(Received for publication, August 8, 1996, and in revised form, October 9, 1996)
From the ¶ Department of Molecular Biophysics and
Biochemistry, Yale University, New Haven, Connecticut 06520-8114, Department of Chemistry, Faculty of Science, University
of Zagreb, 10000 Zagreb, Croatia, and § Rudjer
Bo
kovi
Institute, 10000 Zagreb, Croatia
The active site of class II aminoacyl-tRNA
synthetases contains the motif 2 loop, which is involved in binding of
ATP, amino acid, and the acceptor end of tRNA. In order to characterize
the active site of Saccharomyces cerevisiae seryl-tRNA
synthetase (SerRS), we performed in vitro mutagenesis of
the portion of the SES1 gene encoding the motif 2 loop.
Substitutions of amino acids conserved in the motif 2 loop of
seryl-tRNA synthetases from other sources led to loss of
complementation of a yeast SES1 null allele strain by the
mutant yeast SES1 genes. Steady-state kinetic analyses of
the purified mutant SerRS proteins revealed elevated
Km values for serine and ATP, accompanied by
decreases in kcat (as expected for
replacement of residues involved in aminoacyl-adenylate formation). The
differences in the affinities for serine and ATP, in the absence and
presence of tRNA are consistent with the proposed conformational
changes induced by positioning the 3-end of tRNA into the active site,
as observed recently in structural studies of Thermus
thermophilus SerRS (Cusack, S., Yaremchuk, A., and Tukalo, M. (1996) EMBO J. 15, 2834-2842). The crystal structure of
this moderately homologous prokaryotic counterpart of the yeast enzyme
allowed us to produce a model of the yeast SerRS structure and to place
the mutations in a structural context. In conjunction with structural
data for T. thermophilus SerRS, the kinetic data presented
here suggest that yeast seryl-tRNA synthetase displays tRNA-dependent amino acid recognition.
The formation of aminoacyl-tRNA, catalyzed by aminoacyl-tRNA
synthetases, is a crucial step in maintaining the fidelity of protein
biosynthesis. This family of enzymes can be partitioned into two
classes of 10 enzymes each, based on conserved sequences (1) and
structural motifs (2). All members of class I contain a common loop
with the signature sequence KMSKS (3) and a region of homology with the
HIGH peptide (4) as part of a Rossmann dinucleotide binding fold of
parallel -sheets (5). Class II synthetases have a different topology
of dinucleotide binding based on antiparallel
-sheets (2, 6, 7). The
three common signature motifs of class II synthetases are found in this
domain. Motif 1 forms part of the conserved inter-subunit interface of homodimeric (6, 8) and heteromeric (9) synthetases. Motifs 2 and 3 contain many of the active-site residues important for ATP, amino acid,
and tRNA acceptor stem recognition (10, 11, 12, 13, 14, 15). The elucidation of the
crucial role of sequence motifs in substrate binding have resulted from
the solution of several crystal structures of enzymes and
enzyme-substrate complexes from both class I and class II (16) and
numerous biochemical studies involving mutant synthetases (17, 18, 19, 20, 21, 22).
The evolution of tRNA recognition systems has recently gained much
attention (23, 24, 25, 26, 27). The primary structure of several prokaryotic and
eukaryotic seryl-tRNA synthetases (Ref. 23; see also the legends to
Fig. 1 and 3), including the enzyme that probably functions in yeast
mitochondria (28), have been determined. While two prokaryotic enzymes,
from Escherichia coli and Thermus thermophilus,
have been crystallized in different complexes with substrates and
subjected to biochemical analysis in order to identify the domains
important for tRNA and amino acid binding (23), information on
structure/function relationships in eukaryotic seryl-tRNA synthetases
is still scarce by comparison. In contrast to their prokaryotic
counterparts, these enzymes contain C-terminal extensions abundant in
basic amino acids, which may be important for both stability and
optimal substrate recognition in eukaryotic
SerRS,1 as recently shown for yeast SerRS
(29). To gain further insight into the mechanisms of substrate
recognition employed by yeast SerRS and to identify catalytically
important residues in the active site, we have replaced a number of
amino acids in the motif 2 loop and analyzed the resulting mutants
in vivo and in vitro. The altered kinetic
parameters of the mutant SerRS proteins correlate with predictions
based upon structural studies of the prokaryotic system.
[3H]Serine (30.0 Ci/mmol) was from Amersham Corp. [14C]Serine (166.1 mCi/mmol) and tetrasodium [32P]pyrophosphate (3.09 Ci/mmol) were purchased from DuPont NEN.
Plasmids and StrainsIn vitro mutagenesis of the
SES1 gene was carried out (see below) in pBluescript SK(),
where the BamHI cassette containing SES1 (30) was
inserted, giving pSKSES1. The resulting mutant SES1 genes, named SES1mut, were
recloned as 1.4-kilobase pair BamHI fragments into pVTL-100
behind the ADH (alcohol dehydrogenase) promoter and pCJ11
behind the GAL promoter for complementation of the
null-allele strains, or in pET3 behind the T7 RNA polymerase promoter
for overproduction of mutated proteins in E. coli (29, 30).
Construction of the Saccharomyces cerevisiae SES1 disruption strain BR2727
SES1 (MATa ade2-1 arg4-9
his4 leu2-3, 112 lys2 trp1 ura3-1 SES1::LYS2) has been
described (29). The deletion could be rescued by supplying a functional
SES1 gene on a centromeric plasmid pUN70 (29).
Selected haploids with a disrupted SES1 gene
(Lys+ phenotype), where SerRS function was provided
in trans (from pUN70SES1 plasmid), were used for
complementation experiments via plasmid shuffling involving
pVTL-100SES1mut. Since the parental strain
BR2727 was not fully galactose-inducible, as shown by its inability to
ferment galactose after streaking the colonies on bromthymol blue
indicator plates (1% yeast extract, 2% peptone, 2% agar, 2%
galactose, and 80 mg/liter bromthymol blue; Ref. 31), pUN70SES1 transformants of BR2727
SES1 were
crossed with S. cerevisiae S2088 (MAT
ura3-52 trp 1 lys2-801 leu2
1 his3-
200
pep4::HIS3 prb-
1.6R can 1 GAL). Diploids were
sporulated, tetrads dissected, and haploid segregants of the
pUN70SES1 transformant with a disrupted SES1 gene
(Lys+ phenotype) and the ability to ferment galactose
(produce a yellow halo on indicator plates) were selected. The strain
was named SD816 and used for complementation experiments via
plasmid exchange with pCJ11SES1mut.
Saturation mutagenesis was performed
with a 5:1 molar ratio of synthetic phosphorylated oligonucleotides to
U-containing single-stranded pSKSES1 DNA. A mutagenic
oligonucleotide, a 50-mer complementary to the codons corresponding to
amino acids 278-294, was synthesized doped to 4% with an equal
mixture of A/C/G/T at all positions. A new NcoI restriction
site was introduced internally by silent mutations to allow screening
by restriction analysis. Annealing, extension, and ligation of the
mutagenic mixture was as described (32). After transformation of
E. coli DH5, plasmid DNA was isolated from a large pool
of individual transformants. The pertinent regions of all
NcoI-containing pSKSES1 plasmids were subjected to sequencing in order to determine the mutational changes. Multiple mutations in the SES1 gene were separated by exchanging
either 880-base pair PstI/NcoI or 560-base pair
NcoI/EagI fragments with the corresponding
fragments of pSKSES1wt.
The overproduction of mutated SerRS enzymes was achieved
by transformation of E. coli strain BL21(DE3) with
pET3SES1mut constructs, followed by induction of
the cultures with 2 mM
isopropyl-1-thio--D-galactopyranoside. The enzymes were
purified by a two-step chromatographic procedure on FPLC MonoQ and
MonoS columns (Pharmacia Biotech Inc.) as described previously
(30).
The extent of adenylate formation catalyzed by the purified enzymes was assessed by pyrophosphate exchange at 30 °C (33). The reaction was performed in a buffer consisting of 100 mM Hepes/KOH, pH 7.2, 10 mM MgCl2, 0.5 mM [32P]PPi (1-2 cpm/pmol), 10 mM KF, containing various amounts of the substrate studied (serine or ATP), while the other substrate was kept saturating (serine at 5 mM, Mg-ATP at 2 mM (molar ratio 1:1)). Typical concentrations of the substrate studied were from 0.1 to 10 times Km. The concentration of wild-type enzyme was 45 nM, while the mutants were used in a concentration range from 45 to 450 nM.
Aminoacylation was done at 30 °C as described previously (29) in reaction mixtures containing 50 mM Tris-HCl, pH 7.5, 15 mM MgCl2, 4 mM dithiothreitol. For all kinetic assays, the concentrations of the substrate studied ([3H]serine, Mg-ATP, or tRNASer) were varied from 0.1 to 10 times Km. Saturating concentrations of other substrates were: 5 mM ATP, 12 mg/ml unfractionated brewer's yeast tRNA (Boehringer Mannheim), which corresponds to approximately 20 mM tRNASer, and 1 mM 3H- or 14C-labeled serine. To correct for the low counting efficiency of the free [3H]Ser in comparison to that of [3H]Ser-tRNA, total charging assays with either 3H- or 14C-labeled Ser were performed in parallel so that the conversion factor could be calculated. Enzyme concentrations were 94 nM for wild-type SerRS and 94-470 nM for mutants. Kinetic parameters were determined from Hanes-Woolf plots. All values represent the average of three independent determinations, which varied by less than 10%.
Sequence Analysis and Protein ModelingMultiple alignment of protein sequences was performed using CLUSTALW (34). The PHYLIP package (35) was used to calculate protein distance matrices and to construct phylogenetic trees. Percent similarities between proteins were calculated by the method of Myers and Miller (36). Theoretical models of the yeast SerRS three-dimensional structure were produced and optimized using the PROMOD program suite (37).2 Images were produced using RasMol Molecular Renderer version 2.6.
There is significant similarity among the primary structures of all known SerRS proteins. Pairwise similarities range from 26% between S. cerevisiae cytoplasmic and a recently identified S. cerevisiae putative second SerRS (28) to 78% between E. coli and Haemophilus influenzae SerRS. The second yeast SerRS is most likely a mitochondrial enzyme. This is based on its position in the phylogenetic tree, the presence of a putative N-terminal mitochondrial targeting sequence (as diagnosed by multiple sequence alignment, and with high confidence by the method of Nakai and Kanehisa; Ref. 38)3 and the lack of a C-terminal extension (29). The phylogenetic tree showing evolutionary relationships among seryl-tRNA synthetases from different organisms is presented in Fig. 1. The cytoplasmic enzymes of eukaryotes form a separate group, and the divergence pattern of eubacterial enzymes is in accord with the contemporary phylogenetic trees (39). Surprisingly, an archaeal SerRS (Haloarcula marismortui) clusters with the enzymes of B. subtilis and the three Gram-negative bacteria regardless of the method used to construct the tree. This suggests that, at some point in evolution, a eubacterial gene for SerRS replaced the ancestral H. marismortui gene by horizontal transfer.
In spite of primary sequence similarity, immunological cross-reactivity
has been observed neither between eukaryotic and prokaryotic seryl-tRNA
synthetases nor between the cytoplasmic and organellar proteins (40,
41). As revealed by the crystal structures of two prokaryotic
seryl-tRNA synthetases from E. coli (2) and T. thermophilus (42), each subunit of the homodimeric enzymes is
composed of two domains: an N-terminal "helical arm" comprising a
60-Å solvent-exposed, antiparallel coiled-coil, which binds the
variable arm of cognate tRNASer (12, 43) and the catalytic
domain based on a seven-stranded antiparallel -sheet. The mode of
seryl-adenylate binding in the active site is also known from
crystallographic and solution studies (10, 11, 13). However, none of
the eukaryotic seryl-tRNA synthetases have yet been crystallized, nor
is there any information on substrate binding domains and catalytic
sites from biochemical experiments. In our search for the active site
of yeast SerRS, we have performed saturation in vitro
mutagenesis of the portion of the SES1 gene encoding the
domain that in the model structure based upon the coordinates of SerRS
from T. thermophilus (Fig. 2), corresponds to
the active site loop in motif 2 (residues 278-294 in yeast SerRS)
(Fig. 2). As shown by primary sequence alignment (Fig.
3), this is the most conserved region among all
SerRS proteins.
In Vivo Analysis of SerRS Mutants
From the many SerRS mutants
isolated, which carry one to five amino acid substitutions in region
278-294 of the motif 2 loop, we wanted to select those with the most
severely affected substrate binding or catalytic properties. This was
first assessed in vivo by the ability of the mutants to
complement S. cerevisiae SES1 null-allele strains via
plasmid shuffling (Table I). The haploid strain
BR2727SES1 (where viability of the cell is ensured by a
wild-type SES1 gene on the URA3-containing
plasmid pUN70SES1), was transformed with
pVTL-100SES1mut constructs (which carry the LEU2-selectable marker and SES1mut
behind the ADH promoter). Four of the motif 2 loop mutants
were found to have lost the ability to complement the
SES1 null allele. They were named
SES1mut2, SES1mut3,
SES1mut4, and SES1mut5.
The type and the position of mutational changes in the resulting SerRS
proteins are presented in Fig. 3B. In order to check whether
the function of wild-type SerRS can be restored by the
overproduction of mutated SerRS proteins, the complementation of
S. cerevisiae SD816 (which also carries an inactivated
SES1::LYS2 allele but has a normal galactose
uptake), was performed via plasmid shuffling with
pCJ11SES1mut constructs. After induction of the
GAL promoter by switching the cultures of
SD816/pCJ11SESmut from glucose to galactose
medium, significant amounts of mutant proteins accumulated in the cell,
as detected by SDS-polyacrylamide gel analysis of protein extracts
(not shown). pUN70SES1 was cured by growing induced double
transformants in the presence of uracil, followed by plating the cells
on selective galactose plates. Lys+Leu+
colonies were replica-plated to 5-fluoro-orotic acid-containing medium
(31), which counterselects against colonies containing URA3
plasmids carrying the wild-type S. cerevisiae SES1 gene. Some very slow growing transformants of
pCJ11SES1mut4 and
pCJ11SES1mut5 appeared on 5-fluoro-orotic acid
plates after several days of incubation at 30 °C, while all the
cells containing pCJ11SES1mut2 or
pCJ11SES1mut3 as the only source of seryl-tRNA
synthetase activity were nonviable (Table I). It is thus apparent that
several amino acid substitutions between residues 278 and 294 of yeast
SerRS are responsible for enzyme inactivation and consequently for the phenotypic alteration.
|
To determine the steps of the aminoacylation reaction at which the altered enzymes are defective, noncomplementing (SES1mut2 and SES1mut3) or weakly complementing mutants (SES1mut4 and SES1mut5) which normally accumulate in yeast cells (Table I) were overexpressed in E. coli, purified to apparent homogeneity as described under "Materials and Methods," and characterized in vitro. As discussed previously (29), there are some indications that S. cerevisiae SerRS purified from bacterial overproducing strains may differ in modification and/or conformation compared to the native protein. However, since the kinetic parameters for overexpressed yeast SerRS are very much alike regardless of the overproducing system used (29, 30), the comparison of the substrate binding and catalytic constants for the wild-type and mutant yeast SerRS was performed with proteins isolated from E. coli.
Since the aminoacylation assay did not reveal significant changes in the affinity for tRNA (Table II), but instead showed impaired ATP and serine binding capacities resulting in elevated Km values for both substrates accompanied by decreased turnover rates, the impact of the mutations on the steady-state kinetic parameters were independently determined in the amino acid activation reaction. Mutants SerRS2 (R279C/E281Q/G283A/D288V/W290L) and SerRS3 (R279P/A282R) have no detectable in vitro enzyme activity and could not be used for kinetic studies. This is in agreement with the inability of SES1mut2 and SES1mut3 to substitute for the wild-type SES1 function in vivo. In both mutants Arg279, which aligns with an invariant arginine in the class II synthetases, was replaced with other amino acids. This change probably causes the major inactivating effect in SerRS2 and SerRS3, since this position corresponds to Arg256 in T. thermophilus SerRS, which is in direct contact with the phosphate of the seryl-adenylate. Superposition of the models of the wild-type and SerRS2 motif 2 loops, with the seryl-adenylate analog positioned as in the crystal structure of T. thermophilus SerRS (10), is shown in Fig. 2. In order to determine the contribution of particular amino acid alterations, to enzyme function, the five mutations found in SerRS2 were separated. This was achieved by reconstruction of pSKSES1mut2, carrying the unique NcoI site introduced in SES1 as silent mutations at the position of the Gly260 and Ser261 codons. The PstI-NcoI and NcoI-EagI fragments of pSKSES1mut2 were individually replaced with corresponding fragments excised from pSKSES1wt. The new constructs, named SES1mut6 and SES1mut7, recloned as BamHI cassettes behind the T7 promoter of pET3, were used for the overproduction of SerRS6 (R279C/E281Q/G283A) and SerRS7 (D288V/W290L), respectively. As expected, the mutation of three absolutely conserved residues in motif 2 (see below), render the mutant SerRS6 totally inactive in both the amino acid activation and aminoacylation reactions. The kinetic analysis of SerRS7 showed slightly reduced specificity constants for both ATP and serine, relative to the wild-type enzyme. The affinity for tRNA of this mutant was slightly increased. This is of considerable interest, since the occurrence of tryptophan at position 290 is unique to yeast SerRS, and thus could be expected to be involved in tRNA binding. All other seryl-tRNA synthetases contain an arginine or lysine at this position. The exception is SerRS from A. thaliana, which contains a leucine. This indicates that the replacement of one bulky and hydrophobic amino acid (Trp) for another (Leu) is tolerated.
|
Kinetic analysis of the aminoacylation and amino acid activation reactions revealed that substitution of the class II invariant glycine in SerRS4 (G291V) and SerRS5 (E281D/G291A) has the most dramatic effect on ATP and serine binding. The enzymes exhibit more than an order of magnitude elevated Km for ATP and 35-fold decrease in kcat. The parameters for serine were changed to a similar extent, while the Km for the tRNA (for mutant SerRS5) is almost identical to the wild-type value. Gly291 of yeast SerRS occupies a position that is strictly conserved in the primary structures of all prokaryotic and eukaryotic seryl-tRNA synthetases. Its analogue is not exposed in the active site of the T. thermophilus enzyme, thus it is probably involved in maintaining an overall motif 2 loop conformation (10). As recently discussed by Cusack et al. (13), the occurrence of several glycines in the loop, surrounded by other small residues, may provide the necessary flexibility and reduced steric hindrance to facilitate the conformational switch imposed by tRNA binding. In agreement with this structural data, alteration of the nonpolar Ala282 in yeast SerRS (mutant SerRS10) to serine, another small but polar amino acid, causes the least dramatic change in kinetic parameters.
To single out the effect of the E281D replacement, two mutations in
SES1mut5 were separated by the same experimental
procedure as described above for SES1mut2. Two
new mutant enzymes were obtained by expression of the reconstructed
SES1mut8 and SES1mut9
genes: SerRS8 (E281D) and SerRS9 (G291A), respectively.
Glu281 in yeast SerRS is of special interest since it
aligns with Glu258 in T. thermophilus SerRS,
which is a part of the Arg256-X-Glu peptide. The
side chain conformation of this residue is fixed upon binding of ATP or
adenylate, which is believed to be the first step toward the
stabilization of the motif 2 loop, which is also involved in tRNA
acceptor stem interactions (10, 13). Glu258 is severely
reoriented upon tRNA binding, making direct contact with the
discriminator base. In yeast SerRS, Glu281 is also
apparently involved in ATP binding, since its replacement with Asp,
which only differs in side chain length results in a severe elevation
of Km for ATP (from 18- to 8-fold) accompanied by an
approximately 8-10-fold decreased turnover rate, in the aminoacylation
and PPi exchange reactions, respectively (Table II).
Furthermore, while PPi exchange does not reveal a change of
the Km value of SerRS8 with respect to serine,
aminoacylation shows a 3.6-fold elevation of Km
compared to wild-type. The substitution Gly291 Ala in
SerRS9 has a similar impact on the steady-state kinetic parameters.
Crystallographic and biochemical data have shown that the
specific recognition of cognate tRNASer by prokaryotic
SerRS depends on the mutual interaction of two structural elements that
are unique to the serine system: the -helical coiled-coil (helical
arm) in the enzyme, with the long variable arm of the tRNA (12, 13).
The importance of the variable arm as the crucial identity element for
tRNASer recognition has been demonstrated in both
prokaryotic (44, 45, 46) and eukaryotic (47) systems. Since the N-terminal polypeptide of yeast SerRS, comprising 107 amino acids, can be modeled
into a coiled-coil that resembles the corresponding structural element
of T. thermophilus SerRS (Fig. 2), it seems likely that similar interactions govern tRNASer recognition in yeast.
According to the recent experiments of Cusack et al. (13),
the acceptor stem of tRNASer interacts with the motif 2 loop of the active site, causing a conformational switch in this
protein domain. Although the correct positioning of the 3
end of tRNA
in the active site of yeast SerRS may be disturbed by the alteration of
amino acid residues in the putative loop, the kinetic parameters for
tRNA do not change dramatically. Despite their critical role in
aminoacylation, motif 2 loop residues generally contribute only weakly
to the total tRNA-enzyme binding energy, as revealed by the small
changes in Kd for tRNA of the inactive yeast
aspartyl-tRNA synthetase enzymes mutated in this region (14). Possible
effects of mutations in the active site on the rate of amino acid
transfer, which is commonly rate-limiting for aminoacylation (see
below), were reflected in our experiments by reduced catalytic
efficiencies for both ATP and serine (Table II), as would be expected
for substitution of residues involved in the catalysis of
aminoacyl-adenylate formation.
The Km values for wild-type yeast SerRS, with respect to both ATP and serine, are considerably different when determined by the PPi exchange or aminoacylation reactions. The affinity for serine is more than an order of magnitude higher in the reaction performed in the presence of tRNA rather than in the activation step. Furthermore, the apparent affinity of SerRS for serine, as measured during seryl-adenylate formation, is not significantly changed by amino acid replacements. In contrast, an order of magnitude difference in Km values for wild-type and mutant SerRS, with respect to serine, was observed during aminoacylation. These results are consistent with the possibility that new binding sites for serine are created upon tRNA binding. When the kinetic data presented here are considered in conjunction with structural data for T. thermophilus SerRS, it is clear that SerRS displays tRNA-dependent amino acid recognition, as described previously for E. coli glutaminyl- and tryptophanyl-tRNA synthetases (48).
In agreement with the T. thermophilus SerRS structural studies, the apparent affinity for ATP to the yeast enzyme decreases 6.3-fold in the aminoacylation reaction compared to PPi exchange. This may support the idea that conformational changes in the active site are induced by substrate binding, as recently observed for eukaryotic aspartyl- (14) and prokaryotic seryl-tRNA synthetases (13). Mutational changes in the 278-294 region of yeast SerRS may interfere with the proposed structural flexibility essential to the function of the motif 2 loop. It is also likely that the impaired interaction of mutant SerRS enzymes with ATP and serine makes the amino acid activation reaction, instead of the transfer reaction, rate-limiting. This conclusion is based on the comparison of the kcat values for wild-type and mutant enzymes in PPi exchange and aminoacylation. In contrast to wild-type SerRS, which more rapidly turns over ATP and serine in the absence of tRNA, the catalytic rate constants characterizing the most affected mutant proteins (SerRS4 and SerRS5) are comparable in the activation and aminoacylation reactions. In summary, the results of in vivo and in vitro functional analyses of yeast SerRS mutants, combined with structural data for theT. thermophilus and E. coli enzymes which allowed the prediction of yeast SerRS structure, strongly suggest that seryl-tRNA synthetase from S. cerevisiae structurally and functionally resembles its prokaryotic counterparts in the active site.
We are indebted to Nenad Ban, Michael Ibba, and Sanja Sever for critical discussions.