ABSTRACT
Editing of errors in amino acid selection by an aminoacyl-tRNA synthetase prevents attachment of incorrect amino acids to tRNA, thereby greatly enhancing accuracy of translation of the genetic code. Editing of the non-protein amino acid homocysteine, a frequent type of an error-correcting process, involves reaction of the side chain
sulfhydryl group of homocysteine with its activated carboxyl group forming a
cyclic thioester, homocysteine thiolactone. Here, it is shown that isoleucyl-tRNA synthetase (IleRS), which occasionally misactivates homocysteine
in vitro
and
in vivo
, catalyzes reactions of activated isoleucine with organic thiols (analogues of
the side chain of homocysteine). That these enzymatic reactions occur between Ile-tRNA
Ile
or Ile-AMP (bound in the synthetic sub-site) and a thiol (an analogue of the side chain of homocysteine, bound in the editing sub-site), indicates that the two sub-sites are physically close on the surface of IleRS,
forming a single synthetic/editing active site of the enzyme. Although IleRS
-
Val-AMP undergoes thiolysis as efficiently as do IleRS
-
Ile-AMP and IleRS
-
Ile-tRNA
Ile
, IleRS
-
Val-tRNA
Ile
does not react with thiols. These and other data suggest that the mischarged
valine residue in IleRS-Val-tRNA
Ile
is, most likely, positioned off the enzyme.
Proofreading or editing mechanisms are an essential part of biological
information transfer processes, including translation (reviewed in
1
-
3
). The non-protein amino acid homocysteine (Hcy), an obligatory precursor of
methionine in all cells, poses a major accuracy problem for the protein
biosynthetic apparatus. Three closely related class I aminoacyl-tRNA synthetases (AARS) (reviewed in
4
), MetRS, IleRS and LeuRS, misactivate Hcy
in vitro
(
5
,
6
) at a frequency exceeding the frequency of translational errors
in vivo
(
2
). Two other synthetases, class I ValRS (
5
,
7
) and class II LysRS (H. Jakubowski, unpublished), misactivate Hcy less
efficiently. These five enzymes possess an efficient editing mechanism which
prevents misincorporation of Hcy into tRNA (
5
,
8
) by destroying the Hcy-AMP intermediate. The editing pathway involves reaction of the side chain sulfhydryl group of Hcy with its activated carboxyl group yielding a cyclic
thioester, Hcy thiolactone (
5
-
7
). For at least one synthetase, the editing reaction occurs in the same active
site that carries out the synthetic reaction, as demonstrated by structure-function studies of MetRS (
9
). Editing reactions catalyzed by three class I AARS, MetRS (
10
-
13
), IleRS (
14
) and LeuRS (
14
), have been shown to occur
in vivo
.
Recently, a class I synthetase, ArgRS, has been shown to catalyze synthesis of dipeptides Arg-Cys and Arg-Hcy from Arg-tRNA
Arg
and a corresponding thioamino acid (
15
). The mechanism involves a nucleophilic attack of the thiolate of cysteine (or
homocysteine) on the ester bond in Arg-tRNA
Arg
. The resulting thioesters S-arginyl-cysteine and S-arginyl-homocysteine are not observed as discrete intermediates because of their
rapid rearrangement to form a stable peptide bond. The involvement of thioester intermediates in the
synthesis of Arg-Cys and Arg-Hcy is supported by the observation that thioesters of arginine do
form with cysteine derivatives that do not have a free amino group, such as
N
-acetyl-L-cysteine and 3-mercaptopropionate. Although ArgRS does not appear to have an editing mechanism (
15
), its ability to catalyze the formation of thioesters of arginine is
reminiscent of the formation of the thioester homocysteine thiolactone in
proofreading reactions of other AARSs (
5
-
7
).
IleRS, which has an editing mechanism directed against Hcy
in vivo
(
14
) is also capable of forming Ile-Cys from Ile-tRNA
Ile
and cysteine (
15
). However, the mechanism of Ile-Cys formation has not been determined and it is not known whether IleRS
can also catalyze the formation of thioesters of isoleucine. If it occured, the
synthesis of thioesters of isoleucine by IleRS would mimic editing reaction, in
which misactivated Hcy is cyclized to the thioester homocysteine thiolactone by
the enzyme (see Fig.
4
). In addition to editing Hcy (
5
,
14
), IleRS is also capable of editing valine
in vitro
(
16
,
17
). Editing of valine was suggested to involve a site separate from the synthetic
site (
18
). It is not known whether noncognate valine and Hcy are edited in the same or
separate sites of IleRS. One way to answer this question is to compare
reactivities toward thiols of aminacyl ester bonds in cognate IleRS-Ile-tRNA
Ile
and noncognate IleRS-Val-tRNA
Ile
complexes. Similar reactivities with thiols of both complexes would suggest that the Val
and Ile residues (bound to tRNA
Ile
) occupy the same site on the enzyme. If the Val residue in mischarged Val-tRNA
Ile
occupies a different site, the mischarged tRNA should not react with thiols.
Here, it is shown that IleRS catalyses reactions of cognate Ile-tRNA
Ile
and Ile-AMP, as well as noncognate Val-AMP, but not Val-tRNA
Ile
, with organic thiols (analogues of the side chain of Hcy) in reactions
mimicking editing of Hcy by the enzyme. The data indicate that two major
physiologic functions of IleRS, formation of Ile-tRNA and editing of inadvertently misactivated Hcy, reside in one active
site of the enzyme. A misacylated valine residue in an IleRS-Val-tRNA
Ile
complex, which may form transiently during editing (
19
), appears not to be bound on the enzyme but is still hydrolyzed off the tRNA.
Plasmids containing the genes for
Escherichia coli
IleRS (
18
,
20
) and
Bacillus stearothermophilus
ValRS (
21
) were obtained from P. Schimmel and E. Schmidt. Plasmids were over-expressed in
E.coli
strain JM101 and used as a source of AARS. Cells for enzyme purifications were
obtained from overnight cultures (usually 400 ml, yielding ~2 g cells) grown at 37oC in LB medium containing 100 [mu]g/ml ampicillin.
Escherichia coli
IleRS (
20
) and
B.stearothermophilus
ValRS (
21
,
22
) were purified to homogeneity from the overproducing strains using standard
procedures (
20
,
22
).
[
14
C]Ile-tRNA
Ile
was prepared from aminoacylation mixtures (0.1 ml) containing 50 mM HEPES, pH 7.4, 10 mM MgCl
2
, 0.1 mM EDTA, 2.5 mM ATP, 10 [mu]M tRNA
Ile
from
E.coli
(1600 pmol/
A
260
, Subriden RNA), 45 [mu]M [
14
C]isoleucine (306 Ci/mol) (NEN) and 0.1 [mu]M
E.coli
IleRS. [
14
C]Val-tRNA
Ile
was prepared by mischarging (
22
)
E.coli
tRNA
Ile
with 25 [mu]M [
14
C]valine (285 Ci/mol) (Amersham) using
B.stearothermophilus
ValRS (
18
,
21
,
22
). After 15 min at 37oC the charged tRNA was purified by phenol extraction, and recovered by
precipitation with ethanol. The precipitate was washed several times with 70%
ethanol to remove traces of free [
14
C]amino acids, dissolved in 0.1 ml glass-distilled water, and stored at -20oC.
The reactions were carried out at 37oC in 0.1 M K-HEPES, pH 7.4, 10 mM MgCl
2
, 0.2 mM EDTA. In one set of experiments, the disappearance of [
14
C]aminoacyl-tRNA
Ile
was monitored by trichloroacetic acid precipitation. In another set of
experiments in which all forms of [
14
C]amino acids were followed, the aliquots were analyzed by TLC.
IleRS-[
14
C]Ile-AMP was prepared and incubated with thiols in the same mixture. The
reactions were carried out at 37oC in mixtures containing 1 [mu]M Ile-RS, 15 [mu]M [
14
C]isoleucine, 1 mM ATP, 0.1 M K-HEPES, pH 7.4, 10 mM MgCl
2
, 0.2 mM EDTA, 50 mM thiol and 5 U/ml inorganic pyrophosphatase (Sigma). The
reactions were monitored by TLC. Under these conditions, reactions with L-cysteine, cysteamine and dithiothreitol (DTT) were completed within 2 min,
and with D,L-Hcy in 16 min. With 2-mercaptoethanol (2-ME), reactions were 50% completed in 16 min. Similar reactions were carried out with IleRS-[
14
C]Val-AMP prepared in reaction mixtures containing 15 or 150 [mu]M [
14
C]valine instead of [
14
C]isoleucine.
TLC separations were carried out on cellulose plates from Kodak using butanol-acetic acid-water (4:1:1, v/v) as a solvent (
5
,
15
,
23
). Authentic isoleucine and valine (Sigma) standards were co-chromatographed with samples and visualized by staining with ninhydrin.
TLC plates were autoradiographed using Reflectiontm (NEN) autoradiography film.
Most AARS (
2
,
3
,
23
), including IleRS (
19
,
24
,
25
), catalyze deacylation reactions which result in discharging of an amino acid
from aminoacyl-tRNA in the absence of AMP and PP
i
. As shown in Table
1
, enzymatic deacylation of Ile-tRNA was accelerated by L-cysteine and DTT but not by D-cysteine, D,L-Hcy and 2-ME, indicating specificity of the reaction with
respect to which thiols can participate. This also suggests that stimulation is
not merely due to reactivation of essential -SH groups of the enzyme. When reactions were carried out at different
thiol concentrations, saturation kinetics were observed (Fig.
1
). Control experiments have shown that L-cysteine and DTT did not stimulate nonenzymatic deacylation of Ile-tRNA. The half-life for Ile-tRNA
Ile
(pH 7.4, 37oC) in the absence and presence of 0.125 M L-cysteine or DTT was 180 min. Another nucleophile, hydroxylamine, did
not accelerate the enzymatic deacylation (Fig.
1
), but did stimulate nonenzymatic deacylation of Ile-tRNA
Ile
(the half-life for Ile-tRNA
Ile
was 110 min in the presence of 125 mM hydroxylamine). These results suggest the presence on the enzyme of a specific site (`-SH subsite'), which affects enzymatic deacylation of Ile-tRNA.
To analyze products of enzymatic deacylation of Ile-tRNA, aliquots of reaction mixtures containing IleRS, [
14
C]Ile-tRNA
Ile
and various thiols were subjected to TLC. In the absence of thiols, isoleucine
was a major product (lane 1, Fig.
2
A, B and C). In the presence of L-cysteine (lane 4, Fig.
2
A; lane 2, Fig.
2
B), DTT (lane 3, Fig.
2
A), cysteamine (lane 5, Fig.
2
A; lane 3, Fig.
2
B), 3-mercaptopropionate,
N-
acetyl-L-cysteine, L-cysteine methyl ester (lanes 4, 7 and 8 respectively, Fig.
2
B), and D-cysteine (lane 7, Fig.
2
A), new products formed, suggesting that these thiols bind to the -SH subsite and react with Ile-tRNA. Deacylations in the presence of L-cysteine (lane 4, Fig.
2
A; lane 2, Fig.
2
B) and D,L-Hcy (lane 6, Fig.
2
A) also yielded isoleucine. These observations indicate that, in addition to
binding to the -SH subsite in the editing mode common for all thiols (see Fig.
4
B.iv), L-cysteine and Hcy bind also in the synthetic mode as substrates (
5
) with their side chain -SH unable to react with Ile-tRNA
Ile
(Fig.
4
A.i). Cysteine, in addition to Hcy, is a noncognate substrate which is
misactivated and efficiently edited by IleRS (
5
). Control reactions showed that alanine (lane 5, Fig.
2
B), serine (lane 9, Fig.
2
B) and
S
-methyl-L-cysteine (lane 6, Fig.
2
B) did not react with Ile-tRNA
Ile
.
Formation of isoleucyl-dipeptides is consistent with the following mechanism. The ester bond in
Ile-tRNA
Ile
undergoes thiolation by cysteine (or similar thiol amino acid). The resulting
thioester is not observed as an intermediate because of the rapid rearrangement to form a stable peptide bond. Facile intramolecular reaction results from the favorable geometric arrangement of the [alpha]-amino group of cysteine with respect to the thioester bond (
26
). Support for this mechanism comes from the observations that the -SH group is required for the reaction (alanine, serine and
S-
methyl-cysteine do not react with Ile-tRNA
Ile
, Fig.
2
B) and that isoleucine thioesters do indeed form with cysteine derivatives that
do not have a free amino group, such as 3-mercaptopropionate and
N
-acetyl-L-cysteine. A similar mechanism accounts for synthesis of the
dipeptide Arg-Cys by Arg-RS (
15
).
Products of nonenzymatic deacylation of Ile-tRNA in the presence of cysteine and DTT were also analyzed by TLC. The
half life of Ile-tRNA
Ile
(180 min) was not affected by thiols as stated above. More than 95% isoleucine
and <5% Ile-Cys or Ile-DTT formed during these nonenzymatic reactions. These results
further indicate that reactions of thiols with Ile-tRNA
Ile
are catalyzed by IleRS.
To test whether binding sites for isoleucine and thiols interact, thiol-dependent deacylation reactions were carried out in the absence and
presence of exogenous isoleucine. The cysteine- and DTT-dependent deacylations of Ile-tRNA
Ile
were 2-fold slower in the presence of isoleucine than in its absence (Table
1
). As shown in Figure
2
C, exogenous isoleucine prevented formation of [
14
C]Ile-DTT (thio)ester (compare lanes 6 and 5) and [
14
C]Ile-Cys dipeptide (compare lanes 4 and 3) from [
14
C]Ile-tRNA
Ile
. This suggests that free isoleucine displaces the [
14
C]Ile residue in [
14
C]Ile-tRNA
Ile
from the thiol-reactive site on IleRS either to another site on the enzyme (perhaps the
editing site for valine) or off the enzyme. Because [
14
C]Ile-tRNA
Ile
is deacylated by IleRS even in the presence of free isoleucine or valine (
24
; see also Table
1
), it follows that under these conditions the [
14
C]Ile residue in an [
14
C]Ile-tRNA
Ile
-IleRS complex is positioned off the enzyme: excess of free isoleucine or valine is expected to prevent binding of the [
14
C]Ile residue in [
14
C]Ile-tRNA
Ile
-IleRS to the enzyme. Alternatively, free isoleucine might bind to a second Ile-binding site which in turn would either displace a thiol or affect
reactivity of the ester bond in IleRS-bound Ile-tRNA
Ile
. However, IleRS, a monomeric enzyme (
20
), is very unlikely to possess two Ile-binding sites. That thiols do not bind at the isoleucine binding site of
IleRS is indicated by the inability of thiols to inhibit the aminoacylation
reaction catalyzed by IleRS (see below).
The rate of IleRS-dependent deacylation of [
14
C]Val-tRNA
Ile
, in contrast with the rate of deacylation of [
14
C]Ile-tRNA
Ile
, was not accelerated by DTT or L-cysteine (Table
1
). Saturating concentrations of Hcy and valine, noncognate substrates which are misactivated and edited by IleRS (
5
,
16
), did not prevent enzymatic deacylation of [
14
C]Val-tRNA
Ile
(Table
1
). TLC analyses demonstrated that [
14
C]valine was the only product of the deacylation reaction in the presence or
absence of 100 mM L-cysteine, D,L-Hcy and DTT (not shown). The sensitivity of these analyses was such
that <= 5% of Val-Cys or Val-DTT would have been detected if present. These results suggest
that the charged valine and isoleucine residues in Val-tRNA
Ile
-IleRS and Ile-tRNA
Ile
-IleRS complexes respectively, are positioned differently. The [
14
C]isoleucine residue in [
14
C]Ile-tRNA
Ile
-IleRS is positioned on the enzyme and therefore able to react with thiols; excess of free
isoleucine or valine displaces the [
14
C]isoleucine to the off enzyme position thereby preventing reaction with thiols.
The valine residue in Val-tRNA
Ile
-IleRS is, most likely, positioned off the enzyme and is therefore
unreactive towards thiols. Because neither exogenous valine nor isoleucine significantly inhibits deacylation of Val-tRNA
Ile
by IleRS (Table
1
), it is unlikely that the valine residue in Val-tRNA
Ile
-IleRS is positioned on the enzyme. A proposed second site for editing of mischarged valine by
IleRS (
18
) would be functionally equivalent to the off enzyme position.
Reactions of IleRS
Ile-AMP and IleRS
Val-AMP with thiols
Thiols reacted also with IleRS-bound Ile-AMP or Val-AMP, yielding thioesters and peptides. For example, IleRS-Ile-AMP reacted with DTT and 2-ME to yield (thio)esters Ile-DTT and Ile-(2-ME). Similarly, IleRS-Val-AMP reacted
with DTT to yield (thio)ester Val-DTT. Peptide bond formation occurred with cysteine to yield either Ile-Cys or Val-Cys. The reactions with adenylates proceeded at about the same
rates as enzymatic reactions with Ile-tRNA
Ile
and exhibited similar thiol specificity: cysteine and DTT reacted faster than 2-ME (not shown). Similar reactivities of IleRS-bound aminoacyl-adenylates and Ile-tRNA
Ile
towards thiols indicate that the amino acid residues of Ile-AMP, Val-AMP and Ile-tRNA
Ile
occupy the same sub-site in the active site of the enzyme.
Reactions of thiols with isoleucine occurred also in complete aminoacylation mixtures (Fig.
3
C). This led to transient formation of Ile-tRNA followed by its enzymatic thiolysis in reaction mixtures containing 20 mM L-cysteine or DTT (Fig.
3
A) or 200 mM D-cysteine (Fig.
3
B). Although 2-ME reacted with isoleucine in complete aminoacylation mixtures to some
extent (compare lanes 2 and 6 in Fig.
3
C), the reaction was very inefficient and did not lead to transient
aminoacylation kinetics (Fig.
3
A and B). Importantly, the rate of the aminoacylation reaction was not inhibited
by thiols, even when present at 0.2 M (Fig.
3
A and B), indicating that thiols and isoleucine bind to different (sub)sites on
the enzyme.
The ability of IleRS to catalyze synthesis of thioesters of isoleucine is a consequence of the editing function of the enzyme and can be rationalized
by the following model (Fig.
4
). When misactivated Hcy is in the active site of IleRS, its side chain -SH group occupies a sub-site (an -SH sub-site) next to its carboxyl carbon (Fig.
4
A.ii). This sub-site is presumed to be vacant when a cognate amino acid is in the active
site. Filling the -SH sub-site by providing the -SH function in
trans
, i.e., on another molecule (Fig.
4
B.iv), leads to the formation of a thioester of a cognate amino acid (Fig.
4
B.v.). That this reaction occurs between Ile-tRNA
Ile
or Ile-AMP (bound in the synthetic sub-site) and an organic thiol, an analogue of the side chain of Hcy
(bound in the editing sub-site), indicates that the two sub-sites are intimately close on the surface of the enzyme, forming a
single synthetic/editing active site. Editing of valine can also be
accommodated in the same site: however, while a mischarged valine residue
(attached to tRNA
Ile
) that undergoes editing appears to be positioned off the surface of the enzyme,
the aminoacyl bond is still accesssible to hydrolysis by the enzyme. Thus,
editing occurs when an activated amino acid is no longer bound in the synthetic
sub-site or when the editing sub-site is occupied by a thiol.
The model shown in Figure
4
is also consistent with the data obtained by others (
18
). Implication of this model is that any mutation that affects binding of the
side chain of the amino acid substrate during synthetic reaction will not
affect editing reaction: the side chain of an amino acid that undergoes editing
is not bound in the synthetic sub-site. This explains the behavior of a G56A mutant of IleRS that has lost
the ability to discriminate between isoleucine and valine in the synthetic
reaction (mostly due to the loss of efficient binding of the side chain of the
substrate isoleucine) but was completely active in editing of valine. A related
mutant, G56P, inactive in the synthetic reaction with either isoleucine or
valine, was active in editing of valine (
18
). The lack of synthetic activity of the G56P IleRS mutant is, most likely, due
to a total loss of the ability to bind amino acid substrates; this, according
to the model shown in Figure
4
, will not affect editing activity of the enzyme. Another mutant IleRS, F570S,
has lost the ability to efficiently bind isoleucine and valine during the
synthetic reaction, but was still unimpaired in the editing reaction (
18
), again consistent with the model shown in Figure
4
.
I thank Paul Schimmel and Eric Schmidt for plasmids containing the genes for
IleRS and ValRS. This work was supported by grants from the American Cancer
Society (NP-904) and National Science Foundation (MCB-9218358). I am grateful to Manny Goldman for his critical
discussions.
REFERENCES


