs a control, each group of mutants was tested in parallel with the wild-type transcript. n.m., not measurable.

Lvalues for duplicates varied by at most 30%.

Table H. Detection of additive, cooperative and anti-cooperative effects between identity nucleotides of tRNAAsP in the aspartylation reaction catalysed by yeast AspRS

tRNAASP variant

Multiple mutants

Single mutants

L exp (X-fold)a (1) (2) (3) (4)

Anticodon/acceptor stem: G35/A73 C34/A73

1 2 3 4 S

G35/U73 C34/U73 A34/U36/U73

6 A35/A36/A73 7 C34/A35/U36/A73 D-stem/acceptor stem: 8 C25/U73 D-stem/anticodon: 9 C25/A34 Anticodon/anticodon: 10 A35/A36 11 A34/U36 12 A34/A36 13 C34/A35/U36 aSee Putz et al. b

AAGtcalc

= E

AAGtcalc L exp (kcal/mol)b (X 103-fold)

3 64 0.7 14.4 182 12 000 2 300 000

4.9 6.7 4.0 5.8 7.4 9.9 13.1

320 830 16.7 100 n.m. n.m. n.m.

7.7 8.3 5.4 7.0

(2)

19 400 19 400 71 530 400

160 160 36 36 71 150 530

8

36

0.3

3.5

8

71

0.6

150

79.5 5 11 15 000

36 160 71

71

150 530

71

AAGtC

160

n.m. n.m.

110 13 10 7 (>5) -

2.8 1.6 1.4 1.2 (> 1.0) _

n.m.

-

-

4

5.0

13

1.6

3.9

35

6.4

58

2.5

6.8 5.2 5.6 10.0

20 10 29 3.3

6.0 5.6 6.2 4.9

0.25 2 3 2 x 10-4

-0.8 0.4 0.6 -5.2

(1991) AAGtsm

n.m.: not measurable Errors on calculatedL- values for multiple mutants are estimated to be 30-60%; errors on the coupling

individual single mutants (sm), AAGtam, and of an increment AAGtc that corresponds to the free energy brought by the coupling of single mutations. Thus (2) AAGtmm = E AAGtsm + AAGtc = with AAGtC 0 when the effects of mutations are additive (independent contribution of tRNA identity nucleotides to aminoacylation efficiency without affecting transition state

2952

rk

(fexp/Lcaic) (kcal/mol)

(4)

(1)

530 71 71 400

(3)

AAGtexp (kcal/mol)

Lcalc (x 103-fold)

factorjtand AAGtC are estimated to be - 50%.

by indirect effects), AAGtc > 0 when they are cooperative, and AAGtC < 0 when they are anti-cooperative. To compare experimental values of kcatlKm directly, equation (2) can be rewritten as:

Lfpji

=

PA7-I Lam

(3)

(see Materials and methods for details) where L values represent losses of catalytic efficiency of multiple and single

..

Relationship between tRNA identity nuclootides

mutants, and where ft is a coupling factor where AAGtC =--RTln I/t from equation (2), with ft = 1 for additive effects, t > 1 for cooperative ones and ft < 1 for anticooperative ones. Our calculated data, obtained for the multiple tRNAAsP mutants, are displayed in their thermodynamic and kinetic forms and compared with experimental values in Table II. For single mutants, only partial kinetic data (Lvalues) defined previously are presented (Piitz et al., 1991). Differential kinetic effects between identity nucleotides To evaluate more directly aspartylation kinetics for the variants, we first examine experiments in terms of Lf and ft values. Double mutants with pairwise substitutions in the anticodon and acceptor stem gave significant differences between experimental and calculated values (mutants 1-4). These mutants show t factors clearly superior to 1 (ft = 7-110) and thus exhibit cooperativity between these sites. The strongest cooperativity with ft 110 is found for the G35/A73 variant (mutant 1) whereas a related variant G35/U73 (mutant 3) exhibits a much weaker cooperativity (t = 10). The triple mutant A34/U3WU73 (mutant 5), in the case of additivity, should possess a measurable activity ( f1 = 1.82 x 105, see Table II); no aminoacylation activity could be measured for this variant. Since our assay conditions are sensitive to activities at least 5-fold lower (see aspartylation activity of mutant C34/A73) it can be calculated that this mutant exhibits cooperativity with k > 5. Cooperativity is also found between identity nucleotides located either in the D- and acceptor stems (mutant 8) or in the D-stem and anticodon (mutant 9) since the 13 and corresponding coupling factors are very high ( 58 respectively). Strong anti-cooperativity is observed for the triple anticodon variant C34/A35/U36 (mutant 13) in which the aspartate anticodon GUC was replaced by the methionine anticodon. This variant possesses a residual methionine acceptance (Senger et al., 1992) but should be completely inactive in the case of additivity of individual mutations for aspartylation. Among single mutations in tRNAAsP, it contains mutations in the first and second position of the anticodon which have the strongest effects on aspartylation (L = 400 and 530, respectively). Thus, these two mutations alone should lead to a 2 x 105-fold decrease of aspartylation efficiency. The mutation at position 36 should reduce this activity by another factor of L = 71 and thus Lcalc should be 400 x 530 x 71 = 1.5 x 107. This low level of aspartylation cannot be observed by our approach. However, this variant is significantly charged (Table II). In fact, of all the multiple mutants tested in this work, it presents the highest aspartylation activity ( L of only 3300). As a consequence, its coupling factor is very low ( t = 2 x 10-4), which means that the three mutations C34, A35 and U36 together lead to a 20 000-fold increase of activity compared with the case where their effects would occur

independently. A quadruple mutant was constructed (mutant 7) in which not only all three nucleotides from the anticodon have been replaced by the CAU methionine anticodon, but where the discriminator base has also been changed from G73 to A73. Based on the anti-cooperative behaviour of the CAU anticodon triple mutant, it can be calculated that this

10,0

-

Lcaic(x-fold) 1 2 .i10

1 6

10 4 ,

10. ,

,10,.,

Cooperativity

10 6

2

1

8,0 . ct

6,0

64 D

3
U > A > C) and the comparison of the chemical groups on the bases suggested the importance of Ni and 06 residues of G73 for aspartylation. The complete array of these specific contacts between identity nucleotides and their amino acid counterparts on AspRS (Cavarelli et al., 1993) cannot form with tRNA variants mutated at these positions. For single mutants, the loss in transition state stabilization corresponds to 1.3-3.6 kcal/mol (calculated from data in Piitz et al., 1991). It is likely that a significant contribution to this low energy is the loss of hydrogen bonds upon mutation. However, the free energy of transition state stabilization contains contributions from hydrogen bonds, electrostatic interactions, van der Waals contacts, hydrophobic effects and hydration; it is impossible to break down this free energy into contributions from each type of interaction for the limited set of mutants presented here. Steric effects and reorganization of interactions are probable in the interactions of mutant transcripts with AspRS. Reorganization of interactions may be aided by the conformational flexibility of tRNAs; large conformational changes of tRNAASP are observed when it is complexed to the synthetase (Ruff et al., 1991; Rudinger et al., 1992). The effect of identity base pair G10-U25 on aminoacylation kinetics reflects this 2954

contormational flexibility. Although no direct contact of this base pair with the protein was found by X-ray crystallography, mutation of U25 to C25 led to an increase of Km (Piitz et al., 1991). This base pair is involved in tertiary interaction with G45 in the variable loop of the tRNA and mediates the functional conformation of the tRNA to allow the optimal presentation of the important chemical groups of the anticodon and discriminator bases to the synthetase. The limited loss of activity (AAGt = 2-3 kcal/mol) observed for single tRNAASP variants contrasts with the greater effect (AAGt = 5-6 kcal/mol) of single mutations in other tRNA aminoacylation systems [e.g. in methionine or valine systems as shown by Schulman and Pelka (1988) and Florentz et al. (1991), respectively]. As discussed above, specificity is mainly achieved in the transition state for aspartylation and the increase in Km (- 10-fold) on mutation corresponds to < 1 kcal/mol. In systems where identity determinants for specific aminoacylation are strong (reviewed in Giege et al., 1993), mutations may lead to losses of contacts without compensatory structural rearrangements of the complexes, in contrast with what would occur in systems where these identity elements have weak effects on aminoacylation. Interestingly, footprinting experiments on systems with strong determinants did not detect conformational changes on the tRNA upon complexation with the synthetase (e.g. Florentz and Giege, 1986). The results on specific tRNA -synthetase interaction should be compared with two types of DNA -protein complexes. In operator-repressor complexes, single base pair mutations in the DNA represent 2-6 kcal/mol of binding energy. This level of discrimination is achieved in the transition state of the aminoacylation reaction. In contrast, for DNA -protein complexes that involve catalytic steps, such as restriction enzymes (e.g. for DNA-EcoRI complexes as measured by Lesser etal., 1990) AAGt values can be up to 10 kcal/mol for single base pair changes. In both EcoRI and EcoRV considerable substrate discrimination is achieved in the transition state for DNA cleavage (Winkler, 1992). As with tRNA/synthetase discrimination, EcoRV discriminates extremely poorly among substrates on binding (Km for mutants are the same as for wild-type DNA) (Taylor et al., 1991). tRNA-synthetase interactions are characterized by protein recognition of spatially dispersed regions of the tRNA. To understand the discrimination mechanisms and the contribution of identity elements to specificity of aminoacylation, tRNAASP variants with pairwise or multiple mutations were analysed. Three different contributions can be expected: additivity, anti-cooperativity and cooperativity. Additivity indicates that mutations contribute independently to specificity and that a mutation at one residue does not perturb interactions with the other mutated positions. This is reflected by the absence of significant coupling energies between mutated positions. Anti-cooperativity and cooperativity indicate that mutations are coupled, either by partly overcoming the negative effects of the single mutations or by enhancing them. Hence, coupling energies will be negative in the former case and positive in the latter. Detailed mutational analyses of proteins have been used extensively to analyse protein -ligand interactions and protein folding (e.g. Ackers and Smith, 1985; Wells, 1990; Mildvan et al.,

Relationship between tRNA identity nucleotides

1992). TyrRS from Bacillus stearothermophilus has been extensively studied by mutational analysis of the catalytic site (e.g. Carter et al., 1984; Fersht et al., 1985; Wells and Fersht, 1986; Fersht, 1987). Three examples involving RNAs have been described that demonstrated additivity for identity nucleotides in yeast tRNATYr (Bare and Uhlenbeck, 1986) and tRNAPhe (Sampson et al., 1992), and cooperativity and anti-cooperativity in the tRNA-like domain from turnip yellow mosaic virus RNA aminoacylated by wheat germ ValRS (Dreher et al., 1992). For double mutants of the anticodon of tRNAASP, additive effects on aspartylation kinetics were observed. In Figure 4, which is the collective plot of the sum of the AAGt values for single mutants versus the corresponding AAGt for multiple mutants, anticodon double mutants A35/A36, A34/U36 and A34/A36 are distributed near to the diagonal, indicating AAGt values not significantly different from the predicted ones. In these mutants, specific RNA-protein contacts are disrupted with a well defined 5-stranded 3-barrel domain in AspRS that recognizes the aspartate anticodon (Cavarelli et al., 1993). The additivity of the mutational effects on aspartylation activity suggests that contacts between AspRS and the individual anticodon positions are independent. Interestingly, the additivity observed in yeast tRNATYr and tRNAPhe also occurs for anticodon identity nucleotides pairwise mutated (Bare and Uhlenbeck, 1986; Sampson et al, 1992). Additivity of double mutations within DNA operator sequences is an apparently general phenomenon in specific DNA-protein interactions (Sarai and Takeda, 1989; Takeda et al., 1989). The loss of specificity of double mutants in the anticodon can be analysed in terms of specific contacts lost on mutation. Comparing double and single mutants, the loss of specificity is roughly proportional to the number of specific contacts lost between the tRNA variants and the synthetase. However, a more quantitative comparison of anticodon double mutants illustrates the need for more sophisticated analysis. For mutant A35/U36, two hydrogen bonds are lost, and this corresponds to a loss of 5.6 kcal/mol. For mutant A34/A36, where three potential hydrogen bonds are lost, the energetic loss is 6.0 kcal/mol. Thus, even for positions that act additively, compensatory effects may occur. They are probably the result of conformational adaptation between AspRS and the tRNA variants. Increasing the number of mutations in the anticodon region may raise the number of perturbations at specific contacts with the (-barrel domain of AspRS. Therefore it is expected that enhanced compensatory effects in such mutants will lead to significant deviations from additivity. This was actually observed for triple anticodon variant C34/A35/U36, which exhibits anticooperativity. The anti-cooperativity of the triple anticodon mutant C34/A35/U36 is characterized by a negative coupling energy of 5.2 kcal/mol; this corresponds to a catalytic efficiency 4500-fold better than predicted from single mutant data. Compared with all other multiple mutants studied in this work, this mutant exhibits the highest aspartylation efficiency. There are two explanations for anti-cooperative behaviour. First, a mutation of one anticodon position may lead to disruption of contacts to neighbouring nucleotides, which cannot be 'counted twice' on further mutation. However, the results of double mutation data are not consistent with this interpretation alone. Second, strong anti-

cooperativity can be explained by the creation of novel interactions between mutated tRNA and identity amino acids in AspRS, which partially compensate for the loss of four specific hydrogen bonds. Compensatory changes are facilitated by the structural flexibility of the interacting domains. On complex formation, the anticodon region of tRNAASP changes conformation by unstacking of anticodon bases, whereas these nucleotides are stacked in the free tRNA (Ruff et al., 1991). Such conformational flexibility may also characterize the (-barrel domain of AspRS, but the crystallographic structure of the free synthetase is unsolved. The anti-cooperative effect of the triple mutation is sequence specific, since the other triple anticodon mutant studied in this work (variant A34/A35/A36) is completely inactive. Thus, new contacts may have been created with the chemical groups of the CAU anticodon that will not form with those present in an AAA anticodon. Anti-cooperative behaviour has been observed for a double mutant in the anticodon loop of the turnip yellow mosaic viral tRNA-like domain for its valylation by wheat germ ValRS (Dreher et al., 1992). Interestingly, this active double mutant (mutations at positions equivalent to nucleotides 34 and 36 of classical tRNA) is naturally generated in vivo. Indeed, in host cells, genomic viral RNA strongly inactivated by the presence of one of these mutations becomes active after the occurrence of the second site mutation (Tsai and Dreher, 1992). tRNAASP variants with two delocalized mutations (in anticodon loop and acceptor stem; anticodon loop and Dstem; acceptor stem and D-stem) exhibit cooperativity with coupling energies between mutations of 1.2-2.8 kcal/mol. This indicates that the alteration of an identity nucleotide in one domain affects the functioning of an identity position located far away in another domain of the tRNA. The anticodon and the discriminator base contact two well separated domains of AspRS, one near to the catalytic centre and the other far away (Cavarelli et al., 1993). Single mutation of anticodon or the discriminator nucleotides result in significant changes in kcat for aspartylation. Thus, longrange transfer of chemical information from the anticodon nucleotides is altered during the transition state of aspartylation. The cooperativity of double mutations in the anticodon and discriminator positions may be another manifestation of this long-range contact. We have previously shown that single mutations at the discriminator position do not alter contacts of AspRS in the anticodon, based on footprinting patterns of mutant transcripts with AspRS (Rudinger et al., 1992). These footprinting experiments, which suggest independent recognition of the anticodon and acceptor stem of tRNAASP, are consistent with the mutational studies performed here. Footprinting experiments probe only the initial complex between the tRNA and AspRS, whereas the cooperativity between the anticodon and 3 '-end of tRNAASP is observed in the transition state for aspartylation. The cooperativity of the two variants with alterations in the D-stem (C25) and in the anticodon (A34) or acceptor region (U73) probably represents a different mechanism since identity pair G1O-U25 does not contact AspRS. In these cases, the conformational change within tRNA triggered by the mutation of U25 to C25 affects the contact between AspRS and identity positions 73 and 34 in tRNA. This view is supported by the enhanced reactivities towards iodine observed in the hinge region of tRNAASP variants mutated

2955

J.Putz et a/.

at either U73 or A34 in complexes with AspRS (Rudinger et al., 1992). These footprinting data indicate that mutations at both extremities of the tRNA influence the positioning of the D-stem and variable region in the central core of the tRNA when complexed with AspRS. Thus, cooperativity results from the combination of direct and indirect effects leading to conformational adaptation within the tRNA - synthetase complex. Cooperativity is a standard regulatory mechanism in biological systems. In classic examples, such as binding of oxygen to haemoglobin, binding of ligand at one binding site increases the binding affinity for that ligand at (in many cases) a symmetrically related binding site (Perutz, 1990). In the case of tRNAASP interacting with AspRS, the ligand is a macromolecule and the two 'binding sites' per monomer unit involve interaction with different portions of the tRNA: the anticodon and the acceptor arm. However, as in the case of small molecule ligands, binding (in the transition state) of one 'ligand', the anticodon, improves binding of the second 'ligand', the 3 '-acceptor arm. Most cases of cooperativity and allostery in biological interactions involve multi-subunit proteins. The long-range transfer of information in these systems involve changes in subunit contacts. Similar mechanisms of cooperativity may be involved in the interaction of tRNAASP with AspRS. Information transfer may occur within a single subunit of protein upon conformational changes communicated between the two protein domains that contact the two domains anticodon and acceptor stem-of a single tRNA. AspRS is a dimeric protein with two symmetrically related tRNAbinding sites. The active site domain of one subunit makes extensive intersubunit contacts with the anticodon binding domain of the second subunit. This orientation of the two tRNA binding sites suggests an alternative mechanism of information transfer between subunits from the anticodon of one tRNA to the active site of a second bound tRNA. Cooperativity might be an efficient way to enhance specificity in systems in which individual identity elements have relatively weak effects on activity. These systems often have their identity elements scattered over the entire structure of the tRNA, in contrast to systems having strong determinants (e.g. valine) in which a small number of major identity nucleotides are clustered in one region of the tRNA. The additive (or anti-cooperative) effect of multiple anticodon mutations would be favourable for specific recognition by a single synthetase of multiple isoacceptors. The more complex behaviour of systems with weak determinants may result from particular structural characteristics of the synthetases, like the well defined multidomain structure of AspRS. The strong anti-cooperative behaviour of certain combinations of mutations as found in tRNAMsP (this work) and in a tRNA-like structure (Dreher et al., 1992) illustrates that alternative solutions exist for making specific and efficient tRNA substrates for a synthetase. This is well known for viral tRNA-like structures that mimic canonical tRNAs (Mans et al., 1991; Florentz and Giege, 1993). As discussed above, the existence of functional mimicry among canonical tRNAs is not necessarily detrimental to aminoacylation systems. Indeed, for the anti-cooperative triple mutant of tRNAAsP with a CAU methionine anticodon, the mutations only confer a low methionine acceptance to the variant, although the methionine anticodon is known to be a strong identity element for methionylation

2956

(Senger et al., 1992). In that case, negative effects brought by antideterminants prevent efficient aminoacylation of the tRNAAsP variant by yeast MetRS. Sequence context plays an important role for conferring the biologically significant specificity to a tRNA. To conclude, we would like to emphasize how complex and subtle are the mechanisms leading to specificity in tRNA aminoacylation systems. Although new concepts are emerging from the mutational analysis of tRNA mutants, deeper understanding of the structural and mechanistic solutions retained by evolution for ensuring specific tRNA recognition and aminoacylation by synthetases requires further investigation of various complexes involving wildtype or variant tRNA and synthetase.

Materials and methods Materials Yeast AspRS and T7 RNA polymerase were purified as described by Lorber et al. (1983) and Wyatt et al. (1991) respectively. Oligonucleotides were synthesized on an Applied Biosystems 381A DNA synthesizer using the phosphoramidite method and purified by HPLC on a Nucleosil 120-5 C18 column (Bischoff Chromatography, Zymark-France, Paris). T4 polynucleotide kinase and the restriction enzyme BstNI were obtained from New England Biolabs (Beverly, MA, USA) and T4 DNA ligase from Boehringer-Mannheim (France SA, Meylan). Ribonucleotides were purchased from Sigma (St Louis, MO, USA), and Amersham France (Les Ulis) provided L-[3H]aspartic acid.

Preparation of tRNA transcripts Mutants were obtained by in vitro transcription of synthetic tRNAAsP genes under the control of the polymerase promoter. Ten synthetic DNA oligomers (15- to 25-mers) containing the T7 RNA polymerase promoter and the desired tRNAASP sequences were ligated into plasmid pTFMA, derived from pUC18 according to established procedures (Perret et al, 1990b; Putz et al., 1991). Sequences of the mutated tRNAASP genes were confirmed by dideoxynucleotide sequencing of single stranded DNA using the M13 universal primer and the T7 DNA polymerase (Tabor and Richardson, 1987). Unmodified tRNA transcripts were obtained by transcription of 50 yg of BstNI-linearized plasmid by T7 RNA polymerase. Transcription reactions were incubated at 37°C for 3 h in 40 mM Tris-HCl pH 8.1, 22 mM MgCl2, 1 mM spermidine, 5 mM dithiothreitol, 0.01% Triton X-100, 4 mM each nucleoside triphosphate, 16 mM GMP, 40 U RNasin (Promega, Madison, WI, USA) and the appropriate amount of T7 RNA polymerase. Full length transcripts were systematically separated from unincorporated nucleotides and abortive transcription products by denaturing 12% PAGE. Only the transcripts of the correct length were electroeluted from the gel using a Biotrap electroelution apparatus (Schleicher & Schilil, Dassel, Germany).

Aminoacylation assays Aminoacylation tests were performed in 0.1 M HEPES-KOH pH 7.5, 30 mM KC1, 10 mM MgCl2, 5 mM ATP and 52 AM L-[3H]aspartic acid as described previously (Perret et al., 1990b). 0.3-5 /%M tRNAASP transcript and 0.16-0.32 AM AspRS were used. We used the GF-C72 mutant as a control, which shows equivalent aspartylation parameters to those of fully modified tRNAASP and U1-A72 transcript (Perret et al., 1990a). All aspartylation kinetics were performed under standard conditions previously used, except for the mutants A34/U36/U73, A35/A36/A73 and C34/A35/U36/A73 where we also used conditions allowing misacylation (Giege et al., 1972). kcat/Km values for replicate experiments varied by at most 15%. Transformation of kinetic data The thermodynamic equation (2) (see Results) can be easily transformed into its kinetic equivalent, using equation 1. Thus -RT In (kcat/Km)rel mm which simplifies to

=

-RT In rI

(kcat/Km)rej

sm

-

RT ln k

'(4)

R' HI (kcat/Km)rel. sm (4') with k' being a factor that takes into account the coupling effects of mutations. Since the relative (kcat/Km) value for each single mutant can be

(kcat/Km)rel.

mm =

Relationship between tRNA identity nucleotides measured individually, the product of the relative specificity constants for single mutants can be calculated and compared with the corresponding experimental values of the multiple mutants. This comparison yields 1' and consequently allows judgements to be made about additivity, cooperativity or anti-cooperativity in the expression of identity nucleotides. Since relative specificity constants are inverses of specificity losses [ L = l/(kcat/Km)reI], equation 4' is readily transformed into

(3) L.= kIl L sm with R = 1/R'. Similarly, as explained above for relative kcatIKm ratios for multiple mutants, L. is accessible either experimentally, or theoretically from values obtained for single mutants according to L nm(caic) = fI Lsm(exp). Thus, the coupling factor R between multiple mutations corresponds to the ratio between experimental and calculated L values (R - Lexp/ Lcc). Using this formalism, cooperativity implies that Lexp/Lcal, > 1 and anti-cooperativity that Lexp/ L.ac < 1. Notably, St factors are related to the free energy changes brought by the coupling of mutations (AAGtc) according to equation (5):

AAGtC = -RT ln I/R

(5)

Acknowledgements This work was supported by grants from the Centre National de la Recherche Scientifique (CNRS), Ministare de la Recherche et de l'Espace (MRE), Universite Louis Pasteur, Strasbourg, and by the Human Frontier Science Program (HSFP) Organization. We thank F.Fasiolo and his colleagues for providing mutants 7 and 13 and D.Moras and his colleagues for crystallographic data on the aspartate complex before publication. J.D.P. was the recipient of an EMBO long term fellowship and J.P. was partly supported by the HSFP organization.

References Ackers,G.K. and Smith,F.R. (1985) Annu. Rev. Biochem., 54, 597-629. Bare,L.A. and Uhlenbeck,O.C. (1986) Biochemistry, 25, 5825-5830. Bonnet,J. and Ebel,J.-P. (1972) Eur. J. Biochem., 31, 335-344. Carter,P.J., Winter,G., Wilkinson,A.J. and Fersht,A.R. (1984) Cell, 38, 835-840. Cavarelli,J., Rees,B., Ruff,M., Thierry,J.-C. and Moras,D. (1993) Nature, 362, 181-184. Dietrich,A., Kern,D., Bonnet,J., Giegd,R. and Ebel,J.-P. (1976) Eur. J. Biochem., 70, 147-158. Dreher,T.W., Tsai,C.-H., Florentz,C. and Giegd,R. (1992) Biochemistry, 31, 9183-9189. Ebel,J.-P., Giege,R., Bonnet,J., Kern,D., Befort,N., Bollack,C., Fasiolo,F., Gangloff,J. and Dirheimer,G. (1973) Biochimie, 55, 547-557. Fersht,A. (1985) Enzyme, Structure and Mechanism. Freeman, New York. Fersht,A.R. (1987) Biochemistry, 26, 8031-8037. Fersht,A.R., Shi,J.P., Knill-Jones,J., Lowe,D.M., Wilkinson,A.J., Blow,D.M., Brick,P., Carter,P., Waye,M.M.Y. and Winter,G. (1985) Nature, 314, 235-238. Florentz,C. and Giege,R. (1986) J. Mol. Biol., 191, 117-130. Florentz,C. and Giege,R. (1993) In S611,D. and RajBhandary,U.L. (eds), Transfer RNA, American Society of Microbiology, Washington, DC, in press. Florentz,C., Dreher,T.W., Rudinger,J. and Giegd,R. (1991) Eur. J. Biochem., 195, 229-234. Gangloff,J., Keith,G., Ebel,J.-P. and Dirheimer,G. (1971) Nature NB, 230, 125- 127. Garcia,A. and Giege,R. (1992) Biochem. Biophys. Res. Commun., 186, 956-962. Garcia,A., Giege,R. and Behr,J.P. (1990) Nucleic Acids Res., 18, 89-95. Giege,R., Kern,D. and Ebel,J.-P. (1972) Biochimie, 54, 1245-1255. Giegd,R., Puglisi,J.D. and Florentz,C. (1993) Progr. Nucleic Acid Res. Mol. Biol., 45, 129-206, in press. Lapointe,J. and Giegd,R. (1991) In Trachsel,H. (ed.), Translation in Eukaryotes. CRC Press Inc., Boca Raton, FL, pp. 35-69. Lesser,D.R., Kurpiewski,M.R. and Jen-Jacobson,L. (1990) Science, 250, 776-786. Lorber,B., Kern,D., Dietrich,A., Gangloff,J., Ebel,J.-P. and Giege,R. (1983) Biochem. Biophys. Res. Commun., 117, 259-267. Mans,R.M.W., Pleij,C.W.A. and Bosch,L. (1991) Eur. J. Biochem, 201, 303-324. Mildvan,A.S., Weber,D.J. and Kuliopoulos,A. (1992) Arch. Biochem. Biophys., 294, 327-340.

Moras,D., Comamond,M.-B., Fischer,J., Weiss,R., Thierry,J.-C., Ebel,J.-P. and Giegd,R. (1980) Nature, 288, 669-674. Muramatsu,T., Nishikawa,K., Nemoto,F., Kuchino,Y., Nishimura,S., Miyazawa,T. and Yokoyama,S. (1988) Nature, 336, 179-181. Musier-Forsyth,K. and Schimmel,P. (1992) Nature, 357, 513-515. Normanly,J. and Abelson,J. (1989) Annu. Rev. Biochem., 58, 1029-1049. Pallanck,L. and Schulman,H. (1992) In Hatfield,D.L., Lee,B.J. and Pirtle,R.M. (eds), Transfer RNA in Protein Synthesis. CRC Press Inc., Boca Raton, FL, pp. 279-318. Perret,V., Garcia,A., Grosjean,H., Ebel,J.-P., Florentz,C. and Giege,R. (1990a) Nature, 344, 787-789. Perret,V., Garcia,A., Puglisi,J.D., Grosjean,H., Ebel,J.-P., Florentz,C. and Giegd,R. (1990b) Biochimie, 72, 735-744. Perutz,M. (1990) Mechanisms of Cooperativity and Allosteric Regulation in Proteins. Cambridge University Press, Cambridge. Putz,J., Puglisi,J.D., Florentz,C. and Giegd,R. (1991) Science, 252, 1696-1699. Romby,P., Moras,D., Bergdoll,M., Dumas,P., Vlassov,V.V., Westhof,E., Ebel,J.-P. and Giege,R. (1985) J. Mol. Biol., 184, 455-471. Rould,M.A., Perona,J.J., S6ll,D. and Steitz,T.A. (1989) Science, 246, 1135-1142. Rould,M.A., Perona,J.J. and Steitz, T.A. (1991) Nature, 352, 213 -218. Rudinger,J., Puglisi,J.D., Puitz,J., Schatz,D., Eckstein,F., Florentz,C. and Giegd,R. (1992) Proc. Natl Acad. Sci. USA, 89, 5882-5886. Ruff,M., Krishnaswamy,S., Boeglin,M., Poterszman,A., Mitschler,A., Podjarny,A., Rees,B., Thierry,J.-C. and Moras,D. (1991) Science, 252, 1682-1689. Sampson,J.R., Behlen,L.S., DiRenzo,A.B. and Uhlenbeck,O.C. (1992) Biochemistry, 31, 4164-4167. Sarai,A. and Takeda,Y. (1989) Proc. NatlAcad. Sci. USA, 86, 6513-6517. Schimmel,P. (1989) Biochemistry, 28, 2747-2759. Schulman,L.H. (1991) Prog. Nucleic Acid Res. Mol. Biol., 41, 23-87. Schulman,L.H. and Pelka,H. (1988) Science, 242, 765-768. Schulman,L.H. and Pelka,H. (1989) Science, 246, 1595-1597. Senger,B., Despons,L., Walter,P. and Fasiolo,F. (1992) Proc. NatlAcad. Sci. USA, 89, 10768-10771. Soll,D. (1991) Experientia, 46, 1089-1096. Tabor,S. and Richardson,C.C. (1987) Proc. Natl Acad. Sci. USA, 84, 4767. Takeda,Y., Sarai,A. and Rivera,V.M. (1989) Proc. NatlAcad. Sci. USA, 86, 439-443. Taylor,J.D., Badcoe,I.G., Clarke,A.R. and Halford,S.E. (1991) Biochemistry, 30, 8743-8753. Tsai,C.H. and Dreher,T.W. (1992) J. Virol., 66, 5190-5199. Wells,J.A. (1990) Biochemistry, 29, 8509-8517. Wells,T.N.C. and Fersht,A.R. (1986) Biochemistry, 25, 1881-1886. Westhof,E., Dumas,P. and Moras,D. (1985) J. Mol. Biol., 184, 119-145. Wilkinson,A.J., Fersht,A.R., Blow,D.M. and Winter,G. (1983) Biochemistry, 22, 3581-3586. Winkler,F. (1992) Curr. Opin. Struct. Biol., 2, 93-99. Wyatt,J.R., Chastain,M. and Puglisi,J.D. (1991) BioTechniques, 11, 764-769. Yarus,M., Knowlton,R. and Soll,L. (1977) In Vogel,H.J. (ed.), Nucleic Acid-Protein Recognition. Academic Press, Inc., New York, pp. 391-408. Received on Febnrary 15, 1993

2957