Protein Refolding - American Chemical Society

---

Chapter 4

Structure and Stability of Cytochrome c Folding Intermediates Gülnur A. Elöve and Heinrich Roder

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

Department of Biochemistry and Biophysics, University of Pennsylvania, Philadelphia, PA 19104-6059

Hydrogen exchange labeling and nuclear magnetic resonance (NMR) approaches were used to elucidate the folding mechanism of horse cytochrome c (cyt c). The development of hydrogen bonded structure during refolding was observed by pulse labeling at variable refolding times, and the degree of protection from N H exchange was probed by systematic variation of the labeling pH. The results show that the folding reaction involves both sequential and parallel pathways. About 50% of the molecules follow a sequential pathway where a partially folded intermediate is formed in a 20 ms folding phase. This intermediate has two native-like helices near the chain termini, but lacks stable Η-bonded structure in other parts of the molecule. Amide sites on either helix are not only protected at the same rate, but they also exhibit the same degree of protection, confirming previous evidence that association of the N - and C-terminal helices is an important early event in cyt c folding. Subsequent folding events on the 100 ms time scale involve replacement of a non-native histidine heme ligand by the native methionine ligand. In addition, there is evidence for a minor species of very rapidly folding molecules that form native-like structure within the mixing dead-time, as well as slow-folding forms that take several seconds to fold. Much has been learned in recent years about the folding process of small globular proteins, especially through application of new experimental approaches such as two dimensional nuclear magnetic resonance (2D NMR) and site-directed mutagenesis (1-4). However, we are still far from a detailed understanding of how the three-dimensional structure of a protein is encoded in its ammo acid sequence. Once mainly of academic interest, the protein folding problem has found new motivation from recent developments in protein engineering and biotechnology. For a mechanistic understanding of protein folding, it is necessary to determine the sequence of intermediate states between the fully unfolded and fully folded forms of a polypeptide chain (i.e. the folding pathway) and to describe the major intermediates in structural and kinetic terms. One of the

0097-6156/91/0470-0050$06.00/0 © 1991 American Chemical Society

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

4.

ELOVE& RODER

Structure and Stability of Cytochrome c

51

challenges encountered in the effort to characterize folding intermediates is the cooperativity of the folding transition, which makes it difficult to study partially folded intermediates at equilibrium. Folding intermediates are more likely to become populated under non-equilibrium conditions (i), but kinetic intermediates are usually too short lived to be studied directly by structurally powerful, but intrinsically slow techniques such as NMR. The limited time resolution of NMR can be overcome by combining it with hydrogen exchange labeling and rapid mixing methods. The feasibility of this approach was first demonstrated in a study of early folding events m bovine pancreatic trypsin inhibitor (BPTI) (5), using one-dimensional N M R in conjunction with a competition method (6). Later, a more powerful pulse labeling method developed independently in this laboratory and that of Robert Baldwin, has been applied to study the folding kinetics of cytochrome c (7),ribonucleaseA (8,9), and more recently to barnase (10) and ubiquitin (Briggs and Roder, m preparation). The hydrogen exchange labeling methods rely on backbone amide protons as probes for the formation of hydrogen bonded structure during refolding and uses *H NMR to observe protons that have been protected from exchange in the refolded protein. The intrinsic exchange behavior for the amide protons of an unstructured polypeptide chain is well known from studies of model peptides and unfolded proteins (11,12). Since the exchange reaction is dominated by base catalysis above pH \ pH can be used to vary the rate over a time scale ranging from tens of minutes (at pH 3) to milliseconds (at pH>9). On the other hand, the exchange rates for hydrogen bonded amide protons in folded proteins can be retarded by many orders of magnitude (13-17). The large majority of these slowly exchanging amide protons in several proteins were found to be involved in internal hydrogen bonds (15,16), suggesting that Η-bond formation is the dominant factor in retarding the rate of N H exchange in native proteins. For example, the 42 most slowly exchanging amide protons in oxidized cyt c (17) are all involved in crystallographically defined hydrogen bonds. The exchange rates of these protons are retarded by factors of lCr to >10 relative to model peptides, which makes them sufficiently long-lived for 2D NMR analysis. In a typical pulse labeling experiment, the protein is unfolded in D 0 and refolding is initiated by rapid dilution of a denaturing agent. The time course of structure formation is probed by briefly exposing the partially refolded protein to H 0 under rapid exchange conditions (basic pH) so that amide sites in unstructured parts of the polypeptide chain become completely protonated while sites involved in Η-bonded structure remain deuterated. After the labeling pulse, the protein is allowed to refold at mildly acidic pH where exchange is slow and the labels become trapped in the refolded protein. The proton labeling pattern imprinted by the pulse is then analyzed by using 2D NMR to measure the amount of proton label trapped at protected amide sites (e.g. at the 42 slowly exchanging N H sites of cyt c). In a series of such experiments at different refolding times, one can thus measure the time course of structure formation in terms of individual hydrogen bonds for those regions of the protein where N H sites become protected from exchange upon folding. This chapter describes the application of hydrogen exchange labeling and 2D NMR methods to study the folding pathway of oxidized horse heart cyt c. The results illustrate the capability of this approach to provide a detailed picture of the structural events that take place during folding. We also show that additional information on the mechanism of folding and the stability of folding intermediates can be obtained by systematic variation of the exchange conditions at certain points in folding. Cytochrome c is among the best characterized proteins in terms of structure, folding and NMR, and is thus an ideal model 8

2

2

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

52

PROTEIN REFOLDING

protein for such studies. In particular, high-resolution x-ray structures have been reported for several bacterial and mitochondrial c-type cytochromes (18-20), including horse heart cyt c (21\ Çyt c has also been the subject of numerous studies on thermodynamic ana kinetic aspects of folding (22-31). Moreover, essentially complete *H NMR assignments have recently been reported for both oxidation states of horse cyt c (32-33).

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

Pulse Labeling Strategy The rapid mixing procedures of a typical pulse labeling experiment are illustrated in Figure 1, using our previous studies on oxidized cyt c (7) as an example. The protein is initially unfolded in a D , 0 solution containing 4.5 M GuHCl and is allowed to reach equilibrium so that all the exchangeable amide protons are replaced by deuterons. In the first mixing chamber, refolding is initiated by dilution of the dénaturant with excess D 0 buffer or F^O under slow-exchange conditions (e.g. pH 6 where N H exchange is slow compared to the rate of folding). After variable refolding periods (t ), deuterium-nydrogen exchange is started by mixing the partially folded protein with excess HjO buffer at pH 9.5 (second mixer). The labeling conditions are chosen such that exposed amide sites become protonated within a few milliseconds while sites protected in hydrogen bonded structure remain deuterated. After about 50 ms, the exchange reaction is quenched by injecting the mixture into a cold buffer solution at p H 5. Under these conditions, cyt c refolds rapidly compared to further N H exchange so that many of the proton labels introduced by the pulse become trapped in the refolded protein. The protein is then concentrated and 2D J-correlated NMR spectra (COSY) are recorded as previously described (7). Proton occupancies at individual N H sites are determined by normalizing the intensities of resolved NH-QzH COSY cross peaks (or, in some cases, resolved N H resonances in the ID NMR spectrum), using the spectrum of a fully labeled control sample as a reference. A series of pulse labeling experiments at different refolding times provides kinetic curves that reflect the degree of protection from exchange at many N H sites throughout the protein. In the interpretation of pulse labeling data, it is important to keep the following points in mind, (i) For an amide group to remain deuterated during the 50 ms labeling pulse, its exchange reaction must be retarded by at least 50-fold compared to its intrinsic exchange rate, which is about 10 s" at pH 9.5. Therefore, only relatively stable structure gives rise to measurable protection, (if) Amide sites protected to this extent are most likely involved in Η-bonded interactions. However, their H bond acceptors cannot be identified directly. Nevertheless, examination of the patterns of protection for different amide sites in comparison with the native structure can lead to a qualitative picture of the secondary structure present in a folding intermediate, (hi) The method can only monitor amide sites that are protected in the native state. The transient formation of a non-native H-bond during folding could not be detected if this site becomes exposed in the folded state. However, the presence of non-native structural elements in a folding intermediate could still be deduced from the patterns of protection. For example, if a segment of the chain is part of a β-sheet early in folding, but forms an α-helix in the native structure, one would expect protection of every other amide site at an early phase and complete protection of all helical N H sites after refolding. On the other hand, if the structure formed early in folding is a native-like α-hehx, all the helical amide sites are protected rapidly. 2

f

3

1

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

ELOVE & RODER

Structure and Stability of Cytochrome c

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

I

Sf

H

Figure 1. Schematic diagram of the rapid mixing apparatus used for pulse labeling experiments at variable refolding times (t = 3 ms - 10 s) and labeling pulse (t = 50 ms): SI: Horse heart cytochrome c dissolved in 4.5 M GuHCl in D 0 , pH 6.0. S2: Refolding buffer (100 mM Acetate, pH 6.0). S3: Pulse butter (100 mM Glycine, pH 9.5). Q: Quench buffer (300 mM Citrate, 50 mM Ascorbate, pH 5.3). f

2

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

54

PROTEIN REFOLDING

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

Structural and Kinetic Patterns The time course of protection for a selection of amide protons is shown in Figure 2, using a logarithmic scale for the refolding time. Although different relative amplitudes are observed for different amide protons, one can recognize three common kinetic phases: a fast phase with a tune constant of about 20 ms, an intermediate phase of a few hundred milliseconds and a slow phase with a time constant of 5 to 10 sec. Figure 3 shows that the folding kinetics observed by tryptophan fluorescence is also characterized by three kinetic phases with similar time constants, but different relative amplitudes compared to the pulse labeling results. The fluorescence signal is due to a single tryptophan residue, Tip 59, which becomes quenched upon refolding via energy transfer with the heme group. The fact that different experimental probes (42 N H sites and one fluorescence probe) exhibit distinct kinetic behavior demonstrates that partially folded intermediate states are populated on the folding pathway of cyt c. The majority of amide probes for which the kinetics of protection was measured fall into one of two distinct classes. (ιΊ For a group of protons located near the chain termini (e.g. Val 11, Cys 14, Leu 98, Lys 100), the occupancy drops from 1 to about 0.4 in the first 30 ms of refolding (Figure 2), indicating that they become 60% protected in the fast folding phase. In the intermediate phase, they acquire an additional 20% of protection and become fully protected from exchange in the slow phase. This type of behavior is observed for most of the H bonded amide protons near the N-terminus up to residue 18 and the C-terminal segment between residues 91 and 100. (if) For most of the remaining slowly exchanging amide protons, the occupancy remains nearly constant between 0.75 and 0.85 out to about 100 ms, indicating that these sites remain largely accessible for exchange in the early folding phase. The occupancy decreases to about 0.5 in the intermediate phase and drops to zero in the slow folding phase. In Figure 4, the proton occupancies measured at the end of the fast phase (t =30 ms) are mapped onto the native cyt c structure. The rapidly protected rotons (class 1) with occupancies below 0.6 are all localized in the contact region etween the N- and C-terminal helices, while the more accessible class 2 protons are distributed throughout the rest of the structure. Class 2 protons with occupancies over 0.8 mclude residues in other helical segments of the native structure (the 60's helix and a short 70's helix) as well as amide sites involved in non-helical Η-bonds (reverse turns and irregular tertiary H-bonds). The labeling pattern seen in Figure Vindicates that the formation of stable Η-bonded structure in the fast folding phase is limited to the N-terminal α-helix (N-helix), the N H of His 18 and the C-terminal α-helix (C-helix). That the structure in these regions corresponds to native-like helices is indicated by the observation of continuous segments of protected protons which is inconsistent with any other type of structure. Furthermore, the following evidence strongly suggests that in the early folding intermediate the two helices already form a hydrophobic helix-helix contact similar to the helix pairing site of the native structure: (i) The time course of protection is virtually identical for amide protons on either helix (Figure 2), indicating simultaneous formation of both helices, (ii) Isolated helices are known to be only marginally stable (eg. 35, 36), and one would not expect strongly protected amide protons without stabilizing tertiary contacts, (iii) The pulse variation results presented below show that protons on either side of the helix-helix interface are protected to the same degree, suggesting again that the stability of the helices depends on their mutual stabilization. The high labeling levels observed outside the N - and C-terminal regions indicate that the intervening segment of the polypeptide chain (residues 19 f

E

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

ELOVE& RODER

Structure and Stability of Cytochrome c

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

4.

TIME

(SEC)

Figure 2. Folding kinetics of oxidized cyt c at pH 6.0,10°C, observed by pulse labeling and 2D NMR, using the protocol outlined in Fig. 1. Results are shown for representative amide protons, including examples from the three major helices of cyt c and the indole N H of Trp 59 which forms a tertiary Η-bond with a heme propionate side chain. (Reproduced with permission from ref. 7. Copyright 1988 Macmillan Magazines Ltd.) Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

55

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

56

PROTEIN REFOLDING

TIME

(SEC)

Figure 3. Folding kinetics of cvt c at pH 6.2, 10°C, monitored by the tryptophan fluorescence (350 nm) on a Hi-Tech stopped-flow apparatus (excitation at 280 nm). Oxidized cvt c was unfolded in 4.2 M G u H C l pH 6.0 and refolded at pH 6.2 in the presence of 0.7 M GuHCl. (Reproduced with permission from ref. 7. Copyright 1988 Macmillan Magazines Ltd.)

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

ELOVE & RODER

Structure and Stability of Cytochrome c

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

4

Figure 4. N H exchange labeling pattern observed in pulse labeling experiments at t = 30 ms (c.f. Fig. 2) coded into the native cyt c structure. The shading reflects the proton occupancies for 35 N H sites determined by 2D N M R as described in the text. f

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

57

58

PROTEIN REFOLDING

through 90) does not contain stable Η-bonded structure at this early stage in folding. The lack of strongly protected amide sites is particularly striking for the 60's and 70's region which contains nine slowly exchanging amide protons forming helical Η-bonds in the native structure. In the early intermediate, these H-bonds are either not formed yet or so unstable that the amides become protonated during the labeling pulse. However, we cannot exclude the presence of other types of interactions, such as hydrophobic side chain contacts or interactions involving the heme group. It seems surprising that association of two helical segments near opposite ends of the chain is preferred over other more local contacts, especially since in the native cyt c structure (21) the contact area between the 60's helix and the C helix is comparable to that between the N-helix and the C-helix. However, the airing site between the N-helix and the C-helix appears to be tighter than that etween the 60's helix and C-helix. The closest approach occurs near Gly 6 and Leu 94. Interestingly, Gly 6 is one of the few completely conserved residues among all known c-type cytochromes, and the hydrophobic character of the other contact residues is also highly conserved. The helix pairing site is a prominent feature of all known cyt c structures, even of the very distant prokaryotic relatives. Thus, it is tempting to speculate that the conservation of this structural feature reflects constraints imposed by the folding mechanism. Another possible explanation for the late formation of the 60's helix is related to the involvement of heme ligation in cyt c folding. A number of observations (Elôve & Roder, to be published) indicate that Met 80 which is ligated to the neme iron in the native cyt c structure is replaced by another side chain ligand, probably His 26 or His 33, in the unfolded state. This non-native axial ligand becomes trapped in the partly folded early intermediate, and its presence may interfere with the proper folding of the 60's region and other parts of the structure. The non-native ligand is replaced by the native methionine in the intermediate phase on the 100 ms time scale (34; Elôve & Roder, in preparation). It is interesting to note that the fluorescence of Tip 59 is partially quenched in the early folding phase (Figure 3), while the indole N H of the same tryptophan only begins to be protected in the intermediate phase by forming an Η-bond with one of the heme propionate side chains (Figure 2). Apparently, the early intermediate formed in 20 ms is sufficiently compact for thefluorescenceto become partially quenched via energy transfer with the heme, but a stable H bond is not yet present.

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

Ε

Variation of the Pulse Labeling Conditions The preferential protection of N - and C-terminal amide sites in the fast kinetic phase is a clear indication of a sequential folding reaction with a partly folded intermediate on the folding pathway. On the other hand, the pulse labeling results also show indications of heterogeneous behavior with several species refolding along parallel pathways. For instance, N - and C-terminal sites acquire about 60% protection in the fast phase, suggesting that 60% of the molecules have undergone the helix pairing reaction while the remaining 40% are still unfolded after 30 ms. Similarly, the 50% occupancy observed for amide protons in the 60's helix at t =500 ms is consistent with a heterogeneous population in which about half of the molecules are completely folded while 30-40% of the molecules are structured only at the chain termini and 10-20% are completely unfolded (Figure 2). However, we have to consider another possibility, namely that the system is homogeneous, but amide sites are only partially protected. For example, a proton occupancy of 0.5 for a particular N H could anse if the degree of protection is such that the amide becomes 50% protonated during the labeling f

Georgiou and De Bernardez-Clark; Protein Refolding ACS Symposium Series; American Chemical Society: Washington, DC, 1991.

Structure and Stability of Cytochrome c

4. ELÔVE & RODER

59

Downloaded by FUDAN UNIV on January 7, 2017 | http://pubs.acs.org Publication Date: July 30, 1991 | doi: 10.1021/bk-1991-0470.ch004

pulse. These two possibilities can be distinguished by a modified pulse labeling protocol involving systematic variation of the labeling conditions. Furthermore, this approach can provide qualitative information on the stability of folding intermediates. A similar method has been applied toribonucleaseA (9). The analysis of pulse variation experiments is based on a structural unfolding model which has been used extensively in the interpretation of H exchange data from native proteins (e.g. 13,15-16). While some of the details of the hydrogen exchange mechanism are still unclear, there is general agreement that under destabilizing conditions where the equilibrium between folded molecules (F) and unfolded forms (U) is well approximated by a two-state model, the strongly protected protons exchange via transient unfolding (4). This situation can be described by the following scheme: F(H)

2^

U(H)

U(D)

F(D)

(1)

D 0 2

where represents the equilibrium constant for unfolding (disruption of intramolecular Η-bonds) and k is the exchange rate of a given amide site in the unfolded state. Under most conditions, k is rate limiting (EXj exchange mechanism), so that the measured exchange rate, k , is given by c

c

e

2

K-KK-

()

N H exchange rates can thus provide a measure of the free energy of unfolding, A G , according to the relationship u

AG^-RTln^).

(3)

The validity of Equation (3) has been demonstrated in a study of the effect of destabilizing solvent additives on N H exchange rates in native BPTI (4), and it is expected to be applicable to transient folding intermediates as well. N H exchange rates in folding intermediates can in principle be measured by pulse labeling experiments at constant folding time and vanable pulse time (t ), but due to the limited life time of a folding intermediate it is advantageous to vary pulse pH at constant t . For a homogeneous situation, the resulting pH dependent proton occupancy, P(pH), is given by P(pH) = l-exp[-K k (pH)t ]. u

c

p

(4)

Above pH 3, the exchange reaction is determined by base catalysis (11-13), and k can be expressed as c

k (pH) = k c

H

Q H

lo(P ^),

(5)

where k „ is the second order rate constant for base catalyzed exchange. In the absence of pH dependent conformational transitions (constant K J , Equation (4) describes a sigmoidal curve with P=0 at low pH and P=1 at high pH, as shown in Figure 5 (left panel). For an exposed amide proton (1^=1) under the conditions used in this study (t =30 ms, UrC), the labeling profile is expected to be centered near pH 8 (where the half life for exchange matches t ). With increasing stability (K