Báo cáo Y học: Ribosome-associated factor Y adopts a fold resembling a double-stranded RNA binding domain scaffold potx

Distribution of the conserved residues on the protein surface highlights a positively charged region towards the C-terminal segments of both a helices, which most probably constitutes an

Ribosome-associated factor Y adopts a fold resembling

a double-stranded RNA binding domain scaffold

Keqiong Ye1, Alexander Serganov1, Weidong Hu1, Maria Garber2and Dinshaw J Patel1

1

Cellular Biochemistry & Biophysics Program, Memorial Sloan-Kettering Cancer Center, New York, USA;

2

Institute of Protein Research, Moscow Region, Pushchino, Russia

Escherichia coliprotein Y (pY) binds to the small ribosomal

subunit and stabilizes ribosomes against dissociation when

bacteria experience environmental stress pY inhibits

trans-lation in vitro, most probably by interfering with the binding

of the aminoacyl-tRNA to the ribosomal A site Such a

translational arrest may mediate overall adaptation of cells

to environmental conditions We have determined the 3D

solution structure of a 112-residue pY and have studied its

backbone dynamic by NMRspectroscopy The structure

has a babbba topology and represents a compact

two-layered sandwich of two nearly parallel a helices packed

against the same side of a four-stranded b sheet The 23

C-terminal residues of the protein are disordered

Long-range angular constraints provided by residual dipolar

coupling data proved critical for precisely defining the

position of helix 1 Our data establish that the C-terminal

region of helix 1 and the loop linking this helix with strand b2 show significant conformational exchange in the ms–ls time scale, which may have relevance to the interaction of pY with ribosomal subunits Distribution of the conserved residues

on the protein surface highlights a positively charged region towards the C-terminal segments of both a helices, which most probably constitutes an RNA binding site The observed babbba topology of pY resembles the abbba topology of double-stranded RNA-binding domains, despite limited sequence similarity It appears probable that functional properties of pY are not identical to those of dsRBDs, as the postulated RNA-binding site in pY does not coincide with the RNA-binding surface of the dsRBDs Keywords: backbone dynamics; dsRBD; molecular align-ment; residual dipolar couplings; RNA-binding domain

Protein expression is finely tuned in the cell allowing for

adaptations to various environmental changes The

regula-tion strategies adopted by the cell encompass almost every

aspect of the protein production process, including mRNA

transcription, translation in the ribosome and protein

degradation Ribosome activity constitutes an important

target in the control of protein expression Protein Y, the

product of yfiA gene, is a ribosome-associated protein

present in E coli and many other bacteria [1–3] pY was

earlier assigned to the r54modulation protein family based

on the observation that mutations in the related

down-stream r54gene cause an increase in the level of expression

from r54-dependent promoters [4]

pY (also called ribosome associated inhibitor, RaiA) [2]

was detected in the ribosome fraction during environmental

stress, as a consequence of either low temperature [2] or

excessive cell density [2,3] Furthermore, pY binds to the small ribosomal subunit in an Mg2+-dependent manner and becomes less exposed to solvent upon association of the small and large ribosomal subunits This suggests an intersubunit position for pY in the 70S ribosome and may explain its ribosome stabilization effect [1] pY inhibits translation most probably by interfering with the binding of the aminoacyl-tRNA to the ribosomal A site [2] It has been proposed that A-site blocking by pY arrests protein synthesis during cold shock or at the stationary phase of cell culture, thereby mediating overall adaptation of bacteria to environmental stress pY shows about 40% identity with the protein encoded by E coli yhbH gene The association of pY and 70S ribosomes contrasts with the preferential binding of YhbH and the dimerized 100S ribosomes in the stationary phase of cell growth [3] The authors suggested that these related proteins play a stabilization role during ribosome storage The protein Y is widely dispersed in bacteria Homologs of

pY are also found in two plants, Arabidopsis thaliana and Spinacia oleracea Interestingly, the nuclear encoded distinct homolog from spinach (formerly known as CS-5, PSrp-1, S22, or S30 according to the nomenclature of Schmidt et al [5]) was found in the stroma of chloroplasts as well as in the small ribosomal subunits [6,7] In spite of the larger size and relatively low homology with its E coli counterpart, the ribosome binding site of S30 most probably resembles the binding site of pY because the S30 protein can be incorpor-ated into E coli ribosomes in vivo [8] Another pY homolog, protein LtrA from cyanobacteria Synechococcus PCC 7002,

is expressed only in the dark [9], suggesting that members of the pY family can generally be involved in the process of adaptation to the environmental conditions

Correspondence to D J Patel, Cellular Biochemistry and Biophysics

Program, Memorial Sloan-Kettering Cancer Center, New York,

NY 10021, USA Fax: + 1212 7173066, Tel.: + 1212 6397207,

E-mail: pateld@mskcc.org

Abbreviations: dsRBD, double-stranded RNA binding domain;

HSQC, heteronuclear single quantum coherence; P, protection factor;

pY, protein Y; R1, longitudinal relaxation rate; R2, transversal

relaxation rate; RDC, residual dipolar coupling; r.m.s.d., root mean

square deviation.

Note: The PDB ID code for the pY structures is 1L4S The access

number of NMRassignments at the Biomolecular Magnetic

Resonance Databank is 5315.

(Received 20 May 2002, revised 22 August 2002,

accepted 30 August 2002)

In the present study, we have determined the solution

structure of pY from E coli by NMRspectroscopy using

NOE and residual dipolar coupling restraints The protein

has a babbba arrangement of secondary elements and a

disordered 23-residue C-terminal tail An analysis of the

surface conserved residues in the pY structure suggest a

potential RNA-binding site residing within the C-termini of

two a helices A comparison with the recently published

similar structure of the protein HI0257 from Haemophilus

influenzaeis also presented [10]

M A T E R I A L S A N D M E T H O D S

Sample preparation

The gene encoding 112 residues of pY was amplified by

PCRwith the genomic E coli DNA as the template, and

after purification and digestion by restriction enzymes, was

subcloned into NdeI and EcoRV sites of the pET29b vector

(Novagen) The protein was overproduced in E coli

BL21(DE3) strain Uniformly 15N-enriched protein was

prepared by growing cells in M9 minimal medium

contain-ing15NH4Cl Uniformly15N,13C-labeled protein was

puri-fied from cells grown in a 1 : 1 mixture of Celton CN

medium (Martek Biosciences) and M9 medium with

15NH4Cl and 2 gÆL)1 [13C]glucose To obtain 10% 13

C-labeled protein, cells were grown in the M9 medium

containing 10% of [13C]glucose All minimal media were

supplemented with basal medium eagle vitamin solution

(Life Technologies Gibco BRL) and trace element

supple-ments [11] The purification procedure was as follows: cells

producing pY were disrupted by sonication in the buffer

containing 20 mM Tris/HCl, pH 7.0, 50 mM NaCl, 2 mM

MgCl2, and protease inhibitors Cell debris was removed by

low-speed centrifugation (30 000 g) for 30 min, and

ribo-somes were precipitated by ultra-speed centrifugation

(150 000 g) for 3 h The pH of the supernatant was adjusted

to 5.5 with 1.0Msodium acetate, pH 5.0 The precipitate

was removed by low-speed centrifugation, and the

super-natant was loaded onto an SP Sepharose (Amersham

Pharmacia Biotech Inc.) column equilibrated with the

buffer containing 20 mM sodium acetate (pH 5.5) and

50 mM NaCl The protein was eluted using an NaCl

gradient The protein containing fractions were combined,

concentrated in an Amicon cell, and diluted 10 fold with

10 mM Tris/HCl, pH 9.0 The protein was loaded onto a

MonoQ column (Amersham Pharmacia Biotech Inc.)

equilibrated with buffer containing 50 mM Tris/HCl

(pH 9.0) and 50 mM NaCl, and eluted using the NaCl

gradient Finally, the protein was purified by gel-filtration

on a HiLoad 16/60 Superdex 75 column (Amersham

Pharmacia Biotech Inc.) in buffer containing 10 mMsodium

phosphate (pH 5.7) and 50 mMNaCl

Standard NMRsamples were prepared containing

approximately 1–2 mM pY protein in gel-filtration buffer

supplemented with 0.3 mM NaN3in either 93% H2O/7%

2H2O or 100%2H2O

NMR spectroscopy

NMRspectra were recorded at 31.5C on a Varian Unity

plus600 MHz spectrometer equipped with a

triple-reson-ance probe and Z-axis gradient Spectra were processed

using theFELIX98 (MSI Inc.) software and analyzed using NMRVIEW5.0 software [12]

The1H,15N,13C¢, and13Cabackbone resonances were assigned using triple resonance CBCA(CO)NH, HNCACB, HNCO, and HN(CA)CO experiments collected on a

15N,13C-labeled sample [13] The side-chain resonances were assigned from H(CCO)NH, C(CO)NH, HCCH-TOCSY (2H2O),15N-TOCSY and 2D-TOCSY (2H2O) experiments The b-methylene stereospecific assignments were obtained

by a qualitative comparison of intensities in the 3D HNHB, HACAHB-COSY,15N-NOESY and13C-NOESY spectra v1 torsion angle restraints (including Ile, Thr and Val) were introduced in the form of the common staggered confor-mations (v1¼) 60, 60, 180), with an error of ±20

All methyl groups in Leu and Val residue were stereospecifically assigned using a 10%13C-labeled sample [14] Qualitative

3JHNHacoupling constants were measured from an HNHA [15] experiment using a correction factor of 0.9 to compen-sate the different relaxation properties of the diagonal and cross peaks.3JHNHavalues were directly used in structural refinement [16] Additionally, dihedral angle constraints of )65(±25) and )120(±30) for / were set for 3JHNHa

<5.8 Hz and >8.0 Hz, respectively

Distance restraints were determined using NOEs derived from the 3D 15N-NOESY (100 ms), 3D 13C-NOESY (2H2O, 80 ms), and from the aromatic region in a 2D NOESY (2H2O, 140 ms) Upper bounds of distance con-straints were classified into the ranges of 2.7, 3.3, 4.0, 5.0, and 6.0 A˚ based upon relative NOE volumes of the cross peaks The lower bounds were set to 0.0 A˚ in order to use ambiguous NOEs in the calculation [17]

Molecular alignment with bacteriophage Pf1 was not successful because of protein association with phage [18] Weak alignment of the protein (7.3 Hz splitting of the deuterium signal) was finally obtained in a 3.2% (w/v) bicelle [19] solution in 20 mMpotassium phosphate buffer (pH 6.6) and 1 mMNaN3at 40C Bicelles were prepared

by mixing dimyristoylglycerophosphocholine, dihexanoyl-phosphatidylcholine, and cetyltrimethylammonium bro-mide at a molar ratio of 30 : 10 : 2 Premixed dimyristoylglycerophosphocholine/dihexanoylphosphatidyl choline was purchased from Avanti Cetyltrimethylammo-nium bromide was added to stabilize bicelles [20] The neutral pH required for bicelle stability affected the positions of some peaks in the1H-15N- heteronuclear single quantum coherence (HSQC) spectrum, although peaks in

1H-13C-HSQC were much less shifted Only peaks that could be reliably tracked were analyzed One-bond coupling constants for 1JN-NHand 1JCa-Ha were measured using a J-modulated intensity based method [21] Estimated errors were less than 1 Hz Isotropic data were recorded under standard conditions Residual dipolar couplings were determined as the difference between coupling constants

in the aligned and isotropic conditions, and were incorpor-ated at the late stage of structure calculation One-bond

1DCH values were converted to equivalent 1DNH values Data for residues with1H-15N heteronuclear NOE values less than 0.6 were excluded from the structure calculations The axial component (Da) and rhombicity (R) of the alignment tensor were estimated from the powder pattern of the residual dipolar coupling histogram [22], and from singular value decomposition analysis of the well defined secondary structures [20] The values were then optimized

using a grid search to minimize the energy of the lowest

energy structure Small deviations from the optimal

align-ment tensor had no significant effects on the total energy

and calculated structures The final values of Daand R were

19 Hz and 0.3, respectively The bond length between

pseudo-tetraatoms representing the principle order system

in the structure calculation was modified from the default

value of 1 A˚ to 10 A˚ Such modification resulted in lower

overall CNS energy and increased convergence rate

Hydrogen exchange rates were measured from a series

of 1H-15N-HSQC spectra recorded at 31.5C after the

addition of lyophilized protein in2H2O The spectra were

recorded at 11, 19, 27, 35, 43, 51, 59, 67 and 75 min The

exchange rate kex was obtained by fitting the intensity

decay to a single exponential function The protection

factor P was calculated as krc/kex, where krc is the

exchange rate for amino acid in random coil

conforma-tion [23] Potential hydrogen bond constraints set to each

observed proton in the first recorded spectrum were used

in the later stages of calculation, when a unique hydrogen

bond acceptor could be identified from the structure

model Every hydrogen bond constraint was represented

by two distance constraints of 1.7–2.3 A˚ (between H and

O) and 2.7–3.3 A˚ (between N and O) across the N–HO

hydrogen bond alignment

15N relaxation parameters (R1, R2 and NOE) were

measured by standard methods [24] The R1 relaxation

times were set to 5.4, 54.1, 162.4, 270.6, 378.8, 487.1, 595.3,

703.6, 811.8, 920.0, 1028.3 and 1136.5 ms The R2

relaxa-tion time was set to 8, 16, 32, 48, 64, 80, 96, 112, 128, 144,

160 and 176 ms The peak intensities from different

relaxation times were fit to a single exponential curve using

the program GNUPLOT

(ftp://ftp.darmouth.edu/pub/gnu-plot) Errors were reported by the fitting program.1H-15N

steady-state NOE values were obtained by recording spectra

with and without 1H saturation of 3-s duration, and by

calculating the ratios of the peak intensities The NOE

errors were estimated by the root-mean-square value of the

background noise

Structure calculations Structures were calculated using the simulated annealing method with torsion angle dynamics implemented in CNS 1.0 program [25] Employed energy functions were: quadratic harmonic potential for covalent geometry, soft square-well quadratic potentials for the experimental distance and torsion angle restraints, harmonic potentials for the 3JHNHA coupling constants and dipolar coupling constants, and a quadratic van der Waals repulsion term for the nonbonded contacts Ambiguous NOEs were assigned by NMRVIEW in an ARIA-like manner [17] and were used throughout the structure calculations Final NOEs constraints were assigned by NMRVIEW with a Nilges cut-off value 0.85, and by manual inspection Constraints for nonstereospecifically assigned prochiral protons were set to the loosest values as described by Fletcher et al [26] Only constraints from well-ordered residues (1–90) were used in structure calculation Figures were prepared usingMOLMOL [27] andGRASP[28]

R E S U L T S A N D D I S C U S S I O N

Assignments Mass spectrometry analysis of the unlabeled pY revealed that the N-terminal methionine was completely removed from the protein, which was overproduced in E coli cells using the T7 phage RNA polymerase-based overexpression system [29] The 1H-15N-HSQC spectrum of uniformly labeled pY (Fig 1) shows all the features of a well-folded, conformationally homogeneous protein, although some peaks are overlapped and exhibit weaker intensity Most of the backbone and side-chain resonances were assigned by a standard set of through-bond correlation experiments (See Materials and methods) N-Terminal Met2 and the stretch

of residues Lys28, Gln30, His32, Leu33, and Ile34 were not observed in the1H-15N-HSQC spectrum Nevertheless, the side-chain resonances of Lys28, Gln30, and Ile34 could be

Fig 1.1H-15N-HSQC spectrum of pY recor-ded at 600 MHz and 31.5 °C The protein was

in 10 m M sodium phosphate buffer, pH 5.7, and 50 m M NaCl Assignments are indicated for backbone and side-chain resonances Question marks correspond to the unassigned Gln and Asn side chains.

partially assigned in the H(CCO)NH and C(CO)NH

spectra, which correlate their resonances to successive

residues in the sequence Resonances of the methyl groups

of Leu33 were assigned to two unassigned peaks within the

Leu characteristic region in the 1H-13C-HSQC spectrum

The Hd1and He2of His32 were assigned in a similar way

using the 2D-TOCSY spectrum All these assignments were

self-consistent in the HCCH-TOCSY and 13C-NOESY

spectra After this, we concluded that three strips of weak

peaks observed in the 3D15N-NOESY spectrum belong to

the spin systems of Gln30, Leu33 and Ile34, and their HN

and N signals were assigned despite invisibility in the

1H-15N-HSQC spectrum Finally, we obtained about 97%

of both backbone and side-chain assignments The

C-terminal portion of the protein can degrade over time,

resulting in appearance of major and minor degradation

products corresponding to residues 1–98 and 1–95,

respect-ively The1H-15N-HSQC spectrum recorded after

degrada-tion of the C-terminal part of pY showed that virtually all

peaks corresponding to the residues 2–90 were unchanged,

suggesting that the core protein structure is not affected by

the C-terminal region

Structure determination

The 3D structure of pY was determined using torsion angle

simulated annealing protocol with theCNS1.0 program [25]

from a total of 2207 constraints (Table 1), with an average

of 24 constraints per residue for amino acids 1–90 Twenty

lowest-energy structures were chosen from 30 structures for

further analysis The structural statistics are shown in

Table 1 The superposed backbone coordinates of the 20

lowest energy structures are well aligned for residues 1–26

and 36–90 (Fig 2A) Thus, the core of pY (except residues

27–35) is very well defined by the NMRdata The 23

C-terminal residues are intrinsically disordered as follows from

their relaxation properties and their constraints were not

included in structure calculations The overall agreement of

the structures with the experimental data is very good, as

demonstrated by the small violations (distance violations

< 0.5 A˚, no dihedral angle violations > 5) Excluding the

poorly defined region 27–35 and terminal residues, the root

mean square deviation (r.m.s.d.) is 0.48 ± 0.14 A˚ for

backbone heavy atoms (N, Ca, C¢), and 1.19 ± 0.15 A˚

for all heavy atoms Ramachandran analysis was used to

confirm the good quality of the structure Of the residues,

81.4% have backbone conformation in the most favored

regions, and most residues in less favorable conformations

are from the poorly defined region (27–35)

Use of residual dipolar coupling in the structural

calculation

The inclusion of residual dipolar couplings (RDCs) has

been shown to improve the accuracy, as well as the precision

in NMR-based structure determination of proteins [30]

Here we also incorporate RDC information in our attempts

to accurately define the orientation of helix a1

As many NMRsignals in the loop connecting a1-b2 were

weak or invisible, only a few NOEs could be observed, this

lead to the poor definition of this loop segment Moreover,

the disorder in this region affected preceding helix a1,

resulting in a poorly defined orientation for this helix In the

20 structures calculated without RDC constraints, the C-terminus of a1 flipped away from the b-sheet, such that the interhelical angle between a1 and a2 varied over the range of 22.3 ± 2.4 After incorporating 80 RDC constraints, the interhelical angle became less variable and its value decreased to 16.2 ± 1.4 As a consequence, the backbone r.m.s.d of residues 2–26 and 36–88 were also reduced by 0.18 A˚ This calculation provides a good example of how long-range orientational constraints can help define aspects

of the global geometry under conditions where short-range NOE constraints do not provide sufficient local distance information

The Z-axis of the principal alignment tensor orients along the longest axis of the protein (Fig 2), consistent with the prediction, which requires the alignment to be mainly

Table 1 Structural statistics for 20 structures of pY.

CNS energies (kcalÆmol–1)a

r.m.s.d from idealized geometry

r.m.s.d from experimental constraints b

Residual dipolar coupling (Hz) 0.347 ± 0.019 Ramachandran analysis for 1–90 region (%)c

Residues in most favored regions 81.4 Residues in additional allowed regions 16.2 Residues in generously allowed regions 1.3

Pairwise r.m.s.d for residues 2–26, 36–87 (A˚)

a These values were estimated using CNS 1.0 The final values of the force constants used for the calculations are as follows: 1000 kcal mol –1

ÆA˚)2 for bond lengths; 500 kcal mol –1

Ærad)2 for bond angles and improper torsions; 4 kcalÆmol –1 ÆA˚)4 for the van der Waals term with the atomic radii set to 0.8 times of their CHARMM values; 50 kcalÆmol–1ÆA˚)2 for NOE-derived and hydrogen-bonding distance restraints; 200 kcalÆmol –1 Ærad)2 for dihedral angle restraints; 1 kcalÆHz)2for3J HNHA restraints; and 0.8 kcalÆmol–1ÆHz)2 for residual dipolar coupling restraints. bThe distance restraints include 408 ambiguous and 1507 unambiguous NOEs, as well as 72 hydrogen-bonding restraints for 36 hydrogen bonds Unambiguous NOEs comprising 648 intraresidue, 337 sequential (|i–j| ¼ 1), 193 medium-range (1 < |i–j| < 5) and 329 long-range (|i–j| > 4) NOEs The dihedral angle restraints involve

48 / and 27 v1 There are 653J HNHA restraints, 291D NH and 51 1

D CH residual dipolar coupling restraints c The values were cal-culated for residues 1–90 using PROCHECK [48].

induced by steric interaction with the bicelle disc [31].

Because dipolar coupling data cannot define a unique

orientation of the interatom vectors, there are four possible

orientations of each structure in the ensemble with respect to

the principal order axis

Description of structure

Figure 2(B) shows the ribbon representation of the

calcu-lated pY structure and Fig 3(A) outlines the pY sequence

with secondary structure elements The protein represents a

compact two-layered a/b structure with a babbba topology

The first layer consists of a four-stranded b-sheet, and the

second one comprises two a helices The b1 and b2 strands

are parallel, while b2, b3 and b4 are wound in an antiparallel

fashion The secondary structure elements are defined as

follows: b1, Asn3–Ser6; b2, His37–Glu43; b3, Gly46–Thr55;

b4, Gly58–Gly64; a1, Thr13–Leu25; a2, Asp69–His88

(Fig 2) Four turns were identified in the well-defined

regions: Glu43–Gly46 and Thr55–Gly58 form type I

b-turns; His66–Met69 forms a type VIII b-turn; and Ser6–

Met9 adopts a type IV turn Conformations of all X-Pro

peptide bonds were determined to be trans based on the

observation of the strong Ha

i  1–Hd

i NOEs

Potential hydrogen bonds were classified based on the

identification of slowly exchanging amide protons All

hydrogen bond acceptors, except for Ile15 and Val61, are

proposed to involve backbone oxygens from the regular

secondary structure The amide proton of Ile15, located in

the beginning of helix a1, can make a potential hydrogen

bond to Odof Thr12 Because the amide proton of Val61

points outward towards solvent, it was set to pair with the

spatially proximate side-chain oxygen Oe1of Gln83 This

proposed hydrogen bond alignment is supported by many

NOEs between the side-chain protons of Gln83 and the

amide proton of Val61

Backbone mobility

In order to investigate the dynamic properties of pY, we

collected15N relaxation data for the uniformly15N-labeled

protein The longitudinal (R1) and transversal (R2)

relaxa-tion rates, and1H-15N heteronuclear NOE values are plotted

in Fig 4 as a function of residue number No data were

obtained for the residues, which did not show amide

resonances (Met2, Lys28, Gln30, His32, Leu33, Ile34), show extremely weak intensities (Met9), or involve significant overlapping residues (Ile4, Glu10, Lys42, Lys65, Ala97, Asp102, Val109) in the 1H-15N-HSQC spectrum The dynamic properties of pY are also outlined in the structure

in Fig 5

Most parts of the protein are fairly rigid Their1H-15N NOE values are just below the theoretical maximum of 0.82 (at a 15N frequency of 60 MHz) (Figs 4 and 5) Some residues in loop b2-b3, around loop b3-b4, and in the N-terminus exhibit fast motions on the ps-ns timescale, as evident by their lower1H-15N heteronuclear NOE values (<0.6) The 23 C-terminal residues show significantly lower

1H-15N heteronuclear NOE values and decreased R2 (Fig 4), indicating that they are completely disordered in the solution

Residues around the long loop a1-b2 and the C-terminal region of helix a1 show higher than average R2 values, making many residues in this region invisible This suggests that there is significant local conformational exchange on the ms-ls timescale Such motions may have biological relevance, potentially aiding binding to the ribosome by decreasing the energy cost in the conformation rearrange-ment accompanying the binding process (see role of conserved residues below) Motions spanning a similar range were also reported for the regions responsible for interaction of Bacillus subtilis regulatory protein Spo0F with proteins KinA, Spo0B and RapB [32] A recent study has directly correlated increased motional flexibility with higher RNA-binding affinity in the double-stranded RNA-binding domain 1 (dsRBD1) of human interferon-induced kinase PKRwhen compared to dsRBD2 [33] Slow motions are also found at residues in loop b1-a1, in the turn between b4 and a2, and at residues 52–53 in b3 (Fig 5) Amino acids 52–53 are spatially close to the loop a1-b2 with distances of 4.6 A˚ and 6.2 A˚ from Caatoms of Thr52 and Ile53 to Ca atom of Pro36, respectively, and are probably affected by slow motions within this loop

The slowly exchanging hydrogens were generally found within the regular secondary structure elements that do not display fast or slow motion (Fig 5) The highest protection

of hydrogen exchange (P > 103) was observed for residues

in helix a2, indicative of this being the most stable structural element For example, residue Ile77 has the only amide proton still visible after one week of exchange into2HO

Fig 2 Structure of pY (A) Best-fit superpo-sition of the backbone atoms (N, Ca, C¢) of the

20 best structures determined for pY residues 1–90 (B) Ribbon representation of a single representative structure The orientation of the optimized principle order tensor is shown

as a frame axes.

(P > 104) Helix a1, connected to the b-sheet by two

flexible loops is less stable than helix a2, as evident from its

lower protection factor Moreover, the exchange protection

in this helix shows a bipartite pattern The amide protons at

the buried side of helix a1 are more protected, while the

solvent exposed side of helix a1 is less or not protected This

is probably due to the different solvent accessibility and

stability of the two faces of helix a1

Sequence conservation

The sequence of the E coli pY protein can be compared

with over 50 homologs found in known bacterial genomes,

and in Arabidopsis thaliana and Spinacia oleracea plant

sequences A search using FASTA [34] and BLAST [35]

revealed only a few hits with significant identity associated

with two strains of Salmonella enterica [36,37] (91%

identity), Yersinia pestis [38] (79.5% identity), and Pasteu-rella multocida[39] (69.2% identity) Most of the remaining homologs show lower sequence similarity (21–35% identity) with the E coli pY protein (summarized in Fig 3A) By contrast, a sequence identity of better than 40% is observed between homologs of ribosomal proteins

The pY-related proteins vary in length and can be grouped into short (78–142 residues) and long (174–229 residues) protein families The long proteins are similar to

pY in their N-terminal segments, and most probably adopt folds similar to the E coli pY structure In addition, moderately conserved (25–40% identity) C-terminal exten-sions within the long protein family probably contain additional ribosome binding site(s), as the isolated C-terminal region of the chloroplast-specific pY homolog from Spinacia oleracea can bind E coli ribosomal particles [8]

Fig 3 Distributionof the conserved residues

inpY (A) Conserved residues in the sequence

of pY Upper and lower sequences show

identical and similar residues, respectively,

among 44 pY-related sequences from bacteria.

Red, green and yellow residues are

100%, > 80% and > 50% conserved,

respectively Alignment was carried out by

FASTA [34] using score matrix BLOSUM 62.

Shading was carried out by GENE DOC software

(K.B Nicholas and H.B Nicholas, Jr.,

un-published data) using following similarity

groups: (1) D, E; (2) N, Q; (3) S, T; (4) K, R;

(5) F, Y, W; (6) L, I, V, M (B) Alignment of

pY with its homolog from plastids of Spinacia

oleracea Residues in red and blue are identical

and similar amino acids, respectively.

(C) Electrostatic surface of pY Red and blue

colors represent negative and positive

poten-tials, respectively Conserved charged amino

acids located on the surface are indicated.

(D) Conserved residues (> 80% similar) are

represented by cyan balls Others side chains

are shown by sticks colored blue for positive,

red for negative, and yellow for polar residues.

Role of conserved residues

Most of the conserved residues are located in the N-terminal

region of pY (residues 2–26) and towards the end of the well

structured segment of the protein (residues 67–90),

comprising strand b1, loop b1-a1, helix a1, and helix a2, respectively (Fig 3A) Many of these conserved amino acids (> 80% similar), as well as a few residues from other regions of the protein, participate in the formation of an extensive hydrophobic core between the inner surfaces of the b-sheet and the two a helices (Fig 3D) These residues include Ile4 in b1; Met9 and Ile11 in the loop b1-a1; Ile15, Val19 and Leu26 in a1; Leu33 in loop a1-b2; Ile38 and Leu40 in b2; Ile53 in b3; Leu60 and Ala62 in b4; Met69, Ile73, Leu76, Leu80 and Leu84 in a2 Conserved polar amino acids, such as Thr12 and Gln83, may be important for maintaining local conformation, as their side chains are probably involved in potential hydrogen bonds with slow exchanging amide protons: the Odof Thr12 can make a bond with HN of Ile15 in a1, and the Oe1of Gln83 with HN

of Val61 in b4

Figure 3(C) shows the molecular surface of pY color coded for electrostatic potential, along with a listing of the conserved solvent accessible residues Conserved surface features are predominantly localized within the a-helical segments of the protein The C-terminal parts of the a-helices display a conserved positively charged surface encompassing residues Arg22, Lys25, Lys28, Lys79, Arg82, Lys86 and Lys90 This basic patch can potentially target rRNA upon pY binding to the ribosomal subunit, and the most conserved Lys25, Arg82, Lys86 and Lys90 residues may represent good candidates for involvement in specific RNA recognition Unlike the C-terminal segments of the a-helices, the N-terminal segment of a2 and the adjacent turn display negatively charged surface patch comprised of Glu75, Glu67 and conserved Asp68 residues A stretch of six aspartates and glutamates is also present in the unstructured C-terminus of the protein The unstructured region of pY (C-terminal residues 91–112) does not appear

to be important for binding to ribosomes, as E coli protein YhbH, a homolog of pY, which lacks the 18 C-terminal amino acids, binds equally well to ribosomal subunits [3]

In contrast with the a-helix side of the protein, the solvent exposed face of the b-sheet side of pY consists predomin-antly of polar (Asn3, Thr5, Asn35, His37, Ser41, Thr52, Asn54, Ser63) and a few hydrophobic (Ile34, Ile39, Val48, Val59, Val61) residues (Fig 3D) Although these residues are poorly conserved, the polar/hydrophobic property of the surface seems to be retained in many Y proteins, for instance, in the distant relative, plastid-specific pY homolog from spinach (Fig 3B), which can be incorporated into

E coliribosomes during its in vitro synthesis in bacteria [8] Remarkably, such a noncharged surface is preserved in the spinach protein despite a doubling of the total number of charged residues This implies that the solvent exposed face

of the b-sheet might be buried within the interior upon binding to ribosomal subunits and contribute to the binding affinity

Comparison with other structures Many proteins with diverse functions exhibit folds related to the two-layered a/b structure of E coli pY, in which two a-helices pack against a four-stranded b-sheet However, the

E colipY structure demonstrates a unique babbba topo-logy, not presented in the SCOP database [40] Potential structural homologs of pY in the PDB were searched with the program [41] The search revealed more than 20

Fig 4 15 N R1 (A), R2 (B) an d 1 H- 15 N heteronuclear NOE values (C) as

a function of the residue number The secondary structure elements are

indicated for reference.

Fig 5 Backbone flexibility in the pY structure Residues with1H-15N

heteronuclear NOE < 0.6 are red The residues with fast transverse

relaxation rate (R 2 > 16 Hz) are shown in green, and other amino

acids are gray Slow exchanging backbone amides are shown as balls

colored according to their protection factor: navy for P > 1000, purple

for 1000 > P > 100, sky blue for 10 < P < 100.

proteins with structural similarity (z score > 2) Half of

them are nucleic acid- or nucleotide-binding proteins, with a

predominance of RNA-binding proteins amongst them

The best score of 12.2 was found for the recently determined

structure of E coli pY homolog from H influenzae [10] (see

comparison below) Most of the other proteins with

structural similarity to the pY proteins can be grouped as

follows: (a) proteins with similarity to the double-stranded

RNA-binding domain and (b) proteins with similarity to the

C-terminal domain of glycyl-tRNA synthetase

The first group includes exclusively RNA-binding

pro-teins These include the dsRBD of Xenopus laevis

RNA-binding protein A (Xlrbpa protein) (1 di2, z score 5.4) [42] in

complex with 16 bp dsRNA, and dsRBD1 of human

interferon-induced protein kinase PKR(1qu6, z score 5.2)

[43], which are most similar to pY proteins Despite the lack

of very strong sequence similarity (< 16% identity), the

r.m.s.d values are 3.7 A˚ (for 64 pairs of Ca atoms of

Xlrbpa) and 3.2 A˚ (for 70 pairs of Caatoms of PKR) In

addition, the NMRstructure of the third dsRBD from

DrosophilaStaufen protein bound to 18 nucleotide RNA

stem-loop [44], which was not picked by DALI program,

also exhibits a small r.m.s.d of 2.4 A˚ (for 63 pairs of Ca

atoms) The dsRBD is a 70–80 amino acid long motif, which

mediates selective but nonsequence specific double-stranded

RNA recognition in a variety of proteins [45] Proteins can

contain a single copy or multiple copies of dsRBDs

Superposition of the pY topology upon the folds of Xlrbpa

(Fig 6), Staufen and PKRrevealed good structural

simi-larity of pY with these classical dsRBDs, especially in the b

sheet and helix a2 segments Despite the above aspects of

overall structural similarity, structural alignment of pY with

dsRBDs (Fig 7) did not reveal high sequence similarity pY

shares with each dsRBDs about 11–16% identical and 6–

11% similar amino acids When combined, a total of 24%

of the pY residues can be found in either of three dsRBDs under comparison and 9% represent similar substitutions However, only 9% of the amino acids (E43, F47, I53, G64, A72, and L84) match conserved positions of the dsRBD family (Fig 7), and 11% hit the consensus sequence of dsRBD (not shown) [45] In turn, most of the conserved residues in pY (68%) do not match amino acids in any of the dsRBD sequences presented here

The pY structure also has a few features that are distinct from dsRBD architecture First, classical dsRBD has abbba topology, whereas pY has an extra N-terminal b strand Second, the a helices are virtually parallel in pY, but in dsRBD they make a 25–47 angle Third, the first a helix in dsRBD is moved along the axis in the C-terminal direction, placing its N-terminal tip in the range of the middle part of the second a helix Fourth, the part of the b sheet near the C-terminal regions of a helices is shorter in dsR BD, resulting in the a helices extending over the b sheet Fifth, the b1-b2 loop is longer in dsRBD than the corresponding b2-b3 loop in pY Consequently, despite overall structural and some sequence similarity between pY and dsRBDs, we are unable to directly address issues related to the common evolutionary origin of the two folds and their potential early divergence

The second group of proteins shares structural similarity with the C-terminal anticodon-binding domain of glycyl-tRNA synthetase This similarity could have functional implications for pY, given experimental support for pY-mediated translation inhibition of aminoacyl tRNA-ribo-somal A site recognition [2] However, the C-terminal anticodon-binding domain of glycyl-tRNA synthetase is made of five b strands and three a helices, and therefore, its overall topology is rather different from pY Moreover, the

a helix, which according to the structure of the complex [46] should make most of the contacts to tRNA, has no counterpart in pY

Fig 6 Structural similarity betweenpY an d dsRBD The protein

structures are shown in the same orientation as on the left panels of

Fig 3(C,D) (A) Superposition of pY (gray) with the dsRBD of

Xlrbpa protein (purple) [42] performed with DALI [41] Protein

regions and residues contacting RNA in dsRBD are indicated Cyan

colored amino acids are the same, while orange colored amino acids

are different in the complexes of Xlrbpa and Staufen proteins with their

RNA targets (B) Electrostatic surface of the Xlrbpa protein Red and

blue colors represent negative and positive potentials, respectively.

Fig 7 Structural alignment of pY (amino acids 1–89) and dsRBDs (Ec), protein pY; (Xl), fragment of Xlrbpa [42]; (Pk), the first dsRBD

of the human interferon-induced protein kinase PKR[43]; (St), the third dsRBD from Drosophila Staufen [44] Structure superposition of

pY with Xlrbpa and PKRwas performed with DALI [41] Superpo-sition with Staufen was carried out based on homology of the three dsRBDs Red and blue colors indicate identical and similar residues Conserved residues (> 50% of identity) of the pY and dsRBD families are shaded Conservation of pY is as on the Fig 3(A) Conservation of dsRBDs is adapted from FSSP database [49] Only residues with C a atoms closer than 4 A˚ are shown Note that to present the pY sequence without interruption some unaligned internal amino acids in dsRBDs are omitted in the scheme.

Structure–function relationships

Comparison of functional properties of the dsRBD and pY

proteins could provide clues towards unraveling their

relative structure–function relationships Two available

structures of dsRBD/RNA complexes have revealed details

of intermolecular protein-RNA recognition These involve

an X-ray structure of the Xlrbpa fragment complexed with

16 bp dsRNA [42] and an NMR structure of the third

dsRBD from Drosophila Staufen bound to a 18 nucleotide

RNA stem-loop [44] There is a minimal change in the

conformation of these dsRBDs upon binding their RNA

targets Even disordered loops become only partially ordered

in the Staufen complex dsRBD-dsRNA contacts are

mediated by two conserved loops between b1-b2 and b3-a2

in these complexes (Fig 6A) The first loop recognizes

predominantly 2¢-OH groups in the minor RNA groove of

dsRNA and the second loop interacts with phosphodiester

backbone in the major RNA groove In addition,

noncon-served helix 1 makes important contacts with the minor

groove in the Xlrbpa complex and with the UUCG tetraloop

in the Staufen complex Noteworthy is half of the amino

acids contacting RNA are different in both structures

Despite the apparent resemblance between structures of

dsRBDs and pY, their mode of RNA recognition should be

different for the reasons outlined below The longer length

of the a1 helix in pY relative to its counterpart in dsRBD

proteins makes potential RNA-binding surface less

acces-sible in pY for interaction with dsRNA The shorter length

of the b2-b3 loop segment within the pY fold,

correspond-ing to the b1-b2 loop in dsRBD, results in the absence of

amino acids in pY, which are critical for recognition by

dsRBD proteins of their RNA targets (Fig 6A and 7) In

addition, according to the relaxation data, the b4-a2 region

is very rigid in the pY, although corresponding b3-a2

segment in dsRBDs is flexible in Staufen [44] and PKR [33]

Therefore, it cannot easily undergo necessary motion,

required for binding of pY to RNA Further, there is little

similarity in residue distribution within the a helices of the

pY and dsRBD proteins Indeed, the conserved positively

charged residues located within the C-terminal regions of

the two a helices in the pY protein have no counterparts in

dsRBD proteins (Fig 6B) and conversely, residues K163

and K167 in Xlrbpa, which are rather conserved in dsRBDs

from various proteins and appear to be important for RNA

recognition [47], are missing in the Y protein Thus, the pY

structure most probably represents a novel functional fold

despite elements of structural and sequence similarity and

possibly common origin with dsRBD

Related structural research

While our work was approaching completion, a similar

NMR-based solution structure was published for protein

HI0257, a homolog of pY from Haemophilus influenzae [10]

The proteins pY and HI0257 have 64% sequence identity

and not surprisingly show very similar structures with the

mean backbone r.m.s.d of 1.5 A˚ in the core (residue 1–90)

A major difference in these two structures lies in the loop

a1-b2 In our structure, the loop a1-b2 shows large

amplitudes of motions in the ms-ls timescale Limited

structural constraints cause poor definition of this region in

the structure No severe intermediate motion was reported

in the HI0257 structure, although some residues, for instance Ile35, show clearly reduced peak intensities in the HSQC spectrum Therefore, intermediate motion in the loop a1-b2 probably exists in HI0257, but to a lesser extent Variations in protein sequences probably cause such different dynamic properties between the two highly homologous structures As expected, all positively charged conserved residues have been found in the structure of HI0257 and, despite larger number of negatively charged residues in the helices of HI0257, both proteins have similar electrostatic surfaces in the presumed RNA-binding site

A C K N O W L E D G M E N T S

We thank Soo Park and Anna Polonskaia for technical assistance, Aizhuo Liu for help in NMRexperiments and other members of our laboratory for fruitful discussions The work was supported by NIH grant CA46778 to DJP.

R E F E R E N C E S

1 Agafonov, D.E., Kolb, V.A., Nazimov, I.V & Spirin, A.S (1999)

A protein residing at the subunit interface of the bacterial ribo-some Proc Natl Acad Sci USA 96, 12345–12349.

2 Agafonov, D.E., Kolb, V.A & Spirin, A.S (2001) Ribosome-associated protein that inhibits translation at the aminoacyl-tRNA binding stage EMBO Rep 2, 399–402.

3 Maki, Y., Yoshida, H & Wada, A (2000) Two proteins, YfiA and YhbH, associated with resting ribosomes in stationary phase Escherichia coli Genes Cells 5, 965–974.

4 Merrick, M.J & Coppard, J.R (1989) Mutations in genes downstream of the rpoN gene (encoding r 54 ) of Klebsiella pneu-moniae affect expression from r 54 -dependent promoters Mol Microbiol 3, 1765–1775.

5 Schmidt, J., Srinivasa, B., Weglohner, W & Subramanian, A.R (1993) A small novel chloroplast ribosomal protein (S31) that has

no apparent counterpart in the E coli ribosome Biochem Mol Biol Int 29, 25–31.

6 Zhou, D.X & Mache, R (1989) Presence in the stroma of chloroplasts of a large pool of a ribosomal protein not structurally related to any Escherichia coli ribosomal protein Mol Gen Genet.

219, 204–208.

7 Johnson, C.H., Kruft, V & Subramanian, A.R (1990) Identi-fication of a plastid-specific ribosomal protein in the 30S subunit

of chloroplast ribosomes and isolation of the cDNA clone encoding its cytoplasmic precursor J Biol Chem 265, 12790– 12795.

8 Bubunenko, M.G & Subramanian, A.R (1994) Recognition of novel and divergent higher plant chloroplast ribosomal proteins

by Escherichia coli ribosome during in vivo assembly J Biol Chem 269, 18223–18231.

9 Tan, X., Varughese, M & Widger, W.R (1994) A light-repressed transcript found in Synechococcus PCC 7002 is similar to a chloroplast-specific small subunit ribosomal protein and to a transcription modulator protein associated with sigma 54 J Biol Chem 269, 20905–20912.

10 Parsons, L., Eisenstein, E & Orban, J (2001) Solution structure of HI0257, a bacterial ribosome binding protein Biochemistry 40, 10979–10986.

11 Jansson, M., Li, Y.C., Jendeberg, L., Anderson, S., Montelione, B.T & Nilsson, B (1996) High-level production of uniformly15 N-and13C-enriched fusion proteins in Escherichia coli J Biomol NMR 7, 131–141.

12 Johnson, B.A & Blevins, R.A (1994) NMRView: a computer program for the visualization and analysis of NMRdata J Bio-mol NMR 4, 603–614.

13 Bax, A & Grzesiek, S (1993) Methodological advances in protein

NMR Acc Chem Res 26, 131–138.

14 Neri, D., Szyperski, T., Otting, G., Senn, H & Wu¨thrich, K.

(1989) Stereospecific nuclear magnetic resonance assignments of

the methyl groups of valine and leucine in the DNA-binding

domain of the 434 repressor by biosynthetically directed fractional

13

C labeling Biochemistry 28, 7510–7516.

15 Vuister, G.W & Bax, A (1993) Quantitative J correlation: a new

approach for measuring homonuclear three-bond J HNHA coupling

constants in 15 N-enriched proteins J Am Chem Soc 115, 7772–

7777.

16 Garrett, D.S., Kuszewski, J., Hancock, T.J., Lodi, P.J., Vuister, G.W.,

Gronenborn, A.M & Clore, G.M (1994) The impact of direct

refinement against three-bond HN-CaH coupling constants on

protein structure determination by NMR J Magn Reson B 104,

99–103.

17 Nilges, M., Macias, M.J., O’Donoghue, S.I & Oschkinat, H.

(1997) Automated NOESY interpretation with ambiguous

dis-tance restraints: the refined NMRsolution structure of the

pleckstrin homology domain from beta-spectrin J Mol Biol 269,

408–422.

18 Hansen, M.R., Mueller, L & Pardi, A (1998) Tunable alignment

of macromolecules by filamentous phage yields dipolar coupling

interactions Nat Struct Biol 5, 1065–1074.

19 Tjandra, N & Bax, A (1997) Measurement of dipolar

contribu-tions to 1J CH splittings from magnetic-field dependence of J

modulation in two-dimensional NMRspectra J Magn Reson.

124, 512–515.

20 Losonczi, J.A., Andrec, M., Fischer, M.W.F & Prestegard, J.M.

(1999) Order matrix analysis of residual dipolar coupling using

singular value decomposition J Magn Reson 138, 334–342.

21 Tjandra, N & Bax, A (1997) Direct measurement of distances

and angles in biomolecules by NMRin a dilute liquid crystalline

medium Science 278, 1111–1114.

22 Clore, G.M., Gronenborn, A.M & Bax, A (1998) A robust

method for determining the magnitude of the fully asymmetric

alignment tensor of oriented macromolecules in the absence of

structural information J Magn Reson 133, 216–221.

23 Bai, Y., Milne, J.S., Mayne, L & Englander, S.W (1993) Primary

structure effects on peptide group hydrogen exchange Proteins 17,

75–86.

24 Farrow, N.A., Muhandiram, R., Singer, A.U., Pascal, S.M.,

Kay, C.M., Gish, G., Shoelson, S.E., Pawson, T., Forman-Kay,

J.D & Kay, L.E (1994) Backbone dynamics of a free and

phos-phopeptide-complexed Src homology 2 domain studied by 15 N

NMRrelaxation Biochemistry 33, 5984–6003.

25 Bru¨nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P.,

Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M.,

Pannu, N.S et al (1998) Crystallography and NMRsystem: a

new software suite for macromolecular structure determination.

Acta Crystallogr D 54, 905–921.

26 Fletcher, C.M., Jones, D.N.M., Diamond, R & Neuhaus, D.

(1996) Treatment of NOE constraints involving equivalent or

nonstereoassigned protons in calculations of biomacromolecular

structures J Biomol NMR 8, 292–310.

27 Koradi, R., Billeter, M & Wu¨thrich, K (1996) MOLMOL: a

program for display and analysis of macromolecular structures.

J Mol Graph 14, 51–55.

28 Nicholls, A., Sharp, K.A & Honig, B (1991) Protein folding and

association: insights from the interfacial and thermodynamic

properties of hydrocarbons Proteins 11, 281–296.

29 Studier, F.W., Rosenberg, A.H., Dunn, J.J & Dubendorff, J.W.

(1990) Use of T7 RNA polymerase to direct expression of cloned

genes Methods Enzymol 185, 60–89.

30 Prestegard, J.H., al-Hashimi, H.M & Tolman, J.R (2000) NMR

structures of biomolecules using field oriented media and residual

dipolar couplings Q Rev Biophys 33, 371–424.

31 Zweckstetter, M & Bax, A (2000) Prediction of sterically induced alignment in a dilute liquid crystalline phase: aid to protein structure determination by NMR J Am Chem Soc 122, 3791– 3792.

32 Feher, V.A & Cavanagh, J (1999) Millisecond-timescale motions contribute to the function of the bacterial response regulator protein Spo0F Nature 400, 289–293.

33 Nanduri, S., Rahman, F., Williams, B.R & Qin, J (2000) A dynamically tuned double-stranded RNA binding mechanism for the activation of antiviral kinase PKR EMBO J 19, 5567–5574.

34 Pearson, W.R & Lipman, D.J (1988) Improved tools for biological sequence comparison Proc Natl Acad Sci USA 85, 2444–2448.

35 Altschul, S.F., Gish, W., Miller, W., Myers, E.W & Lipman, D.J (1990) Basic local alignment search tool J Mol Biol 215, 403– 410.

36 Parkhill, J., Dougan, G., James, K.D., Thomson, N.R., Pickard, D., Wain, J., Churcher, C., Mungall, K.L., Bentley, S.D., Holden, M.T et al (2001) Complete genome sequence of a mul-tiple drug resistant Salmonella enterica serovar Typhi CT18 Na-ture 413, 848–852.

37 McClelland, M., Sanderson, K.E., Spieth, J., Clifton, S.W., Latreille, P., Courtney, L., Porwollik, S., Ali, J., Dante, M., Du, F.

et al (2001) Complete genome sequence of Salmonella enterica serovar Typhimurium LT2 Nature 413, 852–856.

38 Parkhill, J., Wren, B.W., Thomson, N.R., Titball, R.W., Holden, M.T., Prentice, M.B., Sebaihia, M., James, K.D., Churcher, C., Mungall, K.L et al (2001) Genome sequence of Yersinia pestis, the causative agent of plague Nature 413, 523–527.

39 May, B.J., Zhang, Q., Li, L.L., Paustian, M.L., Whittam, T.S & Kapur, V (2001) Complete genomic sequence of Pasteurella multocida, Pm70 Proc Natl Acad Sci USA 98, 3460–3465.

40 Murzin, A.G., Brenner, S.E., Hubbard, T & Chothia, C (1995) SCOP: a structural classification of proteins database for the investigation of sequences and structures J Mol Biol 247, 536–540.

41 Sander, C & Schneider, R (1991) Database of homology-derived protein structures and the structural meaning of sequence align-ment Proteins 9, 56–68.

42 Ryter, J.M & Schultz, S.C (1998) Molecular basis of double-stranded RNA–protein interactions: structure of a dsRNA-bind-ing domain complexed with dsRNA EMBO J 17, 7505–7513.

43 Nanduri, S., Carpick, B.W., Yang, Y., Williams, B.R & Qin, J (1998) Structure of the double-stranded RNA-binding domain of the protein kinase PKRreveals the molecular basis of its dsRNA-mediated activation EMBO J 17, 5458–5465.

44 Ramos, A., Grunert, S., Adams, J., Micklem, D.R., Proctor, M.R , Freund, S., Bycroft, M., St Johnston, D & Varani, G (2000) RNA recognition by a Staufen double-stranded RNA-binding domain EMBO J 19, 997–1009.

45 St Johnston, D., Brown, N.H., Gall, J.G & Jantsch, M (1992) A conserved double-stranded RNA-binding domain Proc Natl Acad Sci USA 89, 10979–10983.

46 Sankaranarayanan, R., Dock-Bregeon, A.C., Romby, P., Caillet, J., Springer, M., Rees, B., Ehresmann, C., Ehresmann, B & Moras,

D (1999) The structure of threonyl-tRNA synthetase-tRNA (Thr) complex enlightens its repressor activity and reveals an essential zinc ion in the active site Cell 97, 371–381.

47 Bycroft, M., Grunert, S., Murzin, A.G., Proctor, M & St John-ston, D (1995) NMRsolution structure of a dsR NA binding domain from Drosophila staufen protein reveals homology to the N-terminal domain of ribosomal protein S5 EMBO J 14, 3563– 3571.

48 Laskowski, R.A., Rullmannn, J.A., MacArthur, M.W., Kaptein, R.

& Thornton, J.M (1996) AQUA and PROCHECK-NMR: pro-grams for checking the quality of protein structures solved by NMR J Biomol NMR 8, 477–486.

49 Holm, L & Sander, C (1996) Mapping the protein universe Science 273, 595–603.