NMR Solution Structure of the Pathogenesis-related Protein...

NMR Solution Structure of the Pathogenesis-relatedProtein P14a

CeÂsar FernaÂndez1, Thomas Szyperski1, Thierry BruyeÁre2, Paul Ramage3

Egon MoÈ singer2 and Kurt WuÈ thrich1*

1Institut fuÈ r Molekularbiologieund Biophysik, EidgenoÈssischeTechnische Hochschule-HoÈnggerberg, CH-8093 ZuÈrichSwitzerland2Sandoz Agro AGAgrobiology, CH-4108Witterswil, Switzerland3Sandoz Pharma AG, CH-4002Basel, Switzerland

The nuclear magnetic resonance (NMR) structure of the 15 kDa pathogen-esis-related protein P14a, which displays antifungicidal activity and isinduced in tomato leaves as a response to pathogen infection, was deter-mined using 15N/13C doubly labeled and unlabeled protein samples. Inall, 2030 conformational constraints were collected as input for the dis-tance geometry program DIANA. After energy-minimization with theprogram OPAL the 20 best conformers had an average root-mean-squaredeviation value relative to the mean coordinates of 0.88 AÊ for the back-bone atoms N, Ca and C0, and 1.30 AÊ for all heavy atoms. P14a containsfour a-helices (I to IV) comprising residues 4 to 17, 27 to 40, 64 to 72 and93 to 98, a short 310-helix of residues 73 to 75 directly following helix III,and a mixed, four-stranded b-sheet with topology �3x, ÿ2x, �1, contain-ing the residues 24-25, 53 to 58, 104 to 111 and 117 to 124. These regularsecondary structure elements form a novel, complex a � b topology inwhich the a-helices I, III and IV and the 310-helix are located above theplane de®ned by the b-sheet, and the a-helix II lies below this plane. Thea-helices and b-strands are thus arranged in three stacked layers, whichare stabilized by two distinct hydrophobic cores associated with the twolayer interfaces, giving rise to an ``a-b-a sandwich''. The three-dimen-sional structure of P14a provides initial leads for identi®cation of the sofar unknown active sites and the mode of action of the protein, which isof direct interest for the generation of transgenic plants with improvedhost defense properties.

# 1997 Academic Press Limited

Keywords: plant pathogenesis-related proteins; NMR; protein structure;protein dynamics; a-b-a sandwich*Corresponding author

Introduction

In response to pathogen attack or other biotic andabiotic stresses, plants accumulate a number ofantifungal proteins. These include thionins, ribo-some-inactivating proteins, 2S-storage albumins,defensins and a variety of pathogenesis-related(PR) proteins (Van Loon, 1985; Bol et al., 1990). PRproteins were ®rst detected in tobacco plants thatare hypersensitive to tobacco mosaic virus (VanLoon & Van Kammen, 1970; Gianinazzi et al.,1970). Subsequently, several other PR proteinshave been detected and characterized in monocoty-ledonous and dicotyledonous plant species, andtheir important roles in the stress response ofplants has been well documented (Ryals et al.,1994). It is typical for PR proteins to be highly re-sistant to digestion by proteolytic enzymes,suggesting that high intrinsic stability enables

Abbreviations used: P14a, pathogenesis-relatedprotein 14a from tomato leaves; PR proteins,pathogenesis-related proteins; NMR, nuclear magneticresonance; NOE, nuclear Overhauser effect; 2D,2-dimensional; 3D, 3-dimensional; NOESY, NOEspectroscopy; COSY, correlation spectroscopy; 2QF-,two quantum ®ltered; TOCSY, total correlationspectroscopy; ct, constant time; ppm, parts per million;3JHNa, vicinal spin-spin coupling constant between thebackbone amide proton and the a-proton; 3JNb, vicinalspin-spin coupling constant between the backboneamide nitrogen and one of the b-protons; 3Jab, vicinalspin-spin coupling constant between the a-proton andone of the b-protons; 3Jcacd, vicinal spin-spin couplingconstant between the a-carbon atom and a d-carbonatom; r.m.s.d., root-mean-square deviation; T1,longitudinal relaxation time; T2, transverse relaxationtime; T1r, relaxation time in the rotating frame; REDAC,use of redundant dihedral angle constraints; CD,circular dichroism.

J. Mol. Biol. (1997) 266, 576±593

0022±2836/97/080576±18 $25.00/0/mb960772 # 1997 Academic Press Limited

JMB MS 2838 [10/2/97]

survival of these proteins in harsh natural environ-ments, such as those in vacuolar compartments orintercellular spaces. On the basis of serologicalproperties and sequence homologies, the PR pro-teins have been grouped into seven families, PR-1to PR-7{. While chitinase and b-1,3-glucanase ac-tivity could be assigned to PR-3 and PR-2 proteins,respectively (Joosten & De Wit, 1989; Fischer et al.,1989), only little is known about the biologicalfunction of proteins in the group PR-1, which werethe ®rst PR proteins discovered and which includeP14a. The 135 residue protein P14a, which is thesubject of the present investigation, becomes themost abundant acid-extractable tomato leaf proteinupon infection with pathogens. P14a has also beenfound in trace amounts in healthy plants at theonset of bloom (Fraser, 1981), and during naturalaging of tomato leaves (Camacho Henriquez &SaÈnger, 1982). Similar to other PR proteins, P14acan be induced in tomato leaves by treatment withsalicylate (Christ & MoÈsinger, 1989; Linthorst,1991), which may play the role of a second messen-

ger during systemic acquired resistance (Ryals et al.,1994).

Alexander et al. (1993) showed that transgenictobacco plants which constitutively express the PR-1a gene exhibit increased tolerance to the fungalpathogens Phytophthora parasitica var. nicotianae andPeronospora tabacina, thereby providing the ®rst,albeit indirect evidence that PR-1 proteins exhibitantifungicidal activity. Direct antifungicidal ac-tivity of tomato P14a has recently been documen-ted by Niderman et al. (1995) in an in vitro testmeasuring inhibition of Phytophthora infestans zoos-pore germination, and in an in vivo leaf-disc assayin which variations of P. infestans-infected leaf sur-face was scored. Differential activity was foundbetween the acidic isoforms (tobacco PR-1a andPR-1b) and the basic proteins (tomato P14c andtobacco PR-1g), where the basic proteins exhibitedthe highest antifungicidal activities.

To establish a structural basis for future researchon the molecular mechanism by which proteins ofthe PR-1 family exert their antifungicidal activity,we describe a high-quality NMR solution structuredetermination of P14a and further investigate theinternal dynamics of the protein by measurementsof backbone amide 15N spin relaxation times and

Figure 1. Contour plots of (a) a 2D [15N, 1H]-COSY spectrum of uniformly 15N-labeled P14a in 90% H2O/10% 2H2Oand (b) a 2D [13C, 1H]-COSY spectrum of uniformly 15N/13C-labeled P14a in 2H2O solution 1H-frequency 600 MHz,protein concentration 1.5 mM, pH 5.5, 30�C. The chemical shifts are relative to internal 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt (Wishart et al., 1995). In (a) the cross-peaks are labeled with the sequence positions of the corre-sponding amino acid residues. A cross identi®es the position of the very weak cross-peak of Gly88, which wasobserved when plotting lower contour levels. The pairs of cross-peaks identi®ed with primed and doubly primednumbers that are connected by a horizontal line, correspond to side-chain amide protons of Asn and Gln. The cross-peaks marked R and W belong to side-chain protons He of Arg and He1 of Trp, respectively. Unassigned side-chain15N±1H cross-peaks are indicated with asterisks. The backbone cross-peak of Asp128 at o1 = 107.7 ppm, o2 = 3.6 ppmand an unassigned side-chain amide cross-peak at o1 = 105.3 ppm, o2 = 5.2 ppm are located outside of the displayedspectral region. The cross-peaks of the backbone amide group of Arg100 and of all Arg He protons have been foldedalong o1 and therefore have negative intensity in this plot, as indicated by the use of broken contour lines.

{ With the availability of additional sequence data, aclassi®cation into 11 families has more recently beenconsidered (T. Niderman, personal communication).

NMR Structure of P14a 577

JMB MS 2838 [10/2/97]

steady-state 15N{1H}-NOEs. P14a was chosen forthe structure determination because it is, from itsamino acid sequence and its physiological role,representative of the proteins that have beengrouped together in the PR-1 family. A high yieldoverexpression system in Escherichia coli was avail-able for P14a at the outset of this project.

Results

The thermal denaturation of P14a was monitoredat pH 4.0 and pH 6.0 by circular dichroism (CD)

spectroscopy at 222 nm. A melting point of 51�Cwas obtained at both pH values. Based on thesedata, the NMR structure determination was per-formed at pH 5.5 and 30�C, using uniformly15N/13C doubly labeled, 15N-labeled and unlabeledprotein samples. Figure 1 shows 2D [15N, 1H]-COSY and 2D [13C, 1H]-COSY spectra of P14a,which demonstrate that the non-aromatic protonsas well as the 1H-bound 13C and 15N exhibit goodchemical shift dispersion, and that homogeneousprotein preparations were available for the NMRstructure determination.

Figure 2. Amino acid sequence of P14a and survey of the sequential connectivities and additional data collected forsecondary structure identi®cation. In the rows sCaN and sCbN, a bar indicates that the sequential 13C±15N scalar coup-ling connectivities were identi®ed using the 3D CBCANHN and 3D Ha/bCa/b(CO)NHN experiments. The sequentialNOE connectivities daN, dNN and dbN are indicated with thick, medium or thin black bars for strong, medium andweak NOEs, respectively. Sequential dad and dbd NOEs for Xxx-Pro are indicated by open bars, and a daa NOE isindicated by a cross-hatched bar. The medium-range connectivities daN(i, i � 3) and dab(i i � 3) are shown by linesstarting and ending at the positions of the residues related by the NOE. In the row �d(13Ca) � d(13Ca)obs ÿ d(13Ca)rc,where d(13Ca)obs and d(13Ca)rc denote the observed and the random coil chemical shifts, ®lled and open circles indicateresidues with �d(13Ca) > 1.5 ppm and �d(13Ca) < ÿ1.5 ppm, respectively. In the row 3JHNa, ®lled and open circlesdenote residues with 3JHNa < 5.0 Hz and 3JHNa > 8.0 Hz, respectively. In the row kNH ®lled squares identify residueswith suf®ciently slow amide proton exchange rates to enable observation of the 15H±1H cross-peak in a [15N, 1H]-COSY spectrum recorded 15 minutes after dissolving the protein in 2H2O at 30�C. The sequence locations of regularsecondary structure elements are indicated at the bottom, with I, II, III and IV indicating the a-helices, 310 the 310-helix, and A, B, C and D the b-strands.

578 NMR Structure of P14a

JMB MS 2838 [10/2/97]

Resonance assignments

Sequence-speci®c polypeptide backbone assign-ments for P14a were obtained using 3D 15N-resolved [1H, 1H]-NOESY and 3D 15N-resolved[1H, 1H]-TOCSY (Fesik & Zuiderweg, 1988;Messerle et al., 1989) for observation of sequentialNOE connectivities (Billeter et al., 1982; Wagner &WuÈ thrich, 1982; WuÈ thrich, 1986), and 3D Ha/bCa/

b(CO)NHN (Szyperski et al., 1994a), 3DCBCANHN (Grzesiek & Bax, 1992) and 3DCOHNNCA (Szyperski et al., 1995) for identi®-cation of intraresidual and sequential heteronuclearscalar coupling connectivities (Bax & Grzesiek,1993). Combination of the two approaches yieldednearly complete assignments for the backbone 1HN,15N and 13C�O resonances, and the aCH±bCHn

fragments of the individual residues. The onlymissing backbone assignments are those of the N-terminal amino group, as usual, the NH moiety ofSer49, which could not be observed due to linebroadening arising from slow conformationalexchange (see below), the 13C�O resonances of the®ve residues that precede Pro, i.e. Ser3, Gly21,Arg75, Asp124 and Arg133, and of Asp26, Ile47,His48, Gly60, Val86, Gly87, Cys91, Arg92 andIle130. Figure 2 presents a survey of the sequentialand medium-range connectivities identi®ed forP14a. At least one sequential NOE or sequentialscalar connectivity is observed for each pair ofneighboring residues, with the sole exceptions ofthe two dipeptide segments Ile47-His48 and His48-Ser49. The observation of strong dad NOEs indi-cated that the prolyls 4, 22, 76 and 134 are in atrans-conformation, while a strong daa NOE wasdetected for the dipeptide segment Asp124-Pro125,showing that it is in a cis-conformation (WuÈ thrich,1986).

As starting points for the assignment of the side-chain resonances we used the 13Ca/b and 1Ha/b

chemical shifts obtained from the sequentialassignment routines, and the analysis of the 3DHCCH-COSY spectrum (Kay et al., 1990) wasaided by reference to a 3D 13C-resolved [1H, 1H]-NOESY spectrum (Ikura et al., 1990). Complete 1Hand 13C resonance assignments for the aliphaticCHn moieties were thus obtained. Among thelabile side-chain protons, the amide groups of 15out of 22 Asn and Gln residues were assigned, aswell as the e-proton resonances of six out of 11 Argresidues. For the remaining NH moieties the cross-peaks were observed in the 2D [15N, 1H]-COSYspectrum (Figure 1(a)), but no NOEs with aliphaticprotons could be detected in 3D 15N-resolved [1H,1H]-NOESY. The NMR solution structure willshow that all side-chains with unassigned NHgroups are solvent-exposed and disordered (seebelow). Among the side-chain hydroxyl protons,those of Thr95, Ser102 and Tyr123 could beassigned using a 2D [1H, 1H]-NOESY spectrum(Anil Kumar et al., 1980).

In the course of the 3D structure calculation (seebelow) stereospeci®c assignments were obtainedfor 44 out of 75 bCH2 groups with non-degenerateb-proton chemical shifts, three pairs of glycyla-protons, and 14 pairs of isopropyl methyl groupscorresponding to the ten valyl residues and to fourof the six leucyl residues. The stereospeci®c assign-ment of the leucyl isopropyl methyl groups wasalso supported by analysis of 3Jcacd scalar couplingconstants in a 2D 13C-13C long-range correlationexperiment (Bax et al., 1992).

The aromatic spin systems were identi®ed using2D [1H, 1H]-2QF-COSY (Rance et al., 1983), 2Dclean-[1H, 1H]-TOCSY (Griesinger et al., 1988),2D ct-[13C, 1H]-COSY (Vuister & Bax, 1992) and 2D1H-TOCSY-relayed ct-[13C, 1H]-COSY (Zerbe et al.,1996; and Figure 3) recorded in 2H2O solution. Sub-sequently, sequence-speci®c assignments of thearomatic spin systems were obtained via intraresi-

Figure 3. Contour plots of (a) ct-[13C, 1H]-COSY (Vuister & Bax,1992) and (b) 2D 1H-TOCSY-relayed ct-[13C, 1H]-COSY spectra(Zerbe et al., 1996) used for iden-ti®cation of the aromatic spin sys-tems (solvent 2H2O, 1H-frequency600 MHz, protein concentration1.5 mM, pH 5.5, 30�C). Continuousand dotted contour lines representpositive and negative peaks, re-spectively. In (a) the cross-peaks ofTyr7, Phe62 and Trp129 have beenidenti®ed, and in (b) the assign-ments of the aromatic spin systemsof Tyr7, Phe62 and Trp129 are indi-cated with continuous lines.


JMB MS 2838 [10/2/97]

dual NOEs between aromatic and aliphatic protons(WuÈ thrich, 1986) observed by 2D [1H, 1H]-NOESYor 3D 13C-resolved [1H, 1H]-NOESY, or via scalarconnectivities in a 2D (Hb)Cb(CgCd)Hd spectrum(Yamazaki et al., 1993). Combination of the resultsfrom the two approaches yielded complete 1H and13C resonance assignments for all 18 aromatic ringsin P14a except that the He protons of His48 andHis93 were not individually assigned because theNOEs between the rings and the aCH-bCH2 moi-eties could not be detected. The NMR solutionstructure (see below) will show that both imidazolerings are solvent-exposed and conformationallydisordered.

Ring current calculations of the type described byPerkins & WuÈ thrich (1979) showed that all protonchemical shifts in P14a that deviate by more than1.5 ppm from the corresponding random coilvalues (WuÈ thrich, 1986) include large ring currenteffects. For the side-chain methyl and methyleneprotons, the calculated ring current shifts matchthese large observed conformation-dependentshifts to within better than 0.6 ppm. In particular,the ring current calculation revealed that the Ha/b

protons of Pro4 are shifted by the ring of Trp25,Hb2,3 of Asn54 by Trp71, Ha1,2 of Gly58 by Phe62,Hg2,3 of Gln96 by Tyr123, Ha1,2 of Gly106 and HN

of Cys107 by Trp25, and Ha1,2 of Gly127, HN ofAsn128, Ha of Trp129 and Ha of Pro134 by Trp99.

Collection of conformational constraints andstructure calculation

A total of 3433 NOESY cross-peaks was assignedand used for the generation of the input of upper-limit distance constraints for the structure calcu-lation. Of these, 1055 resulted from 3D 15N-resolved [1H, 1H]-NOESY with 65 ms mixing time,1898 from 3D 13C-resolved [1H, 1H]-NOESY with65 ms mixing time, and 480 from 2D [1H, 1H]-NOESY with 50 ms mixing time. In addition, atotal of 349 vicinal scalar couplings were deter-mined, including 111 3JHNa coupling constantsfrom inverse Fourier transformation of in-phasemultiplets in 2D [15N, 1H]-COSY spectra (Szyperskiet al., 1992), 87 3JNb couplings estimated from 3Dct-HNNHB (Archer et al., 1991), 99 3JC0b couplingsestimated from 3D HNN(CO)HB (Grzesiek et al.,1992), 38 3Jab couplings estimated from 3D HCCH-E.COSY (Eggenberger et al., 1992), and 14 3Jabcoupling constants of Val, Thr and Ile extracted byinverse Fourier transformation from 2D [13C, 1H]-COSY recorded in 2H2O (Szyperski et al., 1992). ForSer39, Glu84, Asn101 and Gln132, the 3Jab,

3JC0b and3JNb scalar couplings indicated rotational averagingabout w1 (Nagayama & WuÈ thrich, 1981); therefore,no w1 dihedral angle constraint was derived fromthese couplings, and NOE upper-limit distanceconstraints to the b-protons of these residues werereferred to the corresponding pseudo-atoms(WuÈ thrich et al., 1983). P14a contains three disul-®de bonds (Lucas et al., 1985), but the location ofthese bonds has not been experimentally deter-

mined by chemical methods. A preliminary struc-ture calculation using as input only the NOEupper distance constraints and the dihedral angle

Figure 4. (a) Plot of the number of NOE constraintsper residue, n, used in the calculation of the solutionstructure of P14a versus the amino acid sequence(black, intraresidual; cross-hatched, sequential; verticallyhatched, medium-range; open, long-range). (b) Plotsversus the amino acid sequence of the mean globalbackbone displacements per residue, Dbb

glob, of the 20energy-minimized DIANA conformers relative to themean NMR structure calculated after superposition ofthe backbone heavy atoms N, Ca and C0 of the regularsecondary structures for minimal r.m.s.d. (broken line).The continuous line represents the mean local r.m.s. de-viation RMSDbb

loc, calculated for backbone superpositionof all tripeptide segments along the sequence onto themean NMR structure, with the r.m.s.d values for thetripeptide segments plotted at the positions of the cen-tral residues. (c) Logarithmic plots of amide protonexchange rates versus the sequence. Filled circles, data at30�C, open circles, 22�C. A lower limit of log kNH � ÿ 1,with kNH in units of minutesÿ1, is given for those resi-dues where the exchange was too fast for the N±Hcross-peak to be observed in the experiments used (seeMaterials and Methods). The locations of the a-helices Ito IV, the 310-helix and the b-strands A to D are indi-cated at the top of trace (b).


JMB MS 2838 [10/2/97]

constraints (WuÈ thrich, 1986) showed unambigu-ously that the three disul®de bonds are Cys44-Cys112, Cys85-Cys91 and Cys107-Cys121.

The input for the ®nal DIANA structure calculationcontained 1701 NOE upper-limit distance con-straints (414 intraresidual, 131 sequential backbone,114 medium-range and long-range backbone, and1042 interresidual distance contraints involvingside-chain protons), 311 dihedral angle constraints(114 for f, 114 for c and 83 for w1), and the threedisul®de bond constraints were represented bynine upper and nine lower distance limits(Williamson et al., 1985). Figure 4(a) shows thesequence distribution of the NOEs observed forP14a. Quite uniform high density of NOEs wasobserved for the residues in the regular secondarystructure elements, but relatively few NOEs weredetected for the N-terminal tripeptide segment andthe polypeptide segments comprising residues 39to 52, 58 to 62, 80 to 91, 112 to 116 and 129 to 133.The calculation was performed with 100 random-ized starting structures, and the 20 best conformers

are used to represent the NMR structure. The smallsize and small number of residual constraint viola-tions (Table 1) show that the input data represent aself-consistent set, and that the constraints are wellsatis®ed in the calculated conformers. The highquality of the structure determination is re¯ectedby the global r.m.s.d. values relative to the meancoordinates of 0.45 AÊ and 0.88 AÊ calculated for thebackbone atoms of the regular secondary structureelements, and the entire polypeptide chain, respect-ively (Table 2). A similar precision is obtainedwhen the core side-chains are included (Table 2).Plots of the local backbone r.m.s.d. values and theglobal backbone displacements versus the sequence(Figure 4(b)) further show that all regular second-ary structure elements are well de®ned through-out. Increased local disorder is observed for the N-terminal tripeptide segment and the three poly-peptide segments 41 to 52, 77 to 91 and 113 to117, which link regular secondary elements(Figure 6(b)). The increased disorder is manifestedin both larger local backbone r.m.s.d. values andlarger global backbone displacements (Figure 4(b)),which shows that it is indeed local conformationalfeatures, on the level of dipeptide segments, thatare not precisely de®ned.

The NMR solution structure of P14a

P14a exhibits an a � b tertiary fold with foura-helices, I to IV, consisting of residues 4 to 17, 27to 40, 64 to 72 and 93 to 98, a single turn of 310-helix immediately C-terminal to the a-helix III, anda mixed four-stranded b-sheet with strands A to Dconsisting of residues 24-25, 52 to 58, 104 to 111and 117 to 124 (Figure 5). The two antiparallelb-strands C and D form the central part of theb-sheet, with B attached antiparallel to D and theshort strand A parallel with the N-terminal seg-ment of strand C, yielding an overall topology(Richardson, 1981) �3x, ÿ2x, �1 (Figure 5(b)).

In the three-dimensional fold of P14a (Figure 6(a)),the helix II is part of the right-handed crossoverthat connects the b-strands A and B. In the orien-tation of P14a in Figure 6(a), this helix is located

Table 1. Analysis of the 20 best DIANA conformers of P14a after restrained energyminimization with the program OPAL

QuantityAverage value � standard

deviation Range

DIANA target function (AÊ 2)a 3.13 � 0.70 1.42 . . . 4.00AMBER energy (kcal/mol) ÿ4579 � 73 ÿ4688.9 . . . ÿ4444Residual NOE distance constraint violations

Sum (AÊ ) 9.20 � 0.64 8.15 . . . 10.44Maximum (AÊ ) 0.10 � 0.01 0.09 . . . 0.12

Residual dihedral angle constraint violationsSum (�) 45.02 � 5.10 33.95 . . . 55.21Maximum (�) 2.26 � 0.20 1.83 . . . 2.68

The ®nal structure calculation was started with 100 randomized conformers. The 20 DIANA con-formers with the lowest residual target function values were re®ned by energy minimization andare used to represent the NMR structure.

a Before energy minimization.

Table 2. Average r.m.s.d. values calculated for differentatom selections in the solution structure of P14a

Atoms used for the comparisona

r.m.s.d.(AÊ ) � standard

deviationb

Backbone atoms N, Ca, C0 (1±135) 0.88 � 0.08All heavy atoms (1±135) 1.30 � 0.09Backbone atoms (regular secondarystructures)c 0.45 � 0.08Backbone �molecular cored 0.91 � 0.07

a The numbers in parentheses denote the protein segments con-sidered in the comparison.

b Averages and standard deviations are given for the pair-wise r.m.s.d. values between each of the 20 energy-re®nedDIANA conformers and the mean structure.

c a-Helices 4 to 17, 27 to 40, 64 to 72 and 93 to 98; 310-helix73 to 75; b-strands 24-25, 53 to 58, 104 to 111 and 117 to 124

d Includes the backbone atoms N, Ca and C0 of residues 1 to135 and the side-chain heavy atoms of 51 residues forming thetwo clusters of the molecular core (see the text), i.e. residues 29,30, 33, 36, 37, 44, 46, 55, 106, 108, 110, 112 and 118 of cluster I,and residues 7, 8, 10, 11, 14, 15, 18, 20, 23, 25, 54, 56, 62, 67, 68,70, 71, 72, 78, 80, 85, 91, 94, 95, 96, 97, 98, 99, 102, 105, 107,119, 121, 122, 123, 128, 134 and 135 of cluster II.


JMB MS 2838 [10/2/97]

below the plane de®ned by the b-sheet. The helicesI and III run parallel and are packed against theupper side of the b-sheet, with the helix axes form-ing an angle of approximately 45� relative to theplane of the b-strands. The short helix IV isoriented nearly perpendicular to the helices I andIII and is completely buried between the b-sheetand the two-helix bundle of helices I and III. Thearrangement of the a-helices and the b-strands inthree stacked layers forms an ``a-b-a sandwich''.This tertiary fold is stabilized by the three disul®debonds (Figure 5(a)): Cys44-Cys112 connects theloop linking helix II and the b-strand B with theloop between the b-strands C and D, Cys84-Cys91is located in the polypeptide segment connectingthe helices III and IV, and Cys107-Cys121 bridgesthe directly adjoining b-strands C and D.

The individual helices are stabilized by a variety oflocal structural motifs (Figure 7). In helix I Pro4acts as a helix initiator (Richardson, 1981) and ahydrogen bond between the g-oxygen atom of Ser3and the backbone amide proton of Asp6 isobserved in six out of the 20 DIANA-conformers,which combines with the hydrogen bond betweenthe backbone amide proton of Ser3 and the d-car-boxylate group of Asp6 to form a ``capping box''(Harper & Rose, 1993). The N termini of the a-helices II and III are N-capped (Richardson &Richardson, 1988) by Asp26 and Thr63, respect-ively, which form hydrogen bonds with the back-bone amide proton of the third helical residue. Inthese two helices the N-terminal ends are furtherstabilized by hydrophobic (i, i � 5) interactions

between the residue preceding the N-cap and thefourth helical residue, i.e. in helix II the indole moi-ety of Trp25 is in close contact with the methylgroup of Ala30, and in helix III the aromaticring of Phe62 interacts with the methyl group ofAla67. These (i, i � 5) hydrophobic interactions aredirectly evidenced by side-chain±side-chain (i,i � 5) NOEs and represent typical ``hydrophobic-staple motifs'' (MunÄ oz et al., 1995). Helix I is C-capped by Gly19, which has an aL-conformation,and there are (i, i � 3) and (i, i � 5) hydrogenbonds formed in reverse order (Schellmann, 1980)between the carbonyl oxygen atoms of Ala16 andArg15, and the backbone amide protons of Gly19and Val20, respectively. Helix I is further stabilizedby an (i, i � 4) hydrogen bond between Od ofAsp13 and He2 of Gln17. The C terminus of helix IIis stabilized by a bifurcated hydrogen bond of thecarbonyl oxygen atom of Arg40 with the backboneamide protons of Asp43 and Cys44, and by an(i, i � 4) cation-p interaction (Dougherty, 1996)between the positively charged guanidino moietyof Arg40 and the aromatic ring of Tyr36. The C-terminal end of helix III forms a 310-helix, which isfurther stabilized by a hydrogen bond betweenOHg of Ser77 and the backbone carbonyl oxygenatom of Glu74. There is also a type I tight turnwith an (i, i � 3) hydrogen bond involving the car-bonyl oxygen atom of Arg75 and the backboneamide proton of Tyr78. The dipole of helix IVforms an electrostatically favorable interaction withthe buried guanidino group of Arg15, which islocated directly in front of the C-terminal face of

Figure 5. (a) Regular secondary structures and topology of the b-sheet in P14a. The start and the end of the regularsecondary structure elements and the polypeptide chain ends are marked with their sequence location. White cylin-ders and black arrows represent the a-helices I to IV and the b-strands A to D, respectively, and the location of a 310-helix turn from residues 73 to 75 is also indicated. Polypeptide segments without regular secondary structure are rep-resented by a black line. The sequence positions of the six Cys residues and the disul®de linkages, as identi®ed bythe present NMR study (see the text), are indicated. (b) The b-sheet in P14a, with double arrows indicating the inter-strand NOEs de®ning the arrangement of the strands A to D. Circles identify slowly exchanging amide protons.


JMB MS 2838 [10/2/97]

the helix (Figure 7(d)). These local interactions areall unambiguously de®ned in the NMR structureobtained with the standard protocol for structuredetermination (see the preceding text), except forthe hydrogen bond of the side-chain carboxylategroup of Asp26, which was characterized by thefollowing additional experiments. Inspection of the20 energy-re®ned DIANA conformers suggestedthat this side-chain might form either a bifurcatedhydrogen bond with the backbone amide protonsof Asn28 and Leu29, or a ``time-shared'' hydrogenbond (Szyperski et al., 1994b) with these two amide

groups. Measurement of the pH-dependence of theamide protons (Bundi & WuÈ thrich, 1979) byrecording a series of 2D [15N, 1H]-COSY spectra atdifferent pH values then showed that the side-chain carboxylate group of Asp26 interacts onlywith the amide proton of Leu29.

The a-b-a sandwich fold of P14a includes twospatially separated hydrophobic cores in the twolayer interfaces (Figure 6(b)). A small cluster isformed by amino acid side-chains of the helix II(Leu29, Ala30, Ala33, Tyr36, Ala37), the b-strandsB (Leu55) and D (Phe118, Asn122), the disul®de

Figure 6. (a) Ribbon drawing of the DIANA conformer of P14a with the lowest residual target function value. Thehelices I to IV are shown red and yellow, the b-strands A to D cyan, other polypeptide segments gray, and the poly-peptide chain ends are indicated with N and C. (b) Stereo view of the backbone and the interior side-chains of theNMR solution structure of P14a. For the presentation of the backbone a spline function was drawn through the Ca

positions; the thickness of the cylindrical rod is proportional to the mean of the global displacements, Dbbglob, of the 20

energy-minimized DIANA conformers calculated after superposition of the heavy atoms N, Ca and C0 in the regularsecondary structure elements for minimal r.m.s.d. (see also Figure 4(b)). The helices are red, the b-strands blue, andthe other polypeptide segments gray. Residues that form a large hydrophobic cluster associated with the b-sheet andthe a-helices I, III and IV are shown in yellow, and residues belonging to the smaller hydrophobic cluster at the inter-face of the b-sheet and the a-helix II are green. Polar side-chains buried in the hydrophobic core are identi®ed; His11,Thr95, Gln96 and Asn128 are shown in magenta and Arg15 in dark blue.


JMB MS 2838 [10/2/97]

bridge Cys44-Cys112, Leu46 in the polypeptidesegment connecting helix II and the b-strand B,and the backbone of the b-strand C (Gly106,Gly108, Ala110). There are also van der Waals con-tacts between backbone atoms of the tripeptidesegment Gly106-Cys107-Gly108 in the b-strand Cand backbone atoms of Ala30 and Gln34 in helixII, so that the anchoring of helix II to the b-sheet issupported by hydrophobic side-chain contacts aswell as by direct backbone-backbone interaction.The second hydrophobic cluster involves side-chains from all regular secondary structures excepthelix II. The short helix IV, which is completely

enclosed by the b-sheet and the layer formed bythe helices I and III, plays an integral role in thiscluster. With the sole exception of the solvent-exposed His93, all side-chains of helix IV (Tyr94,Thr95, Gln96, Val97, Val98, Trp99) are located inthe interior of the protein and participate in hydro-phobic contacts within the cluster (Figure 6(b)). Inaddition, the side-chains of Tyr7, Leu8, Val10,His11 and Ala14 in helix I, Val18, Val20 and Met23in the segment connecting the helix I and theb-strand A, Trp25 in the b-strand A, Ala56 inthe b-strand B, Phe62 in the segment linkingthe b-strand B with helix III, Ala67, Val68, Leu70,

Figure 7. Local structure motifs that stabilize individual helices in P14a. (a) Helix I, with the ``helix initiator'' Pro4,the potential capping box involving Ser3 and Asp6, the (i, i � 4) hydrogen bonds Od Asp13 ± He2 Gln17, and thehydrogen bond O0 Ala16 ± HN Gly19 and O0 Arg15 ± HN Val20 formed in ``reversed order'' at the C-terminal end(Schellmann, 1980). (b) Helix II, with Asp26 as the N-capping residue, an (i, i � 5) hydrophobic-staple motif involvingTrp25 and Ala30, a cation-p interaction between the positively charged side-chain of Arg40 and the aromatic ring ofTyr36, and hydrogen bonds O0 Arg40 ± HN Asp43 and O0 Arg40 ± HN Cys44 stabilizing the C-terminal end of thehelix. (c) Helix III, with Thr63 as the N-capping residue, an (i, i � 5) hydrophobic-staple motif involving Phe62 andAla67, and a 310-helix turn and a type I turn stabilizing the C-terminal end. (d) Helix IV, with the positively chargedguanidinium group of Arg15 near the negatively charged side of the helix dipole, and a tight turn further stabilizingthe C-terminal end.


JMB MS 2838 [10/2/97]

Trp71 and Val72 in helix III, Tyr78, Tyr80 andthe disul®de bridge Cys85-Cys91 in the segmentconnecting the helices III and IV, Leu105 in theb-strand C, Ile119 and Tyr123 in the b-strand D,the disul®de bridge Cys107-Cys121 linking theb-strands C and D, Trp129, Pro134 and Tyr135constitute this second cluster. The two hydro-phobic clusters are well de®ned, as can be seenfrom the fact that nearly identical r.m.s.d. valueswere obtained for the backbone heavy atoms, orthe backbone heavy atoms plus the core side-chains (Table 2).

Hydrogen bonds and hydrogen exchange

In all, 97 hydrogen bonds are present in at leasteight of the 20 energy-re®ned DIANA conformers.Of those, 70 are backbone-backbone hydrogenbonds located within or between regular secondarystructure elements, and there are three long-rangebackbone-backbone hydrogen bonds between resi-dues located outside of the regular secondarystructures, i.e. HN Arg65±O0 Gln1, HN Val86±O0Ser77 and HN Arg100±O0 Pro134, the remaining 24hydrogen bonds involve side-chains (Figures 7 and

Figure 8. (a) Long-range hydrogen-bonding interactions in P14a that involve side-chain atoms. For the side-chains weuse a ball-and-stick presentation with the following color code: carbon atoms, gray; nitrogen atoms, blue; oxygenatoms, red; hydrogen atoms, white. Hydrogen atoms are displayed only when they are part of a hydrogen bond. Thehydrogen bonds are represented by green broken-line cylinders. The hydrogen-bonding side-chains are identi®edwith the one-letter amino acid symbol and the sequence number. (Note that the hydrogen bond W99-Q132 is not pre-sent in the particular conformer shown here.) (b) All-atom ball-and-stick presentation of the hydrogen bonds formedbetween the backbone carbonyl oxygen atom of Tyr7 and Hd1 of His11, and between the OHZ of Tyr123 and Ne2 ofHis11. Same color code as in (a). (c) Cross-section along o1 at the o2 frequency of Tyr123 OHZ taken from a 750 MHz2D [1H, 1H]-NOESY spectrum recorded in H2O, showing NOEs that were used for the re®nement of the molecularcore. (d) Exponential decay of the signal intensity of Tyr123 1HZ measured in a series of 1D 1H NMR-spectra afterdissolving fully protonated, lyophilized P14a in 2H2O. The exchange rate was calculated to be 9 � 10ÿ3 minuteÿ1,which represents a protection factor of about 106 (Liepinsh et al., 1992).


JMB MS 2838 [10/2/97]

8). Of special interest are the three hydrogen bondsHd Asn54±OZ Tyr123, HZ Tyr123±Ne2 His11 andHd1 His11±O0 Tyr7 (Figure 8(a) and (b)), which arecompletely buried in the molecular core and con-nect three regular secondary structure elements,i.e. the a-helix I and the b-strands B and D. Like-wise, the hydrogen bonds Hg1 Thr95±Od1 Asn128,Hd2 Asn128±Og1 Thr95 and Hg Ser102±O0 Pro125are buried in the protein core, which is thus exten-sively stabilized by both hydrophobic contacts andhydrogen-bonding interactions.

Amide proton exchange rates were measured at22�C and 30�C in 2D [15N, 1H]-COSY spectrarecorded at variable time intervals after fully proto-nated 15N-labeled P14a had been dissolved in2H2O (Figure 4(c)). There is excellent agreement ofthe exchange data with the hydrogen bond identi®-cation in P14a in that out of a total of 56 backboneamide protons with slowed exchange, all exceptTyr135 form a hydrogen bond. The slowestexchange is observed for the backbone amide pro-tons of the residues that form the larger hydro-phobic cluster, i.e. the C-terminal segments ofthe a-helices I and III, a-helix IV, and the centralb-strands C and D (Figures 5 and 6). Only weakshielding is evidenced for the amide protons in theN-terminal part of helix I, indicating transient fray-ing of this helix. Comparatively weak protection isobserved for the backbone amide protons of the C-terminal part of helix II. Combined with the rapidexchange of the backbone amide proton of Asn54,this behavior of helix II indicates somewhatreduced stability of the smaller hydrophobic clus-ter (green in Figure 6(b)) when compared with thelarger cluster (yellow in Figure 6(b)).

In addition to a large proportion of the amideprotons, the hydroxyl protons of Thr95, Ser102and Tyr123 experience suf®cient protection fromexchange with the solvent water to be observablein a 2D [1H, 1H]-NOESY spectrum recorded withlow-power presaturation of the water signal.Figure 8(c) shows a cross-section taken along o1 ato2(H

Z) of Tyr123, with identi®cation of the long-range NOEs with Tyr123 OH that were used tore®ne the molecular core. Furthermore, due to thelarge down®eld shift at 13.02 ppm, the exchange ofthe hydroxyl proton of Tyr123 could readily bemonitored by 1D 1H NMR spectra recorded afterdifferent time-intervals following the preparationof a fresh solution of fully protonated, lyophilizedP14a in 2H2O (Figure 8(d)). Taking account of theintrinsic exchange rates of tyrosyl hydroxyl pro-tons reported by Liepinsh et al. (1992), a protectionfactor of about 106 was estimated, which is out-standingly large for a side-chain hydroxyl proton.

Internal mobility of P14a inferred from themeasurement of 15N spin relaxation parameters

Using uniformly 15N-labeled P14a, we measuredthe 15N spin relaxation times T1 and T2, and thesteady-state 15N{1H}-NOEs to complement thestructure determination with information about in-

ternal mobility of the P14a molecule. Figure 9shows that T1, T2 and the 15N{1H}-NOEs all show aremarkably uniform distribution over most of theamino acid sequence, implying that overall themolecular architecture of P14a (Figure 6) is ratherrigid. Exceptions include that for the tripeptidesegment 60 to 62 increased T1 and T2 values, anddecreased 15N{1H}-NOEs reveal the presence ofhigh-frequency rate processes with effective corre-lation times in the nanosecond to picosecond time-range. For this segment, which is located directlybefore the start of the a-helix III (Figure 6),increased values are found for the local r.m.s.d.values and the global backbone displacements(Figure 4(b)). In addition, residues 88 to 90 and theN-terminal dipeptide segment exhibit motionalmodes that affect the 15N{1H}-NOEs, but not the T1

and T2 values. The decreased T2 value for His48and the fact that the backbone amide 15N reson-ance of Ser49 is broadened beyond detection impli-cate a low-frequency conformational exchange pro-cess involving the dipeptide segment His48±Ser49,

Figure 9. Relaxation times and steady-state 15N{1H}-NOEs measured for the backbone amide nitrogen atomsof P14a at a proton frequency of 600 MHz. (a) Longi-tudinal relaxation time, T1; (b) transverse relaxationtime, T2; (c) 15N{1H}-NOEs. The ring centers show themean values, and the standard deviations are indicatedby vertical lines through the ring centers. The verticalarrow in (b) indicates that the backbone amide protonof Ser49 is broadened beyond detection. The locations ofthe a-helices I to IV, the 310-helix (shaded bar) and theb-strands A to D are indicated at the bottom.


JMB MS 2838 [10/2/97]

which is located in the loop connecting the a-helixII and the b-strand B (Figures 5(a) and 6). Based onthe assumption that the chemical shifts of the con-formational states involved in this equilibrium arenot degenerate, a lower limit on the frequency ofthis rate process was determined from rotatingframe 15N spin relaxation times, T1r, at a spin-lockpower level of 12,870 rad sÿ1: since the T1r value ofthe backbone 15N spin of His48 was not signi®-cantly increased when compared with T2 of thesame spin this limit was established as >1 � 104 sÿ1

(Szyperski et al., 1993).For the loop of residues 112 to 116, which links the

b-strands C and D and exhibits increased confor-mational disorder (Figures 4(b) and 6(b)), the 15Nspin relaxation data provide no evidence for localmotional processes. We therefore conclude that thereduced precision of the structure determination ofthis polypeptide segment arises from scarcity oflong-range and medium-range NOE distance con-straints (Figure 4(a)), rather than from dynamicdisorder. Finally, assuming that P14a reorients iso-tropically in solution, the T1/T2 ratios measuredfor the backbone amide protons of the residues inthe regular secondary structures indicate that thecorrelation time for overall rotational tumbling isabout 6 ns.

Discussion

In a search for structural similarity between P14aand other, previously investigated proteins, thedistance matrix algorithm implemented in the pro-gram DALI (Holm & Sander, 1993) revealed notopological similarity with any of the 599 proteinstructures currently used to represent the Brookha-ven Protein Data Bank (Holm & Sander, 1994), andin a recent classi®cation of a � b folds (Orengo &Thornton, 1993) no fold was included that re-sembles the a-b-a sandwich of P14a. The so farunique molecular architecture of P14a represents arigid scaffold with high thermal stability andincludes only a small number of short polypeptidesegments with increased local disorder and/orincreased local mobility, as evidenced by the quiteuniformly high precision of the NMR structure de-termination (Table 2, Figures 4(b) and 6(b) and the15N spin relaxation parameters (Figure 9)). Thetight packing of the a-helices on both sides of thecentral b-sheet (Figure 6(a)) results in a compact,bipartite molecular core (Figure 6(b)), which isstabilized by hydrophobic interactions as well asby a buried hydrogen-bonding network includingthree hydrogen bonds with the side-chain hydroxylprotons of Thr95, Ser102 and Tyr123 (Figure 8).Since each interior hydrogen bond is expected tocontribute about 5 kJ molÿ1 to the free energy ofprotein folding (Chen et al., 1993), these hydrogenbonds may make a substantial contribution to theobserved high stability of P14a, which is obviouslyrequired for proper functioning in the harsh extra-cellular environment where P14a is active.

Multiple alignment of the amino acid sequences ofthe PR-1 proteins deposited in the SWISS-PROTdata bank with the sequence of P14a (Figure 10)emphasizes the high level of sequence homology inthis group of proteins, with 35% sequence identityamong all PR-1 proteins in Figure 10 and pairwisesequence identities with P14a between 56% and96%. On the basis of this high degree of sequencehomology, the three-dimensional fold of P14a canbe expected to be common to all PR-1 proteins(Orengo et al., 1994). This hypothesis is further sup-ported by the following local features of sequenceand structure. (1) The six cysteinyl residues arestrictly conserved, so that the arrangement of thedisul®de bridges in P14a can be preserved in allPR-1 proteins. (2) The sequence alignment revealsa striking correlation between the extent of conser-vation of residues and their structural role in themolecular core of P14a (Figures 6(b) and 10). Ofthe 51 amino acid residues located in the proteincore, 30 are strictly conserved and 16 undergo onlyconservative substitutions in the group of PR-1proteins. The ®ve remaining residues are onlyperipherally associated with the core. (3) Pro125,which adopts the cis-conformation in P14a, isstrictly conserved, which suggests a critical role ofthis residue for the proper positioning of the C-terminal decapeptide segment in the PR-1 proteins.(4) The degree of conservation for the local struc-tural motifs that stabilize the helices (Figure 7) iscomparable with that found for the protein core.Thus, most of the amino acid residues in the pos-itions corresponding to the N-caps in P14a (Ser3,Asp26 and Thr63) have a side-chain that can forma hydrogen bond with the backbone amide protonof the third helical residue; Gly19, which is a prere-quisite for the ``Schellman motif'' at the C terminusof helix I (Figure 7(a)), is strictly conserved; the``hydrophobic staple motifs'' stabilizing the N ter-mini of helix II (Trp25 and Ala30 in P14a) andhelix III (Phe62 and Ala67 in P14a) are conserved;the residues Tyr36 and Arg40 in the (i, i � 4) cat-ion-p interaction at the C terminus of helix II(Figure 7(c)) are present in seven PR-1 proteins,and in the remaining PR-1 proteins Arg40 is substi-tuted by Leu, suggesting that the cation-p inter-action is replaced by an (i, i � 4) hydrophobicinteraction; Arg15, which interacts favorably withthe helix IV dipole, is strictly conserved. Overall,the local structural arrangements stabilizing thehelices as well as the hydrophobic and hydrogen-bonding interactions in the protein core are foundto be largely conserved in the PR-1 proteins ofFigure 10, which indicates that the high thermalstability observed for P14a should be common toall PR-1 proteins.

The octapeptide containing the a-helix IV and thetwo succeeding residues Trp99 and Arg100 isstrictly conserved in all PR-1 proteins (Figure 10),and the segment 94 to 99 is completely buriedbetween the b-sheet and the helices I and III(Figure 6(b)). Furthermore, with the sole exceptionof Tyr80, all residues in direct contact with helix IV


JMB MS 2838 [10/2/97]

are strictly conserved, which emphasizes the pivo-tal role of helix IV in the a-b-a sandwich of P14a(Figure 6). Shakhnovich et al. (1996) recentlyrevealed a clear correlation of residue conservationwithin a set of naturally occurring homologs of thechymotrypsin inhibitor 2 (CI2) with their role inprotein folding. In analogy to this work with CI2,it is tempting to speculate that the polypeptide seg-ment 94 to 99 might be the ``folding nucleus'' ofPR-1 proteins.

With the availability of the three-dimensionalstructure of P14a, future research aimed at moreprofound insights into the mode of action of plantdefense systems will probably focus more andmore on the PR-1 proteins. Presently, in theabsence of information on the mode of action ofP14a, we have further tried to obtain a lead foridenti®cation of possible active sites on the follow-ing basis: highly conserved solvent-accessible resi-dues without an obvious role in the architecture of

Figure 10. Multiple alignment of the amino acid sequence of P14a with those of the other PR-1 proteins deposited inthe SWISS-PROT data bank. The amino acid sequences are given in the one-letter code, and the numbering for P14ais indicated above its sequence. For the other proteins the sequence position of the ®rst residue is indicated immedi-ately before the start of the sequence data, and the total number of amino acid residues is given at the end of eachrow. The locations of the regular secondary structure elements identi®ed for P14a are indicated at the top, with boxesfor the helices I to IV and 310, and arrows for the b-strands A to D. In the row labeled core, the residues forming thesmaller hydrophobic core between the b-sheet and helix II are indicated by 1, and those constituting the larger coreassociated with the b-sheet and the helices I, III and IV by 2, with the residues contained in helix IV and those indirect contact with helix IV (Figure 6(b)) identi®ed in bold. In the row surface, � identi®es residues that exhibit morethan 25% solvent accessibility in at least one of the 20 energy-minimized DIANA conformers. Residues that arestrictly conserved in the group of PR-1 proteins are highlighted in red, and residue positions that show exclusivelyconservative mutations among PR-1 proteins are highlighted in yellow. The abbreviations are: PR04 LYCES, PR leaf-protein-4 precursor from tomato (Lycopersicon esculentum); PRB1 TOBAC, basic form of PR protein-1 precursor(PRP-1) from common tobacco (Nicotiana tabacum); PR1B TOBAC, PR protein-1B precursor (PR-1B) from tobacco;PR1A TOBAC, PR protein-1A precursor (PR-1A) from tobacco; PR1 ARATH, PR protein-1 precursor (PR-1) frommouse-ear cress (Arabidopsis thaliana); PR1C TOBAC, PR protein-1C precursor (PR-1C) from tobacco; PR13 HORVU,PR protein PRB-13 precursor from barley (Hordeum vulgare); PR12 HORVU, PR protein PRB-12 precursor from bar-ley; PR1 HORVU, PR protein-1 precursor from barley; PRMS MAIZE, PR protein precursor from maize (Zea mays).The multiple sequence alignment was obtained using the program CLUSTALW (Thompson et al., 1994).


JMB MS 2838 [10/2/97]

the protein are likely candidates for participationin the formation of functionally important sites(Schulz & Schirmer, 1979). Overall, the surface resi-dues in P14a exhibit a low degree of conservation:out of 49 non-glycyl residues that display a solventaccessibility of more than 25% in at least one of the20 energy-re®ned DIANA conformers used to rep-resent the NMR structure, 36 show non-conserva-tive substitutions within the family of PR-1proteins (Figure 10). Only Ser3, Gln5, His48, Ser49,His93, Arg100 and Asn114 are strictly conserved,and Pro21, Leu46, Arg75, Tyr80 and Trp129 ofP14a show exclusively conservative substitutionsamong the PR-1 proteins. Among these conservedresidues, Ser3 is part of the capping box of helix I(Figure 7(a)), and Leu46 and Tyr80 participate inthe formation of the molecular core, which leaves atotal of ten ``non-structural'' highly conserved resi-dues. Among these ten residues, His48, Ser49 andHis93 are in close proximity, so that the conserva-tion of this group of residues could be taken as anindication of it being important as an active site inP14a.

Materials and Methods

Sample preparation

Uniformly 15N/13C and 15N-labeled P14a was overex-pressed in Escherichia coli cells grown on a minimal med-ium containing 15NH4Cl as the sole nitrogen source and[13C6]glucose, or unlabeled glucose, respectively, as thesole carbon source. The resulting P14a inclusion bodieswere isolated using the procedure of Ramage et al.(1995), and then reduced and solubilized in 100 mM gly-cine buffer at pH 3.0 with 100 mM DTT and 8.5 M urea.The unfolded protein was puri®ed by cation-exchangechromatography in the presence of urea. The eluted P14awas dissolved in 50 mM citric acid buffer at pH 5.0 with150 mM NaCl and 8.5 M urea; this protein solution was®rst diluted fourfold with 100 mM acetic acid buffer, andthen further diluted with 50 mM acetic acid buffer con-taining 2.125 M urea until the protein concentration wasabout 50 mg/ml, at which refolding was initiated. Oxi-dation of the disul®de bonds was achieved by additionof oxidized and reduced glutathione at concentrations of1 mM and 0.1 mM, respectively, and the pH wasadjusted to 8.0 by the addition of 5 M NaOH to facilitatedisul®de shuf¯ing. The solution was incubated for 12hours at 4�C, whereupon the pH was reset to 5.0 by ad-dition of acetic acid, the solution was ®ltered and loadedonto a column of SP-Sepharose High Performance.Native P14a was eluted with a linear gradient of NaCl in50 mM sodium acetate (pH 5.0), dialyzed against dis-tilled water and ®nally lyophilized.

NMR spectroscopy

NMR measurements were performed on BrukerAMX600 and Varian U750� spectrometers equippedwith four channels, using either uniformly 13C/15N dou-bly labeled P14a, uniformly 15N-labeled P14a or un-labeled P14a at a concentration of 1.5 mM in 90% H2O/10% 2H2O or in 99.98% 2H2O. Unless stated otherwise,the spectra were recorded at pH 5.5 and at 30�C. Quad-rature detection in the indirect dimensions was achieved

using States-TPPI (Marion et al., 1989). For data proces-sing and spectral analysis we used the programs PROSA(GuÈ ntert et al., 1992) and XEASY (Bartels et al., 1995), re-spectively.

Resonance assignments and the input for thestructure calculation were obtained from the followingexperiments. Using 13C/15N-labeled P14a: 3D Ha/bCa/

b(CO)NHN (Szyperski et al., 1994a). Time domain datasize 76 � 36 � 512 complex points, t1,max(13C) 6.1 ms,t1,max(1H) 10.1 ms, since the 1H chemical shifts werescaled up by a factor of 1.65, t2,max(15N) 19.4 ms,t3,max(1H) 65.5 ms. 3D CBCANHN (Grzesiek & Bax,1992). Time domain data size 56 � 40 � 512 complexpoints, t1,max(13C) 6.5 ms, t2,max(15N) 21.6 ms, t3,max(1H)65.5 ms. 3D COHNNCA (Szyperski et al., 1995). Timedomain data size 52 � 110 � 512 complex points,t1,max(15N) 33.3 ms, t2,max(13C � O) 18.5 ms, t2,max(13Ca)9.25 ms, since the 13Ca chemical shifts were scaled downby a factor of 2, t3,max(1H) 77.8 ms. 2D [13C, 1H]-COSY(Bodenhausen & Ruben, 1980). Time domain data size206 � 1024 complex points, t1,max(13C) 19.4 ms, t2,max(1H)131.1 ms. 3D HCCH-COSY (Kay et al., 1990). Timedomain data size 100 � 32 � 512 complex points,t1,max(1H) 16.8 ms, t2,max(13C) 10.2 ms, t3,max(1H) 74.8 ms.3D 13C-resolved [1H, 1H]-NOESY (Ikura et al., 1990).Time domain data size 150 � 31 � 512 complex points,t1,max(1H) 25.2 ms, t2,max(13C) 9.2 ms, t3,max(1H) 74.8 ms,tm � 64 ms. 2D ct-[13C, 1H]-COSY for aromatic spin sys-tems (Vuister & Bax, 1992). Time domain data size68 � 1024 complex points, t1,max(13C) 26 ms, t2,max(1H)149.5 ms. 2D (Hb)Cb(CgCd)Hd (Yamazaki et al., 1993).Time domain data size 100 � 2048 complex points,t1,max(13C) 26 ms, t2,max(1H) 205 ms. 2D 1H-TOCSY-relayed ct-[13C, 1H]-COSY (Zerbe et al., 1996). Timedomain data size 68 � 1024 complex points, t1,max(13C)26 ms, t2,max(1H) 149.5 ms, tm � 20 ms. Using 15N-labeledP14a: 3D 15N-resolved [1H, 1H]-TOCSY (Fesik &Zuiderweg, 1988). Time domain data size 160 � 32 � 512complex points, t1,max(1H) 27.2 ms, t2,max(15N) 17.3 ms,t3,max(1H) 65.5 ms, tm � 50 ms. 3D 15N-resolved [1H, 1H]-NOESY (Fesik & Zuiderweg, 1988). Time domain datasize 160 � 29 � 512 complex points, t1,max(1H) 27.2 ms,t2,max(15N) 15.7 ms, t3,max(1H) 65.5 ms, tm � 65 ms. Usingunlabeled P14a: 2D [1H-1H]-DQF-COSY (Rance et al.,1983). 2D clean-[1H, 1H]-TOCSY (tm � 50 ms) (Griesingeret al., 1988). 2D [1H, 1H]-NOESY (tm � 50 ms) (AnilKumar et al., 1980). Time domain data size 360 � 1024complex points, t1,max(1H) 61.2 ms, t2,max(1H) 120.8 ms.2D [1H, 1H]-TOCSY (tm � 50 ms) and 2D [1H, 1H]-NOESY (tm � 50 ms) in H2O solution. Time domain datasize 512 � 1024 complex points, t1,max(1H) 87 ms,t2,max(1H) 102.4 ms in 2H2O solution. The carrier positionwas set to 4.67 ppm for 1H, 117.8 ppm for 15N, 56 ppm,43 ppm and 144 ppm for 13Ca, 13Caliph and 13Carom, re-spectively, and 176.8 ppm for carbonyl carbon atoms.The 1H chemical shifts are relative to internal 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt (DSS).15N and 13C chemical shifts are relative to DSS using theconversion factors that have been reported by Wishartet al. (1995).

Vicinal 3JHNa coupling constants were determined byinverse Fourier transformation of in-phase multiplets(Szyperski et al., 1992) from a 2D [15N, 1H]-COSYspectrum recorded with a data size of 192 � 2048 com-plex points, t1,max(15N) 103.7 ms and t2,max(1H) 208.9 ms.Vicinal 3JNb scalar coupling constants were estimatedfrom a 3D ct-HNNHB spectrum (Archer et al., 1991)with a time domain data size of 160 � 27 � 512 com-plex points, t1,max(1H) 27.2 ms, t2,max(15N) 14.6 ms and


JMB MS 2838 [10/2/97]

t3,max(1H) 65.5 ms. Vicinal 3Jab scalar coupling constantswere estimated from a 3D HCCH-E.COSY spectrum(Eggenberger et al., 1992) with a time domain data sizeof 128 � 42 � 512 complex points, t1,max(1H) 21.5 ms,t2,max(15N) 13.4 ms and t3,max(1H) 74.8 ms. Vicinal 3JC0bscalar couplings were estimated from a 3DHNN(CO)HB experiment (Grzesiek et al., 1992) with atime domain data size of 40 � 85 � 512 complex points,t1,max(15N) 21.6 ms, t2,max(1H) 10.0 ms and t3,max(1H)65.5 ms. Vicinal 3JCaCd scalar coupling constants in leucylresidues were estimated from a 2D long-range 13C-13Ccorrelation experiment (Bax et al., 1992) with a timedomain data size of 150 � 1024 complex points, witht1,max(13C) 12 ms and t2,max(1H) 131.1 ms.

For the determination of amide proton exchange rates, asample of 450 ml of a fully protonated 1.5 mM solution ofuniformly 15N-labeled P14a was lyophilized. The proteinwas redissolved in the same amount of 2H2O, and aseries of 2D [15N, 1H]-COSY spectra were recorded witha time domain data size of 192 � 1024 complex points,t1,max(15N) 103.7 ms and t2,max(1H) 104.5 ms. At 22�C theindividual measurements were started 30, 90, 150, 210,300, 420, 540, 660, 840, 1080, 1320, 1560, 1920, 2400 and2880 minutes after preparation of the 2H2O solution, andat 30�C the delays were 15, 45, 110, 140, 170, 220, 280,370, 490, 610, 920 and 1040 minutes. The rate constantswere obtained from a least-squares ®t of a single expo-nential function to the peak volumes. For the pH ti-tration of backbone amide proton chemical shifts, aseries of eight 2D [15N, 1H]-COSY spectra were recordedwith a time domain data size of 120 � 1024 complexpoints, t1,max(15N) 64.8 ms and t2,max(1H) 104.5 ms at thefollowing pH values: 6.0, 5.5, 5.0, 4.6, 4.3, 4.0, 3.6 and3.2.

15N spin relaxation measurements were performed withuniformly 15N-labeled P14a at a 1H resonance frequencyof 600 MHz as described by Szyperski et al. (1993). All2D [15N, 1H]-COSY spectra needed for these measure-ments were recorded with a time domain data size of220 � 1024 complex points, t1,max(15N) 118.8 ms andt2,max(1H) 108.6 ms. For the measurement of longitudinal15N spin relaxation times, T1, a series of 11 spectra wasrecorded with relaxation delays of 15 ms, 100 ms,200 ms, 300 ms, 400 ms, 500 ms, 700 ms, 900 ms, 1.2seconds, 1.5 seconds, 2.0 seconds. Transverse 15N spinrelaxation times, T2, were obtained from a series of 11spectra recorded at 600 MHz 1H resonance frequencywith relaxation delays of 10 ms, 20 ms, 30 ms, 40 ms,60 ms, 80 ms, 140 ms, 180 ms, 220 ms, 280 ms and400 ms. Rotating frame 15N spin relaxation times, T1r,were measured with a series of eight spectra using relax-ation delays of 30 ms, 50 ms, 70 ms, 90 ms, 130 ms,170 ms, 220 ms and 300 ms, and a spin-lock power of12,870 rad secondÿ1. The 15N spin relaxation times wereobtained from a least-squares ®t of single exponentialfunctions to the experimental data.

15N{1H} steady-state heteronuclear Overhauser effectswere determined as described by Kay et al. (1989). Arelaxation delay of six seconds was used in order toreduce effects arising from amide proton exchange. Thespectra were recorded with a time domain data size of220 � 1024 complex points, t1,max(15N) 118.8 ms andt2,max(1H) 108.6 ms. NOE values were determined as theratios of cross-peak volumes measured from the spectraacquired with and without 1H saturation during the re-cycle delay, and the experimental errors of the NOEvalues were estimated from the spectral noise.

Determination of the three-dimensional structure

The input for the distance geometry calculations with theprogram DIANA consisted of upper distances limits de-rived from NOESY cross-peak intensities with the pro-gram CALIBA (GuÈ ntert et al., 1991), and of dihedralangle constraints derived from 3JHNa

3Jab,3JC0b,

3JNb coup-ling constants and local NOEs using the programHABAS (GuÈ ntert et al., 1989, 1991). HABAS also pro-vided a number of stereospeci®c assignments of b-meth-ylene protons. Several rounds of structure calculationswith DIANA (GuÈ ntert et al., 1991) and NOESY cross-peak assignment with the program ASNO (GuÈ ntert et al.,1993) were performed. Backbone dihedral angle con-straints derived from conformation-dependent 13Ca

chemical shifts (LuginbuÈ hl et al., 1995) were initiallyused to accelerate this iterative procedure, but were notused in the ®nal structure calculation. Additional stereo-speci®c assignments were obtained using the programGLOMSA (GuÈ ntert et al., 1991). Upper and lower dis-tance constraints for disul®de bridges were included asdescribed (Williamson et al., 1985). The ®nal round ofDIANA structure calculations was started with 100 ran-domized conformers and included four REDAC cycles,where the maximal target function value per residue forlocally acceptable segments was set to C(1) � 1.5 AÊ ,C(2) � 1.0 AÊ , C(3) � 0.8 AÊ and C(4) � 0.3 AÊ , respectively,followed by a ®nal cycle using only experimental con-straints (GuÈ ntert & WuÈ thrich, 1991). The 20 DIANA con-formers with the smallest target function values weresubjected to restrained energy minimization using theAMBER all-atom force ®eld (Weiner et al., 1986) as im-plemented in the program OPAL (LuginbuÈ hl et al., 1996).The pseudo-potential for NMR constraints was adjustedsuch that violations of 0.10 AÊ for distance constraintsand 2.5� for dihedral angle constraints correspond tokBT/2 at room temperature. The energy minimizationwas carried out in a shell of water molecules with aminimal thickness of 6 AÊ , performing a total of 1500steps of conjugate gradient minimization for each confor-mer. The resulting 20 energy-re®ned conformers areused to represent the solution structure of P14a. Ther.m.s.d. values were calculated with the program XAM(Xia, 1992). All color Figures were generated with theprogram MOLMOL (Koradi et al., 1996).

Acknowledgements

FinancialsupportwasobtainedfromtheSchweizerischerNationalfonds (project 31.32035.91). C. F. is indebted tothe ``Schweizerische Bundesstipendienkommission'' for afellowship. We further acknowledge the Centro Svizzerodi Calcolo Scienti®co for use of the NEC SX-3 computer,and Mrs E. Ulrich for the careful processing of the manu-script.

References

Alexander, D., Goodman, R. M., Gut-Rella, M.,Glascock, C., Weymann, K., Friedrich, L., Maddox,D., Ahl-Goy, P., Luntz, T., Ward, E. & Ryals, J.(1993). Increased tolerance to two oomycete patho-gens in transgenic tobacco expressing pathogenesis-related protein 1a. Proc. Natl Acad. Sci. USA, 90,7327±7331.


JMB MS 2838 [10/2/97]

Anil Kumar, Ernst, R. R. & WuÈ thrich, K. (1980). A two-dimensional nuclear Overhauser enhancement (2DNOE) experiment for the elucidation of completeproton-proton cross-relaxation networks in biologi-cal macromolecules. Biochem. Biophys. Res. Commun.95, 1±6.

Archer, S. J., Ikura, M., Torchia, D. A. & Bax, A. (1991).An alternative 3D NMR technique for correlatingbackbone 15N with side chain Hb resonances in lar-ger proteins. J. Magn. Reson. 95, 636±641.

Bartels, C., Xia, T., Billeter, M. & WuÈ thrich, K. (1995).The program XEASY for computer-supported NMRspectral analysis of biological macromolecules.J. Biomol. NMR, 5, 1±10.

Bax, A. & Grzesiek, S. (1993). Methodological advancesin protein NMR. Acc. Chem. Res. 26, 131±138.

Bax, A., Max, D. & Zax, D. (1992). Measurement oflong-range 13C-13C J couplings in a 20-kDa protein-peptide complex. J. Am. Chem. Soc. 114, 6923±6925.

Billeter, M., Braun, W. & WuÈ thrich, K. (1982). Sequentialresonance assignments in protein 1H nuclear mag-netic resonance spectra: computation of stericallyallowed proton-proton distances and statisticalanalysis of proton-proton distances in single crystalprotein conformations. J. Mol. Biol. 155, 321±346.

Bodenhausen, G. & Ruben, D. (1980). Natural abun-dance 15N NMR by enhanced heteronuclearspectroscopy. Chem. Phys. Letters, 69, 185±188.

Bol, J. F., Linthorst, H. J. M. & Cornelissen, B. J. C.(1990). Plant pathogenesis-related proteins inducedby virus infection. Annu. Rev. Phytopathol. 28, 113±138.

Bundi, A. & WuÈ thrich, K. (1979). Use of amide 1H-NMRtitration shifts for studies of polypeptideconformation. Biopolymers, 18, 299±311.

Camacho Henriquez, A. & SaÈnger, H. L. (1982). Analysisof acid extractable tomato leaf proteins after infec-tion with a viroid, two viruses and a fungus andpartial puri®cation of the `pathogenesis-related' pro-tein P14. Arch. Virol. 74, 181±195.

Chen, Y. W., Fersht, A. R. & Henrick, K. (1993). Contri-bution of buried hydrogen bonds to protein stab-ility. The crystal structures of two Barnase mutants.J. Mol. Biol. 234, 1158±1170.

Christ, U. & MoÈsinger, E. (1989). Pathogenesis-relatedproteins of tomato: 1. Induction by Phytophthorainfestans and other biotic and abiotic inducers andcorrelation with resistance. Physiol. Mol. Plant Pathol.35, 53±65.

Dougherty, D. A. (1996). Cation-p interactions in chem-istry and biology: a new view of benzene, Phe, Tyrand Trp. Science, 271, 163±168.

Eggenberger, U., Karimi-Nejad, Y., ThuÈ ring, H.,RuÈ terjans, H. & Griesinger, C. (1992). Determinationof Ha,Hb and Hb,C

0 coupling constants in 13C-labeled proteins. J. Biomol. NMR, 2, 583±590.

Fesik, S. W. & Zuiderweg, E. R. P. (1988). Heteronuclearthree-dimensional NMR spectroscopy. A strategyfor the simpli®cation of homonuclear two-dimen-sional NMR spectra. J. Magn. Reson. 78, 588±593.

Fischer, W., Christ, U., Baumgartner, M., Erismann,K. H. & MoÈsinger, E. (1989). Pathogenesis-relatedproteins of tomato: 2. Biochemical and immunologi-cal characterization. Physiol. Mol. Plant Pathol. 35,67±83.

Fraser, R. S. S. (1981). Evidence for the occurrence of the`pathogenesis-related' proteins in leaves of healthytobacco plants during ¯owering. Physiol. PlantPathol. 19, 69±76.

Gianinazzi, S., Martin, C. & ValleÂe, J. C. (1970). Hyper-sensibiliteÂ aux virus, tempeÂrature et proteÂines solu-bles chez le Nicotiana Xanti n.c. Apparition denouvelles macromoleÂcules lors de la reÂpression dela syntheÁse viral. C. R. Acad. Sci. Paris, ser. D, 270,2383±2386.

Griesinger, C., Otting, G., WuÈ thrich, K. & Ernst, R. R.(1988). Clean-TOCSY for 1H spin system identi®-cation in macromolecules. J. Am. Chem. Soc. 110,7870±7872.

Grzesiek, S. & Bax, A. (1992). An ef®cient experimentfor sequential backbone assignment of medium-sized isotopically enriched proteins. J. Magn. Reson.99, 201±207.

Grzesiek, S., Ikura, M., Clore, G. M., Gronenborn,A. M. & Bax, A. (1992). A 3D triple-resonance NMRtechnique for qualitative measurement of carbonyl-Hb J couplings in isotopically enriched proteins.J. Magn. Reson. 96, 215±221.

GuÈ ntert, P. & WuÈ thrich, K. (1991). Improved ef®ciencyof protein structure calculations from NMR datausing the program DIANA with redundant dihe-dral angle constraints. J. Biomol. NMR, 1, 447±456.

GuÈ ntert, P., Braun, W., Billeter, M. & WuÈ thrich, K.(1989). Automated stereospeci®c 1H NMR assign-ments and their impact on the precision of proteinstructure determinations in solution. J. Am. Chem.Soc. 111, 3997±4004.

GuÈ ntert, P., Braun, W. & WuÈ thrich, K. (1991). Ef®cientcomputation of three-dimensional protein structuresin solution from nuclear magnetic resonance datausing the program DIANA and the supporting pro-grams CALIBA, HABAS and GLOMSA. J. Mol. Biol.217, 517±530.

GuÈ ntert, P., DoÈ tsch, V. & Wider, G. (1992). Processing ofmulti-dimensional NMR data with the new soft-ware PROSA. J. Biomol. NMR, 2, 619±629.

GuÈ ntert, P., Berndt, K. D. & WuÈ thrich, K. (1993). Theprogram ASNO for computer-supported collectionof NOE upper distance constraints as input for pro-tein structure determination. J. Biomol. NMR, 3,601±606.

Harper, E. T. & Rose, G. D. (1993). Helix stop signals inprotein and peptides: the capping box. Biochemistry,32, 7605±7609.

Holm, L. & Sander, C. (1993). Protein structure compari-son by alignment of distance matrices. J. Mol. Biol.233, 123±138.

Holm, L. & Sander, C. (1994). The FSSP database ofstructurally aligned protein fold families. Nucl.Acids Res. 22, 3600±3609.

Ikura, M., Kay, L. E., Tschudin, R. & Bax, A. (1990).Three-dimensional NOESY-HMQC spectroscopy ofa 13C-labeled protein. J. Magn. Reson. 86, 204±209.

Joosten, M. H. A. J. & De Wit, P. J. G. M. (1989). Identi-®cation of several pathogenesis-related proteins intomato leaves inoculated with Cladosporium fulvum(Syn. Fulvia fulva) as b-1,3-glucanases andchitinases. Plant Physiol. 89, 945±951.

Kay, L. E., Torchia, D. A. & Bax, A. (1989). Backbonedynamics of proteins as studied by 15N inversedetected heteronuclear NMR spectroscopy: appli-cation to staphylococcal nuclease. Biochemistry, 28,8972±8979.

Kay, L. E., Ikura, M. & Bax, A. (1990). Proton-protoncorrelation via carbon-carbon couplings: a three-dimensional NMR approach for the assignment ofaliphatic resonances in proteins labeled with car-bon-13. J. Am. Chem. Soc. 112, 888±889.


JMB MS 2838 [10/2/97]

Koradi, R., Billeter, M. & WuÈ thrich, K. (1996). MOL-MOL: a program for display and analysis of macro-molecular structures. J. Mol. Graph. 14, 51±55.

Liepinsh, E., Otting, G. & WuÈ thrich, K. (1992). NMRspectroscopy of hydroxyl protons in aqueous sol-utions of peptides and proteins. J. Biomol. NMR, 2,447±465.

Linthorst, H. J. M. (1991). Pathogenesis-related proteinsin plants. Crit. Rev. Plant Sci. 10, 123±150.

Lucas, J., Camacho Henriquez, A., Lottspeich, F.,Henschen, A. & SaÈnger, H. L. (1985). Amino acidsequence of the `pathogenesis-related' leaf proteinp14 from viroid-infected tomato reveals a new typeof structurally unfamiliar proteins. EMBO J. 4,2745±2749.

LuginbuÈ hl, P., Szyperski, T. & WuÈ thrich, K. (1995). Stat-istical basis for the use of 13Ca chemical shifts inprotein structure determination. J. Magn. Reson. ser.B, 109, 229±233.

LuginbuÈ hl, P., GuÈ ntert, P., Billeter, M. & WuÈ thrich, K.(1996). The new program OPAL for moleculardynamics simulations and energy re®nements ofbiological macromolecules. J. Biomol. NMR, 8, 136±146.

Marion, D., Ikura, M., Tschudin, R. & Bax, A. (1989).Rapid recording of 2D NMR spectra withoutphase cycling: application to the study of hydro-gen exchange in proteins. J. Magn. Reson. 85, 393±399.

Messerle, B. A., Wider, G., Otting, G., Weber, C. &WuÈ thrich, K. (1989). Solvent suppression using aspin-lock in 2D and 3D NMR spectroscopy withH2O solutions. J. Magn. Reson. 85, 608±613.

MunÄ oz, V., Blanco, F. J. & Serrano, L. (1995). The hydro-phobic-staple motif and a role for loop-residues ina-helix stability and protein folding. Nature Struct.Biol. 2, 380±385.

Nagayama, K. & WuÈ thrich, K. (1981). Structural in-terpretation of vicinal proton-proton couplingconstants 3Jab in the basic pancreatic trypsin in-hibitor measured by two-dimensional J-resolvedNMR spectroscopy. Eur. J. Biochem. 115, 653±657.

Niderman, T., Genetet, I., BruyeÁre, T., Gees, R., Stintzi,A., Legrand, M., Fritig, B. & MoÈsinger, E. (1995).Pathogenesis-related PR-1 proteins are antifungal.Plant Physiol. 108, 17±27.

Orengo, C. A. & Thornton, J. M. (1993). Alpha plus betafolds revisited: some favoured motifs. Structure, 1,105±120.

Orengo, C. A., Jones, D. T. & Thornton, J. M. (1994).Protein superfamilies and domain superfolds.Nature, 372, 631±634.

Perkins, S. J. & WuÈ thrich, K. (1979). Ring current effectsin the conformation dependent NMR chemicalshifts of aliphatic protons in the basic pancreatictrypsin inhibitor. Biochim. Biophys. Acta, 576, 409±423.

Ramage, R., Cheneval, D., Chvei, M., Graff, P.,Hemmig, R., Heng, R., Kocher, H. P., Mackenzie,A., Memmert, K., Revesz, L. & Wishart, W. (1995).Expression, refolding, and autocatalytic proteolyticprocessing of the inteleukin-1b-converting enzymeprecursor. J. Biol. Chem. 270, 9378±9383.

Rance, M., Sùrensen, O. W., Bodenhausen, G., Wagner,G., Ernst, R. R. & WuÈ thrich, K. (1983). Improvedspectral resolution in COSY 1H NMR spectra ofproteins via double quantum ®ltering. Biochem. Bio-phys. Res. Commun. 117, 479±485.

Richardson, J. S. (1981). The anatomy and taxonomy ofprotein structure. Advan. Protein Chem. 34, 167±339.

Richardson, J. S. & Richardson, D. C. (1988). Amino acidpreferences for speci®c locations at the ends of ahelices. Science, 240, 1648±1652.

Ryals, J., Uknes, S. & Ward, E. (1994). Systemic acquiredresistance. Plant Physiol. 104, 1109±1112.

Schellmann, C. (1980). In Protein Folding (Jaenicke, R.,ed.), pp. 53, Elsevier/North-Holland, New York.

Schulz, G. E. & Schirmer, R. H. (1979). Principles of Pro-tein Structure, Springer-Verlag.

Shakhnovich, E., Abkevich, V. & Ptitsyn, O. (1996). Con-served residues and the mechanism of proteinfolding. Nature, 379, 96±98.

Szyperski, T., GuÈ ntert, P., Otting, G. & WuÈ thrich, K.(1992). Determination of scalar coupling constantsby inverse Fourier transformation of in-phasemultiplets. J. Magn. Reson. 99, 552±560.

Szyperski, T., LuginbuÈ hl, P., Otting, G., GuÈ ntert, P. &WuÈ thrich, K. (1993). Protein dynamics studied byrotating frame 15N spin relaxation times. J. Biomol.NMR, 3, 151±164.

Szyperski, T., Pellecchia, M. & WuÈ thrich, K. (1994a). 3DHa/bCa/b(CO)NHN, a projected 4D NMR exper-iment for sequential correlation of polypeptide1Ha/b, 13Ca/b and backbone 15N and 1HN chemicalshifts. J. Magn. Reson. ser. B, 105, 188±191.

Szyperski, T., Antuch, W., Schick, M., Betz, A., Stone,S. R. & WuÈ thrich, K. (1994b). Transient hydrogenbonds identi®ed on the surface of the NMR solutionstructure of Hirudin. Biochemistry, 33, 9303±9310.

Szyperski, T., Braun, D., FernaÂndez, C., Bartels, C. &WuÈ thrich, K. (1995). A novel reduced-dimensional-ity triple-resonance experiment for ef®cient poly-peptide backbone assignment, 3D CO HN N CA.J. Magn. Reson. ser. B, 108, 197±203.

Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994).CLUSTALW: improving the sensitivity of progress-ive multiple sequence alignment through sequenceweighting, position-speci®c gap penalties andweight matrix choice. Nucl. Acids Res. 22, 4673±4680.

Van Loon, L. C. (1985). Pathogenesis-related proteins.Plant Mol. Biol. 4, 111±116.

Van Loon, L. C. & Van Kammen, A. (1970). Polyacryl-amide disk electrophoresis of the soluble leaf pro-teins from Nicotiana tabacum var. `Samsun' and`Samsun NN' II. Changes in protein constitutionafter infection with tobacco mosaic virus. Virology,40, 199±211.

Vuister, G. W. & Bax, A. (1992). Resolution enhance-ment and spectral editing of uniformly 13C-enrichedproteins by homonuclear broadband 13C de-coupling. J. Magn. Reson. 98, 428±435.

Wagner, G. & WuÈ thrich, K. (1982). Sequential resonanceassignments in protein 1H nuclear magnetic reson-ance spectra: basic pancreatic trypsin inhibitor.J. Mol. Biol. 155, 347±366.

Weiner, P. K., Kollman, P. A., Nguyen, D. T. & Case,D. A. (1986). An all atom force ®eld for simulationsof proteins and nucleic acids. J. Comp. Chem. 7,230±252.

Williamson, M. P., Havel, T. F. & WuÈ thrich, K. (1985).Solution conformation of proteinase inhibitor IIAfrom bull seminal plasma by 1H nuclear magneticresonance and distance geometry. J. Mol. Biol. 182,295±315.

Wishart, D. S., Bigam, C. G., Yao, J., Abildgaard, H. J. D.,Markley, J. L. & Sykes, B. D. (1995). 1H, 13C and


JMB MS 2838 [10/2/97]

15N chemical shift referencing in biomolecularNMR. J. Biomol. NMR, 6, 135±140.

WuÈ thrich, K. (1986). NMR of Proteins and Nucleic Acids,Wiley, New York.

WuÈ thrich, K., Billeter, M. & Braun, W. (1983). Pseudo-structures for the 20 common amino acids for usein studies of protein conformations by measure-ments of intramolecular proton-proton distance con-straints with nuclear magnetic resonance. J. Mol.Biol. 169, 949±961.

Xia, T. (1992). Software for determination and visual dis-play of NMR structures of proteins: the distancegeometry program DGPLAY and the computergraphics programs CONFOR and XAM. PhD thesis9831, ETH ZuÈ rich, Switzerland.

Yamazaki, T., Forman-Kay, J. D. & Kay, L. E. (1993).Two-dimensional NMR experiments for correlating13Cb and 1Hd/e chemical shifts of aromatic residuesin 13C-labeled proteins via scalar couplings. J. Am.Chem. Soc. 115, 11054±11055.

Zerbe, O., Szyperski, T., Ottiger, M. & WuÈ thrich, K.(1996). 3D 1H-TOCSY-relayed ct-[13C, 1H]-HMQC

for aromatic spin system identi®cation in uniformly13C labeled proteins. J. Biomol NMR, 7, 99±106.

Edited by P. E. Wright

(Received 27 August 1996; received in revised form31 October 1996; accepted 6 November 1996)

http://www.hbuk.co.uk/jmb

Supplementary material, comprising three Tables,is available from JMB Online.


JMB MS 2838 [10/2/97]

NMR Solution Structure of the Pathogenesis-related Protein...

Documents

Transcript of NMR Solution Structure of the Pathogenesis-related Protein...