Protein Science (1997), 6:618427. Cambridge University Press. Printed in the USA. Copyright 0 1997 The Protein Society
Fast folding of cytochrome c
MICHAEL M. PIERCE
AND
BARRY T. NALL
Department of Biochemistry, University of Texas Health Science Center, San Antonio, Texas 78284-7760 (RECEIVEDAugust 20, 1996; ACCEPTEDNovember 22, 1996)
Abstract Native iso-2 cytochrome c contains two residues (His 18, Met 80) coordinated to the covalently attached heme. On unfolding of iso-2, the His I8 ligand remains coordinated to the heme iron, whereas Met 80 is displaced by a non-native heme ligand, His 33 or His 39. To test whether non-native His-heme ligation slows folding, we have constructed a double mutant protein in which the non-native ligands are replaced by asparagine and lysine, respectively (H33N,H39K iso-2). The double mutant protein, which cannot form non-native histidine-heme coordinate bonds, folds significantly faster than normal iso-2 cytochrome c: T = 14-26 ms for H33N,H39K iso-2 versus T = 200-1,100 ms for iso-2. These results with iso-2 cytochrome c strongly support the hypothesis that non-native His-heme ligation results in a kinetic barrier to fast folding of cytochrome c. Assuming that the maximum rate of a conformational search is about 10” s”, the results imply that the direct folding pathway of iso-2 involves passage through on the order of lo9 or fewer partially folded conformers.
Keywords: histidine; iso-2 cytochrome c; non-native heme ligands; protein folding; yeast
Native cytochrome c contains two residues (His 18, Met 80) coordinated to the covalently bound heme prosthetic group (Fig. 1). Upon denaturation near neutral pH, His 18 remains coordinated to the heme iron, whereas Met 80 is displaced by one of the remaining histidine residues. The presence of non-native heme ligands in unfolded cytochrome c is supported by spectral data clearly indicating the low spin state of the heme moiety, and suggesting that the non-native heme ligands in the unfolded protein are histidine residues (Babul & Stellwagen, 1971). The identity of the nonnative heme ligands was verified using NMR spectroscopy to monitor imidazole displacement of histidines for species variants of unfolded cytochrome c differing in histidine content. The NMR experiments identified the non-native heme ligands in unfolded forms of two yeast isozymes: His 26, His 33, and His 39 for iso-1; and His 33 and His 39 for iso-2 (Muthukrishnan & Nall, 1991). Similar experiments indicate that His 26 and His 33 make nonnative coordinate bonds to the heme in unfolded horse cytochrome c (K. Muthuknshnan & B.T. Nall, unpubl.). Thus, the unfolded states of yeast iso-2 and horse cytochrome c are similar because both proteins have two histidines that form non-native interactions with the heme: His 33 and His 39 for iso-2; versus His 26 and His 33 for the horse protein. The hypothesis that non-native heme ligation slows folding of cytochrome c is supported by experiments with the horse protein (Brems & Stellwagen, 1983; Eloveet al., 1994; Sosnick et al., 1994). The strategy is to compare the amplitudes and rates of ~
-
_
_
Reprint requests to: Barry T. Nall, Department of Biochemistry, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, Texas 78284-7760; e-mail:
[email protected].
folding under two sets of refolding conditions: low pH, where histidine protonation weakens coordination to the heme iron; and high pH, where histidines bind strongly to the heme. Under both sets of conditions, more than 80% of the folding occurs in two fast folding phases, with time constants in the 10-20-ms and 100500-ms time ranges. It is proposed that formation of native or native-like species of horse cytochrome c proceeds by a slower (“three-state”) pathway involving misfolded intermediates when refolding is from unfolded species with non-native histidine-heme ligands (Sosnick et al., 1996). However, at low pH, where histidine coordination is disfavored, the fraction of folding that occurs on the fastest timescale (10-20 ms) is greatly increased. This shift in amplitude to the fastest folding phase suggests that elimination of non-native histidine-heme interactions enables the majority of the molecules to fold on the fastest (10-20 ms) timescale (Elove et al., 1994; Sosnick et al., 1994). Additional evidence for the involvement of non-native His-heme ligation was provided by the addition of extrinsicligands to therefolding reaction (Brems & Stellwagen,1983). In the presence of imidazole, horse cytochrome c exhibited very rapid refolding ( I O ms), with 100% of the folding amplitude attributed to this fast kinetic phase. Originally, these results were interpreted as coupling of the denatured state (presence or absence of heme ligands) to peptide proline isomerization. With recent evidence, it seems likely that the presence of imidazole blocks the formation of a stable intermediate that presents a kinetic barrier to cytochrome c folding. Similar experiments with yeast iso-2 cytochrome c appear to _ _ give very different results (Nall et al., 1988). For iso-2 folding near neutral pH, the fastest ( I O ms) phase is missing: all fast folding occurs in a 100-1,100-ms time range. The number of iso-2 fast
618
Folding barriers
619
(2) mechanistic differences in folding result in different effectsof folding phases is uncertain. Absorbance-detected refoldingin the non-native heme ligationfor iso-2 and horse cytochromec. To test 100-1,100-mstimerange is well described by a single kinetic phase. However, better fits are obtained for fluorescence-detected whether incorrect heme ligation slows down folding of yeast iso-2 folding using two kinetic phases (2f and 2s) with time constants cytochrome c, we have constructed a mutant protein (H33N,H39K differing by only threefold. Fast folding is only weakly dependent iso-2) in which both potential non-native histidine-heme ligands on refolding conditions known to affect histidine-heme ligation. are replaced. His 18, a native heme ligand, is the only potential For refolding experiments ending at pH values between pH 3-9, histidine-hemeligandinunfoldedH33N,H39Kiso-2, so nonnative histidine-heme ligation is not possible. If non-native heme thefraction of fluorescence-detectedfastfoldingchanges very ligation is a kinetic trapfor cytochrome c folding, then the doublelittle (0.65-l), and the time range of fast folding is slower (80histidine mutant should refold more quickly than normal iso-2. The 600 ms) (Nall et al., 1988). Moreover, for iso-2, the fraction and rate of fast folding to native or native-like species is independent results show that H33N,H39Kiso-2 folds faster than iso-2by more of pH changes that alter non-native histidine-heme ligation in the than an order of magnitude. This strongly supportsthe hypothesis initial unfolded state of the protein (Osterhout & Nall, 1985). For that incorrect histidine-heme ligation slows down fast folding, but both iso-2 and horse cytochromec, very slow phases in folding in raises the question of mechanistic differences in folding of iso-2 and horse cytochrome c. the 10-100-s time range make up a final 0-35%of the kinetically detected folding (Ikai et al., 1973; Tsong, 1976; Nall & Landers, 1981; Nall, 1983). The very slow phases are known to involve Results proline isomerization (Linet al., 1988; Wood et al., 1988a, 1988b; Veeraraghavan & Nall, 1994). Equilibrium unfolding The results summarized above show that the folding reactions of iso-2 and horse cytochrome c respond very differently to changes Fluorescence of cytochrome c is related more directly to overall that affect heme ligation, despite the fact that both have unfolded chain dimensions thanfor most other proteins. Cytochromec constates with non-native histidine-heme ligands and three-dimensional tains a fluorescence donor, the Trp residue at position 59 (mamstructures with almost identical backbone folding (Bushnellet al., malian numbering), and a fluorescence quencher, the heme attached to Cys 14 and Cys 17. The presence of the Trp 59-heme donor1990; Murphy et al., 1992). There are two alternate explanations for the differences in folding: (1) the hypothesis that non-native quencher pair results in Trp fluorescence from the protein that is histidine-heme ligation interferes with fast foldingis incorrect, or quenched by Foster resonance energy transfer to the heme. The
Fig. 1. Ribbon diagramof yeast iso-2 cytochrome c (Murphy et al., 1992; Murphy, 1993). The covalently bound heme prosthetic group is shown in red. Covalent attachment of the heme to the polypeptide chain is by two thioether bonds from Cys 14 and Cys 17, which are shown in blue. The two residues coordinated to the heme iron in native cytochrome c, Met 80 and His 18, are shown in green and yellow, respectively. Residues targeted for site-directed mutagenesis,His 33 and His 39, are shown in yellow. In unfolded cytochromec at neutral pH, the Met 80-heme bond is broken when His 33 or His 39 forms a coordinate bond to the heme, displacing Met 80.
620
M.M. Pierce and B.7: Nall
quenching efficiency, Q,for a static distance, r, between randomly oriented donor-quencher pairs is given by Equation 1 (Cantor & Schimmel, 1980):
Table 1. Thermodynamic stability of iso-2 and H33N,H39K iso-2 a G d H 2 0 )
L! = R;/(RB
+ r6),
(1)
where Ro, estimated to be 30 8, in water (Stryer, 1959), is the static distance that gives 0.5 of the fluorescence of an unquenched Trp. Folded cytochrome c, with a Trp-heme distance of about I O A, has almost no fluorescence. Unfolding leads to a large increase in fluorescence as the Trp-heme distance increases and fluorescence quenching is reduced (Tsong, 1974, 1976). Note that quantitative prediction of the quenching efficiency for an unfolded protein requires an ensemble average of Q over the distribution of donor-quencher distances, r, whereas prediction of the fluorescence of the folded protein requires knowledge of the relative orientations of the donor-quencher transition dipoles. Gdn-HCI-induced unfolding of iso-2 and H33N,H39K iso-2 gives an increase in fluorescence to about 0.5 of that of an unquenched Trp, showing that the Trp-heme distance in both unfolded proteins is of the order of Ro = 30 A (Fig. 2). Thetransitions are similar for the mutant and normal proteins, indicating similar stabilities. One difference is that the total fluorescence change for unfolding at equilibrium may be slightly larger for H33N,H39K iso-2. This is expected because non-native His-heme ligation will reduce the donor-quencher distance in unfolded iso-2. but not in H33N,H39K iso-2. Two-state analysis of the transitions using the linear extrapolation model (Pace, 1986; Santoro & Bolen. 1988) gives unfolding free energies in the absence of denaturant, hCi'~(H20), of 4.5 f 0.4 kcal mol" and 6.3 f 0.4 kcal mol" and transition midpoints, C,, of 1.5 -t 0.2 M and 1.5 f 0.2 M, for iso-2 and H33N,H39K iso-2, respectively (Table I ) . The most notable difference in the transitions is that the m-values, which measure the dependence of the free energy on Gdn.HCI concentration, differ significantly. The m-value for iso-2 is 3.0 f 0.2 kcal mol - I M-I, but that for
"O
(kcal mol")
Protein ______
~ ~ _ _ _ _ _ _
Iso-2 H33N.H39K iso-2
Fig. 2. Gdn-HC1 equilibrium unfolding transitions for iso-2 (filled squares) and H33N,H39K iso-2 cytochrome c (filled circles) at 20°C. Fluorescence intensities of protein solutions at 350 nm following excitation at 285 nm are measured relative to an equal molar concentration of NATA and plotted as a function of the Gdn-HC1 concentration. Protein concentration is 5 X IO-' M in 0.1 M sodium phosphate, pH 6.0. Solid lines represent nonlinear least-squares fits according to a two-state linear extrapolation model. AGE(H20) and m-values obtained from the fits are given in Table 1.
(M)
3.0 rf- 0.2 4.3 c 0.2
1.5 rf- 0.2 1.5 f 0.2
~
4.5 + 0.4 6.3 rf- 0.4
'Parameters obtained from fitting the fluorescence-detected equilibrium unfolding transitions shown in Figure 2 to a two-state linear extrapolation model. Conditions are 0.1 M sodium phosphate, pH 6.0, 20°C.
H33N,H39K iso-2 is 4.3 k 0.2 kcal mol M". The m-value is believed to measure changes in solvent-accessible surface on unfolding (Myers et al., 1995), and the lack of non-native His-heme ligation in H33N.H39K iso-2 may lead to a more solvent-accessible unfolded state. The larger m-value for H33N,H39K iso-2 results in a larger standard state stability, AG6(H20), for the mutant protein than for normal iso-2, even though both proteins have the same transition midpoints, C,. The removal of non-native histidines and subsequently the removal of some residual structure in unfolded cytochrome c may destabilize the unfolded form relative to the nativeconformation,providing an explanation for the larger AGC,(H20) observed for H33N,H39K iso-2. Charucterization of heme ligation states in the unfolded proteins
Heme ligation states of unfolded forms of cytochrome c have been inferred from comparisons of the visible absorbance spectra of unfolded cytochromes c and the absorbance spectra of model hemecontaining peptides (Babul & Stellwagen, 197 1 , 1972). The heme ligand states determined by absorbance spectrophotometry have been verified and extended to the yeast is0 cytochromes c by proton NMR-detected imidazole titrations of the unfolded proteins (Muthukrishnan & Nall, 1991). Thus, visible absorbance spectrophotometry is known to be a reliable means of assessing heme ligation states in unfolded cytochromes c. Figure 3 compares heme absorbance spectra of unfolded forms of iso-2 and H33N,H39K iso-2 at neutral and acid pH. The results aresummarized in Table 2, which assigns the heme absorbance maxima to the heme ligation states. Both proteins lack intrinsic protein heme ligands at pH 2, as shown by high-spin heme absorbance spectra and the Soret maximum at 398 nm. At pH 6, unfolded H33N,H39K iso-2 has the 620-nm absorbance band of the high-spin heme expected for a protein with insufficient histidines to fill all heme ligand sites, but the partially red-shifted Soret maximum (401 nm) indicates that His 18 is a heme ligand. The addition of imidazole to unfolded H33N,H39K iso-2 at pH 6 induces a spin-state transition as observed by the disappearance of the 620-nm absorbance band and the shift of the Soret maximum from 401 nm to 407 nm. This simulates the unfolded absorbance spectrum of iso-2, where the fifth and sixth coordination sites of the heme iron are occupied by strong field ligands. Iso-2 at pH 6 has a low-spin heme absorbance spectrum (eg., no 620-nm band) and a Soret maximum at 407 nm, which shows that histidines are bound at both the fifth and sixth coordination sites in the unfolded protein (Babul & Stellwagen, 1971). Proton NMR-detected imidazole titrations have shown that differences in heme ligation lead to at least two unfolded forms of iso-2 at neutral pH (Muthukrishnan & Nall, 1991): one with His 33
1
[Gdn.HCI] (M)
C,
m (kcal mol" M")
5
62 1
Folding barriers Refolding rates and non-native His-heme ligation 1.5x105
1 .0x105
5 . 0 ~O4 1
1 . , . . . . . 1
0.0
380
390
400
41 0
420
4.0~10~
3.0~10~
2.0~10~
1 .ox1o3
650
625
600
0.0 575
700
Wavelength (nm) Fig. 3. Visible absorbancespectrafor unfolded iso-2and unfolded H33N,H39K iso-2 cytochrome c at 25 "C. A: Unfolded absorbance spectra showing the Soret maxima for 15 p M protein solutions in 4 M Gdn.HC1, 0.1 M sodium phosphate. Starting with the uppermost spectrum at 390 nm, the spectra are: iso-2, pH 2; H33N,H39K iso-2, pH 2; H33N,H39K iso-2, pH 6; iso-2, pH 6; H33N,H39K iso-2 with 1 0 0 mM imidazole, pH 6. B: Unfolded absorbance spectra of the 620-nm band for 60 p M protein solutions in 4 M Gdn*HCI, 0.1 M sodium phosphate. Starting with the uppermost spectrum at 625 nm, the spectra are: iso-2, pH 2; H33N,H39K iso-2, pH 2; H33N,H39K iso-2, pH 6; iso-2, pH 6; H33N,H39K iso-2 with 1 0 0 mM imidazole, pH 6.
and His 18 heme ligands, and the other with His 39 and His 18 heme ligands. Thus, in unfolded iso-2 near neutral pH, His 18 is always a ligand, whereas His 33 and His 39 take turns as the other heme ligand.
Removal of the histidines capable of forming non-native ligand bonds has a dramatic effect on the rates of refolding measured by stopped-flow mixing. As shown in Figure 4 and Table 3, the dominant kinetic phase in refolding, phase 2f, is much faster forfolding of H33N,H39K iso-2. This kinetic phase is the fastest kinetically resolved phase in folding, and accounts for more than 70% of the fluorescence-detectedsignal change for equilibrium unfolding. Phase 2f in folding of H33N,H39K iso-2 is about 12-fold faster than the analogous phase 2fin folding of iso-2 (Table 3). The fast 1526-ms folding phase measured for H33N,H39K iso-2 was observed for both fluorescence and absorbance-detected folding. The time constant for phase T~~ was similar (Table 3, and data not shown) for absorbance-detected folding measured by probes of heme environment (418 nm and 394 nm) and spin state (620 nm). Folding monitored by absorbance at 695 nm, characteristic of Met 80 heme ligation, was slightly slower compared to other folding probes ( T =~ 26 ~ ms at 695 nm compared to T~~ = 14-18 ms for all other probes). Even so, the value of T~~ for H33N,H39K iso-2 folding followed at 695 nm is 8-42-fold faster than the T~~ phase observed for iso-2 cytochrome c by any of the optical probes (Table 3). Two additional minor phases are also detected. A minor phase (T*\) in the 330-830-ms time range makes up only 0.03 of the total kinetically detected signal change. This phase may be a remnant of the fast phases in iso-2 folding ( T or~ T~~ \ )that occur in the 2001,100-ms time range. It could also be a phase present in folding of normal iso-2 ( T ~ that ~ ) is revealed more clearly in folding of H33N,H39K iso-2 because the major fast phase ( T ~ is~ shifted ) to a faster time range. Alternatively, the minor T~~ phase may be unique to folding of H33N,H39K iso-2. The slowest phase ( T ~ in ) folding of H33N,H39K iso-2 is in the 13-32-s time range, and has a small relative amplitude,asdoes phase T~ in iso-2 folding (Table 3). The slow ( T ~ phases ) in absorbance-detected ( T ~ and ~ ) fluorescence-detected ( T ~ , , ) folding of iso-2 are known to involve proline isomerization (Veeraraghavan & Nall, 1994). Absolute amplitudes for fluorescence-detected folding reactions were determined relative to an N-acetyltryptophanamide (NATA) standard and compared to the fluorescence signal change observed for equilibrium unfolding (Table 3). The sum of the absolute amplitudes for all three kinetically detected phases gives the total kinetically detected signal change. For iso-2, the total kinetically
Table 2. Comparison of unfolded absorbance spectra at high and low pH for iso-2 and H33N,H39K iso-2 cytochrome ca Protein (conditions) Unfolded Unfolded Unfolded Unfolded Unfolded
iso-2 (pH 6) H33N,H39K iso-2 (pH 6) iso-2 (pH 2) H33N,H39K iso-2 (pH 2) H33N,H39K iso-2 (100 mM Imidazole, pH 6 )
Soret Amax enmax (nm) (M" cm") 407 401 398 398 407
130 X 10' 144 X I O 3 135 X lo3 152 X IO3 123 X 10'
emnrn (M" cm") 1.0 X 2.7 X 2.8 X 2.9 X 7.0 X
IO' IO' 10' IO' 10'
Spin state
Heme ligands
Low Mixed High High Low
His 18/His 33 or His 39 His 18/H20 HzO/HzO HzO/H20 His 18/Imidazole
"Absorbance spectra were taken as described in Figure 3. All protein solutions were in 4 M Gdn.HC1, 0.1 M sodium phosphate, at the indicated pH. For spectra in the Soret region (380-420 nm), protein concentrations were 15 pM. For spectra in the region 575-700 nm, protein concentrations were 60 p M .The spin state and heme ligation were determined from A,,, of the heme Soret peak and the presence or absenceof the 620-nm absorbance band characteristic of high spin-state hemes (Babul & Stellwagen, 1971). Mixed spin refers to a distribution of high and low spin states with a Soret A,, intermediate between spin states.
622
M.M. Pierce and B.7: Nall
I
0.6 I
I
I
0.4
I
I
Q,
0.2
I
I
c Q, I.2
0.0
u
J 0.0
!
- .o~ - - " ~ " - - H33N.H39K is02
" " "
0.5
1
1.5
2.0
esting to compare these results to horse-heart cytochrome c, which exhibits a significant loss of amplitude within the dead time of stopped-flow mixing for folding at low Gdn.HC1 concentrations. The original interpretation of the results with horse cytochrome c suggested that a folding intermediate, stabilized at lower Gdn-HCI concentrations, formed very rapidly (Elove et al., 1992). Because the burst phase amplitude (ao) is similar for H33N,H39K iso-2 and iso-2, the main difference between the observed kinetics in the presence and absence of non-native His-heme ligation is in the with little rate of the fastest kinetically resolved folding phase, rZf, or no change in the amplitude (ao) of the (kinetically unresolved) "burst" phase.
0
3
i i
p H Dependence of refolding rates
0.10
U
a
N .-
2b
0.05
z 0.00 25
100 50
75
Time (sec) Fig. 4. Comparisonoffoldingkineticphases for iso-2(uppertraces in each panel) and H33N,H39K iso-2 (lower traces in each panel). A: First 2 s of folding, in which 80-90% of the kinetically detected folding occurs. B: First 100 s of folding, in whichslowerfoldingreactionsinvolving proline isomerization occur. A IO-ms/point sampling rate and IO-ms analoguefilterwereusedfortheuppertrace in A. Thelowertrace in A combines two traces collected simultaneously at two different time bases: the first 0.4 s of folding are sampled at 0.5 ms/point with a 0.2-111s filter, and the remainder (to 2.0 s) is sampled at I O ms/point with a IO-ms filter. Bothtraces shown in B are obtained by combiningtwotracescollected simultaneously at two different time bases. For both upper and lower traces of B, the first I O s of folding are sampled atI O ms/point with a 10-ms filter, andtheremainder (to 1 0 0 s) issampledat 1 0 0 ms/point with a 100-ms filter. Refoldingwas initiated by a 1.9 dilution of an unfolded protein stock (250-300 p M in 4 M Gdn-HCI,0.1 M sodium phosphate, pH 6.3) with refolding buffer (0.1 M sodium phosphate, pH 6.0). Final conditions were 25-30 p M protein, 0.4 M Gdn-HCI, pH 6.0. Fluorescence emission was measured through a 360-nm bandpass filterwith excitation at 285 nm. For all kinetic traces, data were normalized to the first measured data point.
detected signal change for refolding (4.0 M Gdn-HCI initial to 0.4 M Gdn-HC1 final) was 0.64 f 0.05 of the signal change observed from the Gdn-HC1 equilibrium unfolding transition. The sum of the absolute amplitudes for the kinetically detected folding phases for H33N,H39K iso-2 was 0.80 f 0.10 of the equilibrium change in fluorescence. Systematic errors in measuring absolute amplitudes are difficult to estimate, and may be large. Total errors could exceed the statistical errors given in Table 3, and the small difference in the total kinetically detected amplitude may not be significant. Thus, the mutations have little or no effect on the amount of kinetically detected folding observed by stopped-flow mixing. These observations suggest that the primary effect of the mutations is that amplitude shifts from the 200-1,100-ms fast phase 7 2 f andlor rzS observed for iso-2 to the 15-26-ms phase r 2 f observed for H33N,H39K iso-2. Small decreases in the amplitudes of the other kinetic phases (phases r2sand 7,) also occur. It is inter-
Figure 5 shows the effects of varying the final pH on the amplitude and rate of the fastest phase detected by stopped-flow mixing (phase 2f). Over the pH range investigated, the refolding rates for H33N,H39K iso-2 and iso-2 have a similar pH dependence, although the rate of folding of the double-mutant protein remains 6-15-fold faster than iso-2. The folding rates for both proteins decrease slightly (2-5-fold) as pH is raised from pH 5 to pH 7. The pH dependence of rZfappears to be slightly greater for iso-2 than for H33N,H39K iso-2. Previously, the final pH dependence of folding of iso-2 has been studied over a wide pH range and folding monitored by both absorbance and fluorescence probes (Nall et al., 1988). The results showed that the rate of fast folding reached a minimum near neutral pH and increased slightly as pH is raised above pH 7. The pH dependence was interpreted previously as indicating the involvement of ionizable groups that alter the stability of folding intermediates. This interpretation is compatible with the present results, but at least part of the pH dependence must result from groups other than His 33 and His 39.
Discussion Equilibrium unfolding The Gdn-HCI-inducedunfolding transitions for H33N,H39K iso-2 and iso-2 coincide in the pretransition region, but, on passing through the transition region, the transitions diverge in a manner that shows that H33N,H39K iso-2 is more sensitive to Gdn-HCI than iso-2. Although H33N,H39K iso-2 is slightly less stable than iso-2 in the transition region, the linear extrapolation model (Pace, 1986; Santoro & Bolen, 1988) shows that the mutant protein is more stable than iso-2 in the absence of denaturant. The greater stability results from a stronger dependence of ACu on Gdn.HC1 (larger m-value; Fig. 2; Table 1) in the transition region. Regardless, both proteins are well into the unfolded protein baseline region by 4.0 M Gdn-HC1, the initial unfolding condition for the kinetic experiments, and both proteins are fully folded at 0.4 M Gdn-HC1, the final conditions of the refolding kinetic experiments (Table 3; Fig. 4). It is likely that the removal of non-native heme ligands from iso-2 cytochrome c destabilizes the unfolded protein, thus increasing the unfolding free energy. The larger m-value measured for H33N,H39K iso-2 supports this hypothesis. Because the m-value is believed to be a measure of the change in solventaccessible surface on unfolding, removal of non-native heme ligands may further ''loosen'' the unfolded protein, increasing the amount of solvent-exposed surface. Similar mutations have been constructed recently for rat cytochrome c, which, like horse cytochrome c, contains histidine res-
623
Folding barriers
Table 3. Kinetic parameters for cytochrome c folding a Kinetic phaseb,‘ probe
Protein
[a01
Iso-2 F(350 nm)
70
a2f
72f
azs
72s
a1
71
(ms)
[azfl
(ms)
[a2s3
(4
[a1
I
(4