Structural Characterization of Unfolded States ofApomyoglobin using Residual Dipolar Couplings

Ronaldo Mohana-Borges, Natalie K. Goto, Gerard J. A. KroonH. Jane Dyson and Peter E. Wright*

Department of MolecularBiology and Skaggs Institute ofChemical Biology, The ScrippsResearch Institute, 10550North Pines Road, La Jolla, CA92037, USA

The conformational propensities of unfolded states of apomyoglobin havebeen investigated by measurement of residual dipolar couplings between15N and 1H in backbone amide groups. Weak alignment of apomyoglobinin acid and urea-unfolded states was induced with both stretched andcompressed polyacrylamide gels. In 8 M urea solution at pH 2.3, con-ditions under which apomyoglobin contains no detectable secondary ortertiary structure, significant residual dipolar couplings of uniform signwere observed for all residues. At pH 2.3 in the absence of urea, a changein the magnitude and/or sign of the residual dipolar couplings occurs inlocal regions of the polypeptide where there is a high propensity forhelical secondary structure. These results are interpreted on the basis ofthe statistical properties of the unfolded polypeptide chain, viewed as apolymer of statistical segments. For a folded protein, the magnitude andsign of the residual dipolar couplings depend on the orientation of eachbond vector relative to the alignment tensor of the entire molecule,which reorients as a single entity. For unfolded proteins, there is no globalalignment tensor; instead, residual dipolar couplings are attributed toalignment of the statistical segments or of transient elements of secondarystructure. For apomyoglobin in 8 M urea, the backbone is highlyextended, with f and c dihedral angles favoring the b or PII regions.Each statistical segment has a highly anisotropic shape, with the N–Hbond vectors approximately perpendicular to the long axis, and becomesweakly aligned in the anisotropic environment of the strained acrylamidegels. Local regions of enhanced flexibility or chain compaction are charac-terized by a decrease in the magnitude of the residual dipolar couplings.The formation of a small population of helical structure in the acid-denatured state of apomyoglobin leads to a change in sign of the residualdipolar couplings in local regions of the polypeptide; the population ofhelix estimated from the residual dipolar couplings is in excellent agree-ment with that determined from chemical shifts. The alignment modeldescribed here for apomyoglobin can also explain the pattern of residualdipolar couplings reported previously for denatured states of staphylococcalnuclease and other proteins. In conjunction with other NMR experiments,residual dipolar couplings can provide valuable insights into the dynamicconformational propensities of unfolded and partly folded states of proteinsand thereby help to chart the upper reaches of the folding landscape.

q 2004 Elsevier Ltd. All rights reserved.

Keywords: apomyoglobin; unfolded protein; NMR; residual dipolarcoupling; conformational ensemble*Corresponding author

Introduction

Although major advances have been made inelucidating the folding mechanism of several pro-teins, detailed understanding of the molecularevents that occur during the folding process

0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

E-mail address of the corresponding author:wright@scripps.edu

Abbreviations used: apoMb, apomyoglobin; NMR,nuclear magnetic resonance; NOE, nuclear Overhausereffect; HSQC, heteronuclear single quantum coherence;IPAP, in-phase–antiphase spectra; RDC, residual dipolarcoupling.

doi:10.1016/j.jmb.2004.05.022 J. Mol. Biol. (2004) 340, 1131–1142

remains limited. It is currently thought that proteinfolding occurs by a hierarchical process thatinvolves progressive collapse of the polypeptidein a rough, funnel-like energy landscape.1,2 Muchremains to be learned about the nature of the partlyfolded states that accompany chain compactionduring the folding process. Many proteins areintrinsically unstructured in their biologicallyfunctional states;3–6 the intrinsic conformationalpropensities of such proteins have direct relevanceto their biological function. Also, the variousamyloid diseases involve the formation ofpathogenic aggregates from partly folded or mis-folded polypeptides.7 Structural characterizationof unfolded and partly folded proteins is thereforeof central importance for understanding the mech-anisms of protein folding and the role of unstruc-tured states in normal biological processes and indisease. Due to their high flexibility and disorderednature, such states are unsuitable for structuralcharacterization by X-ray crystallography, but canbe studied by nuclear magnetic resonance (NMR)spectroscopy, which can provide detailed struc-tural and dynamic information at the level of indi-vidual residues.8

Apomyoglobin (apoMb) has been considered aparadigm in protein folding because it forms anumber of partly folded states of varying compact-ness and secondary structure content that can bestudied at equilibrium, providing insights into itsfolding landscape.9–12 The protein folds by way ofan on-pathway kinetic intermediate, which hasextensive helical structure.13–16 The heme-contain-ing holoprotein adopts a compact structurecontaining eight a-helices (designated A to H);although the apoprotein is less stable, it retains anextensive hydrophobic core and most of thesecondary structure, as well as specific tertiaryinteractions within the hydrophobic clusters.17,18

Acid denaturation of apoMb occurs in two distinctstages,19,20 forming a compact molten globule atpH 4.1 with a helix content of ,35%, and then amore highly denatured state at pH 2.3 with a smal-ler content of helix. The transient helical secondarystructure formed in the pH 2.3 state10 is disruptedin the presence of 8 M urea.12

While chemical shifts, and short and medium-range NOEs provide valuable insights into thesecondary structural propensities of the poly-peptide backbone in unfolded and partly foldedproteins,21 it has generally proved difficult toobserve long-range NOEs for unfolded or partlyfolded states. To circumvent this problem, severalgroups have utilized site-specific nitroxide spinlabeling to probe transient long-range interactionsin disordered proteins.22–25 An alternativeapproach that has recently been introducedinvolves the measurement of residual dipolarcouplings (RDC) in unfolded proteins that areweakly oriented in strained polyacrylamide gels.26

Dipolar couplings contain information on theorientation of internuclear vectors and havebecome an important adjunct to traditional struc-

tural constraints in refinement of NMR structuresof globular proteins.27,28 RDCs are measured byweakly aligning a macromolecule in slightly aniso-tropic nematic liquid crystalline media, such asdetergent bicelles or filamentous phages, or inanisotropically compressed gels that interfere withthe isotropic tumbling of the macromolecule in thesolution.29–32 The small degree of alignment result-ing from the anisotropic environment leads toincomplete averaging of the dipolar couplingbetween magnetic nuclei close in space. The mag-nitude of the RDC is dependent on the orientationof an internuclear vector relative to the alignmenttensor of the protein as a whole. In other words,bond vectors such as 15N-1H or 13C-1H can beoriented relative to a global alignment tensor fixedin the molecular frame, regardless of their locationin the molecule.29 Alternatively, the informationcan be interpreted in terms of angular relationsbetween pairs of bond vectors that are indepen-dent of the intervening distance. The magnitudeof the dipolar coupling DNH between 15N and itscovalently bound 1H is given by equation (1):

DNH ¼ 2S gN gH=r3NH ½Aað3cos

2u2 1Þ

þ 3=2Arðsin2u cos2fÞ� ð1Þ

where S is the generalized order parameterdescribing internal motions of the N–H bondvector, gN and gH are the gyromagnetic ratios of15N and 1H, Aa and Ar are the axial and rhombiccomponents of the alignment tensor, and u and fare the angles relating the orientation of the N–Hbond vector to the alignment tensor.

Experimentally, the limited pH range, tempera-ture range, and solution conditions over whichbicelles and phages form nematic liquid crystalsmake them of limited applicability for studies ofdenatured proteins. In contrast, strained alignedpolyacrylamide gels are a useful alternative align-ment medium because they are stable and inertover a broad range of conditions. Recently, non-zero RDCs were observed for the denatured stateof truncated staphylococcal nuclease (D131D) andfor short unstructured peptides weakly aligned inpolyacrylamide gels.26,33–35 The observation ofmeasurable RDC in highly flexible polypeptides iscounterintuitive because internal motions and con-formational averaging might be expected to resultin vanishingly small residual dipolar couplings. Inthe case of denatured staphylococcal nuclease, theRDCs were interpreted in terms of persistentnative-like topology and structure, even in 8 Murea.26,33,34

Here, we report the measurement of RDCs inunfolded states of apoMb at pH 2.3 and in 8 Murea at pH 2.3. The local and long-range confor-mational propensities and dynamics of these stateshave been characterized in detail using NMRmethods.9,10,12,25 We show that the RDC data forthese unfolded states of apomyoglobin can beinterpreted on the basis of local conformational

1132 Dipolar Couplings for Apomyoglobin Unfolded States

propensities and the well-known physicochemicalproperties of flexible polypeptide chains, ratherthan models that invoke the persistence of native-like structure under denaturing conditions. Wealso discuss the potential role of residual dipolarcouplings for mapping the conformationalensemble of unfolded states of proteins.

Results and Discussion

Polyacrylamide gels as weak alignment mediafor unfolded proteins

Neither bicelles nor filamentous bacteriophagesare suitable for measurement of residual dipolarcouplings for apomyoglobin, even under nativeconditions. The protein resonances are broadenedbeyond detection in the presence of these media,indicating significant interactions betweenapomyoglobin and the alignment medium itself.However, by using strained polyacrylamidegels31,32 it was possible to obtain high-resolution1H-15N heteronuclear single quantum coherence(HSQC) spectra of apomyoglobin in severalunfolded or partly folded states. Figure 1 shows1H-15N HSQC spectra of apoMb in 8 M urea at pH2.3, acid-unfolded (pH 2.3), and native (pH 6.1)states, both in isotropic solution (black) and in10% (w/v) polyacrylamide gels (red). Clearly,there are no significant changes in any of thespectra when the protein solutions are cast intothe gel. Since 1H-15N cross-peaks are very sensitiveto even slight changes in their chemical micro-environment, these observations strongly suggestthat the conformational ensembles of apoMb inthe various unfolded states are not perturbedsignificantly by the presence of the gel matrix.

Furthermore, 15N R2 relaxation rates measured forthe highly denatured state in 8 M urea at pH 2.3are very similar in isotropic solution and in thepolyacrylamide gel (data not shown).

DNH of unfolded states of apoMb

Experimentally, dipolar couplings can beextracted from 2D in-phase–antiphase (IPAP)-HSQC spectra recorded under both anisotropicand isotropic conditions.36 A representative regionof the IPAP-HSQC spectrum of apomyoglobin in8 M urea at pH 2.3 is shown in Figure 2. For

Figure 1. Superposition of 1H-15N HSQC spectra of apoMb samples in solution (black cross-peaks) and cast into 10%polyacrylamide gel (red cross-peaks). A, Urea-denatured state, in 8 M urea at pH 2.3; B, acid-unfolded state, pH 2.3;C, native state, pH 6.1. The band of peaks near 7.0 and 7.7 ppm arises from the amide groups of the polyacrylamidegel.31

Figure 2. Region of the 2D IPAP-HSQC spectrum ofapomyoglobin in 8 M urea at pH 2.3, showing the15N-1H coupling for His82 in isotropic solution (black)and in a stretched polyacrylamide gel (red).

Dipolar Couplings for Apomyoglobin Unfolded States 1133

unfolded states of apomyoglobin, however, the 1Hchemical shift dispersion is very poor, resulting inoverlap of many cross-peaks in the HSQC spec-trum; we therefore used the 3D HNCO-IPAPexperiment37 to separate the 15N-1H cross-peaks ina third dimension according to differences in 13COchemical shifts. There are two ways in which poly-acrylamide gels can be strained to create aniso-tropic environments. A compressed gel can beformed by subjecting a cylindrical gel with adiameter smaller than the internal diameter of theNMR tube to axial compression.31,32 Alternatively,a gel can be stretched in the axial direction (com-pressed radially) by casting it in a cylindrical tubethat is wider than the NMR tube, then squeezingit into a bottomless NMR tube.38 These two typesof strained gels differ in the orientation of themolecular alignment tensor of the solute moleculeswith respect to the applied magnetic field. In thestretched gel, the long axis of the protein solutealigns parallel with the magnetic field, whereasthe alignment axis is orthogonal to the field for acompressed gel. This approximately doubles themagnitude of the dipolar couplings relative tothose obtained in a stretched gel of identical den-sity and degree of compression (i.e. the sameaspect ratio) and leads to a reversal in sign.31,38

This is illustrated in Figures 3 and 4, which showthe residual dipolar couplings measured in com-pressed gels and stretched gels for apoMb in 8 Murea, pH 2.3 and at pH 2.3 without urea.

For urea-denatured apoMb in the compressedgel, the DNH value is positive for all residues forwhich data could be obtained (Figure 3A), whilein the stretched gel, larger values of opposite signare observed (Figure 3B). The DNH values for mostresonances of acid-unfolded apoMb at pH 2.3 alsoshow the expected change in sign between thecompressed and stretched gels (Figure 4). How-ever, in contrast to the urea-denatured state, theresidual dipolar couplings are not uniformly ofthe same sign. The DNH values for each residue inthe compressed and stretched gels are highly corre-lated (Figures 3C and 4C). The negative slope ofthe correlation plots reflects the expected reversalof the sign of DNH between stretched and com-pressed gels. The observed correlations confirmthat even these highly flexible unfolded statesundergo partial alignment within the pores of thestrained polyacrylamide gel.

According to equation (1), the magnitude of theresidual dipolar coupling observed should be pro-portional to S, the generalized order parameter,and (3cos2u 2 1), which describes the orientationof the internuclear vector with respect to theapplied magnetic field. Given the intrinsicflexibility and high degree of disorder of the urea-denatured and acid-denatured states of apomyo-globin, it is at first surprising that RDCs can beobserved at all. Intuitively, one would expect themagnitude of the RDC in an unfolded state to bereduced to zero by averaging of (3cos2u 2 1) overall possible orientations of the internuclear vector

in the conformational ensemble. However, weclearly observe RDCs of significant magnitude forapomyoglobin in two highly unfolded states.

Non-zero RDCs have previously been observedfor denatured states of staphylococcal nucleaseand eglin C, and for short peptides.26,33–35,39,40 TheRDCs for urea-denatured staphylococcal nucleaseand eglin C were interpreted as evidence of long-range ordering, with persistence of native-likeglobal topology in the denatured state.26,40 For apo-myoglobin, there is no correlation between theRDCs measured in denatured states and those forthe native folded protein (Figure 5). Our data sup-port a different interpretation, that the non-zeroRDCs observed for highly unfolded states ofapomyoglobin (and, by extrapolation, for other

Figure 3. 15N-1H residual dipolar couplings as a func-tion of residue number for apomyoglobin in 8 M urea(pH 2.3). A, Compressed polyacrylamide gel; B,stretched polyacrylamide gel; C, correlation betweenresidual dipolar couplings in stretched and compressedgels. RDCs could be measured for all residues exceptthe four proline residues (37, 88, 100, 120) and six resi-dues (32, 99, 104, 105, 130, 138) for which the NH reson-ances were overlapped and could not be resolved. Nodata were obtained for the NH of Lys140 in the stretchedgel medium.

1134 Dipolar Couplings for Apomyoglobin Unfolded States

proteins) arise from the intrinsic properties of theunfolded polypeptide chain. We base thisinterpretation on the following logic. Firstly, chemi-cal shifts, 15N relaxation data, and spin labelingexperiments provide incontrovertible evidencethat the unfolded state of apoMb in 8 M urea atpH 2.3 is highly flexible and is devoid of long-range order.12,25 Secondly, there are no significantchanges in the 1H-15N HSQC spectra of the proteinin the presence or absence of the polyacrylamidegel (Figure 1). The cross-peaks are neither shiftednor broadened and there are no significant changesin 15N R1 or R2 relaxation rates, arguing that the gelmedium does not change the structure ordynamics of the conformational ensemble in anyway. Thus, the conformational propensities andbackbone dynamics previously mapped for ureaand acid-denatured states of apomyoglobin in freesolution10,12 should be applicable to the strainedpolyacrylamide gels. Thirdly, RDCs obtained inthe presence and absence of urea (Figures 3 and 4)can be interpreted in a straightforward manner onthe basis of the structural propensities anddynamics determined previously for unfoldedstates of the protein in isotropic solution.The structure and dynamics of the conformation-

al ensemble formed under these denaturing con-ditions have been investigated in great detail inour laboratory by analysis of chemical shifts,NOEs, backbone 15N spin relaxation rates, and theresonance broadening effects of paramagnetic spinlabels.9,10,12,25 These data show that the polypeptidechain is highly flexible in both the urea and acid-denatured states, with large amplitude backbonefluctuations on picosecond and nanosecond time-scales. However, the chain does not behave as afree-flight random coil in either denaturant. In theacid-unfolded protein at pH 2.3, chemical shiftmeasurements indicate a small but significantpropensity for helical structure in the regions cor-responding to the A, D/E and H helices in thefolded protein, which is abolished by addition of8 M urea.9,10,12 In the urea-unfolded state,deviations of 13Ca, 13CO, and 1Ha chemical shiftsfrom random coil values, obtained under identicalconditions and corrected for local sequence

Figure 4. 15N-1H residual dipolar couplings as a func-tion of residue number for acid unfolded apomyoglobinat pH 2.3. A, Compressed polyacrylamide gel;B, stretched polyacrylamide gel; C, correlation betweenresidual dipolar couplings measured in stretched andcompressed gels. Due to overlap or broadness of theresonances, no data could be obtained for residues 8, 9,10, 12, 14, 15, 16, 63, 78, 101, 119 and 148 in the com-pressed gel and 8, 10, 12, 14, 15, 78, 101 and 148 in thestretched gel.

Figure 5. Scatter plots of 15N-1Hresidual dipolar couplings for myo-globin in the native and denaturedstates. A, Plot of DNH values forholomyoglobin versus DNH ofapomyoglobin in 8 M urea atpH 2.3. B, Plot of DNH values forholomyoglobin versus DNH of acid-denatured apomyoglobin at pH 2.3.The residual dipolar couplings forholomyoglobin at pH 5.6 and 35 8C,were measured in compressed 8%polyacrylamide gels. The residualdipolar couplings shown for thedenatured apoprotein are also forcompressed gels.

Dipolar Couplings for Apomyoglobin Unfolded States 1135

effects,41,42 reveal that most regions of the poly-peptide have a slight preference for extendedbackbone conformations, i.e. there is a higher prob-ability that backbone dihedral angles fall in thebroad minimum encompassing the b and poly-proline II conformations rather than the a-regionof f,c space.

It is striking that all of the RDC values for apo-myoglobin in 8 M urea at pH 2.3 have the samesign, that is, in the compressed gel all of the RDCsare positive (Figure 3A), while in the stretched gelnearly all of the DNH values are negative(Figure 3B). For the acid-unfolded protein, most,but not all, of the DNH values in the compressedgel are positive (Figure 4A), while in the stretchedgel the majority are negative (Figure 4B); theregions of the polypeptide that are “fullyunfolded” give rise to RDCs of a particular sign,positive in compressed gels and negative instretched gels.

RDC in unfolded states reflects local not long-range structure

A consideration of these results in terms of theproperties of polymers rather than those of foldedproteins provides an explanation not only of theamplitudes of the RDCs but also of their signs andthe changes of sign that occur when a propensityfor helix is present. It has been predicted theoreti-cally and demonstrated experimentally that theoverall shape of a random-walk polymer is notspherically symmetrical; rather, there is a propen-sity for the chain to adopt extended structureswith highly anisotropic shape.43,44 Real poly-peptides do not behave as ideal random coils inwhich the backbone dihedral angles of each aminoacid residue are independent of its neighbors; inother words, they do not obey the Flory isolatedpair hypothesis.45,46 The behavior of denaturedproteins in solution is usually described usingrotational isomeric state theory, in which the chainis treated as a polymer of jointed statistical seg-ments that are randomly oriented with respect toeach other.45,47 Due to steric restrictions on rotationsabout the backbone f and c dihedral angles, thepolypeptide chain exhibits local stiffness, which isdependent upon the amino acid composition(sequence). The behavior of the chain can also bedescribed in terms of its persistence length(approximately half the statistical segment length,for homopolypeptides47), which is a measure ofthe length over which the chain persists in thesame general direction. The persistence length fordenatured polypeptides, determined experimen-tally from small-angle X-ray scattering and NMRrelaxation measurements,48,49 is five to sevenamino acid residues. Thus, each statistical segmentcomprises several amino acid residues, with apropensity towards extended conformations, andis highly anisotropic in shape with its own align-ment tensor. We propose that residual dipolarcouplings in unstructured polypeptides originate

from transient alignment of local regions of thechain, i.e. from alignment of the statistical seg-ments. A similar explanation has recently beengiven by Louhivuori et al.,50 who described a theor-etical model for the residual dipolar couplingsbased on the local alignment of statistical elementsin random flight chains.

In unfolded polypeptides, the backbone fluctu-ates over an ensemble of conformations with fand c dihedral angles sampling the b, aR, andpolyproline II (PII) minima on the Ramachandranf,c energy surface.45,51 The experimentallyobserved residual dipolar coupling will be apopulation weighted average over all low-energybackbone conformations in the ensemble. Thedominant backbone conformation in unstructuredpeptides in water as well as in chemicallydenatured proteins is extended, with f,c anglespreferentially populating the b or PII energywells.51–56 The backbone N–H vectors of a statisti-cal segment constrained to conformations withinthe b and PII wells will on average be oriented ina direction that is approximately perpendicular tothe chain direction, i.e. perpendicular to the princi-pal axis of the local alignment tensor (Figure 6Aand B). Thus, in a stretched gel, the N–H vectors

Figure 6. Schematic diagrams showing the backboneconformations of polypeptide statistical segments (fiveto seven residues) that preferentially populate dihedralangles in the (A) b, (B) polyproline II, and (C) aR minimaon the f,c energy surface. The black arrows indicate thedirection of the principal axis of the alignment tensorfor each statistical segment.

1136 Dipolar Couplings for Apomyoglobin Unfolded States

will be preferentially oriented perpendicular to theapplied magnetic field, the average value of(3cos2u 2 1) will be , 2 1 and hence the DNH

value will be negative†. This is precisely what isobserved experimentally for urea-denatured apo-myoglobin, with almost all residues exhibitingnegative DNH values in the stretched gel(Figure 3B). Conversely, for a compressed gel, allDNH values are predicted to be positive in sign, asis observed experimentally (Figure 3A).

For an idealized random-flight polypeptide ofthe length of apomyoglobin, the DNH values shouldbe similar for all residues, except for those near theends of the chain where there is increased localflexibility and the distribution of conformationsbecomes more spherical.50 In other words, for anideal random coil, a plot of DNH versus residuenumber should form a flattened bell-shapedcurve. Clearly, urea-denatured apomyoglobindeviates from ideality, since despite the fact thatall residual dipolar couplings have the same sign,there is considerable variation in the magnitude ofDNH (Figure 3). From our statistical segmentmodel, one might expect a decrease in the magni-tude of the residual dipolar coupling in regions ofincreased flexibility of the polypeptide chain back-bone, i.e. in regions of decreased chain stiffness.Figure 7 shows a superposition of the DNH valueson the J(0) spectral densities determined from 15Nrelaxation measurements for urea-denaturedapomyoglobin.12 Three distinct minima in J(0) (atresidues 21–25, 60–80 and 121–129) are observedin regions that are rich in Gly and Ala residues,and which function as flexible molecular hinges inthe polypeptide chain. These regions correspondclosely to segments with reduced DNH values(Figure 7), showing that a local decrease in chainstiffness leads to more effective averaging of the

N–H vectors. A scatter plot of J(0) versus DNH (notshown) has a correlation coefficient R ¼ 0:58:Several contiguous segments of the apomyo-

globin chain, encompassing residues 28–42,63–68, 77–89, 95–102, and 111–120, exhibit greaterthan average RDC values. These segments tend tobe rich in b-branched (valine, isoleucine, andthreonine) and charged amino acid residues(lysine, arginine, and histidine at pH 2.3); most ofthem also contain proline (residues 37, 88, 100,120). Proline and the b-branched amino acids pre-dispose the backbone towards dihedral angles inthe b and PII minima, i.e. tend to bias the confor-mational ensemble towards more extended back-bone conformations that would be expected toalign more effectively in the strained polyacryl-amide gels. It is at first surprising to find arginineand lysine clusters associated with these regions,since both amino acids have high helical propensi-ties. However, recent evidence suggests that shortlysine homopeptides, in common with otherpeptides with ionizable side-chains, have a highpropensity to adopt PII conformations52,57 thatwould bias the chain towards extended structures.In addition to the regions of enhanced backbone

flexibility, below-average values of DNH areobserved in some other regions of urea-denaturedapomyoglobin, e.g. in the CD loop (residues43–47), and probably reflect a local propensity forchain compaction. 15N relaxation measurementsprovide evidence for motional restriction of thebackbone in the CD loop region due to local hydro-phobic cluster formation, which persists even in8 M urea.12 A strong propensity for structure for-mation in the CD loop region was also observedin a short synthetic peptide in water solution.58

Taken together with these earlier results, thereduced DNH values strongly support a model inwhich preferential chain compaction occurs forresidues 43–47 (sequence FDRFK) in the urea-denatured state. Increased compaction is alsoindicated for many residues in the segmentbetween Ala53 and Leu61; while the nature ofthe interactions that lead to preferential local

Figure 7. Effect of backbonedynamics and amino acid sequenceon residual dipolar couplings forurea-denatured apoMb (pH 2.3).The histogram shows DNH valuesfor a compressed gel plotted as afunction of residue number. Thelocation of Gly and Ala residues isindicated by green bars, residueswith b-branched side-chains (Val,Ile, Thr) by orange bars, andcharged residues (Arg, Lys, andHis only at pH 2.3) are highlightedin blue. The location of the prolineresidues in indicated. Calculatedvalues of the J(0) spectral density12

are plotted as red data points (tri-angles) and a kernel smoothedcurve.

† In an unfolded polypeptide, ensemble averagingabout the long axis of the statistical segment is expectedto result in an axially symmetric alignment tensor.

Dipolar Couplings for Apomyoglobin Unfolded States 1137

compaction in 8 M urea are unknown, it is ofinterest that this region folds spontaneously intohelical structure in both the acid-denatured proteinand in peptide fragments.10

Local conformational propensities in acid-denatured apomyoglobin

The J(0) spectral density for acid-denatured apo-myoglobin is relatively uniform in all regions ofthe polypeptide10 and therefore the DNH value isunlikely to be modulated significantly by localvariations in backbone dynamics. Rather,variations in DNH should primarily reflectdifferences in the structures that contribute to theconformational ensemble. An exception is parts ofthe A and G helix regions, where exchangebroadening leads to increased values of J(0) andmakes accurate measurement of DNH difficult forseveral residues in these regions; this is the reasonfor some of the missing data points in Figures 4and 8.

The correlation between the RDCs for acid-denatured apoMb in a stretched gel and the devi-ations of a weighted average of the 13Ca and 13COchemical shifts from sequence-corrected randomcoil values is shown in Figure 8A. DNH is negativethroughout the F helix region (residues 81–98) andthe central portion of the E helix (resides 63–69), in

accord with the indications from the secondarychemical shifts that the polypeptide backbone inthese regions preferentially adopts extended chainconformations. In contrast, positive values of DNH

are observed for contiguous residues in the A,D–E, and H helix regions, corresponding closelyto regions of the protein where secondary chemicalshifts indicate 15–20% population of helical struc-tures in rapid dynamic equilibrium with unfoldedstates.10 The sign change can readily be rational-ized utilizing the notion of statistical segmentsand calculating the effect of a local propensity forhelix. For a helical segment, the principal axis ofthe alignment tensor lies along the axis of thehelix and the N–H vectors are oriented parallelwith the axis (Figure 6C). Therefore, in a stretchedgel, the N–H bond vectors will tend to align withthe applied magnetic field (u ¼ 0) so that the aver-age value of (3cos2u 2 1) will be ,2 and DNH willbe positive, in agreement with experimental obser-vations (Figures 4B and 8). In accord with our apo-myoglobin data, a change in the sign of DNH inlocal regions of the polypeptide that exhibit apropensity for helix has recently been observed inthe acid-denatured state of ACBP.59 Further,inspection of the RDC data for the D131D fragmentof staphylococcal nuclease in aqueous solution(data from Supplementary Material from Shortle& Ackerman26) reveals a sign change in regions

Figure 8. Effect of local confor-mational propensities on residualdipolar couplings for acid-denatured apoMb (pH 2.3). A, Plotof DNH values for a stretched gelversus residue number (black histo-gram). Smoothed average devi-ations of 13Ca and 13CO chemicalshifts (Dav(C

aC0)) from sequence-corrected random coil values areplotted as a continuous red line.10

The average shifts were calculatedas (3DdCa þ 4DdC0)/7, weightedaccording to the secondary shiftscorresponding to fully formedhelix, i.e. 2.8 ppm for 13Ca and2.1 ppm for 13C0 (Yao et al.,10 Wishart& Sykes67). Regions of helical pro-pensity are indicated by positivevalues of Dav(C

aC0) extending overseveral residues. The black barsindicate the location of the eighthelices in the structure of nativeholomyoglobin. B, Plot of differencein residual dipolar coupling (DDNH)between acid-denatured and urea-denatured states, superimposed onDav(C

aC0) (red line). The differenceRDC is calculated from thestretched gel data of Figures 3Band 4B as DDNH ¼ DNH

acid 2 DNHurea.

1138 Dipolar Couplings for Apomyoglobin Unfolded States

where chemical shifts and NOE connectivities indi-cate formation of helical structure.60

Additional insights into the nature of the confor-mational ensemble of the pH 2.3 state can beobtained by calculating the difference in theresidual dipolar couplings (DDNH) between theurea-denatured and acid-denatured states(Figure 8B). This shows more clearly the decreasein RDC that occurs for much of the polypeptideon going from 8 M urea at pH 2.3 to the acid-denatured form at the same pH. Many of theregions of large positive DDNH values coincidewith regions of the polypeptide where the 13Ca

and 13CO chemical shifts indicate significant helicalpropensity (Figure 8B). However, the increase inDDNH between residues 26–50 and between resi-dues 103–118 does not appear to be associatedwith helix formation, but correlates with regionsof the polypeptide chain in which relaxationmeasurements indicate restriction of backbonemotions due to local hydrophobic collapse.10

These regions encompass much of the B helix, theCD loop, and the G helix of the native foldedprotein.

The population of helix in the acid-denaturedprotein can be estimated from the magnitude ofthe residual dipolar couplings. In the helical seg-ments, the measured DNH for each residue is apopulation weighted average between the valuefor a fully formed helix and the RDC in theunfolded state. To calibrate the change in RDC forthe transition from a fully unfolded state to a fullyformed helix, we assume that the average RDCs inthe F helix region of the acid-denatured apomyo-globin (kDNH

extendedl ¼ 26.8 and þ5.0 Hz for residues81–98 in the stretched and compressed gels,respectively) represent the limit of DNH for theunfolded (extended chain) conformation. This rep-resents the expected DNH value for u ¼ 908, i.e. for(3cos2u 2 1) ¼ 21. For a fully formed helix, theN–H vectors will be aligned along the principalaxis of the helix (u ¼ 08) and (3cos2u 2 1) ¼ þ2;therefore, the difference in DNH value (DDNH) corre-sponding to formation of 100% population of helixis 3lkDNH

extendedll. The population of helix in the seg-ment between residues i and j is then given by therelationship:

P ¼1

N

Xj

i

ðDi;acidNH 2Di;urea

NH Þ=3kDextendedNH l ð2Þ

where N is the number of residues in the helicalsegment, and the DNH for the unfolded state foreach residue is taken as that for apomyoglobin in8 M urea at pH 2.3†.

Using equation (2), we calculate the following

helical populations from the RDC data for thestretched (compressed) gels, respectively: A helixregion (residues 4–11), 17% (11%); D–E helixregion (residues 52–61), 22% (22%); H helix region(residues 127–147), 21% (22%). These populationsare in striking agreement with those calculatedfrom secondary 13Ca, 13CO, and 1Ha chemical shifts,namely 15%, 22%, and 20% helical populations forthe same segments of the A, D–E, and H helixregions.10 The secondary chemical shifts alsosuggest a slight propensity to populate local helicalstructures between residues 72–75 towards theC-terminal end of the E helix of the folded protein(Figure 8B); DDNH is consistent with this interpret-ation, and the helical population is estimated fromthe RDCs to be 11% between residues 70–77 in theacid-denatured protein. The remarkably goodagreement between the populations of helixobtained from chemical shifts and estimated fromthe residual dipolar couplings provides strong evi-dence that our model for alignment of unfoldedpeptide chains is an appropriate one.

Residual dipolar couplings in unfolded proteins

Residual dipolar couplings for unfolded poly-peptides arise through an intrinsically differentmechanism than those for folded globular proteins.For folded proteins, DNH for each amide is deter-mined by the relative orientation of the 1H-15Nbond vector to the alignment tensor of the entiremolecule and DNH therefore depends upon boththe three-dimensional structure of the protein andits overall hydrodynamic shape. In contrast, align-ment in unfolded proteins is a property of thepolypeptide chain and its constitutive statisticalsegments, rather than the molecule as a whole.The alignment tensor is determined by the shapeanisotropy of the statistical segments, and themeasured residual dipolar couplings at individualresidues reflect local conformational propensitieswithin the statistical segments. All of the resultsreported thus far on unfolded proteins, includingthose on urea-denatured staphylococcal nuclease,eglin C, ACBP, and short peptides,26,33–35,40,59 areconsistent with this interpretation. For example,the RDCs for the D131D fragment of staphylococcalnuclease in 8 M urea all have the same sign (theDNH values are all positive for a compressed gel,once the published values26 are corrected for thenegative gyromagnetic ratio of 15N), consistentwith alignment of statistical segments with f,cangles preferentially in the b and PII regions.There is no need to invoke the presence of native-like structure to explain the observed residualdipolar couplings for denatured states of staphylo-coccal nuclease, apomyoglobin, or any otherproteins for which data have been reported.Instead of providing information on the overallstructure, as is the case for a folded globularprotein, RDC measurements for unfolded states ofproteins give very different, but equally useful,information on the composition of the confor-

†Because of differences in aspect ratio of the stretchedgels, the RDC for the urea-denatured protein was scaledso that kDurea

NH l ¼ kDacidNH l averaged over the extended

chain region, residues 80–98; scaling was unnecessaryfor data obtained from the compressed gels.

Dipolar Couplings for Apomyoglobin Unfolded States 1139

mational ensemble. In combination with chemicalshift and spin relaxation data, residual dipolarcouplings provide important insights into second-ary structural propensities and chain compactionin local regions of the unfolded polypeptide chainand therefore hold great promise for mapping theupper reaches of the protein folding landscape.

Methods

Sample preparation

Protein labeled uniformly with 15N or with 15N/13Cwas expressed in Escherichia coli strain BL21-DE3 usinga slight modification of published methods.10 Theinclusion body pellet was solubilized in 0.1% (v/v)trifluoroacetic acid and acetonitrile was added to 10%(v/v). After HPLC purification in C4-reverse phasecolumn, the protein was lyophilized. For samplepreparation, the protein was resuspended in 10 mMsodium acetate (pH 6.1) and incubated on ice for twohours. The pH was adjusted back to pH 6.1 to facilitatefull protein solubilization. The sample was then spundown for ten minutes at 13,000 g before buffer exchangeon a HiTrap desalting column (Amersham-PharmaciaBiotech) to the final desired pH (10 mM sodium acetateand 10% 2H2O were present in all pH values. Approxi-mately 5 mM HCl was added to obtain pH 2.3 and thebuffer containing 8 M urea was prepared according to apublished method.12 Samples were concentrated (Centri-prep 10) to a final concentration of 200 mM, determinedby UV absorbance in 6.0 GdmCl (20 mM phosphate(pH 6.5)) using an extinction coefficient of1280 ¼ 15,200 M21 cm21 (Edelhoch61).

Polyacrylamide gel preparation and sample casting

Polyacrylamide gels were prepared according to theprotocol described by Sass et al.,31 from a stock solutioncontaining 29.2% (w/v) acrylamide and 0.78% (w/v)N,N 0-methylenebisacrylamide. Gels were cast by placingthe appropriate volume of acrylamide/bisacrylamidestock solution for the desired gel concentration in aplastic tube immediately following initiation ofpolymerization by addition of 0.1% (w/v) ammoniumpersulfate and 0.04% (v/v) N,N,N 0,N0-tetramethylene-diamine (TEMED). The internal diameter (ID) of thetube was 4 mm for compressed and 5 mm for stretchedgels. After polymerization was complete, the gel wasextruded from the tube, cut into lengths of 35 mm (forcompressed gels) and 25 mm (for stretched gels), andwashed with water for several days to remove unreactedreagents. The gel was then dried in an uncovered Petridish at 37 8C for several hours until its diameter shrankto 1–2 mm. To prepare a compressed gel, the dried gelwas placed into a Shigemi tube and the protein solutionat the desired pH value was added; the plunger washeld in position with Parafilm to give 20% vertical com-pression after swelling. Stretched gels were preparedusing the funnel apparatus developed by Chou et al.38

The dried gel was placed in a funnel with the same ID(5 mm) as that used to cast the original gel and was rehy-drated in the protein solution to its initial size(5 mm £ 25 mm). The swollen gel was then transferredto an open bottom NMR tube (New Era Enterprises;4.2 mm ID), which was sealed with a Vespel suscepti-

bility matched plug fitted with a Viton O-ring. In bothcases, the gels were swollen overnight.

NMR spectroscopy

NMR experiments were carried out on a Bruker DRX800 MHz spectrometer with a triple resonance gradientprobe head. For native apoMb, dipolar couplings wererecorded at 35 8C using a 2D IPAP-type 15N-1H HSQCcorrelation experiment,62 with 2048 £ 512 complex pointsand spectral widths of 12019.2 and 2272.7 Hz for the 1Hand 15N dimensions, respectively. For the acid-unfoldedstates, 15N-1H one-bond dipolar couplings weremeasured in 15N,13C-labeled protein using a 3D HNCO-IPAP pulse program37 because of poor 1H resonance dis-persion. The spectra of apoMb at pH 2.3 and with 8 Murea at pH 2.3 were recorded at 25 8C and 20 8C with2048 £ 96 £ 100 complex points and the spectral widthsof 12820.5 Hz, 1404.5 Hz and 2127.6 Hz for 1H, 13C and15N dimensions, respectively. The 1H, 15N and 13C carrierswere set at 4.7 ppm, 118.0 ppm and 173.9 ppm, respect-ively, for all experiments. An external DSS sample in thecorresponding buffer solution was used to determinereference frequency for all nuclei.63

Data analysis

NMRPipe software64 was used to process all data sets.The spectra were apodized with a 548 shifted sine squarewindow function in all dimensions. Time domain data inthe 15N and 13C dimensions were linear predicted toincrease the number of data points by half and zero-filledeither to double the original data size (for the nativestate) or to 256 points (for unfolded and partially foldedstates) before Fourier transformation. The NMRview 5program65 was used to confirm the published assign-ments of backbone chemical shifts for acid-unfolded10

and urea-unfolded12 apoMb. The PIPP/CAPP suite ofprograms66 was used to measure the dipolar couplingsplittings.

Acknowledgements

We thank Drs Michael Lietzow, Roberto DeGuzman, Raphael Stoll and John Chung forstimulating discussions, and Linda Tennant forexpert technical help. R.M.B. is a Pew LatinAmerican Fellow. This project was supported bygrant DK34909 from the National Institutes ofHealth.

References

1. Bryngelson, J. D., Onuchic, J. N., Socci, N. D. &Wolynes, P. G. (1995). Funnels, pathways, and theenergy landscape of protein folding: a synthesis.Proteins: Struct. Funct. Genet. 21, 167–195.

2. Wolynes, P. G., Onuchic, J. N. & Thirumalai, D.(1995). Navigating the folding routes. Science, 267,1619–1620.

3. Wright, P. E. & Dyson, H. J. (1999). Intrinsicallyunstructured proteins: re-assessing the proteinstructure-function paradigm. J. Mol. Biol. 293,321–331.

1140 Dipolar Couplings for Apomyoglobin Unfolded States

4. Dyson, H. J. & Wright, P. E. (2002). Coupling of fold-ing and binding for unstructured proteins. Curr.Opin. Struct. Biol. 12, 54–60.

5. Tompa, P. (2002). Intrinsically unstructured proteins.Trends Biochem. Sci. 27, 527.

6. Iakoucheva, L. M., Brown, C. J., Lawson, J. D.,Obradovic, Z. & Dunker, A. K. (2002). Intrinsicdisorder in cell-signaling and cancer-associatedproteins. J. Mol. Biol. 323, 573–584.

7. Kelly, J. W. (2002). Towards an understanding ofamyloidogenesis. Nature Struct. Biol. 9, 323–325.

8. Dyson, H. J. & Wright, P. E. (1996). Insights intoprotein folding from NMR. Annu. Rev. Phys. Chem.47, 369–395.

9. Eliezer, D., Yao, J., Dyson, H. J. & Wright, P. E. (1998).Structural and dynamic characterization of partiallyfolded states of myoglobin and implications forprotein folding. Nature Struct. Biol. 5, 148–155.

10. Yao, J., Chung, J., Eliezer, D., Wright, P. E. & Dyson,H. J. (2001). NMR structural and dynamic character-ization of the acid-unfolded state of apomyoglobinprovides insights into the early events in proteinfolding. Biochemistry, 40, 3561–3571.

11. Gulotta, M., Gilmanshin, R., Buscher, T. C.,Callender, R. H. & Dyer, R. B. (2001). Core formationin apomyoglobin: probing the upper reaches of thefolding energy landscape. Biochemistry, 40,5137–5143.

12. Schwarzinger, S., Wright, P. E. & Dyson, H. J. (2002).Molecular hinges in protein folding: the urea-denatured state of apomyoglobin. Biochemistry, 41,12681–12686.

13. Tsui, V., Garcia, C., Cavagnero, S., Siuzdak, G.,Dyson, H. J. & Wright, P. E. (1999). Quench-flowexperiments combined with mass spectrometryshow apomyoglobin folds through an obligatoryintermediate. Protein Sci. 8, 45–49.

14. Jamin, M. & Baldwin, R. L. (1998). Two forms of thepH 4 folding intermediate of apomyoglobin. J. Mol.Biol. 276, 491–504.

15. Jennings, P. A., Dyson, H. J. & Wright, P. E. (1994).The folding pathway of apomyoglobin. In StatisticalMechanics, Protein-Structure and Protein-SubstrateInteractions (Doniach, S., ed.), pp. 7–18, PlenumPress, New York.

16. Eliezer, D., Jennings, P. A., Wright, P. E., Doniach, S.,Hodgson, K. O. & Tsuruta, H. (1995). The radius ofgyration of an apomyoglobin folding intermediate.Science, 270, 487–488.

17. Eliezer, D. & Wright, P. E. (1996). Is apomyoglobin amolten globule? Structural characterization byNMR. J. Mol. Biol. 263, 531–538.

18. Lecomte, J. T., Sukits, S. F., Bhattacharjya, S. &Falzone, C. J. (1999). Conformational properties ofnative sperm whale apomyoglobin in solution.Protein Sci. 8, 1484–1491.

19. Griko, Y. V., Privalov, P. L., Venyaminov, S. Y. &Kutyshenko, V. P. (1988). Thermodynamic study ofthe apomyoglobin structure. J. Mol. Biol. 202,127–138.

20. Griko, Y. V. & Privalov, P. L. (1994). Thermodynamicpuzzle of apomyoglobin unfolding. J. Mol. Biol. 235,1318–1325.

21. Dyson, H. J. & Wright, P. E. (2002). Insights into thestructure and dynamics of unfolded proteins fromnuclear magnetic resonance. Advan. Protein Chem.62, 311–340.

22. Gillespie, J. R. & Shortle, D. (1997). Characterizationof long-range structure in the denatured state of

staphylococcal nuclease. I. Paramagnetic relaxationenhancement by nitroxide spin labels. J. Mol. Biol.268, 158–169.

23. Yi, Q., Scalley-Kim, M. L., Alm, E. J. & Baker, D.(2000). NMR characterization of residual structurein the denatured state of protein L. J. Mol. Biol. 299,1341–1351.

24. Teilum, K., Kragelund, B. B. & Poulsen, F. M. (2002).Transient structure formation in unfolded acyl-coenzyme A-binding protein observed by site-directed spin labelling. J. Mol. Biol. 324, 349–357.

25. Lietzow, M. A., Jamin, M., Dyson, H. J. & Wright, P. E.(2002). Mapping long-range contacts in a highlyunfolded protein. J. Mol. Biol. 322, 655–662.

26. Shortle, D. & Ackerman, M. S. (2001). Persistence ofnative-like topology in a denatured protein in 8 Murea. Science, 293, 487–489.

27. Bax, A. (2003). Weak alignment offers new NMRopportunities to study protein structure anddynamics. Protein Sci. 12, 1–16.

28. Prestegard, J. H., Al Hashimi, H. M. & Tolman, J. R.(2000). NMR structures of biomolecules using fieldoriented media and residual dipolar couplings.Quart. Rev. Biophys. 33, 371–424.

29. Tjandra, N. & Bax, A. (1997). Direct measurement ofdistances and angles in biomolecules by NMR in adilute liquid crystalline medium. Science, 278,1111–1114.

30. Hansen, M. R., Mueller, L. & Pardi, A. (1998).Tunable alignment of macromolecules by filamen-tous phage yields dipolar coupling interactions.Nature Struct. Biol. 5, 1065–1074.

31. Sass, H. J., Musco, G., Stahl, S. J., Wingfield, P. T. &Grzesiek, S. (2000). Solution NMR of proteins withinpolyacrylamide gels: diffusional properties andresidual alignment by mechanical stress orembedding of oriented purple membranes. J. Biomol.NMR, 18, 303–309.

32. Tycko, R., Blanco, F. J. & Ishii, Y. (2000). Alignment ofbiopolymers in strained gels: a new way to createdetectable dipole–dipole couplings in high-resolution biomolecular NMR. J. Am. Chem. Soc. 122,9340–9341.

33. Ackerman, M. S. & Shortle, D. (2002). Molecularalignment of denatured states of staphylococcalnuclease with strained polyacrylamide gels and sur-factant liquid crystalline phases. Biochemistry, 41,3089–3095.

34. Ackerman, M. S. & Shortle, D. (2002). Robustness ofthe long-range structure in denatured staphylococcalnuclease to changes in amino acid sequence.Biochemistry, 41, 13791–13797.

35. Ohnishi, S. & Shortle, D. (2003). Observation ofresidual dipolar couplings in short peptides. Proteins:Struct. Funct. Genet. 50, 546–551.

36. Ottiger, M., Delaglio, F. & Bax, A. (1998). Measure-ment of J and dipolar couplings from simplifiedtwo-dimensional NMR spectra. J. Magn. Reson. 131,373–378.

37. Goto, N. K., Skrynnikov, N. R., Dahlquist, F. W. &Kay, L. E. (2001). What is the average conformationof bacteriophage T4 lysozyme in solution? A domainorientation study using dipolar couplings measuredby solution NMR. J. Mol. Biol. 308, 745–764.

38. Chou, J. J., Gaemers, S., Howder, B., Louis, J. M. &Bax, A. (2001). A simple apparatus for generatingstretched polyacrylamide gels, yielding uniformalignment of proteins and detergent micelles.J. Biomol. NMR, 21, 377–382.

Dipolar Couplings for Apomyoglobin Unfolded States 1141

39. Alexandrescu, A. T. & Kammerer, R. A. (2003). Struc-ture and disorder in the ribonuclease S-peptideprobed by NMR residual dipolar couplings. ProteinSci. 12, 2132–2140.

40. Ohnishi, S., Lee, A. L., Edgell, M. H. & Shortle, D.(2004). Direct demonstration of structural similaritybetween native and denatured eglin c. Biochemistry,43, 4064–4070.

41. Schwarzinger, S., Kroon, G. J. A., Foss, T. R., Wright,P. E. & Dyson, H. J. (2000). Random coil chemicalshifts in acidic 8 M urea: implementation of randomcoil chemical shift data in NMRView. J. Biomol.NMR, 18, 43–48.

42. Schwarzinger, S., Kroon, G. J. A., Foss, T. R., Chung,J., Wright, P. E. & Dyson, H. J. (2001). Sequencedependent correction of random coil NMR chemicalshifts. J. Am. Chem. Soc. 123, 2970–2978.

43. Kuhn, W. (1934). The shape of fibrous molecules insolution. Kolloid Z. 68, 2–15.

44. Haber, C., Ruiz, S. A. & Wirtz, D. (2000). Shape aniso-tropy of a single random-walk polymer. Proc. NatlAcad. Sci. USA, 97, 10792–10795.

45. Flory, P. J. (1969). Statistical Mechanics of ChainMolecules, Wiley, New York.

46. Pappu, R. V., Srinivasan, R. & Rose, G. D. (2000). TheFlory isolated-pair hypothesis is not valid for poly-peptide chains: implications for protein folding.Proc. Natl Acad. Sci. USA, 97, 12565–12570.

47. Cantor, C. R. & Schimmel, P. R. (1980). BiophysicalChemistry, W.H. Freeman and Company, New York.

48. Damaschun, G., Damaschun, H., Gast, K.,Misselwitz, R., Muller, J. J., Pfeil, W. & Zirwer, D.(1993). Cold denaturation-induced conformationalchanges in phosphoglycerate kinase from yeast.Biochemistry, 32, 7739–7746.

49. Schwalbe, H., Fiebig, K. M., Buck, M., Jones, J. A.,Grimshaw, S. B., Spencer, A. et al. (1997). Structuraland dynamical properties of a denatured protein.Heteronuclear 3D NMR experiments and theoreticalsimulations of lysozyme in 8 M urea. Biochemistry,36, 8977–8991.

50. Louhivuori, M., Paakkonen, K., Fredriksson, K.,Permi, P., Lounila, J. & Annila, A. (2003). On theorigin of residual dipolar couplings from denaturedproteins. J. Am. Chem. Soc. 125, 15647–15650.

51. Dyson, H. J. & Wright, P. E. (1991). Defining solutionconformations of small linear peptides. Annu. Rev.Biophys. Biophys. Chem. 20, 519–538.

52. Eker, F., Griebenow, K., Cao, X., Nafie, L. A. &Schweitzer-Stenner, R. (2004). Tripeptides with ioniz-able side chains adopt a perturbed polyproline IIstructure in water. Biochemistry, 43, 613–621.

53. Zhang, O. & Forman-Kay, J. D. (1995). Structuralcharacterization of folded and unfolded states of anSH3 domain in equilibrium in aqueous buffer.Biochemistry, 34, 6784–6794.

54. Meekhof, A. E. & Freund, S. M. V. (1999). Probingresidual structure and backbone dynamics on themilli- to picosecond timescale in a urea-denaturedfibronectin type III domain. J. Mol. Biol. 286, 579–592.

55. Frank, M. K., Clore, G. M. & Gronenborn, A. M.(1995). Structural and dynamic characterization ofthe urea denatured state of the immunoglobulinbinding domain of streptococcal protein G by multi-dimensional heteronuclear NMR spectroscopy.Protein Sci. 4, 2605–2615.

56. Shi, Z. S., Woody, R. W. & Kallenbach, N. R. (2002). Ispolyproline II a major backbone conformation inunfolded proteins? Unfolded Proteins, 62, 163–240.

57. Rucker, A. L. & Creamer, T. P. (2002). Polyproline IIhelical structure in protein unfolded states: lysinepeptides revisited. Protein Sci. 11, 980–985.

58. Reymond, M. T., Merutka, G., Dyson, H. J. & Wright,P. E. (1997). Folding propensities of peptide frag-ments of myoglobin. Protein Sci. 6, 706–716.

59. Fieber, W., Kristjansdottir, S. & Poulsen, F. M. (2004).Short-range, long-range and transition-state inter-actions in the denatured state of ACBP from residualdipolar couplings. J. Mol. Biol. 339, 1191–1199.

60. Alexandrescu, A. T., Abeygunawardana, C. &Shortle, D. (1994). Structure and dynamics of adenatured 131-residue fragment of staphylococcalnuclease: a heteronuclear NMR study. Biochemistry,33, 1063–1072.

61. Edelhoch, H. (1967). Spectroscopic determination oftryptophan and tyrosine in proteins. Biochemistry, 6,1948–1954.

62. Ishii, Y., Markus, M. A. & Tycko, R. (2001). Control-ling residual dipolar couplings in high-resolutionNMR of proteins by strain induced alignment in agel. J. Biomol. NMR, 21, 141–151.

63. Wishart, D. S., Bigam, C. G., Yao, J., Abildgaard, F.,Dyson, H. J., Oldfield, E. et al. (1995). 1H, 13C and15N chemical shift referencing in biomolecular.NMR. J. Biomol. NMR, 6, 135–140.

64. Delaglio, F., Grzesiek, S., Vuister, G. W., Guang, Z.,Pfeifer, J. & Bax, A. (1995). NMRPipe: a multi-dimensional spectral processing system based onUNIX pipes. J. Biomol. NMR, 6, 277–293.

65. Johnson, B. A. & Blevins, R. A. (1994). NMRView: acomputer program for the visualization and analysisof NMR data. J. Biomol. NMR, 4, 604–613.

66. Garrett, D. S., Powers, R., Gronenborn, A. M. &Clore, G. M. (1991). A common sense approach topeak picking in two-,three, and four-dimensionalspectra using automatic computer analysis ofcontour diagrams. J. Magn. Reson. 95, 214–220.

67. Wishart, D. S. & Sykes, B. D. (1994). Chemical shiftsas a tool for structure determination. MethodsEnzymol. 239, 363–392.

Edited by M. F. Summers

(Received 1 April 2004; received in revised form 19 May 2004; accepted 25 May 2004)

1142 Dipolar Couplings for Apomyoglobin Unfolded States