doi:10.1016/j.jmb.2006.12.005 J. Mol. Biol. (2007) 366, 1624–1638

A Tightly Packed Hydrophobic Cluster Directs theFormation of an Off-pathway Sub-millisecond FoldingIntermediate in the α Subunit of Tryptophan Synthase,a TIM Barrel Protein

Ying Wu†, Ramakrishna Vadrevu†, Sagar Kathuria, Xiaoyan Yangand C. Robert Matthews⁎

Department of Biochemistryand Molecular Pharmacology,University of MassachusettsMedical School, Worcester,MA 01605, USA

† Y. W. and R. V. contributed equAbbreviations used: ACFILMPVY

tyrosine residues; αTS, α subunit of tequilibrium intermediate of αTS popkinetic intermediate of αTS populatresidues; N, native state; U, unfoldeE-mail address of the correspondi

0022-2836/$ - see front matter © 2006 E

Protein misfolding is now recognized as playing a crucial role in bothnormal and pathogenic folding reactions. An interesting example ofmisfolding at the earliest state of a natural folding reaction is provided bythe α-subunit of tryptophan synthase, a (β/α)8 TIM barrel protein. Themolecular basis for the formation of this off-pathway misfolded interme-diate, IBP, and a subsequent on-pathway intermediate, I1, was probed bymutational analysis of 20 branched aliphatic side-chains distributedthroughout the sequence. The elimination of IBP and the substantialdestabilization of I1 by replacement of a selective set of the isoleucine,leucine or valine residues (ILV) with alanine in a large ILV cluster external-to-the-barrel and spanning the N and C termini (cluster 2) implies tight-packing at most sites in both intermediates. Differential effects on IBP and I1for replacements in α3, β4 and α8 at the boundaries of cluster 2 suggest thattheir incorporation into I1 but not IBP reflects non-native folds at the edges ofthe crucial (β/α)1–2β3 core in IBP. The retention of IBP and the smaller andconsistent destabilization of both IBP and I1 by similar replacements in aninternal-to-the-barrel ILV cluster (cluster 1) and a second external-to-the-barrel ILV cluster (cluster 3) imply molten globule-like packing. The tightpacking inferred, in part, for IBP or for all of I1 in cluster 2, but not in clusters1 and 3, may reflect the larger size of cluster 2 and/or the enhanced numberof isoleucine, leucine and valine self-contacts in and between contiguouselements of secondary structure. Tightly packed ILV-dominated hydropho-bic clusters could serve as an important driving force for the earliest eventsin the folding and misfolding of the TIM barrel and other members of the(β/α)n class of proteins.

© 2006 Elsevier Ltd. All rights reserved.

Keywords: hydrophobic cluster; off-pathway intermediate; sub-millisecondreaction; TIM barrel; misfolding

*Corresponding author

ally to this work., alanine, cysteine, phenylalanine, isoleucine, leucine, methionine, proline, valine andryptophan synthase from E. coli; CD, circular dichroism; HX, hydrogen exchange; I1,ulated at∼3M urea; I2, equilibrium intermediate of αTS populated at∼5M urea; IBP,ed within the stopped-flow burst-phase (<5 ms); ILV, isoleucine, leucine and valined state; WT, wild type.ng author: c.robert.matthews@umassmed.edu

lsevier Ltd. All rights reserved.

Scheme I.

1625Hydrophobic Clusters and Misfolding in αTS

Introduction

The folding reactions of small proteins and proteindomains (<∼100 amino acid residues) are often welldescribed by a simple exponential response thoughtto reflect the conformational diffusion of theunfolded polypeptide across an energy barrierseparating the native and unfolded manifolds ofstructures.1,2 The folding of these proteins is limitedonly by the intrinsic features of the energy surfacedefined primarily by the chain topology3 andmodulated by the amino acid sequence.4,5

By contrast, the folding reactions of largerproteins often display complex kinetics indicativeof a significant role for alternative conformationalstates.1,2,6–9 These “frustrated” folding reactions10

have been ascribed to a variety of sources,including slowly interconverting cis/trans prolylpeptide bond isomers in the unfolded state,11–14

mis-ligated side-chain heme complexes in early fol-ding intermediates for cytochrome c,15 on-pathway,stable intermediates,9,16–18 alternative partiallyfolded states in parallel pathways,19,20 maturationof multimeric intermediates that are pre-cursors fornative multimers,21–23 mandatory non-native con-formers,24–28 misfolded off-pathway interme-diates,29–33 and the sub-millisecond formation ofmarginally stable conformers that are rich insecondary structure.34

The case ofmisfolded off-pathway intermediates isparticularly interesting because, contrary to theexpectations of an optimized energy landscape, theprotein has to unfold these partially structured statesto some extent before proceeding to the nativeconformation. In the case of chemotaxis protein Y(CheY), a β/α sheet protein, the off-pathwayintermediate is thought to reflect the rapid formationof the complete complement of α helical segments.

Figure 1. Ribbon diagrams of αTS (a) showing the elemehydrophobic clusters formed by the I, L and V residues: crespectively, obtained using a 4.2 Å cutoff distance. L7 and L11clusters, are not shown. Coordinates of the α subunit of trypsequence identity with E. coli αTS, were used to generate thPyMOL. Out of 63 ILV residues, 57 are conserved in E. coli; thV52I (β2), I148V (loop before β5), V166I (α5), I197V (α6), V22

Go-model simulations35 and ϕ analysis36 suggestthat the C-terminal helices must unfold to access therate-limiting transition state defined by the N-terminal strands and helices. Apo-flavodoxin, an-other member of this class, also misfolds to an off-pathway intermediate before accessing the nativeconformation; the structural basis for this misfoldingreaction is not known.30,31

The α-subunit of tryptophan synthase (αTS)(Figure 1(a)), a member of the TIM barrel, (β/α)8,sub-class of the (β/α)n fold family, is known to foldto an off-pathway intermediate, IBP, in the sub-millisecond time-scale.18 A simplified folding mech-anism, ignoring a set of four parallel pathways thatcorrespond to several cis/trans prolyl peptidebond isomerizations,13,37,38 is shown in Scheme I:

U/I2 represents a pair of rapidly interconverting,thermodynamically distinct unfolded, U, or un-folded-like, I2, states that are highly populatedunder denaturing conditions. IBP represents atransient misfolded intermediate with substantialsecondary structure and appreciable stability.18,32

I1 is a stable intermediate whose conversion to thenative state, N, is the rate-limiting step in folding.The mandatory unfolding of IBP under stronglyrefolding conditions is an early rate-limiting anddistinguishing feature of an off-pathway kineticintermediate.

nts of secondary structure and (b) highlighting the threeluster 1 (green), cluster 2 (blue) and cluster 3 (orange),, as well as V214 and V220, which form two small isolatedtophan synthase from S. typhimurium, which shares 85%e Figure from a refined version of PDB file 1BKS62 usinge six differences primarily involve I/V switches located at4I (α7) and L245I (loop between α8′ and α8).

‡http://pymol.sourceforge.net/

Figure 2. Far-UV CD spectra of αTS variants grouped by replacements in elements of secondary structure: (a) β1 andβ2, (b) β3 and β4 (c) β5 to β8 and (d) α0, α1, α3 and α8. The far-UV CD spectrum for the wild-type protein is shown ineach panel with a black line, for reference. The gray dash-dot-dot and broken guidelines indicate a 15% and 40% decreasein signal at 222 nm, relative to wild-type αTS, respectively. The protein concentration ranged from 5–7 μM. Bufferconditions: 10 mM potassium phosphate (pH 7.8), 0.2 mMK2EDTA, and 1 mM β-mercaptoethanol. Data were collected at25 °C.

1626 Hydrophobic Clusters and Misfolding in αTS

In a previous study on αTS,32 it was shown thatalanine replacements at three sites where amidehydrogen atoms are extremely slow to exchangewith solvent, I37 (α1), L50 (β2) and L99 (β3),eliminate the misfolded IBP intermediate, whilereplacements at more rapidly exchanging non-polar sites, L127 (β4) and L176 (β6), do not. Apotential explanation for these differential effectscan be found in the crystal structure of αTS, whichshows three distinct and substantive hydrophobicclusters comprised of the large branched aliphaticside-chains isoleucine, leucine and valine (Figure1(b)). I37, L50 and L99 are a part of the largestcluster, cluster 2, which is outside the β barrel andinside the helical shell, spans the N and C termini,and contains 31 ILV side-chains from β1, α1, β2, α2,β3, α3, β4, α8′ and α8. L127 belongs to cluster 1, aninternal-to-the-barrel cluster, which is defined by8 ILV side-chains from β1, β3, β4, β5, β6, β7, β8 andcapping helix α0. L176 belongs to cluster 3, a C-terminal external-to-the-barrel cluster, which isformed by 12 ILV side-chains from β5, α5, β6 andα6. These clusters account for 51 (81%) of the 63 ILV

residues, which together comprise 42% of the non-polar side-chains in αTS and are distributedthroughout the sequence.The dramatically different effects on IBP of alanine

replacements for leucine or isoleucine in cluster 2versus clusters 1 and 3 prompted a comprehensivemutational analysis of the relationship of theseclusters to the formation of IBP and to the stabilityof the equilibrium intermediate, I1.

Results

Cluster assignment

The ILV clusters were visualized with PyMOL bygenerating the 3D map of all van der Waals contactswithin 4.2 Å between heavy atoms of only the ILVside-chains‡. Clusters were defined as the sets ofcontacts involving a minimum of 6 ILV side-chains

1627Hydrophobic Clusters and Misfolding in αTS

arranged in a compact, self-contained structure. Thechoice of the minimum size was based upon MDsimulations in which 30 methane or six isobutylenemolecules were required to form a stable hydro-phobic cluster in water.39 Cluster boundaries wereidentified by the absence of intervening ILV–ILVcontacts. On the basis of these criteria, threediscernable clusters can be identified (Figure 1;Supplementary Data Table 1S). Cluster 1, containingeight residues, is internal-to-the-barrel. Cluster 2,containing 31 ILV residues, is outside the β barrel.Cluster 3, containing 12 residues, is inside the helicalshell. All these contacts are defined as ILV self-contacts, to distinguish them from the contacts ofnon-polar side-chains from alanine, cysteine, methi-onine, phenylalanine and tyrosine and non-polar

Figure 3 (legen

components of otherwise polar side-chains with theI, L and V residues.

Secondary and tertiary structures of the variantsare largely conserved

Far-UV and near-UV CD spectra were used tomonitor the perturbation in the secondary andtertiary structure introduced by the individualreplacement of 20 ILV residues with alanine. Thefar-UV CD spectra of most of the αTS variantsclosely resemble the wild-type spectrum with aminimum near 222 nm, a shoulder near 208 nm, anda maximum near 195 nm (Figure 2). The meanresidue ellipticity at 222 nm for these variants wasreduced by less than 20%, suggesting a relatively

d on next page)

Figure 3. Urea-induced equilib-rium unfolding reactions (●) andthe refolding burst-phase ampli-tudes (▵) from stopped-flow CDmeasurements initiated from theunfolded state at 6 M urea for (a)group I αTS variants and (b) groupII αTS variants. For I37A, L50A,L99A, L127A and L176A, data aretaken fromWu et al.32 The unfoldedbaselines are shown as dash-dotlines. The continuous lines repre-sent the fit of the data to a three-state equilibrium model. The burst-phase amplitudes for all proteinswere fit to a two-state model andare represented by the broken lines.The equilibrium unfolding profileof wild-type αTS and its burst-phase intermediate are shown asthick black continuous and brokenlines, respectively, for reference.Buffer conditions were as describedin the legend to Figure 2.

1628 Hydrophobic Clusters and Misfolding in αTS

small perturbation in the secondary structure.Although the far-UV CD spectra of L25A, L48Aand I232A indicate a greater loss of secondarystructure, all three variants retain the same apparentthree-state equilibrium folding model displayed bywild-type αTS and the rate-limiting unfoldingreactions characteristic of the native conformation(see below). Inspection of the near-UV CD spectra(data not shown) also shows that the packing of thearomatic residues is conserved for the majority ofthe variants. Those for which the aromatic packing isdisrupted to some extent also retain the three-statefolding model and rate-limiting unfolding reactions.Thus, the evident perturbations in structure for asubset of the αTS variants do not preclude theinterpretation of their thermodynamic properties interms of its native and stable intermediates states.

Differential effects on the stability of αTS

To examine the effects of these replacements onthe thermodynamic properties of αTS, the urea-induced unfolding was monitored by CD spectros-copy in the range from 215 nm to 265 nm.Equilibrium unfolding transition curves obtainedby monitoring the urea-dependence of the meanresidue ellipticity at 222 nm for wild-type andvariants are shown in Figure 3. Similar to wild-typeαTS, the equilibrium unfolding reactions of all ofthese variants are best described by a three-statemodel, N⇆I1⇆I2/U. Because the ellipticity of I2 isindistinguishable from the U state, I2/U can betreated as a single apparent thermodynamic statefor titrations monitored by CD spectroscopy.Displacements of the midpoints for both the

1629Hydrophobic Clusters and Misfolding in αTS

N⇆I1 and I1⇆I2/U transitions to lower concentra-tions of urea for almost all of the variants (Figure 3)showed that the replacement of these buriedisoleucine, leucine and valine side-chains withalanine destabilizes both the N and the I1 states.The free energy differences between N and I1 andbetween I1 and I2/U in the absence of denaturantare shown in Figure 4(a) and (b), respectively, forall of the variants. These values and the ureadependence of the free energy differences, the mvalues, as well as the corresponding midpoints, Cm,are provided in Table 1.The decreases in the free energy difference

between the N and I1 states for the variantsrange from 0.14 kcal mol−1 for L50A to 4.86 kcalmol−1 for L48A; I151A was the only replacementthat increased the free energy difference, by1.16 kcal mol−1. The variations in the stability of Nrelative to I1 were similar across the entire sequenceand for replacements in all three hydrophobicclusters (Figure 4(a)). By contrast, the majority ofthe replacements in cluster 2 markedly decreasedthe stability of I1 relative to I2/U, while all of thosein clusters 1 and 3 had a smaller effect (Figure 4(b)).For example, the striking contrast between thesubstantial destabilization by V126A and the min-imal perturbation by L127A appears to be related totheir participation in cluster 2 and cluster 1,

Figure 4. Bar graph of the free energy difference in thIBP⇆I2/U versus I1⇆I2/U transitions for group I (open barsbars on the left refer to the stabilities of the burst-phasmeasurements; the bars on the right represent the stabilitiesexperiments. The green, blue and orange bars represent clufor wild-type αTS are shown in gray, for reference. The selabeled underneath the corresponding variant.

respectively. Excluding the four outliers in cluster2, I41A, L85A, L105A and I111A (which are alllocated in surface α helices), the average decrease instability of the I1 state with respect to the I2/U statefor the cluster 2 variants is 2.61(±0.27) kcal mol−1; theaverage decrease experienced by the cluster 1 andcluster 3 variants is 0.74(±0.25) kcal mol−1. The smalldecrease in the stability of I1 for the four cluster 2outliers, 0.95(±0.46) kcal mol−1, may reflect theirparticipation in surface helices that might individ-ually reorient to minimize the disruption instability. It is possible also that the enhanced helicalpropensity of alanine versus leucine and isoleucinemitigates the decrease in stability at all of thesesites.41

Differential effects on the kinetic foldingproperties of αTS

As previously reported, alanine replacements forleucine and isoleucine residues in α1, β2 and β3have two distinct effects on the properties of theearly folding events in αTS.32 The distinguishingfeatures are the effects on the stability of the productof the burst-phase reaction compared to the on-pathway equilibrium intermediate and the presence(Figure 5(a) and (c)) or absence (Figure 5(b)) of asubsequent unfolding reaction in the hundreds of

e absence of urea for (a) N⇆I1, (b) I1⇆I2/U and (c)) and group II (hatched bars) αTS variants. For (c), thee intermediates (IBP) obtained from stopped-flow CDof the I1 species measured from equilibrium unfoldingsters 1, 2 and 3 αTS variants, respectively. The valuescondary structure assignment for each mutation site is

Table 1. Thermodynamic parameters for urea-induced unfolding of variants and wild-type αTS at pH 7.8 and 25 °C

The equilibrium unfolding data were fit to a three-state model, N⇄I1⇄I2/U, while the refolding burst phase data were fit to a two-state model, N⇄U, assuming a linear dependence of the free energy ofunfolding on the denaturant concentration,ΔG°=ΔG°(H2O)+m [urea].40 ΔG°,ΔG°(H2O),m, and Cm represent the energy of unfolding in the presence of urea, the free energy of unfolding in the absenceof urea, the urea dependence of the free energy of unfolding and the concentration of urea at the midpoint of transition, respectively. Units are as follows:ΔG°(H2O), kcal mol−1;m, kcal mol−1 (M urea)−1;Cm, M urea. Errors for ΔG°(H2O) and m are standard errors from the fits. Errors in Cm are obtained by standard error propagation of the equation: Cm=ΔG°(H2O)/m. Conditions: 10 mM potassiumphosphate (pH 7.8), 0.2 mM K2EDTA, and 1 mM β-mercaptoethanol.a Data taken from Wu et al.32

1630Hydrophobic

Clusters

andMisfolding

inαT

S

Figure 5. The refolding kinetic traces of (a) wild-typeαTS, (b) L99A αTS and (c) L105A αTS monitored bystopped-flow CD for refolding jumps from 6.0 M to 1.0 Murea in standard buffer, described in the legend to Figure 2.

1631Hydrophobic Clusters and Misfolding in αTS

milliseconds time-scale under refolding conditions.A lower stability for IBP relative to I1 and thepresence of this unfolding reaction are indicative ofan off-pathway burst-phase species similar to thatobserved for wild-type αTS. Similar stabilities forthe product of the burst-phase reaction and I1 aswell as the absence of this kinetic phase indicatethat the folding reaction can proceed directly to theon-pathway equilibrium intermediate, I1, withinthe dead-time of the stopped-flow measurement(<5 ms) (Scheme I).On the basis of the kinetic and thermodynamic

responses, the 20 αTS variants in this study can beclassified into two groups: group I eliminates the off-pathway intermediate and group II retains thismisfolded species. Chevron plots show that ten ofthe 20 variants, specifically V23A (β1), L25A (β1),I37A (α1), I41A (α1), L48A (β2), L50A (β2), L85A(α2), I95A (β3), I97A (β3) and L99A (β3), lack thehundreds of milliseconds unfolding reaction understrongly refolding conditions (Figure 6(a)). Further,the denaturation profiles of their CD burst-phaseamplitudes are coincident with those for the unfol-ding of the equilibrium intermediate (Figure 3(a)). Ademonstration of the equivalent stabilities of I1 and

the burst-phase intermediate, relative to the appar-ent I2/U state, for the entire set of group Ireplacements is shown in Figure 4(c) and in Table1. The retention of the unfolding kinetic phase underrefolding conditions and the uniformly lowerstability of the burst-phase species compared to theI1 species for L11A (α0), L105A (α3), I111A (α3),V126A (β4), L127A (β4), I151A (β5), L176A (β6),L209A (β7), I232A (β8), and L255A (α8) (Figure 4(c)and Table 1) define these ten variants as members ofthe group II category. The branched aliphatic side-chains at these positions are not essential for theformation of the misfolded intermediate. Interest-ingly, the stability of the IBP species for the group IIvariants is reduced by a small but consistentamount, 1.16(±0.42) kcal mol-1, relative to the wild-type protein (Figure 4(c)).

Discussion

The role of hydrophobic clusters in the earlymisfolding reaction of αTS

The replacement of a series of isoleucine, leucineand valine residues in αTS with alanine has revealeddistinctively different roles for their resident hydro-phobic clusters in the formation of an early off-pathway misfolded intermediate and in the stabilityof the subsequent on-pathway equilibrium interme-diate. Replacements in cluster 2, in β1, α1, β2, α2and β3, eliminate the off-pathway burst-phasespecies and enable direct access to the on-pathwayintermediate in the sub-millisecond time-scale;replacements in α3, β4 and α8 do not eliminate theburst-phase species. Replacements in both clusters 1and 3, including residues in α0, β4, β5, β6, β7, β8and α8, retain the off-pathway intermediatebut decrease its stability by an average value of1.16(±0.42) kcal mol−1. Thus, the N-terminal regionof cluster 2, formed by branched aliphatic side-chains protruding from five consecutive elements ofsecondary structure defining the β barrel and thesurrounding helical shell, i.e. the (β/α)1–2β3 seg-ment, is crucial for the misfolding reaction of αTS.The alanine replacements for the group I variants

could exert their effects by destabilizing the mis-folded species significantly or by increasing the freeenergy of the barrier between the U or I2 precursorand IBP.

32 Either explanation requires a destabiliza-tion exceeding several kcal mol−1 to eliminate IBP,implying tight packing around all of these side-chains. It has been shown that the replacement ofbranched aliphatic side-chains with alanine in theinterior of staphylococcal nuclease can destabilizethe native conformation by up to 5.8 kcal mol−1.42

Similar results were obtained in a more limitedstudy on barnase;43 where the decrease in stabilitywas equivalent to 1.0–1.6 kcal mol−1 per methylenemolecule removed.44

The small but non-negligible perturbations in thestability of IBP for the group II variants show that theC-terminal region of cluster 2 and both clusters 1

1632 Hydrophobic Clusters and Misfolding in αTS

and 3 have also formed in the millisecond time-frame. The close agreement between the averagedecrease in stability, 1.16(±0.42) kcal mol−1, and thedifferences in the free energy of transfer betweennon-polar and aqueous solvents for isoleucine,leucine and valine versus alanine, 1.3 kcal mol−1,1.3 kcal mol−1and 1.0 kcal mol−1, respectively,42,45

suggest that these clusters resemble molten globule-like states.46

The role of hydrophobic clusters in stabilizingthe on-pathway equilibrium intermediate in αTS

Inspection of the perturbations on the stability ofthe I1 intermediate for the cluster 1, 2 and 3 variants(Figure 4(b) and Table 1) shows a trend that isstrikingly similar to that for the misfolded IBP state.The external-to-the-barrel side-chains in β1, α1 (I37),β2, β3, β4 (V126) and α8 are tightly-packed in I1,while the internal-to-the-barrel side-chains in α0, β4(L127), β5, β7 and β8 and the external-to-the-barrelside-chains in β6 occupy molten globule-like envi-ronments. The average decrease in stability of the I1state for these latter variants is 0.74(±0.25) kcalmol−1.Differentiating I1 from IBP is the implied tight-packing for V126 (β4) and L255 (α8) in I1, both ofwhich participate in cluster 2 but are not members ofthe group I variants.

The role of hydrophobic clusters in stabilizingthe rate-limiting transition state in the folding ofαTS

In striking contrast to the differential effects of theILV replacements on the free energies of IBP and I1,17 of the 20 variants (with the exception of I151A inβ5, I232A in β8 and L255A in α8) display a majorslow unfolding reaction that is indistinguishablefrom that for wild-type αTS (Figure 6). Given theobserved destabilization of the native state in thevariants (Table 1), these results demonstrate thatvery native-like packing is maintained in all threehydrophobic clusters in the transition state for therate-limiting N→I1 unfolding step47 in Scheme I. Byinference from the chevron shape48 for the urea-dependence of the relaxation times for the reversibleN⇆I1 reaction (Figure 3), native-like tight packing isachieved at all of these positions in the transitionstate for the I1→N reaction. Thus, the ILV side-chains have significant roles in stabilizing thistransition state and, thereby, potentially acceleratingthe final step in the folding reaction. The resultsshow also that the transition state is formed by side-chains that are distributed throughout the sequence,a finding that may be related to its rate-limiting rolein folding.

Why is misfolding favored over productivefolding for wild-type αTS?

The similar but not identical responses of IBP and I1to the alanine replacements beg the question of whythe unfolded state of wild-type αTS initially mis-

folds. Why does U/I2 not proceed directly to the on-pathway equilibrium intermediate? The answerappears to reside in the peculiar responses ofV126A and L255A. Both replacements decrease thestability of the I1 state substantially, similar to mostof the group I variants, but retain the off-pathwayburst-phase intermediate, a defining feature of thegroup II variants. The differential effects on the IBPand I1 intermediates imply that both side-chains areexcluded from the hydrophobic cluster crucial to theformation of the off-pathway species but areintimately involved in stabilizing the on-pathwayintermediate. One interpretation of these results isthat the folding of the region from β1 to β3 is drivenby the sub-millisecond formation of a localized andself-contained structure whose boundaries, unlikethose in the native TIM barrel, allow favorableinteractions with the solvent. The putative non-native structure at the boundaries does not providesuitable docking sites for β4 and α8, requiring themisfolded IBP species to unfold at least partiallybefore recruiting both of these elements of secondarystructure into the I1 intermediate. The failure toincorporate α8 into IBP may be related to its distalposition in the sequence, where chain diffusionmight limit access to a receptive (β/α)1–2β3 platform.

Differential packing in hydrophobic clusters inthe off and on-pathway intermediates for αTS

The heterogeneous response of the hydrophobicclusters in αTS to alanine replacements impliestight packing for side-chains in cluster 2 (with theexception of α3, β4 and α8 in IBP) and looser,molten globule-like packing for clusters 1 and 3 inboth IBP and I1 (Table 1). The striking segregationof the ILV side-chains into three distinct clusters(Figure 1(b)) motivated a closer inspection of theseclusters and their interactions with other nearbyside-chains (Table 2). The 31 ILV side-chains incluster 2 make 90 self-contacts, assuming a 4.2 Åcut-off distance between carbon atoms, the 8 ILVside-chains in cluster 1 make 14 self-contacts andthe 12 ILV side-chains in cluster 3 make 20 self-contacts. Thus, the average number of self-contactsper residue is about twice as large for cluster 2 asthat for clusters 1 and 3. This distinction disappearsif other non-polar side-chains from alanine, cyste-ine, methionine, phenylalanine, proline and tyro-sine and non-polar components of otherwise polarside-chains in contact with the ILV residues in theclusters are included in this analysis (Table 2).Thus, the differential properties of the threeclusters do not appear to be related to the densityof van der Waals contacts when all buried side-chains are considered.Two previous observations on αTS suggested

that the differential properties might be related toa special role for leucine and isoleucine instabilizing partially folded states. An HX-NMRstudy revealed that seven of the 11 amidehydrogen atoms most resistant to exchange withsolvent deuterium are leucine or isoleucine located

1633Hydrophobic Clusters and Misfolding in αTS

in α1, β2 and β3.32 A network of L–L and L–Iinteractions between these seven side-chainsappears to stabilize the core of the N-terminalhydrophobic cluster (β/α)1–2β3, which is central to

Figure 6 (legend

both the IBP and I1 intermediates. Further supportfor a unique role for leucine and isoleucine canbe found in a heteronuclear single quantumcoherence NMR spectrum of selectively [15N]Leu

on next page)

Figure 6. Urea-dependence ofthe observed refolding phases mon-itored by manual mixing CD (○,▽,□) and stopped-flow fluorescence(⋄), as well as unfolding phasesmonitored by manual mixing CD(●, ▴) for (a) group I αTS variantsand (b) group II αTS variants. Thecontinuous lines represent observedrelaxation times for wild-type αTS.The diamonds (⋄) indicate thebehavior of the several hundredmillisecond refolding phase corres-ponding to the unfolding of the off-pathway, burst-phase intermediate.The open triangles (▽), open circles(○) and open squares (□) reflect thefast, intermediate and slow refold-ing phases attributed to prolyl iso-merization. The filled circles (●) andtriangles (▴) represent the majorand minor unfolding phases. In thecase of L48A, the observation of asub-millisecond phase (◈) reflectsthe refolding of the I2/U to I1 reac-tion, based on its continuous link-age with the unfolding reaction ofthe I1 species (■).

1634 Hydrophobic Clusters and Misfolding in αTS

or [15N]Ile-labeled αTS at 5 M urea where thepossible I2 precursor to IBP is highly populated (R.V., Y.W. and C. R.M., unpublished results). In theabsence of well-defined secondary structure, ap-proximately three leucine residues, including L99 ofβ3, and approximately four isoleucine residues areexchange-broadened by the formation of a previous-ly identified hydrophobic cluster;49 phenylalanine

and valine do not participate in the dynamicbroadening process. The persistent interactionsbetween these leucine and isoleucine residuesunder strongly denaturing conditions imply thatmutual interactions between large branched aliphaticside-chains in the absence of denaturant may be theprimary source of the differential behavior of thethree clusters in αTS.

Table 2. Summary of the side-chain contacts within 4.2 Å for non-polar side-chains and the aliphatic component of polarside-chains in the three αTS hydrophobic clusters

ILV ACFILMPVY ALLa

No.contacts

No.residues

No. contacts/residue

No.contacts

No.residues

No. contacts/residue

No.contacts

No.residues

No. contacts/residue

Cluster 1 14 8 1.8 42 12 3.5 48 13 3.7Cluster 2 90 31 2.9 182 57 3.2 287 100 2.9Cluster 3 20 12 1.7 85 35 2.4 144 52 2.8Total 124 51 2.4 309 104 3.0 479 165 2.9

As noted in the legend to Figure 1, the side-chain contacts are calculated based on the coordinates of the α subunit of tryptophan synthasefrom S. typhimurium (1BKS),62 which shares 85% sequence identity with E. coli αTS. ILV residues are largely conserved in E.coli, except thesix I/V switches: V52I (β2), I148V (loop before β5), V166I (α5), I197V (α6), V224I (α7) and L245I (loop between α8′ and α8).

a All residues in the sequence except glycine.

1635Hydrophobic Clusters and Misfolding in αTS

Branden and Tooze noted a strong bias towardsILV residues in TIM barrel proteins.50 This biasmay reflect the distinct preference observed for ILVside-chains with ILV neighbors in adjacent parallelstrands in TIM barrel and parallel (β/α)n sheetproteins.51 Simulations designed to optimize side-chain placements in predicted protein structuresfound that aliphatic pairs adopt restricted orienta-tions in actual structures that differentiate themfrom other pairs of non-polar side-chains.52 Alongwith pairing and conformational preferences, theILV side-chains are three of the four side-chains(cysteine is the fourth) that are most effective atscreening the underlying peptide hydrogen bondfrom solvent.53 As a result, the hydrogen bond isincreased in strength and the peptide linkage isdecreased in volume. If the increase in packingdensity is propagated to the attached side-chains,the ILV side-chains might pack more tightly thanother non-polar side-chains. The increase in hydro-gen bond strength and decreased accessibility tosolvent would both be expected to contribute toenhanced protection against exchange of amidehydrogen atoms with solvent.54 This conjecture isconsistent with HX-MS results that show that thestrongest protection against exchange was ob-served for the (β/α)1–3β4 segment under conditionsfavoring the I1 species in αTS.55

Given the unique roles of the (β/α)1–2β3 segmentin the misfolded IBP intermediate and the (β/α)1-3β4and α8 segments in the on-pathway I1 intermediate,it appears that a critical number of ILV residues witha significant fraction of self-contacts is required forthe formation of a tightly packed hydrophobiccluster in these intermediates. Smaller ILV clustersor clusters with a reduced fraction of self-contactsamong the ILV subset form molten globule-likestructures that play a lesser role in defining the freeenergy surface on which the αTS folds.

Branched aliphatic side-chain hypothesis andpartially-folded states in TIM barrel and (β/α)nproteins

Taken together, these observations led to theproposal of the Branched Aliphatic Side-Chain(BASiC) hypothesis, which posits a unique role for

isoleucine, leucine and valine in forming tightlypacked hydrophobic clusters that guide the for-mation of structure in TIM barrel and (β/α)nparallel sheet proteins. On the basis of the resultsfor αTS, these clusters contain >12 ILV residuesand correspond to segments of >80 residues. Fromthe perspective of secondary structure, this re-quirement translates to a minimum βαβαβ mod-ule whose hydrogen bonding network wouldenable, and be strengthened by, the tight packingof the branched aliphatic side-chains between thestrands and the helices. Less than a critical mass ofILV residues or dilution with other residues, so asto diminish the ILV self interactions below acritical level, can result in loosely packed moltenglobule-like structures (Table 2). Although themolten globule-like clusters may not guide thefolding reaction beyond reducing conformationalspace, they could serve to decrease the propen-sity for self-association and aggregation duringfolding.Preliminary application of this hypothesis to two

other TIM barrel proteins, phosphoribosyl anthra-nilate isomerase,56 and yeast triose phosphate iso-merase,57 revealed good agreement between thelocation of tightly packed, ILV-dominated hydro-phobic clusters and stable folding cores. Similarto αTS, the minimal core for both proteins encom-passes a βαβαβ module. Regarding the off-pathway misfolded intermediate in CheY, exami-nation of the structure shows two tightly packedILV clusters, one on either side of the parallel β-sheet, which span the N and C termini. If one orboth of the clusters stabilizes the misfoldedspecies, the C terminus would have to unfold atleast partially to access a transition state known tobe defined by the N-terminal β strands and αhelices.36

The relatively high content of isoleucine, leucineand valine residues in all proteins, ∼20%,45,58

suggests that hydrophobic clusters formed bymutual interactions between these residues mightplay important roles in guiding the early foldingevents of other motifs. These same branchedaliphatic side-chains might be crucial to the devel-opment of extremely potent intermolecular interac-tions that drive the formation of a subset of the

1636 Hydrophobic Clusters and Misfolding in αTS

aggregates and amyloids thought to be involved in ahost of human pathologies.59

Materials and Methods

Chemicals and reagents

Ultra-pure urea was purchased from ICN Biomedicals(Costa Mesa, CA). For all experiments, urea was purifiedfreshly by removal of ionic degradation products using acolumn containing AG® 501-XB (D) resin purchased fromBio-Rad laboratories (Hercules, CA). All other chemicalswere reagent grade.

Oligonucleotide-directed mutagenesis, proteinexpression and purification

Oligonucleotides used for mutagenesis were obtainedfrom Operon (Alameda, CA). Escherichia coli strain CB149served as the host for expressing wild-type and variantαTS proteins. Mutations were introduced by the Strata-gene (La Jolla, CA) Quikchange™ site-directed mutagen-esis method. Mutations were verified by DNA sequenceanalysis. Over-expression and purification of the wild-type and the variant αTS proteins were as described.32,47

Purified protein, which was confirmed by the presence ofa single band on electrophoresis in a sodium dodecylsulfate/15% (w/v) polyacrylamide gel, was stored in 80%(w/v) ammonium sulfate solutions at 4 °C and dialyzedjust before use against the standard buffer used for allfolding experiments (10 mM potassium phosphate (pH7.8), 0.2 mM K2EDTA, 1 mM β-mercaptoethanol). Proteinconcentration was determined from measurement of theUVabsorbance at 278 nm, using an extinction coefficient of12,600 M-1 cm-1 for both wild-type and variant αTSproteins.60 The possible effects of amino acid replacementsnear tyrosine residues on the extinction coefficients wereaccounted for by direct measurements of the extinctioncoefficients as described.61

Equilibrum and kinetic experiments

CD measurements for equilibrium and manual-mixingkinetic experiments were carried out on a Jasco Model J-810 spectropolarimeter equipped with a thermoelectriccell-holder as described.13 The dead-time for manualmixing was 3 s. Unfolding jumps were from 0 M urea tovarious final concentrations of urea (2.0 M–7.0 M, andrefolding jumps were performed from 8 M or 6 M tovarious final urea concentrations (0.3 M–3.4 M). The finalconcentration of protein was 3–5 μM. The stopped-flowCD refolding experiments were performed with an Avivmodel 202SF stopped-flow CD spectrometer, with a 5 msdead-time for mixing. All of the tyrosine fluorescenceexperiments were performed with an Applied Photophy-sics SX-18MV stopped-flow fluorimeter. The excitationpathlength was 10 mm, and the emission pathlength was2 mm. The excitation wavelength was set at 280 nm with a2.32 nm bandwidth. The emission above 320 nm wasdetected with an appropriate cut-off filter.

Data analysis

The refolding burst-phase amplitudes as a function ofurea and the equilibrium data were fit to a two-state and a

three-state model, respectively, as described.32 The freeenergy changes were assumed to depend linearly on theconcentration of denaturant.40 The kinetic data obtainedfrom unfolding or refolding experiments were fit to thesum of a minimum number of statistically significantexponentials and a constant.

Cluster assignment

The distance between side-chain carbon atoms of theselected residue types was computed with an in-houseprogram using a cutoff distance of 4.2 Å. The number ofcontacts between any two residues were counted andexported to PyMOL for visualizing contacts in a three-dimensional format. Crystal structure of the α subunit oftryptophan synthase from Salmonella typhimurium (PDBcode 1BKS, a refined version of 1WSY),62 which shares85% sequence identity with E. coli αTS, was chosen for thecluster calculation, instead of the E. coli structure (PDBcodes 1V7Y and 1WQ5),63 because the crystal contacts inthe E. coli αTS structure result in ambiguous results for thestructure of the 52–77 and 178–192 regions in solution.

Acknowledgements

The authors thank Drs Jill Ann Zitzewitz andOsman Bilsel for critical reviews of the manuscriptand many helpful comments. This work wassupported by the National Institutes of Healththrough grant GM23303 to C.R.M.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2006.12.005

References

1. Bilsel, O. & Matthews, C. R. (2000). Barriers in proteinfolding reactions. Advan. Protein Chem. 53, 153–207.

2. Jackson, S. E. (1998). How do small single-domainproteins fold? Fold. Des. 3, R81–R91.

3. Alm, E. & Baker, D. (1999). Matching theory andexperiment in protein folding. Curr. Opin. Struct. Biol.9, 189–196.

4. Ferguson, N., Capaldi, A. P., James, R., Kleanthous, C.& Radford, S. E. (1999). Rapid folding with andwithout populated intermediates in the homologousfour-helix proteins Im7 and Im9. J. Mol. Biol. 286,1597–1608.

5. Friel, C. T., Beddard, G. S. & Radford, S. E. (2004).Switching two-state to three-state kinetics in thehelical protein Im9 via the optimisation of stabilisingnon-native interactions by design. J. Mol. Biol. 342,261–273.

6. Matthews, C. R. (1993). Pathways of protein folding.Annu. Rev. Biochem. 62, 653–683.

7. Baldwin, R. L. (1996). On-pathway versus off-pathwayfolding intermediates. Fold. Des. 1, R1–R8.

8. Kamagata, K., Arai, M. & Kuwajima, K. (2004).

1637Hydrophobic Clusters and Misfolding in αTS

Unification of the folding mechanisms of non-two-stateand two-state proteins. J. Mol. Biol. 339, 951–965.

9. Cecconi, C., Shank, E. A., Bustamante, C. & Marqusee,S. (2005). Direct observation of the three-statefolding of a single protein molecule. Science, 309,2057–2060.

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

11. Brandts, J. F., Halvorson, H. R. & Brennan, M. (1975).Consideration of the possibility that the slow stepin protein denaturation reactions is due to cis-transisomerism of proline residues. Biochemistry, 14,4953–4963.

12. Maki, K., Ikura, T., Hayano, T., Takahashi, N. &Kuwajima, K. (1999). Effects of proline mutations onthe folding of staphylococcal nuclease. Biochemistry,38, 2213–2223.

13. Wu, Y. &Matthews, C. R. (2002). Parallel channels andrate-limiting steps in complex protein folding reac-tions: prolyl isomerization and the alpha subunit ofTrp synthase, a TIM barrel protein. J. Mol. Biol. 323,309–325.

14. Mello, C. C., Bradley, C. M., Tripp, K. W. & Barrick, D.(2005). Experimental characterization of the foldingkinetics of the notch ankyrin domain. J. Mol. Biol. 352,266–281.

15. Sosnick, T. R., Mayne, L., Hiller, R. & Englander, S. W.(1994). The barriers in protein folding.Nat. Struct. Biol.1, 149–156.

16. Roder, H., Elove, G. A. & Englander, S. W. (1988).Structural characterization of folding intermediates incytochrome c by H-exchange labelling and protonNMR. Nature, 335, 700–704.

17. Raschke, T. M., Kho, J. & Marqusee, S. (1999).Confirmation of the hierarchical folding of RNase H:a protein engineering study. Nature Struct. Biol. 6,825–831.

18. Bilsel, O., Zitzewitz, J. A., Bowers, K. E. & Matthews,C. R. (1999). Folding mechanism of the alpha-subunit of tryptophan synthase, an alpha/betabarrel protein: global analysis highlights the inter-conversion of multiple native, intermediate, andunfolded forms through parallel channels. Biochem-istry, 38, 1018–1029.

19. Wallace, L. A., & Matthews, C. R. (2002). Sequentialvs. parallel protein-folding mechanisms: experimentaltests for complex folding reactions. Biophys. Chem.101–102, 113–131.

20. Gianni, S., Travaglini-Allocatelli, C., Cutruzzola, F.,Brunori, M., Shastry, M. C. & Roder, H. (2003). Parallelpathways in cytochrome c(551) folding. J. Mol. Biol.330, 1145–1152.

21. Gloss, L. M. & Matthews, C. R. (1998). Mechanism offolding of the dimeric core domain of Escherichia colitrp repressor: a nearly diffusion-limited reaction leadsto the formation of an on-pathway dimeric interme-diate. Biochemistry, 37, 15990–15999.

22. Topping, T. B., Hoch, D. A. & Gloss, L. M. (2004).Folding mechanism of FIS, the intertwined, dimericfactor for inversion stimulation. J. Mol. Biol. 335,1065–1081.

23. de Prat-Gay, G., Nadra, A. D., Corrales-Izquierdo,F. J., Alonso, L. G., Ferreiro, D. U. & Mok, Y. K. (2005).The folding mechanism of a dimeric beta-barrel do-main. J. Mol. Biol. 351, 672–682.

24. Rothwarf, D. M. & Scheraga, H. A. (1996). Role of non-native aromatic and hydrophobic interactions in thefolding of hen egg white lysozyme. Biochemistry, 35,13797–13807.

25. Colon, W., Wakem, L. P., Sherman, F. & Roder, H.(1997). Identification of the predominant non-nativehistidine ligand in unfolded cytochrome c. Biochemis-try, 36, 12535–12541.

26. Arai, M., Ikura, T., Semisotnov, G. V., Kihara, H.,Amemiya, Y. & Kuwajima, K. (1998). Kinetic refoldingof beta-lactoglobulin. Studies by synchrotron X-rayscattering, and circular dichroism, absorption andfluorescence spectroscopy. J. Mol. Biol. 275, 149–162.

27. Capaldi, A. P., Shastry, M. C., Kleanthous, C., Roder,H. & Radford, S. E. (2001). Ultrarapid mixingexperiments reveal that Im7 folds via an on-pathwayintermediate. Nature Struct. Biol. 8, 68–72.

28. Cho, J. H., Sato, S. & Raleigh, D. P. (2004). Thermo-dynamics and kinetics of non-native interactions inprotein folding: a single point mutant significantlystabilizes the N-terminal domain of L9 by modulatingnon-native interactions in the denatured state. J. Mol.Biol. 338, 827–837.

29. Munoz, V., Lopez, E. M., Jager, M. & Serrano, L.(1994). Kinetic characterization of the chemotacticprotein from Escherichia coli, CheY. Kinetic analysis ofthe inverse hydrophobic effect. Biochemistry, 33,5858–5866.

30. Fernandez-Recio, J., Genzor, C. G. & Sancho, J. (2001).Apoflavodoxin folding mechanism: an alpha/betaprotein with an essentially off-pathway intermediate.Biochemistry, 40, 15234–15245.

31. Bollen, Y. J., Sanchez, I. E. & Van Mierlo, C. P. (2004).Formation of on- and off-pathway intermediates inthe folding kinetics of Azotobacter vinelandii apoflavo-doxin. Biochemistry, 43, 10475–10489.

32. Wu, Y., Vadrevu, R., Yang, X. & Matthews, C. R.(2005). Specific structure appears at the N terminus inthe sub-millisecond folding intermediate of the alphasubunit of tryptophan synthase, a TIM barrel protein.J. Mol. Biol. 351, 445–452.

33. Butler, J. S. & Loh, S. N. (2005). Kinetic partitioningduring folding of the p53 DNA binding domain.J. Mol. Biol. 350, 906–918.

34. Roder, H., Maki, K. & Cheng, H. (2006). Early eventsin protein folding explored by rapid mixing methods.Chem. Rev. 106, 1836–1861.

35. Clementi, C., Nymeyer, H. & Onuchic, J. N. (2000).Topological and energetic factors: what determinesthe structural details of the transition state ensembleand “en-route” intermediates for protein folding? Aninvestigation for small globular proteins. J. Mol. Biol.298, 937–953.

36. Lopez-Hernandez, E. & Serrano, L. (1996). Structure ofthe transition state for folding of the 129 aa proteinCheY resembles that of a smaller protein, CI-2. Fold.Des. 1, 43–55.

37. Wu, Y. & Matthews, C. R. (2002). A cis-prolyl peptidebond isomerization dominates the folding of the alphasubunit of Trp synthase, a TIM barrel protein. J. Mol.Biol. 322, 7–13.

38. Wu, Y. & Matthews, C. R. (2003). Proline replacementsand the simplification of the complex, parallel channelfolding mechanism for the alpha subunit of Trp syn-thase, a TIM barrel protein. J. Mol. Biol. 330, 1131–1144.

39. Tsai, J., Gerstein, M. & Levitt, M. (1997). Simulatingthe minimum core for hydrophobic collapse inglobular proteins. Protein Sci. 6, 2606–2616.

40. Schellman, J. A. (1978). Solvent denaturation. Biopoly-mers, 17, 1305–1322.

41. Pace, C. N. & Scholtz, J. M. (1998). A helix propensityscale based on experimental studies of peptides andproteins. Biophys. J. 75, 422–427.

1638 Hydrophobic Clusters and Misfolding in αTS

42. Shortle, D., Stites, W. E. & Meeker, A. K. (1990).Contributions of the large hydrophobic amino acids tothe stability of staphylococcal nuclease. Biochemistry,29, 8033–8041.

43. Kellis, J. T., Jr, Nyberg, K. & Fersht, A. R. (1989). Ener-getics of complementary side-chainpacking in a proteinhydrophobic core. Biochemistry, 28, 4914–4922.

44. Shortle, D. (1992). Mutational studies of proteinstructures and their stabilities. Quart. Rev. Biophys.25, 205–250.

45. Creighton, T. E. (1993). Proteins: Structures andMolecular Properties. W. H. Freeman and Company,New York.

46. Arai, M. & Kuwajima, K. (2000). Role of the moltenglobule state in protein folding. Advan. Protein. Chem.53, 209–282.

47. Beasty, A. M., Hurle, M. R., Manz, J. T., Stackhouse, T.,Onuffer, J. J. & Matthews, C. R. (1986). Effects of thephenylalanine-22-leucine, glutamic acid-49-methio-nine, glycine-234-aspartic acid, and glycine-234-lysinemutations on the folding and stability of the alphasubunit of tryptophan synthase from Escherichia coli.Biochemistry, 25, 2965–2974.

48. Matthews, C. R. (1987). Effect of point mutations onthe folding of globular proteins.Methods Enzymol. 154,498–511.

49. Saab-Rincon, G., Gualfetti, P. J. & Matthews, C. R.(1996). Mutagenic and thermodynamic analyses ofresidual structure in the alpha subunit of tryptophansynthase. Biochemistry, 35, 1988–1994.

50. Branden, C. & Tooze, J. (1999). Introduction to ProteinStructure, 2nd edit., Garland Science Publishing, NewYork.

51. Fooks, H. M., Martin, A. C., Woolfson, D. N., Sessions,R. B. & Hutchinson, E. G. (2006). Amino acid pairingpreferences in parallel beta-sheets in proteins. J. Mol.Biol. 356, 32–44.

52. Misura, K. M., Morozov, A. V. & Baker, D. (2004).Analysis of anisotropic side-chain packing in proteinsand application to high-resolution structure predic-tion. J. Mol. Biol. 342, 651–664.

53. Schell, D., Tsai, J., Scholtz, J. M. & Pace, C. N. (2006).Hydrogen bonding increases packing density in theprotein interior.Proteins: Struct. Funct. Genet. 63, 278–282.

54. Englander, S. W. & Kallenbach, N. R. (1983). Hydro-gen exchange and structural dynamics of proteins andnucleic acids. Quart. Rev. Biophys. 16, 521–655.

55. Rojsajjakul, T., Wintrode, P., Vadrevu, R., RobertMatthews, C. & Smith, D. L. (2004). Multi-stateunfolding of the alpha subunit of tryptophansynthase, a TIM barrel protein: insights into thesecondary structure of the stable equilibrium inter-mediates by hydrogen exchange mass spectrometry.J. Mol. Biol. 341, 241–253.

56. Akanuma, S. & Yamagishi, A. (2005). Identificationand characterization of key substructures involved inthe early folding events of a (beta/alpha)8-barrelprotein as studied by experimental and computationalmethods. J. Mol. Biol. 353, 1161–1170.

57. Silverman, J. A. & Harbury, P. B. (2002). Theequilibrium unfolding pathway of a (beta/alpha)8barrel. J. Mol. Biol. 324, 1031–1040.

58. McCaldon, P. & Argos, P. (1988). Oligopeptide biasesin protein sequences and their use in predictingprotein coding regions in nucleotide sequences.Proteins: Struct. Funct. Genet. 4, 99–122.

59. Chiti, F. & Dobson, C. M. (2006). Protein misfolding,functional amyloid, and human disease. Annu. Rev.Biochem. 75, 333–366.

60. Matthews, C. R., Crisanti, M.M., Manz, J. T. & Gepner,G. L. (1983). Effect of a single amino acid substitutionon the folding of the alpha subunit of tryptophansynthase. Biochemistry, 22, 1445–1452.

61. Gill, S. C. & von Hippel, P. H. (1989). Calculation ofprotein extinction coefficients from amino acid se-quence data. Anal. Biochem. 182, 319–326.

62. Hyde, C. C., Ahmed, S. A., Padlan, E. A., Miles, E. W.& Davies, D. R. (1988). Three-dimensional structure ofthe tryptophan synthase alpha 2 beta 2 multienzymecomplex from Salmonella typhimurium. J. Biol. Chem.263, 17857–17871.

63. Nishio, K., Morimoto, Y., Ishizuka, M., Ogasahara,K., Tsukihara, T. & Yutani, K. (2005). Conforma-tional changes in the alpha-subunit coupled to bin-ding of the beta 2-subunit of tryptophan synthasefrom Escherichia coli: crystal structure of the trypto-phan synthase alpha-subunit alone. Biochemistry, 44,1184–1192.

Edited by K. Kuwajima

(Received 27 September 2006; received in revised form 22 November 2006; accepted 3 December 2006)Available online 15 December 2006