determination of multiple φ-torsion angles in proteins by selective and extensive 13c labeling and...

Determination of Multiple f-Torsion Angles in Proteins by Selective

c1

slsiiplttatmuHo21

rqaucs

s

tboTb

psssp

ap-

Journal of Magnetic Resonance139,389–401 (1999)Article ID jmre.1999.1805, available online at http://www.idealibrary.com on

and Extensive 13C Labeling and Two-Dimensional Solid-State NMR

Mei Hong

Department of Chemistry, University of Massachusetts, Amherst, Massachusetts 01003

Received January 27, 1999; revised April 28, 1999

We describe an approach to efficiently determine the backbone desirable to employ more extensive isotopic labeling

dereanbl

nce

anebyacco

it

p frome byp ereo stincts therd -dp tods too entsio olid-s emlp gic-a dl pin-n off eds( tr ongc ys ed icch con-s n as

a einsa ns theb cani ilel dardp sive( ctly-b therb cing

onformation of solid proteins that utilizes selective and extensive3C labeling in conjunction with two-dimensional magic-angle-pinning NMR. The selective 13C labeling approach aims to reduceine broadening and other multispin complications encountered inolid-state NMR of uniformly labeled proteins while still enhanc-ng the sensitivity of NMR spectra. It is achieved by using specif-cally labeled glucose or glycerol as the sole carbon source in therotein expression medium. For amino acids synthesized in the

inear part of the biosynthetic pathways, [1-13C]glucose preferen-ially labels the ends of the side chains, while [2-13C]glycerol labelshe Ca of these residues. Amino acids produced from the citric-cid cycle are labeled in a more complex manner. Information onhe secondary structure of such a labeled protein was obtained byeasuring multiple backbone torsion angles f simultaneously,

sing an isotropic–anisotropic 2D correlation technique, theNCH experiment. Initial experiments for resonance assignment

f a selectively 13C labeled protein were performed using 15N–13CD correlation spectroscopy. From the time dependence of the

5N–13C dipolar coherence transfer, both intraresidue and inter-esidue connectivities can be observed, thus yielding partial se-uential assignment. We demonstrate the selective 13C labelingnd these 2D NMR experiments on a 8.5-kDa model protein,biquitin. This isotope-edited NMR approach is expected to fa-ilitate the structure determination of proteins in the solidtate. © 1999 Academic Press

Key Words: 13C labeling; proteins; torsion angle; resonance as-ignment; solid-state NMR.

INTRODUCTION

Recently solid-state NMR has been increasingly appliehe structure determination of noncrystalline and disordiological molecules. Techniques that yield both distancerientational constraints to high precision are now availahe sensitivity of these NMR experiments is greatly enhay the enrichment of NMR-sensitive isotopes such as13C and

15N. So far, most NMR structural studies of solid peptidesroteins have adopted a specific-labeling approach whermall number of sites in the molecule is labeled in eynthesis and then one or two distances or torsion angletraints are measured in each experiment (1–7). To determinerotein structures more efficiently and comprehensively,

389

todd

e.d

da

hn-

is

roaches and to extract multiple structural constraintsach experiment. The second condition may be fulfillederforming two-dimensional (2D) NMR experiments whne dimension separates and identifies the chemically diites through their isotropic chemical shifts, while the oimension yields the structural parameters (8). The first conition can be fulfilled by uniform13C and 15N labeling of theroteins, which is used routinely now in solution NMRetermine the structure of globular proteins (9, 10). In solid-tate NMR, uniform15N labeling has also been utilizedbtain information such as the orientation of protein segm

n the membrane bilayer (11). However, uniform13C labelingf proteins leads to several spectroscopic difficulties for state NMR. First, the one-bond13C–13C J-couplings and thultispin dipolar couplings that are effective in fully13C-

abeled proteins cause significant line broadening. TheJ-cou-ling-induced line broadening cannot be removed by mangle spinning (MAS) alone (12, 13) while the dipolar-induce

ine broadening is only removed completely at very high sing speeds (.25 kHz) (14). Thus the spectral resolution

ully 13C-labeled proteins is limited under normal MAS spe3–15 kHz). Second, the weak13C–13C dipolar couplings thaeflect long-range13C–13C distances are obscured by the strouplings between directly bonded13C spins. This occurs bpin diffusion within the dense13C spin network and by thipolar truncation mechanism (15, 16). These spectroscophallenges, combined with the high cost of13C labeling, haveindered the routine measurement of multiple structuraltraints in solid proteins from a single experiment and oingle sample.In this paper we explore selective and extensive13C labeling

s a means to facilitate the NMR investigation of solid protnd demonstrate the application off torsion measurement ouch a labeled model protein, ubiquitin. By exploitingiosynthetic pathways of amino acids in bacteria, we

ncorporate13C labels into some sites in the proteins wheaving other sites unlabeled. Easily feasible with the stanrotocol for protein expression, the selective and extenS&E) labeling approach reduces the number of direonded13C spin pairs or larger clusters in the molecule, eiy not labeling one of the carbons in the chain, or by redu

1090-7807/99 $30.00Copyright © 1999 by Academic Press

All rights of reproduction in any form reserved.

the labeling levels of the sites involved. The principal advan-t s int mos iq-u tad eap ueT oss perm ena tionp thefi eN

s enbb e-s aa tals sigm otes dd nta

1

ths eriaU lye acT coy cy( thg bog rting coC vatC ten clew te,s is ap vat do elima inee bonF al fro3 w

l pro-d phatep

elyl diaw .T thel ob-l glu-c umlp leds ableJ cidcd no twos thec ten-s eling

lec-t aysa d. Ther ted bya el

390 MEI HONG

age of this S&E labeling approach for solid-state NMR liehe resolution enhancement that it achieves. This is detrated by the13C spectra of several differently labeled ubitin (M r 5 8565, 76residues). The abundant X-ray crysiffraction (17) and solution-state NMR (18–21) data that arvailable for ubiquitin allow us to verify the13C labelingrotocol and to demonstrate the solid-state NMR techniqhe modest molecular weight of ubiquitin also makes it pible to assess the realistic sensitivities of the NMR exents. Similar selective13C labeling approaches have bepplied in solution NMR for measuring precise relaxaarameters (22, 23); however, to our best knowledge this isrst demonstration of selective13C labeling for solid-statMR investigation of protein structure.Using a13C selectively and15N uniformly labeled ubiquitin

ample, we demonstrate the residue-specific measuremackbone torsion anglesf by correlating the N–H and Ca–Ha

ond orientations (24). The f-angle experiment yielded sitpecific information on the backbone conformation withccuracy of65 to 630°, by comparison with the X-ray crystructure. Since structure determination requires the asent of spectral resonances to the residues in the pr

equence, we applied a15N–13C 2D correlation technique anemonstrate that it is possible to achieve partial sequessignment directly in the solid state.

RESULTS AND DISCUSSION

. Selective and Extensive13C Labeling

The selective and extensive labeling method relies onpecificity of the amino acid biosynthetic pathways in bactnder aerobic conditions, bacteria can utilize glucose or grol as the sole carbon source to synthesize the 20 aminohis is achieved via three major enzymatic pathways: glysis, the pentose phosphate pathway, and the citric-acid25, 26). As an example, we outline the metabolic fates oflucose C1 carbon in Fig. 1. During glycolysis, the six-carlucose is oxidized to the three-carbon pyruvate, convelucose C1 into the pyruvate methyl carbon, C3. Since glu1 and C6 contribute equally and indistinguishably to pyru3, only 50% of pyruvate results from glucose C1. Pyruvaext oxidized to acetyl-CoA, which enters the citric-acid cyhere two key intermediates,a-ketoglutarate and oxaloacetaerve as the precursors for 10 amino acids. Oxaloacetateroduced directly from phosphoenolpyruvate and pyru

hrough several anaplerotic reactions. Botha-ketoglutarate anxaloacetate are reused in the cyclical reactions, thus labultiple carbon sites, particularly Ca, Cb, and Cg, of these 10mino acids. In contrast, amino acids synthesized by the lnzymatic reactions have fewer glucose C1-derived carrom pyruvate C3, the side chain ends of Ala, Val, and Leu

abeled. Gly loses the glucose C1 label in its synthesis-phosphoglycerate, which also generates Ser and Cys

n-

l

s.-i-

t of

n

n-in

ial

e.

c-ids.l-cleeng

seeis,

lsote

ng

ars.

remith

abeled side chain ends. The 3 aromatic amino acids areuced from phosphoenolpyruvate via the pentose phosathway and are labeled at Cb and several ring positions.Therefore, we can obtain a protein that is selectiv

abeled in13C by supplementing the bacterial growth meith 13C specifically labeled glucose such as [1-13C]glucosehe absence of labeling for certain sites will alleviate

ine broadening, spin diffusion, and dipolar truncation prems mentioned above. The chemical symmetry of theose molecule can be exploited to control the maximabeling level to 50%. Thus, the probability of a13C spinair is at most 25% even for two directly bonded labeites. This further reduces the influence of the undesir-couplings. For amino acids produced from the citric-aycle, the labeling levels of many sites are reduced to;25%ue to the loss of some13C labels as CO2 and the distributiof labels between chemically indistinguishable sites byymmetric intermediates (succinate and fumarate) ofycle. The lower labeling levels compensate for the exiveness of labeling in these amino acids. When a lab

FIG. 1. Amino acid biosynthetic pathways in bacteria, illustrating seive and extensive13C labeling of proteins. Most compounds in the pathwre precursors for the 20 amino acids. The fate of the glucose C1 is traceesulting labeled carbons in the amino acid precursors are designasterisks. The smaller asterisks in oxaloacetate anda-ketoglutarate indicat

ower labeling levels.

l d.T ds rgs ves

a epc rifya ed2 lvei wei wn

f Ase men-t tl the[ as

eledw ri-s edt nge ylc nge

traint f-

p la partialc

391TORSION ANGLES IN PROTEINS

evel of 100% is desired, [1, 6-13C]glucose may be usehis selective labeling approach can also be extendepecifically labeled glycerol and pyruvate, which undeimilar enzymatic reactions and have been used in seolution NMR studies of protein dynamics (22, 23).To demonstrate this selective and extensive13C labeling

pproach, we obtained two ubiquitin samples that wereressed from bacterial media supplemented with [1-13C]glu-ose and [2-13C]glycerol as the sole carbon source. To vend compare the distribution of13C-labeled sites, we employD 13C–1H correlation experiments on these samples disso

n aqueous solution (Fig. 2). The labeled carbon sitesdentified by using the solution NMR chemical shifts kno

FIG. 2. Solution NMR 13C–1H heteronuclear correlation spectra of13C1atterns. (a) CH3 region of 13C1-ubiquitin. (b) CH3 region of 13C2-ubiquitin.ssignment is based on the known chemical shifts of this protein (27) and is iomplementary nature of the13C1 and13C2 labeling schemes. Areas of th

tooral

x-

dre

or this protein (27) and are compiled in Tables 1 and 2.xpected, the two labeling schemes exhibit partial comple

arity, in that many sites labeled by [1-13C]glucose are noabeled by [2-13C]glycerol, and vice versa. For example, in1-13C]glucose-derived ubiquitin (heretofore referred to

13C1-ubiquitin), most side chain methyl carbons are labithout directly bonded13C neighbors (Fig. 2a). In compaon, the [2-13C]glycerol-derived ubiquitin (heretofore referro as 13C2-ubiquitin) exhibits poor methyl carbon labelixcept for Ile Cd1 and Thr Ca (Fig. 2b). The isolated metharbon labels are potentially useful for measuring long-ra

13C–15N and 13C–13C distances between residues to conshe protein tertiary structure. The [2-13C]glycerol scheme pre

iquitin and13C2-ubiquitin, verifying the selective and extensive13C labelingCa–Ha region of 13C1-ubiquitin. (d) Ca–Ha region of 13C2-ubiquitin. Spectracated for some peaks to illustrate the selectivity of the labeling and theme type of carbons are enclosed by dashed lines.

-ub(c)ndie sa

e iletu hepa ntsu e-ld

ll bt . N

c s inu notd en thee ng isn

i-ms ng,e f theb eling

TABLE 1

. For amina erl well with thet

LE

. For amina eleda ets), showg d s

392 MEI HONG

rentially labels the13Ca of the linear-chain amino acids, whhe [1-13C]glucose scheme does not. Isolated13Ca sites areseful for measuringf torsion angles to directly probe trotein secondary structure. Those amino acids with both13Ca

nd 13CO labels are suitable forc torsion angle measuremesing the NCCN technique (28, 29) and techniques that corr

ate the 13CO chemical shift anisotropy with the13Ca–1Ha

ipolar coupling (30, 31).Overall, the predicted labeling patterns are borne out we

he solution NMR spectra, as shown in Tables 1 and 2

13C Labeling Patte

Residue

13C labels

Theory Expt.

Gly None NoneCysb b N.A.Ser b bAla b bLeu a, d1, d2 a, d1, d2Val g1, g2 g1, g2His CO1 d2Phe b, d1, d2 b, g, d1, d2Tyr b, d1, d2 b, d1, d2, zTrpb b, d1, d2 N.A.

Note.The theoretically predicted labeled sites are compared with the ecids produced from the citric-acid cycle (right half of the table), two type

abeling level of;25% (square brackets). The experimental peak intensheoretical predictions. For amino acids produced from the citric-acid c

a Carbon sites predicted to be labeled but cannot be confirmed by th13C–b Amino acids not present in ubiquitin. The experimental data are no

TAB13C Labeling Patte

Residue

13C labels

ReTheory Expt.

Gly a aCysb a N. A.Ser a aAla a aLeu COa, b, g b, gVal a, b a, bHis a d2Phe a, g, e1 a, g, e1Tyr a, g, e1 a, g, e1, zTrpb a, d2, z3 N. A.

Note.The theoretically predicted labeled sites are compared with the ecids produced from the citric-acid cycle, three types of labeling levelst;50%, and sites in round brackets are labeled at,25%. The experimentaleneral agreement with the theoretical predictions. For amino acids pro

a Carbon sites predicted to be labeled but cannot be confirmed by th13C–b Amino acids not present in ubiquitin. The experimental data are no

yo

omparison is made for the two missing amino acid typebiquitin and for the unprotonated carbons, which areetectable by the experiment. The good agreement betwexperiment and the prediction indicates that label scrambliegligible. Based on residual signals of glycine Ca in the

15N–13C correlation spectra of13C1-ubiquitin (Fig. 6), we estate that scrambling contributes;10% of the labeling. Low

crambling is crucial for securing the selectivity of labelispecially for amino acids produced from the linear part oiosynthetic pathways. If desired, the more extensive lab

of [1-13C]Glucose

sidue

13C labels

Theory Expt.

Glu [COa], a, b, g, 2 a, b, gln [COa], a, b, g, 2 a, b, gro [COa], a, b, g, 2 a, b, g, 2rg [COa], a, b, g, 2, 2 a, 2, g, 2, 2sp [COa], a, b, [ga] a, bsn [COa], a, b, [ga] a, bet [COa], a, b, [g] a, b, 2hr [COa], a, b, [g] a, b, [g]ys 2, b, [g], d, e 2, b, d, ele [CO1], a, 2, g1 a, 2, g1,

g2, [d1] g2, [d1]

rimentally observed ones obtained from the solution NMR data in Fig. 2aof labeling levels are predicted: the normal level of;50% (no brackets), and a low, classified as strong (no brackets) and weak (square brackets), agree, dashes are used to indicate the unlabeled sites.orrelation experiment due to the lack of directly bonded protons.

ailable (N. A.).

2of [2-13C]Glycerol

ue

13C labels

Theory Expt.

[COa], (a), [b], 2, da [a], [b], 2,[COa], (a), [b], 2, da [a], [b], 2[COa], (a), [b], 2, d [a], [b], 2, d[COa], (a), [b], 2, d [a], [b], 2, d[COa], [a], (b), [ga] [a], [b][COa], [a], (b), [ga [a], [b][COa], [a], (b), [g] [a], 2, [g], e[COa], [a], (b), [g] [a], [b], ga, 2, [g], (d), [e] a, 2, g, [d], e[COa], [a], b, (g1), [d1] [a], [b], [g1], d1

rimentally observed ones obtained from the solution NMR data in Fig. 2bopredicted: the normal labeling level is;100% sites in square brackets are labk intensities, classified as strong (no brackets) and weak (square brack

ced from the citric-acid cycle, dashes are used to indicate the unlabeleites.orrelation experiment due to the lack of directly bonded protons.

ailable (N. A.).

rn

Re

GPAAAMTLI

xpes oitiesyclee1H ct av

rn

sid

GluGlnProArgAspAsnMetThrLysIle

xpearepeadu

e1H ct av

o d bs eda rodu

d bat g( eine cal

iala hisl effic r thr omp

2 ed

r thel rd eledu beleds s afi incea turala sityd er-t f thes ffects

u sitydt uald


f the amino acids from the citric-acid cycle can be avoideupplementing the minimal medium with unlabeled intermtes of the citric-acid cycle or the unlabeled amino acid pcts themselves (32).The selective and extensive13C labeling approach (22, 23)

iffers from other selective labeling methods that requireerial auxotrophs (33–35) and from random fractional labelin36, 37). It is easy to implement with the standard protxpression protocol and is very versatile, since specifi

13C-labeled glucose, glycerol, and pyruvate are commercvailable in many forms. The feasibility and versatility of t

abeling approach, together with the resulting potentialiency in the NMR structure determination, compensate foeduced control in the placement of the isotopic labels cared to site-specific labeling methods.

FIG. 3. Solid-state13C CPMAS spectra of differently labeled ubiquitin sbiquitin, 10 mg, 64 scans. (c) [1-13C]glucose,15N-labeled ubiquitin, 15 mg, 1istributions of spectra (c) and (d) compared to (a) and (b) reflect the se

he effects of selective labeling. Spectrum (b), of the uniformly labeledipolar interaction.

yi--

c-

lyly

-e-

. One-Dimensional13C NMR Spectra of Selectively LabelUbiquitin

To demonstrate the effects of S&E13C labeling on theesolution of the solid-state NMR spectra and to estimateabeling levels, we acquired 1D13C CPMAS spectra of fouifferent ubiquitin samples. The spectrum of the unlabbiquitin (Fig. 3a) serves as a reference to those of the laamples in three ways. First, its intensity distribution ingerprint of the complete carbon skeleton of the protein, sll carbons in the molecule contribute equally at the nabundance (1.1%) level. Deviations from this total intenistribution would thus reflect the preferential labeling of c

ain carbon sites. Second, the homogeneous linewidth opectrum is minimal at the natural abundance, since the e

ples. (a) Unlabeled ubiquitin, 35 mg, 4096 scans. (b) [U-13C6]glucose,15N-labeledscans. (d) [2-13C]glycerol, 15N-labeled, 20 mg, 128 scans. The different intentivity of labeling. Assignments for several resolved signals are indicated to illustrateple, has the lowest resolution due to the13C–13C scalar couplings and the resid

am28lecsam

of the 13C–13C scalar couplings and residual dipolar couplingsa trup f th[ ed

b)e idu

olaa ortrf higs ds( cod orsw pem akr plem

ig.3 umtc wi pp( thesp -t in

nssd t thcu th

rum( ans ofes ).T acb

rper idti atecf eds linwc cer lyr

3

witS 2D

correlation techniques for measuring structural constraints sucha lr a-s uchai uec s( ds,wfsi pic–a ofe edt

.4s ,o ,a met s arep si ide-b r ofip withfmTd hichi heA u ints o-r hec ly47i typeI

–4d,r queh et rareo att n then ds rd-i

w ix

394 MEI HONG

re absent. Third, the signal-to-noise ratio of the specrovides a benchmark against which the labeling levels o

1-13C]glucose and [2-13C]glycerol schemes can be estimatFor U-13C, 15N-labeled ubiquitin, the spectrum (Fig. 3

xhibits significant line broadening, as expected when res13C–13C dipolar and scalar couplings are present. This dipnd scalar-coupling induced line broadening was also repecently for crystalline organic compounds (14, 38). It wasound that to remove the effects of these couplings, verypinning speeds (.25 kHz) and proton decoupling fiel;200 kHz) needed to be employed. These experimentalitions are currently only feasible on the smallest MAS rothich have limited sample volumes and thus reduced exental sensitivity. In addition, such high spinning speeds m

otor-synchronized NMR pulse sequences difficult to iment.Under [1-13C]glucose labeling, the ubiquitin spectrum (F

c) exhibits reduced intensities than the reference spectrhe Ca region between 50 and 65 ppm and at;40 ppm. Inomparison, the [2-13C]glycerol labeling scheme yields lontensities in the side chain region between 15 and 35Fig. 3d). These different intensity distributions reflectelectivity of labeling. For example, the Gly and Val13Ca

eaks are conspicuously missing in the13C1-ubiquitin specrum, while the Ile 13Cg2 carbons are mostly unlabeled

13C2-ubiquitin. (These resonances are assigned based oolution NMR data (27) and are confirmed by the 2D15N–13Cpectra shown later.) In the carbonyl region,13C2-ubiquitinisplays substantially higher intensities, suggesting thaarbonyl carbons are labeled at higher levels than in13C1-biquitin. In general, based on the relative intensities of

13C1- and 13C2-ubiquitin spectra to the reference spectFig. 3a), and taking into account the sample amountsignal-averaging times, we find approximately a factornhancement in the labeling levels of the [2-13C]glycerolcheme (50–90%) over the [1-13C]glucose scheme (20–45%his is consistent with the predictions based on the aminoiosynthetic pathways.Both selectively labeled ubiquitin samples exhibit sha

esonances than the uniformly labeled sample. The linewn spectra (c) and (d) range from 0.8 ppm for the unprotonarbonyl carbons and the mobile methyl carbons, to;1.3 ppmor the CH and CH2 groups. In contrast, the fully labelample under the same experimental conditions exhibitsidths of ;1 ppm for CH3 groups and.2 ppm for CH2

arbons. These linewidth differences confirm the importaneducing directly bonded13C spin pairs for obtaining highesolved solid-state NMR spectra.

. Site-Specific Determination of Multiplef Torsion Anglesby HNCH Correlation

The high spectral sensitivity and resolution achieved&E 13C labeling makes possible a variety of solid-state

me

.

alr-ed

h

n-,ri-e-

in

m

the

e

e

d2

id

rhsd

e-

of

h

s torsion angles (24, 28, 30, 31, 39, 40) and for sequentiaesonance assignment (41–43). We first demonstrate the feibility of residue-specific torsion-angle determination on sselectively labeled protein by measuring backbonef angles

n 13C2-ubiquitin. The recently introduced HNCH techniqorrelates the N–H and the Ca–Ha dipolar coupling tensor24) in order to extract the mutual orientation of the two bonhich reflects thef torsion angle about the N–Ca bond. The-dependent dipolar patterns are manifested in thev1 dimen-ion and separated in thev2 dimension according to the Ca

sotropic chemical shifts of the residues. Due to this isotronisotropic correlation, the HNCH technique is capablextracting multiplef torsion angles simultaneously, provid

hat the13Ca chemical shift dispersion is sufficient.The HNCH spectrum of13C2-ubiquitin is displayed in Fig

. The projection onto thev2 dimension and thev1 crossections of four selected sites are shown. In thev2 dimensionnly carbons directly bonded to nitrogen spins, i.e., Ca and COre observed, since a short15N–13C double- and zero-quantuxcitation time was used (0.47 ms). In thev1 dimension, only

he intensity envelopes connecting the tips of the sidebandlotted in order to emphasize thef-angle-induced difference

n the intensity distribution. The spectra consist of 14 sands and one center band, as determined by the numbet 1

ntervals measured. The dipolar cross section of Val26 Ca (67pm) displays a dip in the central intensity and is best-fit5 260° (Fig. 4b). This corresponds to thea-helix confor-ation, which is in agreement with the crystal structure (17).he dipolar cross section of Thr12 and Thr14 Ca (62 ppm) isominated by a sharp intensity maximum at the center, w

s the signature for theb-sheet conformation (Fig. 4c). Tla46 (53 ppm) cross section exhibits a broad high platea

he center and corresponds tof 5 120° or 1100°, bothuggesting ab-turn structure (Fig. 4d). This is also corrobated by the crystal structure (17). Even in the absence of trystal structure, this result, combined with the fact that Gs the neighboring residue, would strongly suggest a9–III 9 b-turn (44).

The experimental spectra are each best fit (Figs. 4bight) with two f-torsion angles because the HNCH technias an intrinsic double degeneracy (24). However, one of thes

wo angular values can often be ruled out based on itsccurrence in nature (45). The best-fit simulations are found

he minimal root-mean-square deviations (rmsd) betweeormalized experimental spectrumSi ,exp and the normalizeimulated onesSi(f). The rmsd curve is calculated accong to

rmsd~f! 5 H 1

NOi51

N

@Si ,exp2 Si~f!# 2J 1/ 2

, [1]

here the summationi is carried out over the central s

s (T alcl fov rvec a ip gee

( ear-m nearf beeno

mea-s h ther reas).

Pc . Tht ystal srb lack oas


pinning sidebands and the center band in each spectrum46).he outer eight side bands were excluded in the rmsd c

ation since their intensities are low and relatively similararious f angles. The global minima of each rmsd cuorrespond to the best-fit torsion angles. A pair of minimresent for each rmsd curve, which reflects the double deracy of the HNCH technique. For thea-helix (b) andb-turn

FIG. 4. HNCH spectrum of13C2-ubiquitin for determining thef torsionro19, (b) Val26, (c) Thr12 and Thr14, and (d) Ala46. The assignment isorrelation spectra. The root-mean-squared deviations (rmsd) betweenorsion angles (dashed lines) are obtained at the minimum rmsd value.esults are indicated by the solid lines. For cross section (c), two solidest-fit simulations are shown in the right column of (b–d). The effectivemide proton. The sample was spun at 4209 Hz. A15N–13C coherence transfpectrum was acquired in 14 h.

u-r

sn-

d) conformations, an additional pair of angles exhibit ninimum rmsd values. This results from an accidental,

ourfold degeneracy in these angular regions, which hasbserved in 2D (f, f) rmsd plots previously (46).To assess the angular resolution of the technique, we

ured the experimental noise and used its intersection witmsd curve to define an angular uncertainty (shaded a

les. The experimental dipolar cross sections (v1 dimension) are the Ca of (a)tained from solution NMR data (27) and is confirmed by the solid-state15N–13C 2Dexperiment and simulation are displayed in the middle column of (b–d)e best-fite experimental uncertainties are indicated by the shaded areas. The crtructures are shown that correspond to Thr12 (f 5 2120°) and Thr14 (f 5 2101°). TheH dipolar coupling can be extracted from the Pro19 slice (a) due to itsf the

ime of 475ms was used to select carbons directly bonded to the amide15N. The

angob

theTh

lineC–

er t

Based on this criterion, the Val26 spectrum has the highesta cisT essb shel elyh utl .Ti iblt yc s.

eni allu otoi lop(s ledo ioni rsa

s tt -sitr c-t thc theH fp urew thm evew ,t oe e 2H ls . T3 al-a enis

4

s bN cost stram pet o-nT ignm osl redbT –

C .D ansfermct ueN gt

ut trumy lly orc nec-

m es aree typea on thec g pat-t quiredi

396 MEI HONG

ngular resolution, while the Ala46 result is the least prehe accuracy of the NMR measurement can also be assy comparing the NMR-derived torsion angle values (da

ines) with the crystal structure (solid lines). We find relativigh accuracy for thea-helical andb-sheet conformations, b

ow accuracy for the positively valued,b-turn conformationhis may partly result from the fact that the Ca signal of Ala46

s not sufficiently resolved from other sites. It is also posshat the flexibility of theb-loop makes the NMR or the X-rarystal structure an average of a number of conformationAn internal check for the precision of the HNCH experim

s provided by the dipolar cross sections of the chemicnique amino acid, proline. Due to the lack of the amide pr

n proline, its HNCH cross section exhibits a narrow enveFig. 4a) that reflects the effective Ca–Ha dipolar couplingcaled by the homonuclear decoupling sequence. Knowf this effective C–H coupling makes the HNCH simulat

ndependent of any adjustable parameter other than the tonglef.The advantage of the selective and extensive13C labeling

cheme compared to uniform13C labeling lies in the fact thahe uniformly labeled sample has practically no singleesolution in the Ca region to yield any useful HNCH sperum. Although higher spinning speeds could improvehemical shift resolution, they are incompatible withNCH sequence, which requires a sufficient number ot 1

oints as defined by the MREV-8 cycle time to be measithin each rotor period. This slow-spinning requirement isain limitation of the torsion-angle technique, becauseith the enhanced resolution afforded by13C selective labeling

he v2 dimension still does not resolve all Ca resonances. Txtract more torsion angles, it is necessary to extend thNCH technique into a third dimension, where15N chemicahift may be employed to further disperse the resonancesD experiment should not entail prohibitively long signveraging time, since the dipolar dimension of the experim

ncremented in a constant-time fashion (24), requires only amall number of points to be measured.

. Resonance Assignment by15N–13C Correlation

The structure elucidation of extensively labeled proteinMR requires the assignment of the measured structuraltraints to the protein sequence (47). So far, solid-state NMRechniques for resonance assignment have been demonainly on small molecules such as amino acids and oligo

ides (42, 48–51), and few solid-state NMR studies of resance assignment for proteins have been shown (41, 43, 52).he S&E 13C labeling approach facilitates resonance assent by enhancing the spectral resolution. As a first dem

tration of this feasibility, we carried out a 2D15N–13C corre-ation experiment (42) where the coherence is transferetween13C and15N via the recoupled dipolar interaction (53).he spectra yield sequence-specific assignment based oni–

e.ed,d

e

tyne

ge

ion

e

e

den

D

his

t,

yn-

tedp-

-n-

N

ai–Ni11–Ci11

a – and –Ni–COi–Ni11–COi11– connectivitiesue to the distance-dependent nature of this coherence trechanism, one can first identify the N–Ca and COi21–Ni

ross peaks by using a short15N–13C mixing time, then identifywo- and three-bond15N–13C spin pairs such as the intraresid–Cb and the interresidue Ci21

a –Ni peaks using a longer mixinime (42).

We applied this15N–13C correlation technique to13C2-ubiq-itin. Two 15N–13C spectra, acquired with a15N–13C mixing

ime of 0.6 and 2.3 ms, are displayed (Fig. 5). Each specields several dozens of resonances that are either partiaompletely resolved. Based on the chemical shifts, the con

FIG. 5. 15N–13C 2D correlation spectrum of13C2-ubiquitin with a15N–13Cixing time of (a) 0.6 ms and (b) 2.3 ms. Several dozens of resonancither partially or completely resolved in each spectrum. Amino-acidssignment and partial sequential assignment were achieved basedonnectivity patterns, the characteristic chemical shifts, and the labelinerns derived from Fig. 2 and Tables 1 and 2. The spectra were each acn 13 h.

tivity patterns, the labeling pattern specific to this sample, andt ens trua eld

cha ls

CwC tlyb orm bleW aw e Cr tot aP gP ni s tP nnt ea int es st ls idbic dc ns Rc s ae ssim elp er edt ths

n-m plel misiu6 seS roCTt luc

rec exp ep cals tea cle

13C correlation schemes will also be useful for type assignment.E aper( ten-s thatac -p pro-t

c fula pro-t atti onsb allyl n theb ex-p entf theb rbonp elingl romt onsh thes

esa hichf en-h tech-n intsp con-s oteinc eH f them e se-q nts isp fr pec-t ltra-h nol-o ationt nancea le.

hisw ern,Pc len hes om-


he time-dependence of the resonance signals, we can ideveral sequential connectivities. For example, in the speccquired with the short mixing time (a), the six most downfi

15N resonances can be assigned to prolines due to theircteristic chemical shifts. Among these peaks, the signa

13C chemical shifts of;50 ppm can be assigned to prolined,hile those at;62 and;67 ppm are proline Ca. Both Cd anda are labeled by the [2-13C]glycerol scheme and are direconded to the amide15N, thus they are observed at the shixing time. At this point, only type assignment is possihen the longer15N–13C mixing time is used, we observeeak but clear cross peak between two of the three prolina

esonances. This unambiguously assigns the two peakswo consecutive prolines in the ubiquitin sequence, Pro37ro38. Thus the remaining proline signals must belonro19. One can continue along the connectivity patter

dentify the cross peak between Pro37 and Ile36, as well aro19–Ser20 cross peak. Another observed sequential co

ivity is Gly47–Ala46–Phe45 (43). In addition to interresidussignment, the long mixing-time experiment also allows

raresidue assignment to be made to some extent, sincidechain resonances appear in the same15N cross section ahe primary Ca signal but have more upfield13C chemicahifts. One example of the intraresidue assignment is provy the prolines, whose Cb site, two bonds from the amide15N,

s clearly shown in Fig. 5b. Similarly, the Thr Cb’s and Cg’san also be assigned, since Thr Cb’s have uniquely downfielhemical shifts. These assignments were made based oolid-state spectra alone, without utilizing solution NMhemical shift information. The latter, however, is useful aventual measure of the uniqueness of the solid-state aent given the achievable linewidths (41). In these spectra, th

imiting values of the full width at half maximum are;0.8pm for 13C and;2.5 ppm for 15N. This resolution is quiteasonable for an unoriented powder sample and is reflecthe partial separation of the six glycine resonances inpectra.Although the selective13C labeling makes complete assigent from a single sample not possible, with two sam

abeled in a complementary fashion, one can retrieve theng information to some extent. For example, the13C1-ubiq-itin spectrum acquired at the long15N–13C mixing time (Fig.b) exhibits the Cb signals of Ala, Ser, and Pro. Among theer20 Cb can be assigned based on its cross peak with Pa, and Ala46 Cb is identified by its cross peak with Gly47 Ca.hr9 can be assigned by its cross peak with Leu8 Ca. Similarly,

he sequential connectivities Thr66–Ser65 and Pro19–Gan be identified.To assign the13C and 15N spectra of a solid protein mo

ompletely, additional correlation experiments that fullyloit the labeling patterns are required. For example, 3Deriments that correlate the Ca and sidechain carbon chemihifts with the backbone15N are useful for more complemino acid type assignment of the resonances. Homonu

tifym

ar-at

t.

thendtotoheec-

-the

ed

the

ngn-

bye

ss-

,19

18

-x-

ar

xperiments along these lines will be detailed in a future p54). On the other hand, variations of the selective and exive labeling approach can be exploited to label carbonsre either unlabeled or labeled at low levels by the [1-13C]glu-ose and [2–13C]glycerol labeling schemes. This will be imortant to achieve complete resonance assignment of the

eins.

CONCLUSIONS

We have shown that selective and extensive13C labeling inonjunction with high-resolution solid-state NMR is a usepproach to the efficient structure investigation of solid

eins. The selective13C labeling primarily refers to the fact thhe statistical probability of directly bonded13C–13C spin pairss close to 0% while the probability of one of the two carbeing 13C is significant. This is achieved by using specific

abeled glucose or glycerol as the sole carbon source iacterial growth media. We found, both theoretically anderimentally, that the selectivity of labeling is most promin

or the 10 amino acids produced from the linear part ofiosynthetic pathways. In these amino acids, many caairs with 45–90% and 1.1% (i.e., natural abundance) lab

evels are found. In comparison, in amino acids derived fhe more complex citric-acid cycle, many adjacent carbave 50% or lower labeling levels for both sites, thuspin-pair probabilities are below 25%.The selective and extensive13C labeling approach providgeneral method of spectral narrowing and editing, w

acilitates resonance assignment in solid proteins. Theanced spectral resolution makes many solid-state NMRiques practical for measuring multiple structural constraer experiment and per sample. We showed here thattraints on the backbone conformation of such a labeled pran be obtained by measuring thef torsion angles using thNCH technique. The angular resolution and accuracy oeasurement were analyzed. We also showed that thuence-specific assignment of these structural constraiartially feasible by 2D15N–13C correlation. The extent oesonance assignment is still limited at this point by the sral resolution due to the lack of ultrafast spinning and uigh magnetic field. Combining the advance of these techgies, various heteronuclear and homonuclear correl

echniques, and 3D spectroscopy, more complete resossignment of proteins in the solid state is well foreseeab

MATERIALS AND METHODS

The three13C, 15N-labeled ubiquitin samples used in tork were custom-synthesized by VLI-Research (MalvA). The human ubiquitin gene was expressed inEscherichiaoli. Cells were grown in M9 minimal medium, with the soitrogen source being15N-labeled ammonium sulfate, and tole carbon source being one of the following three c

p( ress olp eiT tropT a au tedt asf inlb in lo

orv ukeD efo

c scs

aB anceM 120kC lyr cttd -i de-c pliedo

lycineC

398 MEI HONG

ounds: [1-13C]glucose, [2-13C]glycerol, and [U-13C6]glucoseCambridge Isotope Labs, Inc.). For a typical ubiquitin expion experiment, a 5-liter culture was induced and the saste was purified by column chromatography to homogenhe purity of the final products was analyzed by gel elechoresis and amino acid composition analysis to be.99%.he unlabeled ubiquitin sample was purchased from Sigmsed directly without purification. All samples were hydra

o ;30% (w/w) prior to the NMR experiments. Hydration wound to be important for narrowing the spectral lines, may decreasing the conformational heterogeneity presentphilized samples (55–57).The solution-state13C–1H 2D correlation experiments f

erifying the labeling schemes were carried out on a BrPX-300 spectrometer using a 5-mm inverse probe. A r

FIG. 6. 15N–13C 2D correlation spectrum of13C1-ubiquitin with a15N–13Ca signals, whose presence indicates low-level scrambling in the13C labeling

-idty.-

nd

yy-

r-

used INEPT pulse sequence was used (58). The sampleonsist of 0.7 mM ubiquitin dissolved in a 90% H2O/10% D2Oodium acetate buffer at pH 4.5.All solid-state NMR experiments were performed on

ruker DSX-300 spectrometer using a 4-mm triple-resonAS probe. Typical radiofrequency field strengths were

Hz for 1H decoupling, 65 kHz for13C, and 50 kHz for15N.ross polarization from1H to 13C was achieved with a linear

amped (59, 60) RF field on the13C channel and with a contaime of 0.5 ms. The FIDs were detected under1H TPPMecoupling (61). The evolution period of thef torsion exper

ment employed a MREV-8 pulse train for homonuclearoupling. Otherwise, continuous-wave decoupling was apn the1H channel. Typical recycle delays were 3 s.The 1D 13C CPMAS experiments and the 2D15N–13C cor-

ing time of (a) 0.6 ms and (b) 2.3 ms. Dashed circle indicates the weak gheme.

mixsc

relation experiments were performed under 7 kHz of magic-ab i-m ren( do( citet da

thb Ha thes dob

ee e-ccs ymd d o4 .85m onra ,s .

them ctlmr ni thuf

ANpd thes at5 eas tiow

e aE 373T quis

3. R. R. Ketchem, W. Hu, and T. A. Cross, High resolution conforma-

1

1

1

1

1

1

1

1

1

1

2

2

2


ngle spinning. The spinning speed was controlled to62 Hzy a Bruker MAS controller. The15N–13C correlation experents were carried out using the pulse sequence in refe

42), except that the15N–13C REDOR mixing period consistef a single 13C 180° pulse and multiple15N 180° pulses62, 63). A total of 4 and 16 rotor periods were used to exhe double- and zero-quantum15N–13C coherence of one-bonnd multiple-bond15N–13C pairs, respectively.The f torsion angle experiment was performed using

asic pulse sequence in Ref. (24), except that the apparent N–nd Ca–Ha dipolar couplings were doubled relative topinning speed to achieve high angular resolution. Theling was achieved by moving the 180° pulses on the13C and

15N channels as a function oft 1 during the constant-timvolution period (46). Meanwhile, MREV-8 homonuclear doupling (64) was applied for the entire rotor period on the1Hhannel. The MREV-8 90° pulse length was 2.2ms and thehort-window delay was 0.9ms. They were optimized baximizing the sideband intensities of the leucine13Ca signaletected under MREV-8 decoupling. At a spinning spee209 Hz, 16t 1 points were collected at an increment of 14s, so that the maximum evolution time corresponded to

otor period. Before thet 2 period, a TOSS sequence (65) waspplied to remove the carbonyl sidebands in thev2 dimensionince they overlap with some of the aliphatic resonancesTo eliminate uncertainties in the MREV-8 scaling factor,agnitude of the one-bond C–H dipolar coupling was direeasured using a DIPSHIFT pulse sequence (66–68). The

esult was further corroborated by the proline Ca cross section thef-angle spectra. The one-bond C–H couplings wereound to be 106 1 kHz.

The f-angle spectra were simulated using a FORTRrogram described previously (24). A long-range Ca–H(N)ipolar coupling of 1.2 kHz was taken into account inimulation. The theoreticalf-angle spectra were calculated° increments over the entire 360° range. The root-mquare deviations between the experiment and the simulaere calculated in MATLAB as a function of thef-angle.

ACKNOWLEDGMENTS

This work was supported by the NSF Materials Research Sciencngineering Center at UMass and by a NSF POWRE grant (MCB-9870he author thanks Tauseef Butt for competent expression of the ubiamples and W. Hu for making the MAS simulation program available.

REFERENCES

1. F. Creuzet, A. McDermott, R. Gebhard, K. van der Hoef, M. B.Spijker-Assink, J. Herzfeld, J. Lugtenburg, M. H. Levitt, andR. G. Griffin, Determination of membrane protein structure byrotational resonance NMR: Bacteriorhodopsin, Science 251,783–786 (1991).

2. K.-J. Shon, Y. Kim, L. A. Colnago, and S. J. Opella, NMR studies ofthe structure and dynamics of membrane-bound bacteriophagePf1 coat protein, Science 252, 1303–1305 (1991).

ce

e

u-

f

e

y

s

n-ns

nd).

tin

tion of Gramicidine A in a lipid bilayer by solid-state NMR, Science261, 1457–1460 (1993).

4. L. M. McDowell, C. A. Klug, D. D. Beusen, and J. Schaefer, Ligandgeometry of the ternary complex of 5-enolpyruvvylshikimate-3-phosphate synthase from rotational-echo double-resonance NMR,Biochemistry 35, 5395–5403 (1996).

5. J. Kummerlen, J. D. vanBeek, F. Vollrath, and B. Meier, Localstructure in spider dragline silk investigated by two-dimensionalspin-diffusion nuclear magnetic resonance, Macromolecules 29,2920–2928 (1996).

6. H. W. Long, and R. Tycko, Biopolymer conformational distributionsfrom solid-state NMR: a-helix and 310-helix contents of a helicalpeptide, J. Am. Chem. Soc. 120, 7039–7048 (1998).

7. J. R. Long, J. L. Dindot, H. Zebroski, S. Kiihne, R. H. Clark, A. A.Campbell, P. S. Stayton, and G. P. Drobny, A peptide that inhibitshydroxyapatite growth is in an extended conformation on the crys-tal surface, Proc. Natl. Acad. Sci. USA 95, 12083–12087 (1998).

8. K. Schmidt-Rohr and H. W. Spiess, “Multidimensional Solid-StateNMR and Polymers,” Academic Press, San Diego (1994).

9. G. M. Clore and A. M. Gronenborn, Two-, three-, and four-dimen-sional NMR methods for obtaining larger and more precise three-dimensional structures of proteins in solution, Annu. Rev. Biophys.Chem. 20, 29–63 (1991).

0. J. Cavanagh, W. J. Fairbrother, A. G. P. III, and N. J. Skelton,“Protein NMR Spectroscopy: Principles and Practice,” AcademicPress, San Diego (1996).

1. T. A. Cross and S. J. Opella, Solid-state NMR structural studies ofpeptides and proteins in membranes, Curr. Opinions Struct. Biol. 4,574–581 (1994).

2. M. M. Maricq and J. S. Waugh, NMR in rotating solids, J. Chem.Phys. 70, 3300–3316 (1979).

3. S. K. Straus, T. Bremi, and R. R. Ernst, Resolution enhancement byhomonuclear J decoupling in solid-state MAS NMR, Chem. Phys.Lett. 262, 709–715 (1996).

4. B. A. Tounge, A. K. Mehta, and K. W. Zilm, High field CP MAS offully 13C labeled polycrystalline compounds, 38th ExperimentalNMR Conference, Orlando, 1997.

5. R. G. Griffin, P. R. Costa, J. D. Gross, M. Hatcher, M. Hong, J. Hu,L. Mueller, C. M. Rienstra, S. Fesik, and J. Herzfeld, Structuralstudies with solid-state NMR: Assignments, distances and torsionangles, Experimental NMR Conference, Asilomar, CA, 1998.

6. S. Kiihne, M. A. Mehta, J. A. Stringer, D. M. Gregory, J. C. Shiels,and G. P. Drobny, Distance measurements by dipolar recouplingtwo-dimensional solid-state NMR, J. Phys. Chem. A 102, 2274–2282 (1998).

7. S. Vijay-Kumar, C. E. Bugg, and W. J. Cook, Structure of ubiquitinrefined at 1.8 Å resolution., J. Mol. Biol. 194, 531–544 (1987).

8. D. L. Stefano and A. J. Wand, Two-dimensional 1H NMR study ofhuman ubiquitin: A main chain directed assignment and structureanalysis, Biochemistry 26, 7272–7281 (1987).

9. P. L. Weber, S. C. Brown, and L. Mueller, Sequential proton NMRassignments and secondary structure identification of human ubiq-uitin, Biochemistry 26, 7282–7290 (1987).

0. A. C. Wang and A. Bax, Determination of the backbone dihedralangles f in human ubiquitin from reparameterized empirical Karplusequations, J. Am. Chem. Soc. 118, 2483–2494 (1996).

1. M. Ernst and R. R. Ernst, Heteronuclear dipolar cross-correlatedcross relaxation for the investigation of side-chain motions, J.Magn. Reson. 110, 202–213 (1994).

2. D. M. LeMaster and D. M. Kushlan, Dynamical mapping of E. coli

thioredoxin via 13C NMR relaxation analysis, J. Am. Chem. Soc.

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

sional solid-state NMR spectra of uniformly labeled proteins, J. Bi-

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

6

6

400 MEI HONG

118, 9255–9264 (1996).

3. A. L. Lee, J. L. Urbauer, and A. J. Wand, Improved labeling strategyfor 13C relaxation measurements of methyl groups in proteins,J. Biomol. NMR 9, 437–440 (1997).

4. M. Hong, J. D. Gross, and R. G. Griffin, Site-resolved determinationof peptide torsion angle phi from the relative orientations of back-bone N–H and C–H bonds by solid-state NMR, J. Phys. Chem. B101, 5869–5874 (1997).

5. A. L. Lehninger, D. L. Nelson, and M. M. Cox, “Principles ofBiochemistry,” Worth Publishers, New York (1993).

6. L. Stryer, “Biochemistry,” W.H. Freeman, New York (1995).

7. A. C. Wang, S. Grzesiek, R. Tschudin, P. J. Lodi, and A. Bax,Sequential backbone assignment of isotopically enriched proteinsin D2O by deuterium-decoupled HA(CA)N and HA(CACO)N, J. Bi-omol. NMR 5, 376–382 (1995).

8. P. R. Costa, J. D. Gross, M. Hong, and R. G. Griffin, Solid-StateNMR Measurement of c in Peptides: A NCCN 2Q-HeteronuclearLocal Field Experiment, Chem. Phys. Lett. 280, 95–103 (1997).

9. X. Feng, M. Eden, A. Brinkmann, H. Luthman, L. Eriksson, A.Graslund, O. N. Antzutkin, and M. H. Levitt, Direct determination ofa peptide torsion angle c by double-quantum solid-state NMR,J. Am. Chem. Soc. 119, 12006–12007 (1997).

0. Y. Ishii, T. Terao, and M. Kainosho, Relayed anisotropy correlationNMR: Determination of dihedral angles in solids, Chem. Phys. Lett.256, 133–140 (1996).

1. T. Fujiwara, T. Shimomura, and H. Akutsu, Multidimensional solid-state nuclear magnetic resonance for correlating anisotropic inter-actions under magic-angle spinning conditions, J. Magn. Reson.124, 147–153 (1997).

2. M. Hong and K. Jakes, Selective and extensive 13C labeling of amembrane protein for solid-state NMR investigation, J. Biomol.NMR 14, 71–74.

3. D. C. Muchmore, L. P. McIntosh, C. B. Russell, D. E. Anderson, andF. W. Dahlquist, Expression and nitrogen-15N labeling of proteinsfor proten and nitrogen-15 nuclear magnetic resonance, MethodsEnzymol. 45–73 (1989).

4. D. M. LeMaster, Deuteration in protein proton magnetic resonance,Methods Enzymol. 177, 23–73 (1989).

5. D. S. Waugh, Genetic tools for selective labeling of proteins witha-15N-amino acids, J. Biomol. NMR 8, 184–192 (1996).

6. A. J. Wand, J. L. Urbauer, R. P. McEvoy, and R. J. Bieber, Internaldynamics of human ubiquitin revealed by 13C-relaxation studies ofrandomly fractionally labeled protein, Biochemistry 35, 6116–6125(1996).

7. A. J. Wand, R. J. Bieber, J. L. Urbauer, R. P. McEvoy, and Z. Gan,Carbon relaxation in randomly fractionally 13C-enriched proteins, J.Magn. Reson. 108, 173–175 (1995).

8. M. T. Zell, B. E. Pudden, D. J. W. Grant, and E. J. Munson,Assignment techniques for understanding polymorphism in crys-talline organic compounds with solid-state NMR, 39th Experimen-tal NMR Conference, Asilomar, CA, 1998.

9. D. Weliky and R. Tycko, Determination of peptide conformations bytwo-dimensional magic angle spinning NMR exchange spectros-copy with rotor synchronization, J. Am. Chem. Soc. 118, 8487–8488 (1996).

0. X. Feng, P. J. E. Verdegem, Y. K. Lee, D. Sandstrom, M. Eden, P.Bovee-Geurts, W. J. d. Grip, J. Lugtenburg, H. J. M. d. Groot, andM. H. Levitt, Direct determination of a molecular torsion angle in themembrane protein rhodopsin by solid-state NMR, J. Am. Chem.Soc. 119, 6853–6857 (1997).

1. R. Tycko, Prospects for resonance assignments in multidimen-

omol. NMR 22, 239–251 (1996).

2. M. Hong and R. G. Griffin, Resonance assignment for solid pep-tides by dipolar-mediated 13C/15N correlation solid-state NMR,J. Am. Chem. Soc. 120, 7113–7114 (1998).

3. S. K. Straus, T. Bremi, and R. R. Ernst, Experiments and strategiesfor the assignment of fully 13C/15N-labeled polypeptides by solidstate NMR, J. Biomol. NMR 12, 39–50 (1998).

4. T. E. Creighton, “Proteins: Structures and Molecular Properties,”W.H. Freeman, New York (1993).

5. G. N. Ramachandran and V. Sasisekharan, Conformation ofpolypeptides and proteins, Adv. Protein Chem. 23, 283–437(1968).

6. M. Hong, J. D. Gross, C. M. Rienstra, R. G. Griffin, K. K. Kumashiro,and K. Schmidt-Rohr, Coupling amplification in 2D MAS NMR andits application to torsion angle determination in peptides, J. Magn.Reson. 129, 85–92 (1997).

7. G. Wagner and K. Wuthrich, Sequential resonance assignments inprotein 1H nuclear magnetic resonance spectra, J. Mol. Biol. 155,347–366 (1982).

8. M. Baldus and B. H. Meier, Total correlation spectroscopy in thesolid state. The use of scalar couplings to determine the through-bond connectivity, J. Magn. Reson. A 121, 65–69 (1996).

9. B. Q. Sun, C. M. Rienstra, P. R. Costa, J. Herzfeld, J. Williamson,and R. G. Griffin, 3D 15N–13C–13C chemical shift correlation spec-troscopy in rotating solids, J. Am. Chem. Soc. 119, 8540–8546(1997).

0. A. Lesage, C. Auger, S. Caldarelli, and L. Emsley, Determination ofthrough-bond carbon–carbon connectivities in solid-state NMRusing the INADEQUATE experiment, J. Am. Chem. Soc. 119, 7867–7868 (1997).

1. Z. Gu and S. J. Opella, 1H–15N–13C triple-resonance solid-stateNMR experiments for peptides and proteins, Rocky Mountain Con-ference on Analytical Chemistry, Denver, 1998.

2. A. E. McDermott, T. Polenova, G. Montelione, K. Zilm, E. Paulsen,R. Martin, and G. Coker, Multidimensional solids NMR of a uni-formly labeled protein, 40th Experimental NMR Conference, Or-lando, 1999.

3. T. Gullion and J. Schaefer, Rotational echo double resonanceNMR, J. Magn. Reson. 81, 196–200 (1989).

4. M. Hong, Resonance assignment of 13C/15N labeled proteins bytwo- and three-dimensional magic-angle-spinning NMR, J. Biomol.NMR, in press (1999).

5. P. L. Poole and J. L. Finney, Hydration-induced conformational andflexibility changes in lysozyme at low water content, Int. J. Biol.Macromol. 5, 308–310 (1983).

6. S. D. Kennedy and R. G. Bryant, Structural effects of hydration:Studies of lysozyme by 13C solids NMR, Biopolymers 29, 1801–1806 (1990).

7. R. B. Gregory, M. Gangoda, R. K. Gilpin, and W. Su, The influenceof hydration on the conformation of lysozyme by solid-state 13C-NMR spectroscopy, Biopolymers 33, 513–519 (1993).

8. A. Bax and G. A. Morris, Refocused INEPT, J. Magn. Reson. 42,501 (1981).

9. B. H. Meier, Cross polarization under fast magic-angle spinning:Thermodynamic considerations, Chem. Phys. Lett. 188, 201–207(1992).

0. S. Hediger, V. H. Meier, and R. R. Ernst, Cross polarization underfast magic angle spinning using amplitude-modulated spin-locksequences, Chem. Phys. Lett. 213, 627–635 (1993).

1. A. E. Bennett, C. M. Rienstra, M. Auger, K. V. Lakshmi, and R. G.

Griffin, Heteronuclear decoupling in rotating solids, J. Chem. Phys.

6

6

6

6

NMR spectra in spinning samples, J. Chem. Phys. 77, 1800–1809

6

6

6


103, 6951–6958 (1995).

2. T. Gullion and J. Schaefer, Detection of weak heteronucleardipolar coupling by rotational-echo double-resonance nuclearmagnetic resonance, in “Advances in Magnetic Resonance”(W. S. Warren, Ed.), pp. 57– 83, Academic Press, San Diego(1989).

3. C. A. Michal and L. W. Jelinski, REDOR 3D: Heteronuclear distancemeasurements in uniformly labeled and natural abundance solids,J. Am. Chem. Soc. 119, 9059–9060 (1997).

4. W.-K. Rhim, D. D. Elleman, and R. W. Vaughan, Analysis of multi-ple-pulse NMR in solids, J. Chem. Phys. 59, 3740–3749 (1973).

5. W. T. Dixon, Spinning-sideband-free and spinning-sideband-only

(1982).

6. M. G. Munowitz, R. G. Griffin, G. Bodenhausen, and T. H. Huang,Two-dimensional rotational spin-echo NMR in solids: Correlation ofchemical shift and dipolar interactions, J. Am. Chem. Soc. 103,2529–2533 (1981).

7. A. C. Kolbert, D. P. Raleigh, M. H. Levitt, and R. G. Griffin, Two-dimensional spin-echo nuclear magnetic resonance in rotating sol-ids, J. Chem. Phys. 90, 679–689 (1989).

8. J. E. Roberts, G. S. Harbison, M. G. Munowitz, J. Herzfeld, andR. G. Griffin, Measurement of heteronuclear bond distances inpolycrystalline solids by solid-state NMR techniques, J. Am. Chem.Soc. 109, 4163–4169 (1987).

determination of multiple φ-torsion angles in proteins by selective and extensive 13c labeling and...

Documents