Structure, Vol. 13, 1799–1807, December, 2005, ª2005 Elsevier Ltd All rights reserved.DOI 10.1016/j.str.2005.08.013
Structural Basis for Discrimination
of L-Phenylalanine from L-Tyrosine
by Phenylalanyl-tRNA Synthetase
Olga Kotik-Kogan,1,3Nina Moor,2,3
Dmitry Tworowski,1and Mark Safro1,*
1Department of Structural Biology
Weizmann Institute of Science
2Institute of Chemical Biology
and Fundamental Medicine
Aminoacyl-tRNA synthetases (aaRSs) exert control
over the faithful transfer of amino acids onto cognate
tRNAs. Since chemical structures of various amino
acids closely resemble each other, it is difficult to dis-
criminate between them. Editing activity has been
evolved by certain aaRSs to resolve the problem. In
this study, we determined the crystal structures of
complexes of T. thermophilus phenylalanyl-tRNA syn-
thetase (PheRS) with L-tyrosine, p-chloro-phenylala-
nine, and a nonhydrolyzable tyrosyl-adenylate analog.
The structures demonstrate plasticity of the synthetic
site capable of binding substrates larger than phenyl-
alanine and provide a structural basis for the proof-
reading mechanism. The editing site is localized at the
B3/B4 interface, 35 A˚from the synthetic site. Glub334
plays a crucial role in the specific recognition of the
tural data and tyrosine-dependent ATP hydrolysis en-
hanced by tRNAPheprovide evidence for a preferential
posttransfer editing pathway in the phenylalanine-
tant for the function of living cells and is intimately asso-
ciated with the proper operation of correctly assembled
macromolecules. The aaRSs play a crucial role in the
maintenance of faithful translation, promoting close
control over the two-step aminoacylation reaction and
thus establishing a linkage between the cognate amino
acid and three nucleotides of the tRNA anticodon(s). In
the first step, aaRS activates the amino acid and forms
an aminoacyl-adenylate intermediate. In the second
step, the amino acid moiety is transferred to the 30-
terminal ribose of tRNA. The rate of misincorporation
of noncognate amino acids in vivo is approximately
1 out of 104–105reactions (reviewed in Jakubowski
and Goldman, 1992). This brings up the question of
how such a level of fidelity can be achieved for those
aaRSs that have to distinguish between amino acids
with very similar chemical structures. The problem has
been resolved with the discovery of proofreading activ-
ity associated with a distinct active site at which misac-
tivated aminoacyl-adenylate (pretransfer editing) (Fersht,
1977) or misaminoacylated tRNA (posttransfer editing)
(Eldred and Schimmel, 1972) are hydrolyzed.
The ‘‘double-sieve’’ model for aminoacylation and ed-
iting proposed by Fersht (1977) has been visualized in
synthetic catalytic site, which is suggested to act as
‘‘coarse sieve,’’ binds isosteric or smaller amino acids
and discards those that are larger than or dissimilar to
the cognate one. The smaller substrates form incorrect
aminoacyl-adenylates or aminoacyl-tRNAs hydrolyzed
at the editing site, which acts as ‘‘fine sieve.’’ The avail-
ability of the ‘‘fine sieve’’ for IleRS derives from the de-
tection of a second binding site for Val (Nureki et al.,
1998). The synthetic and editing sites are widely spaced
from each other (for a review, see Jakubowski, 2004).
Even though the double-sieve model allows us to ex-
plain the proofreading in IleRS, ValRS, and LeuRS, it
could not give insight into the editing activities of
MetRS, AlaRS, and PheRS, shown to misactivate amino
acids larger than the correct ones (ethionine, Ser, and
Tyr, respectively) (Tsui and Fersht, 1981; Lin, et al.,
1983; Jakubowski and Goldman, 1992).
Untilrecently, theeditingactivityofclass IIaaRSswas
not characterized sufficiently. Structural and biochemi-
cal studies of ThrRS have revealed that the zinc ion di-
rectly involved in amino acid recognition is responsible
for rejection of isosteric Val at the activation step (San-
karanarayanan et al., 2000). The system-specific N2 do-
main employed to correct misacylation of tRNAThrwith
Ser is 39 A˚away from the synthetic site (Dock-Bregeon
et al., 2000). A seryl-adenylate analog (Ser-AMS) has
been shown to bind in the active site of the isolated N2
domain, although the existence of pretransfer hydrolytic
activity is not evidenced by these data (Dock-Bregeon
et al., 2000, 2004). Based on comparative modeling,
a similar architecture of the editing domain was pre-
dicted for AlaRS (Beebe et al., 2003), shown earlier to
hydrolyze misactivated Gly and Ser (Tsui and Fersht,
1981). In E. coli ProRS, editing activity directed toward
misactivated Alaandmisacylated Ala-tRNAProisassoci-
ated with an INS domain located between signature
motifs 2 and 3 and not maintained in the sequences of
eukaryote/archaeon-like ProRSs (Wong et al., 2003).
The editing domains of class II aaRSs are more diverse
in amino acid sequences and distinguishing features
of their folds in comparison with those of the class I
A double-rejection mechanism of noncognate Tyr, us-
ing pretransfer and posttransfer hydrolysis, has been
suggested for yeast PheRS (Lin et al., 1983, 1984), the
most complex representative of class II aaRSs. During
evolution, the (ab)2subunit organization of cytoplasmic
PheRS is markedly conserved from prokaryotes to eu-
karyotes. The 3D structure of T. thermophilus PheRS
and its complexes with functional substrates (Mosyak
3These authors contributed equally to this work.
et al., 1995; Goldgur et al., 1997; Reshetnikova et al.,
1999; Fishman et al., 2001) reveals that PheRS is both
structurally and functionally an (ab)2heterodimer; the
but it exerts control over the recognition and binding of
cognate tRNAPhe. Three domains of the b subunit (B2,
B4, and B5) do not contact tRNAPhe, and, until recently,
there was no clear notion of their functional role (Safro
et al., 2004). As proposed in early studies of yeast
PheRS, the pretransfer hydrolysis triggered by native
tRNAPhemakes a major contribution to the Tyr rejection
also misactivates Tyr and further corrects such errors
(Roy et al., 2004). Based on structural modeling and an
in vivo genetic screen, a few residues of the E. coli
PheRS b subunit (from the B3/B4 domains) presumably
associated with the editing activity of the enzyme were
identified. The editing activity of E. coli PheRS against
Tyr occurs predominantly at the posttransfer stage.
For yeast PheRS, however, a major role was assigned
to the pretransfer editing stage.
This work presents the structural basis of editing ac-
tivity of T. thermophilus PheRS directed toward misacti-
vated Tyr. The crystal structures of the bacterial enzyme
complexes with noncognate Tyr, an unnatural para-
substituted Phe analog, p-Cl-Phe, and a nonhydro-
lyzable analog of tyrosyl-adenylate (Tyr-AMS) have
been determined. The key feature of the complexes is
binding ofTyrinboth thesynthetic active siteand an ed-
iting site localized on the b subunit. Protein residues in-
volved in the sculpting of the editing site were identified.
Discrimination of the editing site against cognate Phe is
ensured by specific recognition of the p-hydroxyl group
of Tyr. A second key feature is the binding of Tyr-AMS
exclusively in the synthetic site, in close proximity to
the position of phenylalanyl-adenylate. The structural
data combined with a kinetic study of the substrate
properties of Tyr provide evidence for a preferential
posttransfer editing mechanism in PheRS, i.e., deacyla-
tion of the incorrectly charged Tyr-tRNAPhe.
Tyr as Substrate of PheRS
The ability of L-Tyr to act as a substrate of T. thermophi-
noacylation kinetics. No charging of tRNAPhewith the
noncognate (14C-labeled) substrate was detected. To
further study misactivation, the commercially available
Tyr was purified from contaminating Phe and tested
for inhibition activity in the tRNAPheaminoacylation re-
action (Table 1). The inhibition constant of the 4-fold re-
crystallized Tyr is three orders of magnitude higher than
the KMvalue of Phe (1.8 mM). A similar large difference in
the affinities of noncognate and cognate amino acids
was observed for yeast PheRS (Lin et al., 1983).
Activation of Tyr by T. thermophilus PheRS was as-
sayed by ATP hydrolysis. The apparent rate of con-
sumption in definite conditions (with concentrations of
ATP R 50 mM and Tyr R 300 mM) was higher than that
in the presence of cognate substrate (Figure 1). Addi-
tion of inorganic pyrophosphatase significantly (3.7-
fold) enhanced Tyr-dependent ATP hydrolysis, but it
only slightly (1.2-fold) enhanced the Phe-dependent
reaction. Higher sensitivity of the noncognate system
to pyrophosphate concentration was previously re-
vealed for yeast PheRS (Lin et al., 1983) and is explained
by different rates of formation and pyrophosphorolysis
for phenylalanyl- (Phe-AMP) and tyrosyl-adenylates
(Tyr-AMP). tRNAPhe(native T. thermophilus or in vitro-
transcribed E. coli) induced a 1.4-fold enhancement of
the ATP hydrolysis stimulated by Tyr. This is a much
Table 1. Inhibition Properties of L-Tyr and Synthetic Ligands
Kivaluesof competitive inhibitors were determined inthe aminoacy-
lation reaction catalyzed by T. thermophilus PheRS with respect to
the substrate indicated in parentheses. The KMvalue for L-Phe is
1.8 mM, and the value for ATP is 130 mM.
Figure 1. ATP Hydrolysis by T. thermophilus PheRS
(A and B) (A) ATP hydrolysis stimulated by Phe. (B) ATP hydrolysis
stimulated by noncognate Tyr. The reaction mixture contained 100
mM ATP, 0.2 mM PheRS, 300 mM amino acid, and other constituents
as described in Experimental Procedures. The reaction was per-
lus tRNAPhe(2), 4 U/ml inorganic pyrophosphatase (3), tRNAPheand
smaller effect compared to the cognate reaction and is
independent of tRNAPheconcentration (in the range of
0.5–10 mM). The activating effect of pyrophosphatase
on Tyr-stimulated ATP hydrolysis was significantly
suppressed in the presence of tRNAPhe, indicating that
tRNA-dependent ATP consumption is the prevailing re-
action pathway. tRNAPheshortened by one nucleotide
from the 30end and hence deprived of charging activity
failed to activate ATP hydrolysis (data not shown). Thus,
the tRNA-dependent editing activity against Tyr re-
vealed for yeast (Lin et al., 1984) and E. coli PheRSs
(Roy et al., 2004) is also characteristic of the T. thermo-
Three noncognate L-amino acids (Ile, His, and Trp)
were tested additionally for their ability to inhibit amino-
acylation of tRNAPheand act as substrates in activation
and aminoacylation reactions. The Ki value for Trp
(4.6 mM) was found to be closely similar to that of Tyr,
while the other two amino acids showed no inhibition ef-
fect even at 20 mM concentration. None of the three
amino acids stimulated ATP hydrolysis (in the absence
or presence of tRNAPhe) or can be attached to tRNAPhe.
These results suggest that the editing (or hydrolytic) site
of PheRS specifically rejects only Tyr, while the other
naturally occurring amino acids are efficiently discrimi-
nated in the synthetic active site.
The Structure of PheRS Complexed with Tyr
We have determined the crystal structure of PheRS with
Tyr at 2.7 A˚resolution (see Table 2, PDB ID code 2AMC).
A clear electron density (Figure 2A) on the (Fobs2 Fcalc)
map may be unambiguously attributed to the Tyr mole-
cule locatedatthesyntheticsite.Theanchoring ofthea-
NH3+group of Tyr is achieved by its interactions with the
Ogatom of Sera180 and with a well-ordered water mol-
ecule, S6 (see Table S1 in the Supplemental Data avail-
able with this article online), which in turn is located
at an H bonding distance from Ogof Thra179, Ne2of
Glna218, and Oe1of Glua220. The network is completed
by the interaction of Sera180 with Glua220. The
a-carboxylate of the substrate makes H bonds with the
side chains of Trpa149, Hisa178, and the class II invari-
Table 2. Crystallographic Data and Refinement Statistics
Unit cell parameters (A˚)
Completeness (last shell) (%)
Rmergea(last shell) (%)
a = b = 173.20; c = 138.23
a = b = 173.64; c = 139.81
a = b = 172.98; c = 138.51
Resolution range (A˚)
Number of protein atoms
Number of ligand atoms
Number of water atoms
Bond distance (A˚)
Bond angle (º)
hkljI(hkl),whereIj(hkl)and<Ij(hkl)>aretheintensity ofmeasurement jandthemeanintensity forthereflection
with indices hkl, respectively.
bR factor =P(Fobs2 kFcalc)/P
hklFobs, where k is a scale factor, and Rfreeis the R factor for the test set of reflections.
Figure 2. The Synthetic Active Site of the T. thermophilus PheRS with Bound Ligands
(A–C) The T. thermophilus PheRS is orange, and the ligands are slate. The electron density maps calculated as described in Experimental Pro-
cedures are contoured at 2.5s, 3s, and 3s for (A) Tyr, (B) Tyr-AMS, and (C) p-Cl-Phe, respectively. Dashed lines show direct and water-mediated
H bonds. The water molecules are depicted by red/magenta spheres.
Structural Basis of PheRS Editing against Tyrosine
Stereochemical discrimination between the cognate
substrate and Tyr is not easy to achieve, as the latter
has the aromatic side chain of Phe with only one extra
OH group, which makes the phenyl ring less hydropho-
bic than that of Phe. However, being a neutral polar
amino acid, Tyr may participate in various hydrophobic
interactions. More specifically, Tyr participates in aro-
matic-aromatic interactions that characterize the amino
acid binding site of T. thermophilus PheRS. The binding
mode of Tyr, much like that of Phe (Reshetnikova et al.,
1999; Fishman et al., 2001), is also characterized by
‘‘edge-to-face’’ interactions with Phea258 and Phea260.
The displacement of Tyr for about 0.8–1.0 A˚toward the
entrance of the active site cleft as compared to the po-
sition of the Phe moiety in the complex with Phe-AMP
or to that of Phe itself does not cause steric clashes of
the Tyr OH group with the Vala261 and Alaa314. The
Tyr is additionally stabilized by interactions of the OH
group with water S98 and the carbonyl oxygen of
Alaa314. The water molecules have been located reli-
ably, since they are observed independently (i.e., omit-
ted from the starting model) in almost all PheRS com-
plexes and occupy nearly the same positions in these
complexes. The position of S98 is additionally fixed by
interactions with the main chain oxygen of Phea260
and Oe1of Glna183.
A second Tyr was identified at the interface between
the B3 and B4 domains (Figure 3A). The B3 domain re-
sembles an a-b sandwich with a layer of two a helices
packed against a four-stranded antiparallel b sheet.
There are three insertions in this domain; the largest
insertion (residues 265–328) constitutes domain B4. In
complexes with Phe-AMP (Fishman et al., 2001) and
Phe alone (unpublished data), no electron density was
detected in the cavity between domains B3 and B4.
The distinct pocket visualized on the b subunit is spe-
cific for Tyr and is most likely related to the editing activ-
ity ofPheRS. By analogy with the synthetic site, the edit-
ing site is designed to bind the aromatic amino acid: the
phenyl ring of Pheb360 and the side chain of Prob259
provide ‘‘edge-to-face’’ interactions with Tyr. Pheb360
is strictly conserved in eubacterial PheRSs, while, in
several cases, Prob259 is replaced with Ile or Leu capa-
ble of participating in hydrophobic interactions. The an-
choring of the Tyr OH group is achieved by its inter-
actions with the Oe1of Glub334 and the main chain
amide of Glyb315 (Figure 3B). A remarkable structural
peculiarity of the editing site is the appearance of the
invariant hydrophilic Glub334 in a fully hydrophobic en-
vironment formed by Pheb213, Leub215, Pheb263,
Ileb300, Alab314, Glyb315, Alab336, and Pheb338. Con-
sequently, theTyraromatic ring isplacedinto thehydro-
phobic environment. Thus, Glub334 plays a critical role
in specific recognition of the Tyr moiety.
The main chain of Tyr forms both direct and water-
Whereas water molecule S99 bridges the amino group
of Tyr to the hydroxyl of Thrb354 and the main chain am-
interaction with Hisb261 and extensive H bonding medi-
ated by three water molecules (see Table S1). The car-
boxyl group of Tyr interacts through the S14 molecule
with Ogof Thrb249, through S80 with Oe2of Glub323,
and uses S112 to make contact with Od1of Asnb250.
The distance between the Caatoms of two Tyr mole-
cules located in the synthetic and the editing sites is
about 35 A˚. It is similar to the distances that occur in
class I IleRS (34 A˚), ValRS (35 A˚), LeuRS (38 A˚), and class
II ThrRS (39 A˚). No electron density was observed in the
area in which the Tyr moiety may be anchored by the
main chain carbonyl of Leub311 and the side chain hy-
droxyl of Ser/Thrb322, as was expected by comparative
structural modeling (Roy et al., 2004).
The Structure of PheRS Complexed
The crystal structure of the PheRS complex with 50-O-
drolyzable Tyr-AMP analog, determined at 2.6 A˚resolu-
tion (PDB ID code 2ALY) reveals clear density for the
ligand in the synthetic site only (see Table 2 and Fig-
ure 2B). In Tyr-AMS, the sulphamoyl group replaces
Figure 3. View of the Editing Site of T. thermophilusPheRS Showing
the Interactions with L-Tyrosine
(A) PheRS editing site (colored green) with bound Tyr (colored or-
ange). The electron density map is contoured at 2.5s. The protein
residues participating in direct and water-mediated (red spheres)
contacts are shown.
(B) Schematic drawing of the interactions with Tyr. Direct and water
(marked as S)-mediated H bonds are shown by dashed lines. Van
der Waals interactions are depicted by solid arrows.
the phosphate group. The conformation of Tyr-AMS and
its principal interactions within the active site are closely
similar to those observed in the complex of PheRS with
Phe-AMP. The exception is provided with a lack of the H
bond between the carbonyl oxygen of Tyr-AMS and the
Ne1of Trpa149 belonging to a ‘‘helical loop’’ (residues
138–152) of the catalytic a subunit; the two atoms are
separated by w4 A˚. In the complex with Phe-AMP, the
respective atoms are at H bonding distance. The
loop’’ as a whole. This effect was also observed in the
ternary complex of PheRS with tRNAPheand L-phenyla-
laninyl-50-adenylate (PheOH-AMP), a stable analog of
Phe-AMP with a methylene group substituted for the
carbonyl one (unpublished data), and in the binary com-
sity). Universally conserved Phea260 and Alaa314, to-
gether with Vala261 (replaced in most sequences with
isosteric Thr), provide an optimal size for the amino
acid binding pocket. However, the distances between
the OH group of free Tyr or Tyr-AMS, or between the
chlorine atom of p-Cl-Phe (see below) and a pair of
atoms, Cg1of Vala261 and Cbof Alaa314 (see Table
S1), provide evidence for actual broadening of the Phe
Multiple interactions of Phe-AMP and PheOH-AMP
(Reshetnikova et al., 1999; Fishman et al., 2001) are re-
tained in the complex with the noncognate adenylate
analog. Thisis consistent with invitro aminoacylation ki-
netic data, which show Tyr-AMS to efficiently compete
with both Phe and ATP for binding (see Table 1). A
2-fold higher affinity of Tyr-AMS as compared to that
of PheOH-AMP presumably results from the presence
of a carbonyl group (replaced in PheOH-AMP) shown
to make extra contacts with the enzyme.
The Structure of the PheRS Complex with p-Cl-Phe
Recognition of unnatural amino acids by appropriate
aaRSs and subsequent misacylation of the cognate
tRNAs is a feasible way of in vivo incorporation of the
noncanonical amino acids into protein. Ibba et al.
(1994) have demonstrated attachment of the p-Cl-
substituted analog ofPhe totRNAand in vivo incorpora-
tioninto cellular protein bythe E.coli Alaa294Gly mutant
PheRS (Alaa314 in T. thermophilus PheRS) that exhibits
relaxedsubstratespecificity. Incontrast tothewild-type
E. coli PheRS shown to activate only p-F-Phe, the engi-
neered enzyme can activate all four p-halogenated
(F, Cl, Br, and I) Phe derivatives. To elucidate the mech-
ity of the PheRS amino acid binding pocket, the struc-
ture of wild-type PheRS complexed with p-Cl-Phe has
been determined at 2.8 A˚resolution (see Table 2, PDB
ID code 2AKW).
The electron density distinct in shape from those at-
tributed to Phe and Tyr is clearly visible at the synthetic
site (Figure 2C). The characteristic extension of the
density is in agreement with the C-Cl bond distance (in
para-position) of w1.75 A˚observed in various p-Cl-
substituted Phe derivatives (Shu ¨rmann et al., 1999).
The p-Cl-Phe fits well into the electron density, and its
location at the active site is similar to those observed
for Phe, Tyr, and the aromatic moieties of Tyr-AMS,
PheOH-AMP, and Phe-AMP. The appearance of p-Cl-
Phe in the Phe binding pocket suggests the absence
of steric clashes between the Cl atom, and the Cb
(3.5 A˚) and Cg1(3.3 A˚) atoms of Alaa314 and Vala261, re-
spectively. Unlike the Tyr binding mode (see above), the
the Oe2of Glua220 and the Ogof Sera180 (see Table S1).
The change from water-mediated (S6) to direct H bond-
ing contact is dictated by a small rearrangement in the
position of the analog and by a conformational switch
of the carboxylic group of Glua220. From least squares
superpositions of all of the above complexes in pairs
(the referenced complex is PheRS/PheAMP), it follows
that Ogof Sera180 occupies a rather stable position
and makes direct H bonds with the a-NH3+group in all
of the complexes, while the conformation of the
Glua220 side chain varies substantially.
Localization of Phe, Tyr, and Tyr-AMS Binding
Regions in PheRS Using a Docking Approach
To further investigate the regions of PheRS most appro-
priate for pretransfer and posttransfer editing activities,
we employed the AutoDock protocol, which has been
shown to localize ligand binding sites efficiently (Morris
et al., 1998). The docking algorithm treats ligands as
flexible molecules, while keeping protein conformation
restrained. A properly docked ligand structure should
meet the following requirements: the centers-of-mass
of a ‘‘good’’ candidate have to be found in a well-popu-
lated cluster; the rmsd between the docked molecule
revealed two well-defined clusters of docked Tyr mole-
cules placed at the synthetic site area and the interface
between the B3 and B4 domains (Figure S1B). The
denser population of the Tyr binding sites is observed
in the editing site area.
The docking protocol has also been applied for
searching Phe and Tyr-AMS binding regions (Figures
site as the most preferable for accommodation of both
ligands. There is only one potential energy funnel for
Phe. The vast majority of Tyr-AMS docking patterns
are also related to the synthetic site. Two attractive en-
ergy funnels exist for the Tyr molecule. In contrast to
the synthetic site, the editing site prefers the Tyr mole-
cules and ‘‘filters out’’ the Phe molecules. This is further
proof of the limited ability of PheRS’s synthetic active
site to effectively discriminate between the two amino
Specificity of the PheRS Editing Site
Complexes with Tyr, p-Cl-Phe, and Phe provide evi-
dence that the noncognate substrates bind in the amino
acid binding pocket in a manner closely similar to that of
Phe. Moreover, activation of Tyr by the wild-type T. ther-
mophilus PheRS immediately follows from the ATP con-
sumption experiments, as does the close similarity of
the Phe-AMP and Tyr-AMS binding modes in the syn-
thetic site. The direct observation of the presence of
unnatural p-Cl-Phe in the amino acid binding pocket
of T. thermophilus PheRS and the previously revealed
Structural Basis of PheRS Editing against Tyrosine
ability of E. coli to incorporate p-F-Phe into repetitive
polypeptides (Yoshikawa et al., 1994) suggest that not
only the E. coli mutant Gly294PheRS (Ibba et al., 1994),
but also wild-type T. thermophilus PheRS, can activate
p-Cl-Phe. In contrast to the PheRS/Tyr complex, no
electron density was observed in the editing pocket of
PheRS complexed with p-Cl-Phe. The space com-
posed of the invariant Glyb315, Glub334, Pheb360, and
Leub286 together with Pro(Ile/Leu)b259, His(Asn)b261,
and Ala(Ser)b356 is just large enough to accommodate
the Cl extremity in the para position of Phe. The only re-
sult that would impair the appearance of p-Cl-Phe in the
hydrolyticpocket isanunfavorable contactwiththeside
chain of Glub334. The exclusive binding of p-Cl-Phe
ical experiments and is indicative of PheRS editing
activity against Tyr only. Stable attachment of para-
substituted Phe analogs to tRNAPheby PheRS (Ibba
et al., 1994; Kirshenbaum et al., 2002) implies that they
are not substrates for editing.
The all-important steric factor of the editing site inte-
rior is Glyb315 localized at the end of the b hairpin
312–322. The appearance of any residue with a bulky
side chain at this position may interlock the correct po-
sitioning of the Tyr moiety. The signature sequence of
the b hairpin ‘‘l4G/AG44GlxxS/T’’ (where l stands for
small G, A, or S residues, 4 stands for hydrophobic res-
idues, and x stands for any residue) displays a high level
of conservation, strongly suggesting the stability of its
conformation in eubacterial PheRSs. Thus, invariance
of Glub334 and Glyb315, and conformation of the b hair-
pin, adds up to the conservation of the editing site inte-
rior in the prokaryotic PheRSs.
Posttransfer Editing by PheRS
The lack of Tyr-AMS from the editing site suggests that
iting of the translocated Tyr-AMP. On the other hand,
close structural similarity in the positioning of Phe-
tive site bears witness to the absence of certain steric
hindrance for the transfer of activated Tyr onto tRNAPhe
and the subsequent formation of the Tyr-tRNAPheinter-
mediate. Furthermore, the Kivalue determined for Tyr-
AMS from kinetic measurements (see Table 1) implies
thatitbindstightly tothe syntheticsiteofPheRS incom-
plex with tRNAPhe. The data obtained by an ATP con-
sumption assay clearly show that Tyr-stimulated ATP
hydrolysis is enhanced by tRNAPhe, and that the tRNA-
dependent reaction dominates alternative pathways
controlled, in particular, by pyrophosphate dissociation.
The activating effect of tRNAPheis related to its charging
capability. This is further confirmed by detailed analysis
of the triggering role of tRNAPhein the editing activity of
E. coli PheRS (Roy et al., 2004). Both structural and bio-
chemical data are indicative of the posttransfer editing
activity of prokaryotic PheRSs against misactivated
Tyr. Nevertheless, it is pertinent to note that 2 out of 27
docked Tyr-AMS positions withthe lowest binding ener-
gies are visible in the editing site (Figure S1C). It is likely
that calculations based on PheRS structure determined
trum, which, in turn, will bring the docking calculations
into better agreement with experiment.
Interpretation of Site-Directed Mutagenesis
Experiments in Terms of Structural Data
A few replacements of residues within the proposed ed-
iting site of E. coli PheRS, impairing tRNA-dependent
proofreading activity, have been reported (Roy et al.,
2004). Hisb261Ala or Hisb261Leu mutations break
down the entire net of H bonds within the editing site.
First, they destabilize the positioning of Tyr, breaking
the bond between its COO2group and Nd1of Hisb261.
Second, Hisb261 is located within the core of the exten-
sive network of H bonds formed by water molecules
of Thrb249, Asnb250, Serb322, and Glub323 (see Fig-
ure 3B). The rupture of water-mediated contacts in the
vicinity of Tyr prevents a cleavage of the ester bond be-
tween the tRNA and the amino acid. Significant losses in
hydrolytic activity caused by the Glub334Ala mutation
are consistent with the critical role of Glub334 and the
fact that the methyl group of Ala renders anchoring of
the Tyr OH group impossible. A Trp replacement of
Alab356, whose side chain is exposed to the B3/B4 in-
terface area, may result in substantial conformational
changes preventing penetration of Tyr into the editing
site cavity. It is notable that residues with the shortest
side chains (Ala or Ser) appear in the equivalent position
of the b subunit in almost all sequences of eubacterial
site is the only residue whose side chain makes H bond-
placement by Trp would impair the correct positioning
of the substrate in the editing site and, moreover, can
make the pocket beyond the reach of the substrate.
Translocation of Tyr-tRNAPhetoward the Editing Site
Thetwo Tyrmoleculesboundwithin theeditingandsyn-
thetic sites are separated by w35 A˚. This implies that
Tyr-AMP or Tyr-tRNAPhehas to be transferred from the
ceed. A clamp binding mode of tRNAPhe, where the
anticodon arm is sandwiched between the N-terminal
unit (* stands for the symmetry-related heterodimer),
suggests that tRNA and PheRS are specifically fitted to-
it is obvious that there is no alternative ‘‘editing tRNA/
PheRS’’ binding mode, in which misacylated tRNA dis-
sociated from PheRS will realign to the new position
that is favorable for hydrolysis. This observation is in
line with the commonly accepted model of posttransfer
editing that suggests the translocation of the tRNA ac-
ceptor end loaded with the incorrect amino acid (Dock-
the B3/B4 domains is covalently connected by the ex-
tended segment Leub474-Alab482 with the ‘‘catalytic-
cavity wherein the catalytic module (A1-A2) is inserted
(Mosyak et al., 1995). The translocation process that
may substantially displace the B3/B4 domains will dis-
rupt the network of interactions centered on a Mg2+
ion, which is located at the a/b subunit interface (near
tion. Thus, shuttling from the synthetic to the editing site
ified with the noncognate substrate. Based on the
arrangement of Tyr (in both the synthetic and editing
sites) and tRNAPhe(as it was observed in the ternary
complex of PheRS with PheOH-AMP and tRNAPhe), we
have modeled the CCA end to bring A76 to the position
occupied by Tyr in the editing site (Figure 4A). This
model shows how A76 could be placed and stabilized
to facilitate the hydrolysis of misacylated Tyr-tRNAPhe
(Figure 4B). The anchoring of the adenosine moiety is
established by hydrophobic interactions with Ileb242
and Leub286. Ileb242 belongs to a conserved motif,
‘‘I/LN/DxV/LVD/N’’ (positions 242–247), of domain B3.
Leub286 localized on the extremity of the B4 domain
b hairpin 282–292 is an invariant residue in prokaryotic
PheRSs and obviously bears functional loads. It is of in-
terest that neighboring Aspb287 is also highly con-
served. Together with Thrb285 and Argb291, it is impli-
cated in stabilization of the b hairpin conformation. The
plane of the A76 base occupies the position midway be-
tween the above-mentioned structural elements. The
appearance of hydrophobic residues on both sides of
the adenine ring reflects the largely hydrophobic nature
high propensities of hydrophobic residues to interact
with adenine (Moodie et al., 1996).
Structural Aspects of Hydrolytic Activity in PheRS
So far, all known aaRSs catalyze ester bond formation
between the activated a-COOH group of the amino
acid and the 20- or 30-OH group of the tRNA 30-terminal
adenosine. Thus, the posttransfer editing, thought of
as a correction of the erroneously acylated tRNAPheas
a prevalent mode in the Phe system, should be associ-
ated with cleavage of the ester bond between the car-
bonyl carbon of the aminoacyl moiety and the oxygen
atom of the terminal ribose. The presence of Hisb261,
Glub334, Asnb250, Thrb249, and Glub323 in the vicinity
of the ester bond subjected to hydrolysis is reminiscent
of the active site of peptidyl-tRNA hydrolase (PTH) deal-
ing with hydrolysis of the ester bond between tRNA and
the peptide (Schmitt et al., 1997). Residues Asn10,
His20, and Asp93 considered as being crucial for PTH
activity (Schmitt et al., 1997; Goodall et al., 2004) are
similar to the PheRS triad, Asnb250, Hisb261, and
Glub323, on which they can be superimposed with an
rmsd of 1.4 A˚for the Caatoms. However, the general to-
pology of the B3/B4 domains (a-b sandwich and slightly
distorted b barrel) differs from that of PTH (single a/b
ferent structural elements. Obviously, the two-esterase
activities are spanned on the distinct folds. Close in-
spection of the editing site with the built-in Tyr-A76
shows that the nearest position to the cleaved bond is
occupied by water molecule S80. It is conceivable that
substrate hydrolysis may be initiated via nucleophilic at-
tack of water S80 oriented by an activating Glub323. In-
in the apo enzyme (S147, PDB code 1PYS) and in its
complex with Tyr-AMS (S12). It is therefore reasonable
to assume with high probability that a ‘‘catalytic’’ water
molecule coordinated by Hisb261 and Glub323 is pres-
ent in the editing site. The position of S80 is additionally
stabilized by Hbonds with Thrb249 and theamide group
of Alab262 via water molecule S14. The activating role of
Glub323, which is not invariant, may be performed by
Thrb249 (via highly the coordinated water S14 observed
in all available PheRS structures) or Hisb261, which are
both well conserved in eubacterial PheRSs.
Manual fitting of the CCA-Tyr moiety into the cavity of
theB3/B4interfacegives risetomorethan oneplausible
conformation of C74 and C75 nucleotides of the mis-
charged tRNA. This is due to the hydrophilic character
of the interface and the high level of conservation of res-
idues covering the inner surface of the cavity (Asp/
Glub33, Glu/Aspb127, Asp/Glub157, Arg/Lysb240, Glu/
Aspb355, Arg/Lysb362, etc.). Further determination of
PheRS structures with nonhydrolyzable analogs of the
posttransfer substrates is necessary to detail the trans-
location and editing events.
Materials and General Methods
PheRS and native tRNAPhe(isoacceptor I, 1700 pmol of Phe/A260
unit) from T. thermophilus HB8 were isolated and purified as de-
scribed (Ankilova et al., 1988; Stepanov et al., 1998). E. coli tRNAPhe
was synthesized by using runoff transcription of a synthetic gene
Figure 4. Model for tRNAPheTranslocation
(A) Structure of PheRS complexed with two Tyr (colored magenta) in
has been modeled to the editing site (red- and green-colored for do-
mains B3 and B4, respectively) to show its translocation with the
substrate upon the editing reaction. The anticodon arm of tRNA is
clamped between the N-terminal coiled-coil of the a* subunit and
the B8 domain of the b* subunit (colored beige).
(B) The model of the posttransfer substrate (Tyr-A76) in the editing
site. B3/B4 domains are shown in the same colors as in (A). The res-
idues shown are proposed to interact with A76.
Structural Basis of PheRS Editing against Tyrosine
correct length transcript as described (Vasil’eva et al., 2002). Plas-
mid DNA containing a gene of E. coli tRNAPheunder control of the
phage T7 promoter was a gift from O.C. Uhlenbeck. The tRNAPhe
transcript lacking the 30-terminal nucleotide was prepared as de-
scribed (Vasil’eva et al., 2002). Radiochemicals were purchased
from Amersham Biosciences. L-Phe was purchased from Sigma,
and other unlabeled L-amino acids (Ile, His, Trp, and p-Cl-Phe)
were purchased from Fluka. L-Tyr from Merck and Fluka was as-
sayed by AccQTag amino acid analysis for Phe contamination and
purified by 4-fold recrystallization in water as described (Lin et al.,
1983). Tyr-AMS was a gift from M. Tukalo and S. Cusack. PheOH-
AMP was synthesized and purified as described (Lavrik et al., 1978).
Aminoacylation of tRNAPhe(native T.thermophilus orE.colitRNAPhe
transcript) with T. thermophilus PheRS was performed in conditions
described previously (Vasil’eva et al., 2002). Direct attachment of
noncognate amino acids to tRNAPhewas measured in similar condi-
tions by the addition of 100 mM L-[14C]Tyr, L-[14C]Ile, or L-[14C]His or
20 mM L-[3H]Trp to the reaction mixture instead of14C/3H-labeled
Phe. Inhibition of Phe-tRNAPhesynthesis with noncognate amino
acids was measured in the presence of a nonsaturating concentra-
tion of L-[3H]Phe (0.5 mM); concentrations of Ile and His varied in the
range of 0.1–20 mM, and those of Trp and Tyr were in the range of
0.1–4 mM. In all inhibition analyses, conditions were selected to en-
sure linear reaction rates. The Phe concentration used as variable
substrate was 0.4–4 mM, and the concentration of ATP varied in
the range of 0.05–0.6 mM. Concentrations of inhibitors were in the
range of their inhibition constants. Inhibition types were analyzed
from double-reciprocal plots (Cornish-Bowden and Wharton,
1976). The kinetic parameters (kcat, KM, and KMapp= KM[1 + I/Ki]
for competitive inhibitors, where I is concentration of the inhibitor)
were calculated by using the Microcal Origin 4.10 program. The re-
ported KMand Kivalues represent the average of at least two deter-
minations with standard deviations less than 15%.
ATP Consumption Assay
ATP hydrolysis catalyzed by PheRS was measured at 37ºC. A 20 ml
reaction mixture contained 50 mM Tris-HCl (pH 8.5), 9 mM MgCl2,
2–200 mM [3H]ATP, and 2–300 mM Phe or 2–1200 mM Tyr (or other
noncognate amino acids added at 300 mM). A total of 0.5–10 mM
tRNAPhe(native T. thermophilus or E. coli tRNAPhetranscript, cor-
rect-length or lacking the 30-terminal nucleotide) and 4 U/ml yeast
inorganic pyrophosphatase (Sigma) were added when indicated.
Thereactions wereinitiatedbythe addition of 0.2mM PheRS. Atvari-
able incubation times (2–40 min), samples of 1 ml were spotted onto
silicagel sheets (DC-Plastikfolien Kieselgel 60 F254, Merck). Unla-
beled ATP and AMP were applied to the plate at the origin prior to
use. The nucleotides were separated by TLC developed in a mixture
of dioxane, concentrated ammonia, and water (6:1:4, v/v). The spots
were marked in UV light and cut out, and the radioactivity was mea-
sured in a liquid scintillation counter.
Crystal Preparation and Data Collection
The crystals of native T. thermophilus PheRS were grown as de-
scribed previously(Chernaya et al.,1987). The protein concentration
in the drop was 3 mg/ml. Crystals grew within 3 months to maximum
dimensions of 0.3 3 0.3 3 0.2 mm3.
ture containing the crystallization buffer and 1 mM ligand (Tyr, Tyr-
AMS, or p-Cl-Phe). Data were collected at the synchrotron radiation
source at ESRF (beamline ID14), France and on the in-house Rigaku
R-axis IV++ image plate detector.Before flash cooling, soaked crys-
tals were transferred into a cryoprotectant containing 30% (v/v)
glycerol. Diffracted intensities were evaluated and integrated by us-
ing the HKL package (Otwinowski and Minor, 1997) and were scaled
further by using programs from the CCP4 package. A summary of
the data collection statistics is given in Table 2.
Structure Determination and Refinement
The 2.9 A˚refined structure of T. thermophilus PheRS (Mosyak et al.,
1995, 1pys) was used as a starting model. After rigid-body refine-
mentand cycles ofsimulatedannealing andconjugategradientmin-
imization, a difference Fourier map with coefficients (Fobs2 Fcalc)
was calculated. The refinement and map calculation were carried
out with the CNS package (Brunger et al., 1998). A random sample
set containing 5% of the reflections was excluded from the refine-
ment and was used for Rfreecalculation. All the calculations were
performed against the Fobsdata sets at 50–2.7 A˚, 50–2.6 A˚, and
50–2.8 A˚resolution for PheRS/Tyr, PheRS/Tyr-AMS, and PheRS/p-
Cl-Phe complexes, respectively.
The ligands were built into the electron density and manually ad-
justed with the program O (Jones et al., 1991). The statistics of the
refinement are represented in Table 2. Molecular graphics pictures
in Figures 2–4 were prepared with the Pymol (De Lano, 2002).
Docking PheRS with Phe, Tyr, and Tyr-AMS
To generate the input files, AutoDockTools (ADT) (http://www.
scripps.edu/pub/olson-web/doc/autodock/tools.html) was used.
Gasteiger-Hu ¨ckel charges were assigned to ligand atoms. Water
molecules were eliminated from PDB files to suit ADT requirements.
To include binding regions of both the active and editing sites, the
simulation box 50A˚3 50A˚3 50A˚has been chosen. Grids for poten-
tial evaluation had a resolution of 0.45 A˚. The parameters of the
Lamarckian genetic algorithm correspond to the AUTODOCK 3.05
default values. All s bond torsion angles were free to rotate: 3 bonds
for Phe and p-Cl-Phe, 4 bonds for Tyr, and 14 bonds for Tyr-AMS.
Supplemental Data including a graphical representation of the li-
gand binding sites deduced from the AutoDock protocol and a list
of the interactions in the described complexes are available at
Weare gratefultoM.Tukalo andS.CusackforTyr-AMS.WethankV.
Ankilova for purification of T. thermophilus PheRS. This work was
supported by the Kimmelman Center for Biomolecular Structure
and Assemblies and by grants to N.M. from the Russian Foundation
for Basic Research (03-04-48384) and to M.S. from the Israel Sci-
ence Foundation (1034/03).
Received: June 27, 2005
Revised: August 17, 2005
Accepted: August 18, 2005
Published: December 13, 2005
Ankilova, V., Reshetnikova, L., Chernaya, M., and Lavrik, O. (1988).
Phenylalanyl-tRNA synthetase from Thermus thermophilus HB8.
Purification and properties of the crystallizing enzyme. FEBS Lett.
Beebe, K., Ribas De Pouplana, L., and Schimmel, P. (2003). Elucida-
tion of tRNA-dependent editing by a class II tRNA synthetase and
significance for cell viability. EMBO J. 22, 668–675.
Brunger, A., Adams, P., Clore, G., DeLano, W., Gros, P., Grosse-
Kunstleve, R., Jiang, J., Kuszewski, J., Nigles, M., Pannu, N., et al.
(1998). Crystallography & NMR system: a new software suite for
macromolecular structure determination. Acta Crystallogr. D Biol.
Crystallogr. 54, 905–921.
Chernaya, M., Korolev, S., Reshetnikova, L., and Safro, M. (1987).
Preliminary crystallographic study of the phenylalanyl-tRNA synthe-
tase from Thermus thermophilus HB8. J. Mol. Biol. 198, 555–556.
Cornish-Bowden, A., and Wharton, C.W. (1976). Enzyme Kinetics
(Oxford: IRL Press).
DeLano, W. (2002). The PyMOL Molecular Graphics System (San
Carlos, CA: DeLano Scientific).
Dock-Bregeon, A., Sankaranarayanan, R., Romby, P., Caillet, J.,
Springer, M., Rees, B., Francklyn, C.S., Ehresmann, C., and Moras,
D. (2000). Transfer RNA-mediated editing in threonyl-tRNA synthe-
tase: the class II solution to the double discrimination problem.
Cell 103, 877–884.
Dock-Bregeon, A., Rees, B., Torres-Larios, A., Bey, G., Caillet, J., Download full-text
and Moras, D. (2004). Achieving error-free translation: the mecha-
nism of proofreading of threonyl-tRNA synthetase at atomic resolu-
tion. Mol. Cell 16, 375–386.
Eldred, E., and Schimmel, P. (1972). Rapid deacylation by isoleucyl
transfer ribonucleic acid synthetase of isoleucine-specific transfer
ribonucleic acid aminoacylated with valine. J. Biol. Chem. 247,
Fersht, A. (1977). Editing mechanism in protein synthesis. Rejection
of valine by the isoleucyl-tRNA synthetase. Biochemistry 16, 1025–
Fishman, R.,Ankilova, V.,Moor,N.,andSafro,M.(2001).Structure at
2.6 A˚resolution of phenylalanyl-tRNA synthetase complexed with
phenylalanyl-adenylate in the presence of manganese.ActaCrystal-
logr. D Biol. Crystallogr. 57, 1534–1544.
Goldgur, Y., Mosyak, L., Reshetnikova, L., Ankilova, V., Lavrik, O.,
Khodyreva, S., and Safro, M. (1997). The crystal structure of phenyl-
alanyl-tRNA synthetase from T. thermophilus complexed with cog-
nate tRNAPhe. Structure 5, 59–68.
Goodall, J., Chen, G., and Page, M. (2004). Essential role of histidine
20 in the catalytic mechanism of Escherichia coli peptidyl-tRNA hy-
drolase. Biochemistry 43, 4583–4591.
Ibba, M., Kast, P., and Hennecke, H. (1994). Substrate specificity is
determined by amino acid binding pocket size in Escherichia coli
phenylalanyl-tRNA synthetase. Biochemistry 33, 7107–7112.
Jakubowski, H. (2004). Accuracy of aminoacyl-tRNA synthetases:
proofreading of amino acids. In The Aminoacyl-tRNA synthetases,
M. Ibba, C.S. Francklyn, S. Cusack, eds. (George Town, TX: Landes
Bioscience), pp. 384–396.
Jakubowski, H., and Goldman, E. (1992). Editing of errors in selec-
tion of amino acids for protein synthesis. Microbiol. Rev. 56, 412–
Jones, T., Zou, J., Cowtan, S., and Kjeldgaard, M. (1991). Improved
methods for building protein models in electron density maps and
the location of errors in these models. Acta Crystallogr. A 47, 110–
Kirshenbaum, K., Carrico, I., and Tirrell, D. (2002). Biosynthesis of
proteins incorporating a versatile set of phenylalanine analogues.
ChemBioChem 3, 235–237.
Lavrik, O.I., Moor, N.A., and Nevinsky, G.A. (1978). Synthesis of L-
phenylalanyladenylate analogs and studies on their interaction
with phenylalanyl-tRNA synthetase from E. coli. Bioorgan. Khimia
(Russian) 4, 1480–1487.
Lin, S., Baltzinger, M., and Remy, P. (1983). Fast kinetic study of
yeast phenylalanyl-tRNA synthetase: an efficient discrimination be-
tween tyrosine and phenylalanine at the level of the aminoacylade-
nylate?enzyme complex. Biochemistry 22, 681–689.
Lin, S., Baltzinger, M., and Remy, P. (1984). Fast kinetic study of
yeast phenylalanyl-tRNA synthetase: role of tRNAPhein the discrim-
ination between tyrosine and phenylalanine. Biochemistry 23, 4109–
Moodie, S., Mitchell, J., and Thornton, J. (1996). Protein recognition
of adenylate: an example of a fuzzy recognition template. J. Mol.
Biol. 263, 486–500.
Morris, G., Goodsell, D., and Halliday, R. (1998). Automated docking
using a Lamarckian genetic algorithm and an empirical binding free
energy function. J. Comput. Chem. 19, 1639–1662.
Mosyak, L.,Reshetnikova, L.,Goldgur,Y.,Delarue, M., andSafro, M.
(1995). Structure of phenylalanyl-tRNA synthetase from Thermus
thermophilus. Nat. Struct. Biol. 2, 537–547.
Nureki, O., Vassylyev, D., Tateno, M., Shimada, A., Nakama, T., Fu-
kai, S., Konno, M., Hendrickson, T., Schimmel, P., and Yokoyama,
S. (1998). Enzyme structure with two catalytic sites for double-sieve
selection of substrate. Science 280, 578–582.
Otwinowski,Z., and Minor, W. (1997). Processing of X-ray diffraction
data collected processing in oscillation mode. Methods Enzymol.
Reshetnikova, L., Moor, N., Lavrik, O., and Vassylyev, D. (1999).
Crystal structures of phenylalanyl-tRNA synthetase complexed
with phenylalanine and a phenylalanyl-adenylate analogue. J. Mol.
Biol. 287, 555–568.
Roy, H., Ling, J., Irnov, M., and Ibba, M. (2004). Post-transfer editing
in vitro and in vivo by the bsubunitof phenylalanyl-tRNA synthetase.
EMBO J. 23, 4639–4648.
tase. In The Aminoacyl-tRNA Synthetases, M. Ibba, C.S. Francklyn,
S. Cusack, eds. (George Town, TX: Landes Bioscience), pp. 250–
Sankaranarayanan, R., Dock-Bregeon, A.C., Rees, B., Bovee, M.,
Caillet, J., Romby, P., Francklyn, C., and Moras, D. (2000). Zinc ion
mediated amino acid discrimination by threonyl-tRNA synthetase.
Nat. Struct. Biol. 7, 461–465.
(1997). Crystal structure at 1.2 A˚resolution and the active site map-
Shu ¨rmann, M., Silvestri, A., Ruisi, G., Girasolo, M., Barbieri, A.,
Huber, F., and Barbieri, R. (1999). The structure and dynamics of
Cl-substituted tetraphenyl- and tetrabenzyl-tin (IV). J. Organometal.
Chem. 584, 293–300.
Stepanov, V., Moor, N., Ankilova, V., Sukhanova, M., and Lavrik, O.
(1998). A peculiarity of the reaction of tRNA aminoacylation cata-
lyzed by phenylalanyl-tRNA synthetase from the extreme thermo-
phile Thermus thermophilus. Biochim. Biophys. Acta 1386, 1–15.
Tsui, W.C., and Fersht, A. (1981). Probing the principles of amino
acid selection using the alanyl-tRNA synthetase from Escherichia
coli. Nucleic Acids Res. 9, 4627–4637.
Vasil’eva, I., Ankilova, V., Lavrik, O., and Moor, N. (2002). tRNA dis-
crimination by T. thermophilus phenylalanyl-tRNA synthetase at
the binding step. J. Mol. Recognit. 15, 188–196.
Wong, F.C., Beuning, P., Silvers, C., and Musier-Forsyth, K. (2003).
An isolated class II aminoacyl-tRNA synthetase insertion domain is
functional in amino acid editing. J. Biol. Chem. 278, 52857–52864.
Yoshikawa, E., Fournier, M., Mason, M., and Tirrell, D. (1994). Genet-
ically engineered fluoropolymers. Synthesis of repetitive polypep-
tides containing p-fluorophenylalanine residues. Macromolecules
Coordinates have been deposited in the PDB with accession codes
2AMC, 2AKW, and 2ALY.
Structural Basis of PheRS Editing against Tyrosine