J. Mol. Biol. (1995) 253, 505–513

Disulfide Mutants of Barnase II: Changes in Structureand Local Stability Identified by Hydrogen Exchange

Jane Clarke, Andrea M. Hounslow and Alan R. Fersht*

The hydrogen exchange behaviour of two stabilised disulfide mutants ofCentre for Proteinbarnase has been examined using NMR H/2H exchange measurements.Engineering, MRC Unit forThe choice of experimental conditions is crucial in experiments to study theProtein Function and Design

MRC Centre, Hills Road effects of mutations on local and global stability and dynamic behaviourCambridge, CB2 2QH, UK of proteins. If exchange conditions allow both local and global unfolding

events to be examined, then a comparison of three proteins (two mutantsand wild-type) allows the effect of a mutation on the folded state to beascribed to specific increases or decreases in local stability. This methodwas used to examine the effect of an introduced crosslink on the foldedstate of two different disulfide mutants of barnase, and the results arerelated to structural studies. It is found that disulfide bonds can stabiliseor destabilise local structures as well as having an effect upon globalstability. The effects of the mutations on exchange rate constants of protonsare compared with the effects on the structure upon the introduction of thedisulfide bonds. In the protein with a disulfide bond between residues 43and 80, some local exchange rate constants are higher, reflecting localdestabilisation at the site of the disulfide bond, associated with smallstructural rearrangements. In the protein with a disulfide bond betweenresidues 85 and 102, amide protons of the adjacent loop are protected toa considerable extent. This is not associated with a structural rearrangementyet indicates that this disulfide bond has an effect on the stability of thisloop.

7 1995 Academic Press Limited

Keywords: hydrogen/deuterium exchange; mutation; protein stability;disulfide bond; barnase*Corresponding author

Introduction

The H/2H-exchange of amide groups in proteinscan be monitored by NMR (Wagner et al., 1984).Individual protons within a folded protein canexchange with widely differing rate constants,reflecting relative local mobility, local structure,protection of the proton by hydrogen bonding, andsolvent accessibility (Englander & Kallenbach, 1984;Pedersen et al., 1991). Hydrogen exchange has beenused to study the structure and dynamic propertiesof folded proteins and to probe for residualstructure in denatured and partially folded states(Wagner & Wutrich, 1982; Englander & Kallenbach,

1984; Pedersen et al., 1991; Gallagher et al., 1992;Radford et al., 1992). We have previously demon-strated that a quantitative analysis of the exchangebehaviour of a mutant protein with wild-typeenables us to distinguish those parts of the proteinthat exchange through local unfolding events, thosethat exchange through global unfolding and thosethat exchange by a mixture of the two mechanisms(Clarke et al., 1993). A qualitative comparison ofexchange rates of three different proteins, which allhave essentially the same tertiary structure, andwhich have the same hydrogen bonding patterns,allows local effects of a mutation to be distinguishedfrom the overall effects upon protein stability.

The proteins used in this study are wild-typebarnase and two different mutants with engineereddisulfide bonds. Natural disulfide bonds canstabilise proteins to such an extent that manyproteins unfold when their disulfide bonds are

Abbreviations used: 43–80, mutant of barnase with adisulfide bond linking residues 43 and 80; 85–102,mutant of barnase with a disulfide bond linkingresidues 85 and 102.

0022–2836/95/430505–09 $12.00/0 7 1995 Academic Press Limited

H/2H Exchange in Disulfide Mutants of Barnase506

reduced (Creighton & Goldenberg, 1984). While ithas been possible to stabilise some proteins to aconsiderable extent by the introduction of disul-fide bonds, other proteins have been destabilised(Matsumura et al., 1989; Mitchinson & Wells, 1989;Kanaya et al., 1991; Betz & Pielak, 1992; Clarke &Fersht, 1993). It is widely accepted that a disulfidebond stabilises a protein by destabilising theunfolded state by reducing its entropy. Set againstthis is the ‘‘strain energy’’ imposed on the foldedstate by the introduction of the disulfide bond andthe effect of mutagenesis, possibly removingfavourable interactions and introducing unfavour-able ones. Based upon this theory, rational schemesfor the choice of sites for the introduction ofdisulfide bonds have been proposed (Wetzel, 1987;Matsumura & Matthews, 1991). It has beensuggested, alternatively, that the effect of a di-sulfide bond could be largely enthalpic (Doig &Williams, 1991). Clearly, the effect of an introducedcrosslink is complex and the effect of a disulfidebond on the folded state of a protein cannot beignored (Betz, 1993).

Wild-type barnase contains no Cys residues. Inthe accompanying paper (Clarke et al., 1995) wehave reported two mutants of barnase withengineered disulfide bonds that stabilise theprotein. The disulfide bond between residues43 and 80 (43–80) confers a stabilisation of3.2 kcal mol−1 compared with the dithiol form ofthe protein. The disulfide bond between residues85 and 102 (85–102) increases the stability by4.8 kcal mol−1, relative to the corresponding dithiolform (Clarke & Fersht, 1993); this is larger than thepredicted stabilisation, implying that this disul-fide bond may increase the stability of thefolded state. The values of DDGU-F (the differencein free energies of unfolding) of oxidised 43–80and 85–102 relative to wild-type at pH 6.3 are −2.1and −4.3 kcal mol−1, respectively. We suggested, onthe basis of biophysical data, that the disulfidebond at 43–80 causes the folded state to be‘‘strained’’, whereas the disulfide bond at 85–102might stabilise the folded protein in some way(Clarke et al., 1995, accompanying paper). Thecrystal structure of 43–80 revealed some structuralrearrangements at the site of the disulfidebond. Disulfide bond formation in the mutant85–102, however, induces no such structuralrearrangements. The structural data provided noevidence for understanding the basis of theconsiderable stability induced by this disulfidebond.

In this study, we investigate the effect ofdisulfide bonds upon the folded state of theprotein by comparing the kinetics of amide protonexchange, measured by NMR, of the disulfidemutants with wild-type protein. We demonstratethat disulfide bonds can stabilise or destabiliselocal structures as well as having an effect uponglobal stability. These results correlate with themeasured stabilities of the disulfide-bonded mutantproteins.

Results

Over 50% of the amide protons of all threeproteins exchange within the dead time of 15 to20 minutes for the experiment. Most of the pro-tected protons are involved in hydrogen bondsin the crystal structure (Mauguen et al., 1982, Clarkeet al., 1993). The protected protons of barnase fallprincipally in a-helices and b-sheets; very fewprotected protons are in loops, presumably reflect-ing the mobility and solvent exposure of thesestructural elements. All the observed hydrogenbonds are maintained in the mutant structures(Clarke et al. 1995, accompanying paper). Since weare comparing proteins with only very smallstructural differences, it is possible to comparedirectly the exchange rates of the same proton inwild-type and mutant proteins. In the previouspaper we also described a third mutant with adisulfide bond between residues 70 and 92. Sincethis disulfide bond induced considerable structuralchange we could not use this method to do a similaranalysis.

The measured rate constants, displayed in Table 1,vary by five orders of magnitude. In general, therate constants for exchange of protons reflect thestability of the proteins, with the rate constant forexchange in wild-type barnase being greater thanthose for the mutant 43–80, which generallyexchanges more rapidly than 85–102. At p2H 6.3,exchange is effectively complete for wild-typebarnase within five weeks, and for 43–80 within17 weeks. A number of protons in 85–102 do notexchange appreciably over 17 weeks. The rateconstants for wild-type vary within about twoorders of magnitude under these conditions,whereas those for 43–80 and 85–102 vary withinapproximately three and four orders of magnituderespectively. The lowest exchange rate constantwhich could be measured is 5.2 × 10−7 min−1 (NH ofresidue 25 in 85–102). This was estimated by fittingthe initial decay to a linear function, and dividingthe initial rate of decay by the amplitude of theintensity. After 17 weeks, the experiment wasdiscontinued. It was calculated that, at 37°C, theexchange rate for some of the slowest exchangingprotons of 85–102 would have a half life of betweenfive and six years.

All three proteins exchange much more rapidly atp2H 7.5, reflecting the increased intrinsic rate ofexchange, so that it was possible to obtain rateconstants of exchange for all protected protons in allthree proteins. The measured rate constants ofexchange vary within three orders of magnitude.Exchange in wild-type barnase is complete after24 hours, 43–80 exchanges completely in two weeksand 85–102 takes two months to exchangecompletely.

The change of activation energy for exchange onmutation, DDGex, was determined at p2H 6.3 for allresidues (Table 2). As described previously (Clarkeet al., 1993), the values of DDGex can be divided intothree categories: those where DDGex is close to zero

H/2H Exchange in Disulfide Mutants of Barnase 507

(in italics in Table 2), which exchange predomi-nantly by local unfolding; those where DDGex isclose to DDGU-F, the full change in free energyfor unfolding upon mutation (in bold), whichexchange predominantly through global unfolding;and those where DDGex displays an intermediatevalue, and presumably exchanges from a mixture of

local and global unfolding events. The mean valuesfor DDGex of the globally exchanging protons,determined by H/2H exchange are, within error,the same as the values determined by equilib-rium denaturation. (43–80, DDGex = −2.0 kcal mol−1,DDGU-F = −2.1 kcal mol−1; 85–102, DDGex = −4.2 kcalmol−1, DDGU-F = −4.3 kcal mol−1).

Table 1. Rate constants for exchange of N-H protons in barnase, wild-type and disulfide mutantsRate constant for exchange (kex)

p2H 6.3 p2H 7.5Location

hydrogen bonda Wild-type 43–80 85–102 Wild-type 43–80 85–102Residue (donor-receptor) (min−1)b (min−1)b (min−1) (min−1) (min−1) (min−1)

10 a1-a1 9.9 × 10−3 1.7 × 10−2 4.7 × 10−3 Not protected Not protected Not protected11 a1-a1 1.4 × 10−3 1.8 × 10−3 5.3 × 10−4 2.3 × 10−2 3.0 × 10−2 1.3 × 10−2

12 a1-a1 4.0 × 10−3 2.7 × 10−3 2.8 × 10−3 5.0 × 10−2 3.8 × 10−2 5.5 × 10−2

13 a1-a1 8.8 × 10−4 4.9 × 10−4 2.5 × 10−4 1.4 × 10−2 6.4 × 10−3 5.5 × 10−3

14 a1-a1 4.8 × 10−4 2.2 × 10−5 c 7.6 × 10−3 2.9 × 10−4 3.2 × 10−5

15 a1-a1 9.6 × 10−4 1.2 × 10−4 9.1 × 10−5 1.0 × 10−2 1.6 × 10−3 2.5 × 10−3

16 a1-a1 8.3 × 10−3 5.4 × 10−3 7.2 × 10−3 v.fast v.fast v.fast17 a1-a1 3.1 × 10−3 1.9 × 10−3 2.2 × 10−3 4.0 × 10−2 2.4 × 10−2 3.6 × 10−2

18 a1-a1 8.5 × 10−3 5.5 × 10−3 8.0 × 10−3 5.5 × 10−2 Poor data 6.0 × 10−2

19 Buried 2.3 × 10−3 8.8 × 10−5 2.2 × 10−5 1.0 × 10−2 1.1 × 10−3 8.0 × 10−4

25 Tertiary 3.4 × 10−4 1.8 × 10−5 5.2 × 10−7 7.8 × 10−3 2.9 × 10−4 3.6 × 10−5

26 a2-side chain 2.3 × 10−3 1.1 × 10−3 1.2 × 10−3 2.4 × 10−2 1.6 × 10−2 1.2 × 10−3

30 a2-a2 9.5 × 10−4 6.5 × 10−5 3.1 × 10−5 1.0 × 10−2 9.2 × 10−4 1.3 × 10−3

31 a2-a2 3.5 × 10−3 1.7 × 10−3 2.0 × 10−3 5.3 × 10−2 4.3 × 10−2 5.1 × 10−2

32 a2-a2 9.7 × 10−2 5.1 × 10−2 6.6 × 10−2 Not protected Not protected Not protected33 a2-a2 7.9 × 10−4 3.3 × 10−4 3.3 × 10−4 1.6 × 10−2 6.1 × 10−3 8.5 × 10−3

35 Turn-a2 4.3 × 10−3 4.0 × 10−3 4.1 × 10−3 6.9 × 10−2 7.5 × 10−2 5.8 × 10−2

36 Turn-side chain 2.5 × 10−2 4.0 × 10−2 2.7 × 10−2 Not protected Not protected Not protected44 a3-a3 2.2 × 10−2 Not protected 2.0 × 10−2 Not protected Not protected Not protected45 a3-a3 4.1 × 10−3 2.6 × 10−2 4.5 × 10−3 Not protected Not protected 1 × 10−2

46 a3-a3 6.7 × 10−4 1.3 × 10−4 1.0 × 10−4 1.1 × 10−2 2.5 × 10−3 3.4 × 10−3

49 Turn-a3 1.8 × 10−3 1.7 × 10−4 4.8 × 10−5 1.2 × 10−2 2.0 × 10−3 2.1 × 10−2

50 Tertiary 2.5 × 10−3 7.6 × 10−5 1.3 × 10−5 1.3 × 10−2 7.6 × 10−4 5.1 × 10−4

51 Buried 1.4 × 10−3 1.4 × 10−4 4.6 × 10−5 1.2 × 10−2 2.2 × 10−3 1.5 × 10−3

52 Tertiary 1.7 × 10−3 4.4 × 10−5 2.6 × 10−6 1.1 × 10−2 4.2 × 10−4 2.0 × 10−4

53 b1-b2 2.2 × 10−3 2.0 × 10−4 2.0 × 10−4 1.5 × 10−2 2.5 × 10−3 5.0 × 10−3

56 b1-b2 1.0 × 10−3 9.3 × 10−5 6.8 × 10−5 1.1 × 10−2 1.5 × 10−3 2.l × 10−3

71 Side chain-loop 9.5 × 10−3 7.3 × 10−3 7.8 × 10−3 Not protected Not protected Not protectcd72 b2-b3 1.2 × 10−3 4.0 × 10−5 1.9 × 10−6 8.7 × 10−3 2.5 × 10−4 7.4 × 10−5

73 b2-b1 1.3 × 10−3 4.2 × 10−5 8.3 × 10−7 9.6 × 10−3 3.3 × 10−4 6.3 × 10−5

74 b2-b3 4.8 × 10−4 2.2 × 10−5 c 7.6 × 10−3 2.9 × 10−4 3.2 × 10−5

75 b2-b1 1.1 × 10−3 3.7 × 10−5 Overlapping 9.6 × 10−3 2.9 × 10−4 Overlapping76 b2-b3 5.4 × 10−4 8.5 × 10−5 1.0 × 10−5 8.2 × 10−3 1.3 × 10−3 3.0 × 10−4

77 Tertiary 4.9 × 10−2 6.5 × 10−2 1.8 × 10−2 Not protected Not protected Not protected86 b3-loop Not protected Not protected 3.2 × 10−2 Not protected Not protected Not protected87 b3-b4 1.8 × 10−3 9.7 × 10−4 Overlapping 1.3 × 10−2 5.0 × 10−3 Overlapping88 b3-b2 5.4 × 10−4 2.6 × 10−5 c 8.1 × 10−3 2.2 × 10−4 2.8 × 10−5

89 b3-b4 2.7 × 10−4 1.5 × 10−5 c 7.2 × 10−3 1.8 × 10−4 1.8 × 10−5

90 b3-b2 6.2 × 10−4 1.6 × 10−5 c 8.3 × 10−3 2.3 × 10−4 2.8 × 10−5

91 b3-b4 1.8 × 10−3 5.2 × 10−5 1.3 × 10−5 1.0 × 10−2 5.3 × 10−4 4.2 × 10−4

94 b-turn 8.5 × 10−3 6.1 × 10−3 9.0 × 10−3 Not protected Not protected Not protected95 b4-side chain 5.2 × 10−4 2.5 × 10−4 3.3 × 10−4 1.2 × 10−2 4.3 × 10−4 6.8 × 10−3

97 b4-b3 3.1 × 10−4 1.4 × 10−5 c 7.8 × 10−3 1.9 × 10−4 1.9 × 10−5

98 b4-b5 1.1 × 10−3 5.4 × 10−5 1.9 × 10−6 8.7 × 10−3 5.7 × 10−4 9.5 × 10−5

99 b4-b3 2.1 × 10−3 1.7 × 10−4 1.6 × 10−6 1.1 × 10−2 1.3 × 10−3 9.1 × 10−5

101 Loop-side chain Not protected Not protected 1.4 × 10−2 Not protected Not protected Not protected102 Loop-b3 Not protected Not protected 1.6 × 10−2 Not protected Not protected Not protected103 Loop-side chain Not protected Not protected 1.7 × 10−3 Not protected Not protected 5.7 × 10−2

104 Loop-side chain Not protected Not protected 1.9 × 10−3 Not protected Not protected Not protected105 Loop-side chain Not protected Not protected 5.1 × 10−3 Not protected Not protected Not protected107 b5-b4 7.9 × 10−3 4.2 × 10−3 1.3 × 10−3 Not protected 8.4 × 10−2 4.0 × 10−2

a Hydrogen bonds observed in the crystal structure of wild-type barnase, maintained in the mutant crystal structures.b Data taken from Clarke et al. (1993).c Rate constant for exchange < 5.2 × 10−7 min−1. No decrease in intensity of crosspeaks after 17 weeks.

H/2H Exchange in Disulfide Mutants of Barnase508

Table 2. DDGex of N-H protons in barnase, wild-type anddisulfide mutants, determined at p2H 6.3

DDGex (kcal mol−1)a

Residue 43–80b 85–102b

10 0.33 −0.4611 0.15 −0.6012 −0.24 −0.2213 −0.36 −0.7814 −1.90 c

15 −1.28 −1.4516 −0.26 −0.0917 −0.30 −0.2118 −0.27 −0.0419 −2.01 −2.8625 −1.81 −4.4426 −0.45 −0.4030 −1.65 −2.1131 −0.44 −0.3432 −0.40 −0.2433 −0.54 −0.5435 −0.04 −0.0336 0.29 0.0544 −0.0645 1.14 0.0646 −1.01 −1.1749 −1.45 −2.2350 −2.15 −3.2451 −1.42 −2.1052 −2.25 −3.9953 −1.47 −1.4856 −1.46 −1.6671 −0.16 −0.1272 −2.10 −3.9773 −2.11 −4.5574 −1.90 c

75 −2.09 d

76 −1.14 −2.4677 0.17 −0.6287 −0.38 d

88 −1.87 c

89 −1.78 c

90 −2.25 c

91 −2.18 −3.0494 −0.20 0.0495 −0.45 −0.2897 −1.91 c

98 −1.86 −3.9299 −1.55 −4.42107 −0.39 −1.11

a DDGex values fall into three categories, DDGex 1 0 (italics),DDGex 1 DDGU-F (bold), and DDGU-F > DDGex > 0 (normal type).

b For 43–80, DDGU-F = −2.1 kcal mol−1; for 85–102,DDGU-F = −4.3 kcal mol−1.

c No decrease in intensity of crosspeaks after 17 weeks.d In 85–102 crosspeaks of 75 and 87 could not be deconvoluted.

change, upon the breaking of the protectivehydrogen bonds.

Exchange in the b-sheet

As shown in Figure 1, the rate constants ofexchange of protons located in the three centralstrands of the five-stranded b-sheet reflect theincrease in stability of the protein. This is areflection of the protons being largely buried: theprotein will have to unfold considerably, orcompletely, for the hydrogen bonds to break, and forsolvent to penetrate. The exchange rate constants inthe first strand are similar for 43–80 and 85–102,while they are higher in wild-type.

Local effects of disulfide bonds

A detailed comparison of the exchange kinetics ofthe three proteins at p2H 6.3 reveals that there arelocal effects of the disulfide bonds upon exchangebehaviour. These effects are not apparent at p2H 7.5,where most of the protons involved exchange toorapidly for exchange to be observed.

Mutant 43–80

Cys43 is located within a short turn of helix(41–46). It is apparent that a disulfide bondbetween 43 and 80, connecting this helix to anadjacent loop, affects the exchange behaviour ofprotons within that helix. As shown in Figure 2, therate constants for exchange of protons within thishelix are very similar in wild-type and disulfidemutant 85–102, demonstrating that the region isunaffected by an increase in overall stability and soexchanges during local breathing events. In 43–80,however, the protons of residues 44 and 45exchange more rapidly, the proton of 44 beingunprotected in the time scale of this experiment. Incontrast, the NH of residue 36, which is hydrogenbonded with the Od of residue 41, exchanges at asimilar rate in all three proteins. Studies of theintrinsic exchange rates of amide protons in un-structured peptides suggest that the substitution ofAla with a disulfide bond at residue 43 shoulddecrease the local intrinsic exchange rate (Bai et al.,1993).

In 43–80, the stability conferred by the disulfidebond (3.2 kcal mol−1) is not as large as that conferredby the disulfide bond at 85–102, although it enclosesa larger loop. We have suggested that the disulfidebond ‘‘strains’’ the folded structure, there is a smalllocal change in structure at the site of the mutation(Clarke et al., 1995). These exchange data may at leastpartly reflect the structural change upon mutation.Since the small structural rearrangements do notresult in any loss of hydrogen bonds, the resultspresented here suggest that the formation of the43–80 disulfide bond has decreased local stability,resulting in greater local unfolding, characterised byfaster H/2H exchange. The rearrangements ob-served may account for this loss in local stability.

Discussion

Exchange in the helices

The rate constants of exchange of the protons inthe two major helices (6–18, 26–34) are largelyunaffected by the overall stability of the protein(Figure 1) apart from a central residue in each helixwhich serves to anchor the helix into an adjacenthydrophobic core (residues 14 and 30 for helices1 and 2, respectively). During local structuralfluctuations (local ‘‘breathing’’), the relativelysolvent-exposed protons of the helices can ex-

H/2H Exchange in Disulfide Mutants of Barnase 509

Figure 1. Rate constants for exchange of protons in wild-type barnase and disulfide mutants 43–80 and 85–102 atp2H 6.3. The positions of the helices are indicated by open boxes, and the positions of the sheets by arrows. Apart from‘‘anchor’’ residues (14 and 30), the protons within the helices exchange at similar rates in all three proteins. The protonsof the three central strands of the b-sheet exchange at rates which reflect the overall stability of the protein. Note: theexchange rates of residues 75 and 87 could not be determined in 85–102, due to overlapping of crosspeaks.

A close comparison of exchange rates in the firsthelix (Figure 3) shows that the disulfide bond at43–80 has an effect on the local stability of the firstturn of the first helix. A comparison of wild-typeand 85–102 shows that not only are the exchangerates in the helix very similar, but the overallpattern of exchange rates is the same. In 43–80,however, the pattern of exchange rates is different.The exchange rate constants of the amide protons ofresidues 10 and 11, hydrogen bonded to the firsttwo residues of the helix, 6 and 7 respectively, arehigher, than those of wild-type and 85–102, relativeto the other exchange rates within this helix. Bycomparison, the exchange pattern in the secondhelix is the same for all three proteins. Theimplication of this is that the first helix isdestabilised by the disulfide bond at 43–80. Theobserved local destabilisation is associated with apositive DDGex (Table 2). There are no structuralchanges in this first helix in the crystal structure ofthe mutant (Clarke et al., 1995). The disulfide bondand the first turn of helix 1 are not close in thetertiary structure. However, although there is nostructural rearrangement in the first helix in themutant protein, the crystallographic B-factors of the43–80 mutant are higher than those of wild-typeand 85–102 in the regions where destabilisation isidentified by H/2H exchange (Figure 4). It is worthnoting that this observation would not have been asclear if 43–80 had been compared with wild-typealone, and demonstrates the value of comparingthree proteins in this study.

Mutant 85–102

85–102 is stabilised to a very considerable extentby the formation of a disulfide bond (4.8 kcalmol−1). This is more than predicted from thedestabilisation of the unfolded state (Pace et al.,1988). This suggests that there may be some

stabilisation of the folded state by the cross link, orthat the unfolded state is further destabilised(Clarke et al., 1995). The disulfide at 85–102 joins thestart of strand 3 of the five-stranded b-sheet to theend of strand 4. The loop adjacent to the cross link,101–105, has amide protons that are significantlyprotected in 85–102 (Figure 5). These amide protonsare unprotected in both wild-type and 43–80, whichis not surprising since the vast majority of protectedamide protons are within a-helices and b-sheets(Figure 1). Compared with other elements ofsecondary structure, loops are mobile, and struc-tural fluctuations will allow access to solvent.Further, whereas most of the protected amideprotons are involved in main chain–main chainhydrogen bonds, most of the hydrogen bonds thatprotect the amide protons in this 101–105 loop aremain chain–side chain hydrogen bonds. There is norearrangement in this region of the crystal structure(Figure 6; Clarke et al., 1995, accompanying paper).The disulfide bond has significantly increasedthe local stability of this loop, decreasing localunfolding. This change in local stability andincreased life-time of hydrogen bonds may accountfor the increased stability of 85–102. The crystallo-graphic B-factors show an increase in mobility inthis region in the mutant protein 85–102 (Figure 4).This is, however, associated with the loss of His102,which is involved in crystal packing interactions inthe wild-type protein. The decrease in the mobilityof this region, observed by hydrogen exchange,could not be inferred from the crystallographic data.

Previous comparisons of the exchange behaviourof wild-type and mutant proteins have not givensuch clear-cut results. Jandu et al. (1990) comparedhydrogen exchange of wild-type chymotrypsininhibitor 2 with a destabilised mutant. In contrast tothe results reported here, they found that thedestabilisation of the mutant was reflected through-out the protein. Kim et al. (1993) in their study of

H/2H Exchange in Disulfide Mutants of Barnase510

Figure 2. Exchange of NH protons protected byhydrogen bonds in helix 3 for wild-type (A), 85–102 (B),and 43–80 (C). The exchange rates of protons of wild-typeand 85–102 are similar, but the NH protons of residues44 and 45 exchange more rapidly in 43–80. For protonswith relatively slow exchange rates, the first 16 datapoints were not used to fit the data (see Materials andMethods).

Figure 3. Exchange of NH protons protected byhydrogen bonds in helix 1 for wild-type (A), 85–102 (B),and 43–80 (C). The pattern of exchange is the same inwild-type and 85–102, but in 43–80 the NH protons ofresidues 10 and 11 exchange relatively faster.

mutations. Our experiments were carried out inconditions close to those used for other experimen-tal work, which facilitates a direct comparison ofexchange behaviour and other data. The local effectsof the mutations could not be observed at p2H 7.5.

The results presented here demonstrate thatdisulfide bonds, as well as having a structural effect

highly destabilised mutants of BPTI saw no generalcorrelation between exchange rates and stability.The choice of conditions and the choice of mutantsis clearly crucial if one is to use hydrogen exchangestudies to report on structural or dynamic effects of

H/2H Exchange in Disulfide Mutants of Barnase 511

Figure 4. Main chain crystallographic B-factors for wild-type barnase and disulfide mutants 43–80 and 85–102.

upon a folded protein, can also affect its dynamicbehaviour. There can be global and local effects.The local effect of a disulfide bond may act tostabilise or to destabilise the folded protein. Theselocal dynamic effects will not be apparent whenstudying mutant crystal structures, nor whenstudied under conditions favouring exchangethrough a global unfolding mechanism. Thismethod could be extended to allow analysis of theeffect of other mutations upon the folded structureof a protein, by comparing the H/2H exchangebehaviour of mutant with wild-type protein. Atleast three proteins should be included in such astudy, to allow sensitive qualitative comparisons tobe made.

Materials and Methods

Materials

Deuterated chemicals, imidazole, 2HCl, and 2H2Oand 15N ammonium chloride were obtained fromAldrich Chemicals. Other chemicals and reagents wereobtained from BDH, Fisons, or Sigma.

Expression and purification of mutant proteins

The selection of the site of introduction of thedisulfide bonds and the method of expression andpurification of proteins, both unlabelled and uniformly15N-labelled, has been described previously (Clarke &Fersht, 1993; Clarke et al., 1993). The disulfide proteinswere shown to contain no free thiols following puri-fication, and no dimers were detectable by gel electro-phoresis (results not shown). The purified proteins weredialysed against several changes of de-ionised water andlyophilised.

NMR H/2H exchange experiments

The NMR exchange experiments, using wild-type and43–80 have been reported previously (Clarke et al., 1993).For assignment of 85–102 at pH 6.3 and 37°C, 20 mg oflyophilised 15N-labelled protein were dissolved in 50 mMdeuterated imidazole (pH 6.3) containing 90% H2O and10% 2H2O. For assignment of the proteins at pH 7.5, thebuffer was 50 mM deuterated Tris (pH 7.5) dissolved in90% H2O 10% 2H2O. 1H-15N HSQC spectra (Bax et al.,1990) were acquired on a Bruker AMX500 spectrometer,utilising 2 K complex data points over 4000 Hz in the 1Hdimension and 256 increments over 3000 Hz in the 15Ndimension. Both the 1H and 15N chemical shifts of thebackbone amide groups were assigned by comparisonwith 1H chemical shifts of wild-type barnase, and withpreviously assigned spectra of the mutant protein at pH4.5. Where ambiguities arose, these were resolved bytwo-dimensional X-filtered TOCSY and NOESY spectrausing the 15N-labelled sample.

The exchange buffer was 50 mM deuterated imidazolep2H 6.3 or 50 mM deuterated Tris p2H 7.5, containing0.05% sodium azide, dissolved in 100% 2H2O. All sampleswere dissolved on the same stock buffer. Approximately20 mg of lyophilised, uniformly 15N-labelled protein were

Figure 5. Exchange of NH protons in the loop residues100 to 105 in the disulfide mutant 85–102. None ofthese protons are protected in the time scale of theseexperiments in wild-type barnase or 43–80.

H/2H Exchange in Disulfide Mutants of Barnase512

Figure 6. Crystal structure of theloop 100 to 105 in wild-type barnase(red) and the disulfide mutant85–102 (coloured according to atomtype). The hydrogen bonds protect-ing the amide protons which havesignificantly lower exchange rates inthe mutant are shown.

dissolved in the exchange buffer, (final concentration03 mM), centrifuged, transferred to an NMR tube, andallowed to equilibrate in the magnet for ten minutesbefore the start of the first NMR experiment. The firstspectrum was recorded approximately 20 minutesafter dissolving the protein. The sample was kept at 37°C,in the NMR tubes, throughout the study, which lastedover four months. The final p2H, uncorrected, measuredusing a glass electrode, did not vary by more than 0.2 p2Hunits.

1H-15N HSQC spectra were acquired at regularintervals. In the first two hours, ten spectra were acquired(two scans per increment), followed by six spectra in thenext two hours (four scans per increment). Thereafter,spectra with a total acquisition time of 40 minutes (eightscans per increment), were acquired at increasingintervals, ranging from one hour to several weeks. About30 experiments were performed in the first 24 hours.

The spectra were zero-filled to 2 K × 1 K real pointsand were processed with a mild gaussian windowfunction in both dimensions. The volume integrals of thecross-peaks were calculated by the Bruker programUXNMR for each spectrum. The data were transferred toa Macintosh computer and the decays were fitted to asingle exponential, using the program KaleidaGraph(Abelbeck Software). Where two cross-peaks overlap inthe HSQC spectrum, the decay was fitted to a doubleexponential, and a series of 2D X-filtered TOCSY spectrawere run to ascertain which peak was decaying morerapidly. This allowed the assignment of a particularexchange rate to a specific amide proton. The first ten andsecond six data points were omitted when fitting slowdecays, as they were relatively noisy data points, withonly two and four scans per increment, respectively. Thisallowed a more accurate calculation of the rate constantof decay. In cases where the rate of exchange was veryslow, and the decay could not be fitted to an exponential,but where an initial rate could be determined, a rateconstant was calculated from the ratio of the initial rateto the estimated amplitude of the decay, assuming that allcrosspeaks will have the same final proton occupancy.

Quantitative analysis

Following a simple, two-state model of proton exchange(equation (1)) we have previously demonstrated (Perrettet al., 1995) that, at p2H 6.3, 37°C, protons in barnaseexchange by an EX2 mechanism (Hividt & Neilsen, 1966)where the rate of exchange depends upon the equilibriumbetween ‘‘open’’, exchange competent form (O), andclosed, protected form (C) equation (2):

CHghko

kc

OH04kint

D2OODgh

kc

ko

CD (1)

kex = K·kint (2)

where K = [C]/[O], kex is the observed rate constant ofexchange, and kint is the intrinsic exchange rate.

DDGex, the change in activation energy for exchangeon mutation, for each residue was determined usingequation (3):

DDGex = −RT ln kex

k'ex(3)

where kex is the rate constant for exchange of the protonin wild-type and k'ex is the exchange rate constant in themutant protein.

References

Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W. (1993).Primary structure effects on peptide group hydrogenexchange. Proteins: Struct. Funct. Genet. 17, 75–86.

Bax, A., Ikura, M., Kay, L. E., Torchia, D. A. & Tschudin,R. (1990). Comparison of different modes of2-dimensional reverse correlation NMR for the studyof proteins. J. Magn. Reson. 86, 304–318.

Betz, S. F. (1993). Disulfide bonds and the stability ofproteins. Protein Sci. 2, 1551–1558.

Betz, S. F. & Pielak, G. J. (1992). Introduction of adisulfide bond into cytochrome c stabilises acompact denatured state. Biochemistry, 31, 12337–12344.

H/2H Exchange in Disulfide Mutants of Barnase 513

Clarke, J. & Fersht, A. R. (1993). Engineering disulfidebonds as probes of the folding pathway of barnase:increasing the stability of proteins against the rate ofdenaturation. Biochemistry, 32, 4322–4329.

Clarke, J., Hounslow, A. M., Bycroft, M. & Fersht, A. R.(1993). Local breathing and global unfolding inhydrogen exchange of barnase and its relationship toprotein folding pathways. Proc. Natl Acad. Sci. USA,90, 9837–9841.

Clarke, J., Henrick, K. & Fersht, A. R. (1995). Disulfidemutants of barnase I: changes in stability andstructure assessed by biophysical methods and X-raycrystallography. J. Mol. Biol. 253, 493–504.

Creighton, T. E. & Goldenberg, D. P. (1984). Kinetic roleof a meta-stable native-like 2-disulfide species in thefolding transition of bovine pancreatic trypsininhibitor. J. Mol. Biol. 179, 497–526.

Doig, A. J. & Williams, D. H. (1991). Is the hydrophobiceffect stabilising or destabilising in proteins? Thecontribution of disulfide bonds to protein stability.J. Mol. Biol. 217, 389–398.

Englander, S. W. & Kallenbach, N. R. (1984). Hydrogenexchange and structural dynamics of proteins andnucleic acids. Quart. Rev. Biophys. 16, 521–655.

Gallagher, W., Tao, F. & Woodward, C. (1992). Compari-son of hydrogen exchange rates for bovine pancreatictrypsin inhibitor in crystals and in solution.Biochemistry, 31, 4673–4680.

Hividt, A. A. & Neilsen, S. O. (1966). Hydrogen exchangein proteins. Advan. Protein Chem. 21, 287–386.

Jandu, S. K., Ray, S., Brooks, L. & Leatherbarrow, R. J.(1990). Role of argenine 67 in the stabilisation ofchymotrypsin inhibitor 2. Biochemistry, 29, 6264–6269.

Kanaya, S., Katsuda, C., Kimura, S., Nakai, T., Kitakuni,E., Nakamura, H., Katayanagi, K., Morikawa,K. & Ikehara, M. (1991). Stabilisation of Escherichiacoli ribonuclease H by introduction of an artificialdisulfide bond. J. Biol. Chem. 266, 6038–6044.

Kim, K., Fuchs, J. A. & Woodward, C. K. (1993). Hydrogenexchange identifies native state motional domainsimportant in protein folding. Biochemistry, 32,9600–9608.

Matsumura, M. & Matthews, B. W. (1991). Stabilisation offunctional proteins by introduction of multipledisulfide bonds. Methods Enzymol. 202, 336–356.

Matsumura, M., Signor, G. & Matthews, B. W. (1989).Substantial increase of protein stability by multipledisulfide bonds. Nature, 342, 291–293.

Mauguen, Y., Hartley, R. W., Dodson, E. J., Dodson, G. G.,Bricogne, G., Chothia, C. & Jack, A. (1982). Molecularstructure of a new family of ribonucleases. Nature,297, 162–164.

Mitchinson, C. & Wells, J. A. (1989). Protein engineeringof disulfide bonds in subtilisin BPN'. Biochemistry, 28,4807–4815.

Pace, C. N., Grimsley, G. R., Thomson, J. A. & Barnett,B. J. (1988). Conformational stability and activityof ribonuclease T1 with zero, one and twointact disulfide bonds. J. Biol. Chem. 263, 11820–11825.

Pedersen, T. G., Sigurskjold, B. W., Andersen, K. V., Kjaer,M., Poulsen, F. M., Dobson, C. M. & Redfield, C.(1991). A NMR study of the hydrogen-exchangebehaviour of lysozyme in crystals and solution.J. Mol. Biol. 218, 413–426.

Perrett, S., Clarke, J., Hounslow, A. M. & Fersht, A. R.(1995). The relationship between equilibrium amideproton exchange behavior and the folding pathway ofbarnase. Biochemistry, 34, 9288–9298.

Radford, S. E., Buck, M., Topping, K. D., Dobson,C. M. & Evans, P. A. (1992). Hydrogen exchangein native and denatured states of hen egg whitelysozyme. Proteins: Struct. Funct. Genet. 14, 237–248.

Wagner, G. & Wutrich, K. (1982). Amide proton exchangeand surface conformation of BPTI in solution. J. Mol.Biol. 160, 343–361.

Wagner, G., Stassinopoulou, C. I. & Wutrich, K. (1984).Amide-proton exchange studies by two-dimensionalcorrelated 1H NMR in two chemically modifiedanalogs of the basic pancreatic trypsin inhibitor. Eur.J. Biochem. 145, 431–436.

Wetzel, R. (1987). Harnessing disulfide bonds usingprotein engineering. Trends Biochem. Sci. 12, 478–482.

Edited by R. Huber

(Received 24 May 1995; accepted 17 August 1995)