Hydrogen exchange in chymotrypsin inhibitor 2 probed by mutagenesis1

Hydrogen Exchange in Chymotrypsin Inhibitor 2Probed by Denaturants and Temperature

Laura S. Itzhaki, JoseÂ L. Neira and Alan R. Fersht*

MRC Unit for ProteinFunction and DesignCambridge Centre for ProteinEngineering, UniversityChemical Laboratory, Lens®eldRoad, Cambridge, CB2 1EWUK

Hydrogen exchange of chymotrypsin inhibitor 2 has been measured inthe presence of low concentrations of GdmCl and at different tempera-tures. The study of exchange at different temperatures allows us toobtain the activation enthalpies for the local exchange processes, and thechange in enthalpy between the closed, exchange-incompetent, forms andthe open, exchange-competent, forms. From the GdmCl dependence ofexchange, an m-value, which is a measure of the new surface areaexposed to solvent in the equilibrium between open and closed forms,can be determined for individual protons. This parameter therefore pro-vides information about the structural nature of the opening reactions. Inthe absence of denaturant, exchange from native and native-like statesdominates. As GdmCl concentration is increased, opening reactions thatinvolve global unfolding are selectively promoted for the majority ofamide protons. Three classes of protons emerge: for one set of protons,there is a linear and weak dependence on denaturant, indicating that thedominant opening reaction is the same throughout the range of GdmClconcentrations and involves local ¯uctuations with exposure of little newsurface. For another set of protons, the most slowly exchanging residues,a linear, but much stronger, denaturant dependence is observed. Forthese protons, global unfolding dominates, and the m-values are similarto that obtained by equilibrium GdmCl denaturation measured by ¯uor-escence under identical conditions. For the remaining protons, theGdmCl-dependence is weak at low GdmCl concentrations and increasesat higher GdmCl concentrations. No segment of sub-global unfoldingcould be identi®ed. Rather, all protons appear to merge together at highGdmCl concentrations to the global unfolding reaction.

# 1997 Academic Press Limited

Keywords: folding pathway; hydrogen exchange; protein folding; stability*Corresponding author

Introduction

The vast majority of protein molecules are in thenative state under native conditions, but there willbe a very small population in all the higher energystates, according to the Boltzmann distribution.Unfolded and partially folded states are dif®cult todetect under most experimental conditions, be-cause their populations are small and their signals,detected by most biophysical techniques, are hin-

dered by the signals of the abundant native state.Hydrogen exchange has the advantage that thenative state does not always contribute tothe measurement, since it is a hydrogen-exchangeincompetent state for the majority of the amideprotons.

Under conditions where the native state is thestable state, exchange occurs under the so-calledEX2 limit (Hvidt & Nielsen, 1966) and the observedrate constant for exchange is given by theequation:

kex � kop

kclkint � Kopkint �1�

where Kop is the apparent equilibrium constant forthe formation of the open state. Thus, hydrogen ex-

Abbreviations used: BPTI, basic pancreatic trypsininhibitor; CI2, chymotrypsin Inhibitor 2; GdmCl,guanidinium chloride; [GdmCl], concentration ofguanidinium chloride; HSQC, heteronuclear singlequantum coherence spectroscopy; NMR, nuclearmagnetic resonance.

J. Mol. Biol. (1997) 270, 89±98

0022±2836/97/260089±10 $25.00/0/mb971049 # 1997 Academic Press Limited

change rates can be converted directly to an appar-ent change in Gibbs free energy for the equilibriumbetween open and closed forms:

�Gappex � ÿRT ln Kop � ÿRT ln

�kex

kint

��2�

where R is the gas constant and T the temperaturein K. Both localised ¯uctuations of the native state,k1

ex and K1op, and global unfolding, kg

ex and Kgop, can

give rise to exchange competence. This has beenformalised in the two-process model (Woodward& Hilton, 1980):

�Gappex � ÿRT ln �K1

op � Kgop� �3�

The respective values of Kop will determine whichprocess dominates. Opening reactions may involvebreaking of single hydrogen bonds, conformational¯uctuations of segments of the chain, sizeable un-folding and global unfolding (Woodward &Hilton, 1979). Partially folded states may be acces-sible to, and could be probed by, hydrogen ex-change measurements (Miller & Dill, 1995). If theopening reaction involves an unfolding process,then �Gapp

ex should exhibit the standard linear de-pendence on concentration of chemical denaturant(Mayo & Baldwin, 1993; Bai et al, 1994., 1995; Qianet al., 1994):

�Gappex � �Gapp;0

ex ÿm�den� �4�as observed for the global unfolding of proteins(Pace, 1986), where m, the m-value for an individ-ual proton, is proportional to the new surfacearea exposed to solvent during the opening reac-tion of that proton and �Gapp,0

ex is the apparentfree energy of exchange at zero denaturant. Com-bining equations (3) and (4), and assuming thatlocal events in which exchange occurs from thefully folded state and which expose little newsurface area have no denaturant dependence,gives:

�Gappex � ÿRT ln

�K1

op � Kg;0op exp

�m�den�

RT

��5�

We use the term ``apparent change in Gibbs freeenergy'' because the values of kint are obtainedfrom model peptide studies (Bai et al., 1993) andthese may not be a good measure of the rateof chemical exchange if the proton is not fullyexposed to solvent in the state from whichexchange occurs. Regions of substructure in cyto-chrome c (Bai et al., 1995) and RNase H (Cham-berlain et al., 1996) could be identi®ed withindividual values of global �Gapp

ex and m; and itwas suggested that those regions represent coop-erative unfolding units.

Heat can be used instead of chemical denaturantto destabilise the protein. The observed depen-dence of the free energy of unfolding is (Privalov& Potemkhin, 1986):

�GUÿF ��HM ÿ T�SM

��Cp��T ÿ TM� ÿ T ln �T=TM��6�

where TM is the midpoint of the thermal transition(in K), �HM is the enthalpy and �SM is the entropyof denaturation (local or global) at the midpoint,and �Cp is the change in heat capacity. Equation(3) becomes:

�Gappop �ÿ RT ln �exp�ÿ�Gapp;1

op =RT�

� exp�ÿ�Gapp;gop =RT��

�7�

to take into account local and global pathways,where �Gapp,1

op and �Gapp,gop are given by equation

(6). It is usually assumed that �Cp � 0 cal molÿ1

Kÿ1 for the local ¯uctuations, so that the local ¯uc-tuations are given by the ®rst two terms ofequation (6).

Studies on mutants of CI2 have allowed us to as-sign which protons exchange by global unfolding,which by local ¯uctuations, and which by a mix-ture of the two pathways (Neira et al., 1997). Theexchange rates of a small number of protons are af-fected by mutation to the same extent as the globalstability of the protein, indicating that the openingreaction for these protons must require global un-folding. For other protons, the exchange rates arethe same in wild-type and mutants (whose stabilityvaries by up to 1.9 kcal molÿ1). These protonsmust, therefore, be exchanging through ¯uctu-ations of the native state ensemble. Other protonsexhibit behaviour between these two extremes.

Here, the exchange behaviour of wild-type CI2has been studied as a function of GdmCl concen-tration and of temperature, using conditions wellbelow the unfolding transition region. We obtainfrom these studies thermodynamic parameters ofthe hydrogen exchange opening reactions, whichcan be compared with those obtained from equili-brium unfolding studies of the protein. In particu-lar, the individual m-values may provide insightinto the structural nature of the opening reactions.We consider what information about partiallyfolded and unfolded states can be elucidated fromthe approach.

Results

Under some of the conditions used, the amidesignals of residues Lys24 and Leu49, and Asp55and Asn56, respectively, overlapped, and thereforean accurate measurement of kex could not be made.The exchange rates for the protons of Glu15, Asp23and Leu54 could not be measured accurately be-cause of the very small coupling constants.

GdmCl-dependence of hydrogen exchangeof CI2

Residues with measurable exchange rates are lo-cated in the a-helix and the most buried regions of

90 Stability in CI2 Measured by Hydrogen Exchange

the b-sheet. The great majority are hydrogenbonded, although some of them are protected byburial from solvent. In general, protons exchangefaster at higher denaturant concentrations. Theexchange rate constants vary by more than fourorders of magnitude at low GdmCl concen-trations (<10ÿ5 minuteÿ1 to 0.05 minuteÿ1;Figure 1). At 2 M GdmCl, there is a much nar-rower range of rate constants (0.003 minuteÿ1 to0.15 minuteÿ1). Representative examples of the ®tof the exchange data to equation (5) are shownin Figure 2(A). As [GdmCl] is increased, globalunfolding events become more important, andthis is manifested as a greater denaturant depen-dence and a convergence of �Gapp

ex .�Gapp

ex for residues Glu4, Gly10, Val13, Lys17,Ala27, Val34 and Gly64 decreased monotonicallythroughout the range of [GdmCl]. For these resi-dues, values for m and �Gapp,0

ex were obtained by®tting the data assuming a simple linear depen-dence of �Gapp

ex on [GdmCl] (Figure 2(B)). Values

of �Gapp,0ex and m obtained from the ®t are listed in

Table 1.The exchange rates for the amide protons of resi-

dues Lys11, Val19, Ile20, Leu21, Ile30, Leu32,Val47, Arg48, Leu49, Phe50 and Val51 are too slowto determine at the lower GdmCl concentrationused. For these residues, an m-value was deter-mined from a linear ®t of the data at higher[GdmCl] and �Gapp,0

ex was extrapolated from thesedata alone (Figure 2(B), Table 1).

The analysis requires that the mechanism of ex-change is EX2 throughout the whole range of de-naturant concentrations. Recent experiments withbarnase have shown that at 1.5 M GdmCl (306 K,pH 6.8) some protons deviate from EX2 (Clarke &Fersht, 1996), and this change in mechanism can,

Figure 1. �Gappex for CI2 at 33�C, pH 5.3 (acetate buffer)

in 0 M GdmCl, 2.5 M NaCl (upper panel), and 1.25 MGdmCl, 1.25 M NaCl (lower panel). The secondarystructure is indicated using a for an a-helix, b for a b-sheet and t for turn (Ludvigsen et al., 1991). Error barsare shown. Upward and downward-pointing arrows in-dicate residues whose exchange was too slow and toofast, respectively, to measure.

Figure 2. 1H/2H exchange behaviour for some amideprotons as a function of [GdmCl] measured at 33�C, pH5.3 (acetate buffer). (A) Examples of ®tting of theexchange data to equation (5). (B) Representativeexamples of amide protons displaying the three types ofbehaviour observed. 1. Protons that exchange by globalunfolding throughout the range of [GdmCl], for whichexchange rates could be ®tted (squares), show a lineardependence on [GdmCl]. 2. Protons that exchange bylocal ¯uctuations throughout the range of denaturantconcentrations (®lled circles), show a linear dependenceon [GdmCl]. 3. Protons that could be ®t to equation (5)(open circles). Error bars are shown.

Stability in CI2 Measured by Hydrogen Exchange 91

in fact, account for the effects of denaturant. Toaddress this possibility, exchange experiments at1.5 M GdmCl were performed at two values ofpH (pH 5.3 and 6.0). Figure 3 (A) shows a plotof logkex for individual residues at one pHagainst the value at another pH. The slope of theplot is close to 1, indicating that exchange isoccurring under these conditions with an EX2limit. A slope close to 1 is observed also when

residues that exchange by global, and those thatexchange by local or mixed exchange pathways,determined at 0 M denaturant, are plotted separ-ately (Figure 3 (B)).

Temperature-dependence of hydrogenexchange of CI2

Temperature-dependence experiments were per-formed in 0 M GdmCl. As for the hydrogen ex-change reaction at increasing [GdmCl], theexchange rates increase as the temperature is in-creased (Figure 4). Experiments were limited to amaximum temperature of 45�C, because above thistemperature aggregation occurred at the high con-centration used for NMR. This aggregation pro-blem at high temperatures (even in the presence oflow urea concentrations) precluded the accuratedetermination of TM and �HM for overall unfold-ing using CD-monitored thermal denaturation orcalorimetry (C. M. Johnson, J. L. N. & A. R. F., un-published results). The dominant exchange path-way for most protons is the local one throughout,

Table 1. GdmCl denaturation parameters in CI2 from1H/2H-exchange experiments

Residue

Hydrogenbondingacceptorc

Exchangepathway

(0 M GdmCl)d

m(kcal molÿ1

Mÿ1)�Gapp,0

ex

(kcal molÿ1)

Glu4a Mixed 0.3 � 0.1 2.5 � 0.3Trp5 Pro61 Mixed 2.3 � 0.5 4.6 � 0.2Leu8 Trp5 Local 2.8 � 0.8 4.9 � 0.2Val9 Pro6 Local 3.1 � 0.5 6.0 � 0.1Gly10a Ile57 Mixed 0.5 � 0.1 4.5 � 0.3Lys11b Leu8 Global 2.7 � 0.3 9.1 � 0.4Val13a Asp55 Local 0.6 � 0.1 4.4 � 0.2Ala16 Ser12 Local 1.3 � 0.4 4.3 � 0.2Lys17a Val13 Local 0.2 � 0.03 4.0 � 0.4Lys18 Glu14 Local 2.4 � 0.4 5.0 � 0.1Val19b Glu15 Mixed 2.8 � 0.3 9.4 � 0.8Ile20b Ala16 Global 2.8 � 0.3 8.9 � 0.6Leu21b Lys17 Global 3.3 � 0.2 9.4 � 0.4Gln22 Lys18 Local 3.2 � 0.3 5.8 � 0.1Ala27a Lys43 Local 0.5 � 0.06 4.1 � 0.3Gln28 Asp45 Local 1.6 � 0.6 5.0 � 0.1Ile30b Val47 Global 3.6 � 0.02 10.3 � 0.4Leu32b Leu49 Local 3.2 � 0.2 9.9 � 0.4Val34a Val51 Mixed 0.4 � 0.05 3.1 � 0.1Arg46 Arg43 Mixed 2.3 � 0.1 5.5 � 0.2Val47b Gln28 Global 2.8 � 0.03 9.3 � 0.05Arg48b Arg62 Mixed 2.5 � 0.3 8.8 � 0.6Leu49b Ile30 Global 2.7 � 0.5 8.8 � 0.6Phe50b Global 3.1 � 0.4 9.5 � 0.6Val51b Leu32 Global 2.8 � 0.2 8.7 � 0.4Asp52 Asn56 Local 2.9 � 0.5 5.2 � 0.3Leu54 Mixed 0.6 � 0.3 2.7 � 2.0Asn56 Lys53 Local 1.5 � 0.2 5.1 � 0.2Ile57 Lys11 Mixed 2.8 � 0.4 6.0 � 0.4Ala58 Phe50 Local 2.0 � 0.5 5.5 � 0.1Gln59 Phe50 Local 2.4 � 0.1 6.4 � 0.1Arg62 Mixed 2.8 � 0.5 6.1 � 0.3Val63 Thr3 Local 1.8 � 0.2 4.5 � 0.1Gly64a Arg46 Mixed 0.5 � 0.06 3.1 � 0.2

Experiments were perfomed at 33�C in 50 mM sodium acetatebuffer (pH 5.3). The total salt concentration was maintained at2.5 M throughout the range of GdmCl concentrations by addi-tion of NaCl. Values of m and �Gapp,0

ex were calculated by ®t-ting the exchange data at different [GdmCl] to equation (5).Fitting was performed only when kex could be measured formore than six [GdmCl].

a Exchange rates were ®tted assuming a linear dependence of�Gapp

ex on [GdmCl].b Slowly exchanging residues for which kex could not be mea-

sured at the lower [GdmCl]. For these residues a linear depen-dence of �Gapp

ex on [GdmCl] was assumed, and �Gapp,0ex was

obtained by extrapolation to 0 M GdmCl.c From Ludvigsen et al. (1991). All the hydrogen bonds are

between main-chain atoms.d Pathway of exchange at 0 M GdmCl, 33�C, determined by

comparing exchange rates of wild-type and mutant proteins(Neira et al., 1997). ��Gex � ��GU-F indicates global exchange,��Gex � 0 local exchange, and 0 < ��Gex < ��GU-F mixedexchange.

Figure 3. Plots of pH-dependence of exchange rates at33�C in 1.5 M GdmCl and 1. 0 M NaCl. The value of logkex at pH 5.3 is plotted against the value of log kex at pH6.0 for the same residue. (A) All residues; (B) residuesthat exchange by global (®lled squares) and local/mixed(open circles) pathways determined in 0 M denaturantare distinguished. Error bars are shown.


because of the small range of temperatures ex-plored, so thermodynamic parameters for onlythese events can be obtained. Thus, we do not ob-serve the hydrogen exchange curves (Figure 5)merging together at high temperatures as is ob-served at high GdmCl concentrations. The slowestexchanging amide groups, namely those of resi-dues Ile20, Leu21, Ile30, Val47, Leu49, Phe50 andVal51, which exchange by global unfolding, couldnot be investigated because there was insuf®cientdecay of signal to obtain rate constants.

We have therefore applied equations (6) and (7)assuming only local contributions to exchange. Fit-ting of the data was performed only for those resi-dues for which kex could be determined at morethan three temperatures. The �HM and �SM arelisted in Table 2. For residues Glu4, Trp5, Leu8,Glu15, Ala16, Val34 and Asn56, �Gapp

ex was con-stant over the temperature range within exper-imental error, and therefore �SM is close to 0 kcalmolÿ1 Kÿ1 for these protons.

Activation energies were determined by Arrhe-nius plots (Figure 6), using the equation:

ln�kex� � Cÿ EA=RT �8�where C is a constant and EA is the activation en-ergy, which is equal to the activation enthalpy,�H{, plus a factor that is equal to RT (Atkins,1994). The activation enthalpies are listed inTable 2.

Discussion

Relationship between m-values andprotein stability

The experimental conditions determine which oflocal or global exchange pathways dominates. Inthe absence of denaturant, the native state is themost stable state, and the fully unfolded state ispopulated only at a very low level, making littlecontribution to the exchange of most protons; onlyeight protons exchange by global unfolding at 0 Mdenaturant, as identi®ed by mutant studies (Neiraet al., 1997). For the remaining protons, local ex-change pathways dominate, and at low [GdmCl]only a weak denaturant dependence of �Gapp

ex isobserved. As [GdmCl] is increased, three types ofbehaviour emerge.

For one set of protons, those of residues Glu4,Gly10, Val13, Lys17, Ala27, Val34 and Gly64, thedominant opening reaction is the same throughoutthe range of [GdmCl], and involves local ¯uctu-ations with exposure of very little new surface: theplots (Figure 2(B)) of �Gapp

ex as a function of[GdmCl] are linear over the whole range and theslopes, m, are the smallest observed (0.2 to 0.6 kcalmolÿ1 Mÿ1, Table 1).

For another set of protons, those of the mostslowly exchanging residues Lys11, Val19, Ile20,Leu21, Ile30, Leu32, Val47, Arg48, Leu49, Phe50and Val51, a linear plot (Figure 2(B)) is also ob-served over the short range of high [GdmCl] at

Figure 4. �Gappex at 30�C (upper panel); 35�C (lower

panel), 0 M NaCl, pH 5.3. The secondary structure isindicated using a for an a-helix, b for a b-sheet and t forturn (Ludvigsen et al., 1991). Upward and downward-pointing arrows indicate residues whose exchange wastoo slow and too fast, respectively, to measure. Errorbars are shown.

Figure 5. 1H/2H exchange results for selected protons atpH 5.3 (acetate buffer), as a function of temperature.Error bars are shown.


which exchange rates could be measured. Theslopes of these plots were similar (2.6 to 3.6 kcalmolÿ1 Mÿ1), and the highest obtained. For theseprotons, global unfolding dominates.

For the remaining protons, the GdmCl-depen-dence of �Gapp

ex increased at higher [GdmCl],and the data were ®tted to equation (5)(Figure 2(A) and (B)). The m-values thus ob-tained range from 1.3 kcal molÿ1 Mÿ1 for Ala16to 3.2 kcal molÿ1 Mÿ1 for Val9. These m-valuesare de®ned by only a small number of pointsat the highest [GdmCl]. Further, some residuesmay not have reached a limiting slope of theGdmCl-dependence. Therefore, rather than re-¯ecting differential, sub-global stability, it islikely that the range of values of m observedfor these residues re¯ects the problems inherentin the data analysis.

Activation enthalpies of exchange in CI2

The activation enthalpies are in the range of 20to 40 kcal molÿ1. As stated above, only local ex-change processes could be investigated in the tem-perature study because aggregation occurred at the

higher temperatures under which global exchangeis favoured. According to the EX2 limit, the acti-vation enthalpy is given by:

�H zkop ÿ�H zkcl ��Hzkint � �Hop ��Hzkint

Thus, it can be considered as the sum of two con-tributions: one is the activation enthalpy for thechemical exchange reaction (�H{kint), and the otheris the enthalpy difference between the open andclosed forms (�Hop). �Hop obtained by subtractingthe value of �H{kint � 17 kcal molÿ1 (from modelpeptide studies; Bai et al., 1993) from �H{, arelisted in Table 2. These compare well with �HM

obtained from the plot of �Gappex versus temperature

(column 2 in Table 1). Comparison of �Hop with�HM for overall unfolding was not possible be-cause of the aggregation problems at high tempera-tures (C. M. Johnson, J. L. N. & A. R. F.,unpublished results). From model peptides, as thefree energy for formation of the hydrogen bond be-tween a proton and a carbonyl group is of theorder of 4 kcal molÿ1 (Creighton, 1993), we shouldexpect the smallest �Hop to be around 5 kcalmolÿ1. This is similar to the smallest value of�Hop, obtained for Asp23, which is 7 kcal molÿ1.

Table 2. Thermal parameters for CI2 obtained from 1H/2H-exchange experiments in50 mM sodium acetate, pH 5.3

�HlM

a �SlM

a �H{b �H{kop ÿ �H{kclc

Residue (kcal molÿ1) (kcal molÿ1 Kÿ1) (kcal molÿ1) (kcal molÿ1)

Glu4d 19 � 1Trp5d 24 � 7Trp (NH)d 15 � 6Leu8d 23 � 4Val9 14 � 2 0.031 � 0.007 30 � 2 13 � 2Gly10 12 � 4 0.02 � 0.01 28 � 5 11 � 5Val13 18 � 3 0.043 � 0.009 34 � 3 17 � 3Glu15d 20 � 2Ala16d 27 � 4Lys17 18 � 1 0.045 � 0.004 34 � 1 17 � 1Lys18 18 � 1 0.043 � 0.005 37 � 4 20 � 4Val19 16 � 2 0.033 � 0.006 35 � 5 18 � 5Gln22 15 � 5 0.03 � 0.01 31 � 5 14 � 5Asp23 8 � 2 0.014 � 0.001 24 � 2 7 � 2Lys24 11 � 3 0.019 � 0.01 23 � 4 6 � 4Ala27 16 � 3 0.04 � 0.01 32 � 3 15 � 3Gln28 25 � 3 0.06 � 0.01 39 � 1 22 � 1Leu32 22 � 1 0.05 � 0.003 41 � 3 24 � 3Val34d 27 � 4Arg46 21 � 2 0.048 � 0.006 37 � 2 20 � 2Arg48 14 � 1 0.026 � 0.002 30 � 1 13 � 1Asp52 20 � 4 0.05 � 0.01 36 � 4 19 � 4Leu54 25 � 9 0.07 � 0.03 41 � 9 24 � 9Asp55 14 � 2 0.036 � 0.006 30 � 2 13 � 2Asn56d 29 � 3Ala58 13 � 4 0.03 � 0.01 29 � 2 12 � 2Gln59 17 � 3 0.03 � 0.01 35 � 5 18 � 5Arg62 19 � 5 0.04 � 0.01 33 � 3 16 � 3Val63 20 � 2 0.05 � 0.01 36 � 3 19 � 3Gly64 17 � 2 0.04 � 0.005 33 � 2 16 � 2

a From the ®t of the exchange data at different temperatures to equation (6).b Obtained from Arrhenius plots of kex, for those residues for which more than three kex could be

measured.c The difference �H{ ÿ �H{kint, calculated assuming �H{kint � 17 kcal molÿ1 (Bai et al., 1993).d For these residues �Gapp

ex was constant over the temperature range within experimental error(�Sl

M � 0 kcal molÿ1 Kÿ1).


Hydrogen exchange activation enthalpies havebeen obtained for lysozyme (Thomsen & Poulsen,1993) and BPTI (Woodward & Hilton, 1980). Va-lues from 5 to 80 kcal molÿ1 were observed for ly-sozyme, for residues exchanging by both local andby global unfolding pathways. Similar results wereobtained for BPTI, again for both types of residues.

Comparison between ��Gappex and ��GU-F

Figure 7 (A) shows the linear ®t of the exchangedata versus GdmCl for Leu21, Phe50 and Val51.These are representative of the slowest exchangingprotons that were previously identi®ed as exchan-ging by global unfolding protons (Neira et al.,1997). Also shown are the values of �GU-F, in thetransition region, obtained from GdmCl denatura-tion experiments monitored by ¯uorescence in thesame buffer (broken line: Neira et al., 1997). The va-lues of �Gex for these protons coincide with theglobal denaturation data at 2 M GdmCl, which isjust before the transition region for unfolding. Thiscon®rms that the slowest protons do exchange bya global unfolding pathway. However, �Gex tendsto rise above the linearly extrapolated denaturationdata at lower [GdmCl]. The average value of �G0

ex

obtained by linear extrapolation of these data to0 M GdmCl is greater by 2 kcal molÿ1 than �G0

U-F

from the linearly extrapolated ¯uorescence dena-turation data, and approximately 1 kcal molÿ1

higher than �Gex estimated from exchange exper-iments performed in the absence of denaturant andsalt (Neira et al., 1997). The average m-value of 3kcal molÿ1 Mÿ1 obtained for these protons is alsohigher than that obtained through the transition re-gion of the denaturation experiment (1.9 kcalmolÿ1 Mÿ1).

There are two likely origins of these discrepan-cies. Firstly, the GdmCl-dependent exchange ex-periments were performed at constant ionic

strength (2.5 M salt), whereas the exchange exper-iments in the earlier work (Neira et al., 1997) wereperformed in the absence of salt. Secondly, it hasbeen suggested that extrapolation of GdmCl dena-turation data from the transition region to 0 M de-naturant is not linear, but in fact curves upwards(Aune & Tanford, 1969; Nozaki & Tanford, 1970;Shortle et al., 1989; Makhatadze & Privalov, 1992;Johnson & Fersht, 1995). This has been proposed asthe origin of discrepancy observed between �Gex

and �GU-F in cytochrome c (Bai et al., 1994). Inorder to assess these possibilities, we investigatedthe denaturation of a destabilized mutant of CI2,Ala16Gly/Leu49Ala (A. G. Ladurner, L.S.I. &A.R.F. unpublished results). The midpoint of dena-turation of this protein is 1.3 M GdmCl at 33�C,and so its unfolding transition region covers a

Figure 6. Arrhenius plots for selected protons of CI2 atpH 5.3 (acetate buffer), 0 M NaCl. Error bars are shown.

Figure 7. (A) Free energies for the slowest-exchangingprotons as a function of [GdmCl] at 33�C, pH 5.3. Datafrom GdmCl-induced denaturation experiment, moni-tored by ¯uorescence at the same pH and temperatureand in 2H2O (Neira et al., 1997), are also shown (opensquares) through the transition region together with thelinear ®t (broken line). (B) �Gapp

ex values showing differ-ent sets of protons as a function of denaturant concen-tration. Data from GdmCl-induced denaturationexperiment, monitored by ¯uorescence at the same pHand temperature and in 2H2O (Neira et al., 1997) areshown (broken line).


range of GdmCl similar to that over which the ex-change experiments were performed. Both the mid-point of denaturation (1.5 � 0.03 M) and the m-value (2.9 � 0.2 kcal molÿ1 Mÿ1) were greaterwhen the experiment was performed at a constantionic strength of 2.5 M salt than those obtainedwith no additional salt (1.3 � 0.01 M and2.4 � 0.1 kcal molÿ1 Mÿ1) (Figure 8). CI2, therefore,is stabilised by the presence of salt. The m-value atconstant ionic strength is very similar to that ob-tained from the GdmCl-dependence of the hydro-gen exchange experiments. The difference in �G0

U-F

of Ala16Gly/Leu49Ala in high and low salt is1.2 kcal molÿ1, which is similar to the difference in�Gex extrapolated to 0 M GdmCl from the GdmCl-dependent exchange experiments at high salt and�Gex determined in the absence of salt. It is im-portant to note, also, that the m-values forAla16Gly/Leu49Ala and other very destabilizedmutants of CI2 (Jackson et al., 1993; Itzhaki et al.,1995) tend to be larger than that for wild-type.This lends further support to a non-linear depen-dence of �GU-F on denaturant. To conclude, thediscrepancies in the free energies and m-values re-sult from non-linearity of the dependence on[GdmCl] and from differing ionic strengths used inthe experiments.

Relationship between denaturant effects onhydrogen exchange and structure and foldingpathway of CI2

Opening reactions that involve unfolding will bepromoted when [GdmCl] or temperature is in-creased. Subglobal folding units (i.e. residues thatmerge together at lower-lying isotherms beforejoining the higher-energy global unfolding iso-therm) could be identi®ed in cytochrome c (Baiet al., 1995; Bai & Englander, 1996) and RNase H(Chamberlain et al., 1996), using chemical or ther-mal denaturation. These subglobal folding units re-semble various intermediates in the kinetic folding

pathway of cytochrome c and RNase H. For bar-nase, the existence of partially folded states couldnot be addressed by the GdmCl-dependence of ex-change because of deviation from EX2 (Clarke &Fersht, 1996). Moreover, it was pointed out thatequilibrium measurement can provide informationonly about equilibrium stability and not about thepathway between different states, and that identi-cal hydrogen exchange behaviour would be ob-served whether an intermediate was on or off thekinetic folding pathway. Further, the series of sub-global folding units may or may not be sequen-tially related (Bai et al., 1995; Bai & Englander,1996).

Hydrogen exchange measurements can, in prin-ciple, detect very low populations of partiallyfolded states that would be invisible in equilibriumunfolding experiments measured by spectroscopicor calorimetric techniques. It appears, however,that CI2 is relatively cooperatively destabilized bydenaturant. Residues from the a-helix and differentstrands of the b-sheet merge simultaneously intothe global unfolding isotherm (Figure 7 (B)). Theredo not appear to be lower-energy isotherms thatwould indicate regions of sub-global unfolding.The aggregation problems at higher temperaturesprevent us from determining whether the beha-viour is the same when heat rather than chemicaldenaturant is used to destabilise the protein.

There is no intermediates populated either atequilibrium or on the folding or unfolding path-ways of CI2 (Jackson & Fersht, 1991). The single,rate-determining, transition state is weakly struc-tured, with only the a-helix being signi®cantlyformed (Otzen et al., 1994; Itzhaki et al., 1995). Thehydrogen exchange data indicate that, at the resi-due level, CI2 unfolds at equilibrium in a highlycooperative manner. No exchange-competent, par-tially folded, species can be detected from the hy-drogen exchange experiments.

Materials and Methods

Materials

Deuterium oxide (99% atom in 2H2O) was obtainedfrom Goss Scienti®c Instruments Ltd, and TSP from Al-drich Chemical Co. (Gillingham, UK); d3-acetic acid andd3-sodium acetate were purchased from SIGMA.15NH4Cl was from CK gas, UK. Standard suppliers wereused for all other chemicals and water was deionizedand puri®ed with an Elgastat system.

The exchange buffer was 50 mM sodium d3-acetate, in2H2O, pH 5.3. A stock solution of 7 M GdmCl in 2H2Owas prepared and adjusted to pH 5.3 with smallamounts of d3-acetic acid. A stock solution of 4 M NaClin 2H2O was prepared and adjusted to pH 5.3 with asmall amount of d3-acetic acid. The pH of all sampleswas measured at the beginning and end of every exper-iment using a Russell glass electrode. Prior to pHmeasurements, the electrode was calibrated with twostandards from Radiometer Copenhagen. No differencewas found between pH measurements at the beginningand the end of the exchange experiments. Values of pH

Figure 8. Equilibrium GdmCl denaturation of the CI2mutant Ala16Gly/Leu49Ala in 2H2O at 33�C, 50 mMacetate buffer (pH 5.3) in the absence of NaCl (open cir-cles) and in the presence of NaCl to maintain a total saltconcentration of 2.5 M (®lled circles).


reported here for 2H2O solutions represent apparent pHvalues, without correction for isotope effects.

For the exchange experiments at different [GdmCl],NaCl was added to maintain a constant total salt concen-tration of 2.5 M, in order to minimise possible salt-de-pendent effects at low GdmCl concentrations. No saltwas added in the temperature experiments.

Preparation and purification of proteins

15N-labelled wild type CI2 was expressed and puri®edas described (Neira et al., 1997). The mutant Ala16Gly/Leu49Ala was expressed and puri®ed as described (PratGay et al., 1995).

NMR spectroscopy

Samples were prepared and the exchange experimentsacquired as described (Neira et al., 1997). The GdmClexperiments were carried out at 33�C.

Assignment of 1H-15N HSQC experiments of wild-type CI2 under different conditions was carried out bycomparison of the observed chemical shifts with thereported assignments (Kjñr et al., 1987) and our owndata (unpublished results). Base plane differences havebeen neglected in rate determinations.

Hydrogen exchange data analysis

The volume integrals of the cross-peaks were calcu-lated and the decays ®tted as described (Neira et al.,1997). The rates obtained under different [GdmCl] were®tted to equation (5) using the program Kaleidagraph(Abelbeck Software) and are available from the authorson request.

Values of intrinsic rates, kint, for each exchangeableamide proton at different temperature were calculated asdescribed (Neira et al., 1997).

Acknowledgements

L. S. I. was supported by a Beit Memorial fellowshipin Medical Research. J. L. N. was supported by an E. C.Human Capital and Mobility fellowship. We are gratefulto Dr J. Clarke and Dr C. Johnson for advice.

References

Atkins, P. W. (1994). Physical Chemistry, 5th edit OxfordUniversity Press, Oxford.

Aune, K. C. & Tanford, C. (1969). Thermodynamics ofthe denaturation of lysozyme by guanidine hydro-chloride. II. Dependence on denaturant concen-tration at 25�C. Biochemistry, 8, 4586±4590.

Bai, Y. & Englander, S. W. (1996). Future directions inprotein folding: the multi-state nature of proteinstructure. Proteins: Struct. Funct. Genet. 24, 145±151.

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

Bai, Y., Milne, J. S., Mayne, L. & Englander, S. W.(1994). Protein stability parameters measured byhydrogen exchange. Proteins: Struct. Funct. Genet.20, 4±14.

Bai, Y., Sosnick, T. R., Mayne, L. & Englander, S. W.(1995). Protein folding intermediates: native-statehydrogen exchange intermediates. Science, 269,192±197.

Chamberlain, A. K., Handel, T. M. & Marqusee, S.(1996). Detection of rare partially folded moleculesin equilibrium with the native conformations ofRNase H. Nature Struct. Biol. 3, 782±787.

Clarke, J. & Fersht, A. R. (1996). An evaluation of theuse of hydrogen exchange at equilibrium to probeintermediates on the protein folding pathway. Fold.Des. 1, 243±254.

Creighton, T. E. (1993). Proteins, 2nd edit., W. H.Freeman and Co, New York.

Hvidt, A. & Nielsen, S. O. (1966). Hydrogen exchange inproteins. Advan. Protein Chem. 21, 187±386.

Itzhaki, L. S., Otzen, D. E. & Fersht, A. R. (1995). Thestructure of the transition state for folding of chy-motrypsin inhibitor-2 analysed by protein engin-eering methods: evidence for a nucleation-condensation mechanism for protein folding. J. Mol.Biol. 254, 260±288.

Jackson, S. E. & Fersht, A. R. (1991). Folding of chymo-trypsin inhibitor-2. 1. Evidence for a two-statetransition. Biochemistry, 30, 10428±10435.

Jackson, S. E., Moracci, M., elMasry, M., Johnson, C. M. &Fersht, A. R. (1993). Effect of cavity creating mu-tations in the hydrophobic core of chymotrypsininhibitor 2. Biochemistry, 32, 11259±11269.

Johnson, C. M. & Fersht, A. R. (1995). Protein stabilityas a function of denaturant concentration: the ther-mal stability of barnase in the presence of urea. Bio-chemistry, 34, 6795±6804.

Kjñr, M., Ludvingsen, S., Sorensen, O. W., Denys, L. A.,Kindler, J. & Poulsen, F. M. (1987a). Sequencespeci®c assignment of the proton nuclear magneticresonance spectrum of barley serine proteinase in-hibitor 2. Carslberg Res. Commun. 52, 353±362.

Ludvigsen, S., Shen, H., Kjñr,, M., Madsen, J. C. &Poulsen, F. M. (1991). Re®nement of the three-dimensional solution structure of barley serine pro-teinase inhibitor 2 and comparison with the struc-ture in crystals. J. Mol. Biol. 222, 621±635.

Makhatadze, G. I. & Privalov, P. L. (1992). Protein inter-actions with urea and guanidinium chloride. Acalorimetric study. J. Mol. Biol. 226, 491±505.

Mayo, S. L. & Baldwin, R. L. (1993). Guanidinium chlor-ide induction of partial unfolding in amide protonexchange in RNase A. Science, 262, 873±876.

Miller, D. W. & Dill, K. A. (1995). A statistical mechan-ical model for hydrogen exchange in globularproteins. Protein Sci. 4, 1860±1873.

Neira, J. L., Itzhaki, L. S., Otzen, D. E., Davis, B. J. &Fersht, A. R. (1997). Hydrogen exchange in a chy-motrypsin inhibitor 2 probed by mutagenesis. J.Mol. Biol. 270, 99±110.

Nozaki, Y. & Tanford, C. (1970). The solubility of aminoacids, diglycine, and triglycine in aqueous guani-dine hydrochloride solutions. J. Biol. Chem. 245,1648±1652.

Otzen, D. E., Itzhaki, L. S., elMasry, N. F., Jackson,S. E. & Fersht, A. R. (1994). Structure of the tran-sition state for the folding/unfolding of chymotryp-sin inhibitor-2 and its implications for mechanismsof protein folding. Proc. Natl Acad. Sci. USA, 91,10422±10425.

Pace, C. N. (1986). Determination and anlysis of ureaand guanidine hydrochloride denturation curves.Methods. Enzymol. 131, 266±280.


Prat Gay, G. de, Ruiz-Sanz, J., Neira, J. L., Itzhaki, L. S. &Fersht, A. R. (1995). Folding of a nascent polypep-tide chain in vitro: cooperative formation of struc-ture in a protein module. Proc. Natl Acad. Sci. USA,92, 3683±3686.

Privalov, P. L. & Potemkhin, S. A. (1986). Scanningmicrocalorimetry in studying temperature-inducedchanges in proteins. Methods. Enzymol. 131, 4±51.

Qian, H., Mayo, S. L. & Morton, A. (1994). Proteinhydrogen exchange in denaturant: quantitativeanalysis by a two process model. Biochemistry, 33,8167±8171.

Shortle, D., Meeker, A. K. & Freire, E. (1989). Effects ofdenaturants at low concentrations on the reversible

denaturation of staphylococcal nuclease. Arch. Bio-chem. Biophys. 272, 103±113.

Thomsen, N. K. & Poulsen, F. M. (1993). Low energy ac-tivation for amide hydrogen exchange reactions inproteins supports a local unfolding model. J. Mol.Biol. 234, 234±241.

Woodward, C. K. & Hilton, B. D. (1979). Hydrogenexchange kinetics and internal motions in proteinsand nucleic acids. Annu. Rev. Biophys. Bioeng. 8,99±127.

Woodward, C. K. & Hilton, B. D. (1980). Hydrogenisotope exchange kinetics of single protons inbovine pancreatic trypsin inhibitor. Biophys. J. 32,561±575.

Edited by J. Karn

(Received 23 December 1996; received in revised form 20 February 1997; accepted 18 March 1997)


Hydrogen exchange in chymotrypsin inhibitor 2 probed by mutagenesis1

Documents

Transcript of Hydrogen exchange in chymotrypsin inhibitor 2 probed by mutagenesis1