Nonnative isomers of proline-93 and -114 predominate in heat

Neil B. Tweedy , Satish K. Nair , Steven A. Paterno , Carol A. Fierke , and David W. Christianson. Biochemistry 1993 32 (41), 10944-10949. Abstract | ...
0 downloads 0 Views 788KB Size
Biochemistry 1990, 29, 821 1-8216

821 1

Nonnative Isomers of Proline-93 and - 1 14 Predominate in Heat-Unfolded Ribonuclease Ai M a r c Adlert and Harold A. Scheraga* Baker Laboratory of Chemistry, Cornell University, Ithaca, New York 14853- 1301 Received November 3, 1989; Revised Manuscript Received May 25, 1990

ABSTRACT: T h e peptide bonds preceding both Pro-93 and Pro-I 14, which a r e in the cis conformation in

native RNase A, are predominantly in the trans conformation in the heat-unfolded protein. The percentages a r e estimated to be 60% and 63%, respectively, with a standard deviation of &7% in each quantity. These ratios are close t o those found for corresponding sequences in X-Pro-Y peptides. The concentration of the trans proline species was determined from the integrated intensities of resonance peaks of the C*H protons of Tyr-92 and Asn- 1 13, which are well resolved in the 1D proton N M R spectrum of heat-unfolded R N a s e A. The assignments of the resonances were deduced from 2D N O E S Y and DQF-COSY spectra of unfolded RNase A in D20.Furthermore, the CaH protons of both Tyr-92 and Asn-1 13 had an intense NOE cross-peak with the C8H and C6'H of the respective following prolines. For both Pro-93 and Pro-I 14, these NOE cross-peaks would arise only if the X-Pro peptide bond were in the trans conformation. It is generally believed that the rate of refolding of R N a s e A is considerably reduced by nonnative proline isomers, such a s trans Pro-93. Two models for folding RNase A, that are consistent with these new results and the work of previous investigators, are presented here.

w h e n bovine pancreatic ribonuclease A (RNase A)' is unfolded reversibly by chemical denaturants and/or heat, it forms a mixture of three slowly interconverting species called Uf, Ui, and Uf'. The three species account for roughly 20%, 3096, and 5070, respectively, of the total protein, as judged by U V absorbance and inhibitor binding (Tsong et al., 1972; Garel & Baldwin, 1973; Garel et al., 1976; Hagerman et al., 1979; Cook et al., 1979; Schmid & Blaschek, 1981; Lin & Brandts, I983a, I987a). Once refolding is initiated, the three species behave very differently. Uf rapidly regains the native conformation ( 1 8-400 ms). The other two forms of the protein, V i and Uf', refold on a time scale of seconds to minutes. A third slow phase has been identified by Lin and Brandts ( 1 987a). It accounts for 5% of the unfolded species as judged by U V absorbance. Because of its low concentration, it has not been detected in our experiments. Also, any effect that this phase has on our conclusions falls within the limits of accuracy (f770); therefore, the third slow phase will not be discussed any further in this paper. The experimental evidence indicates that unfolded RNase A is a mixture of three conformational isomers, of which two, Ui and Ui', contain at least one nonnative conformation around a key covalent bond. For example, the peptide bond between Tyr-92 and Pro-93 is 100% cis in the fully folded protein. When RNase A is unfolded, this bond undergoes reversible isomerization to form a mixture of cis and trans isomers (Lin & Brandt, 1983b, 1987b; Schmid, 1986; Schmid et al., 1986). It is generally believed (although not conclusively proven) that, once the protein is returned to conditions where the native form is thermodynamically stable, the molecules containing cis Pro-93 refold rapidly. The folding is retarded 'This work was supported by Grant GM-24893 from the National Institute of General Medical Sciences of the National Institutes of Health. Support was also received from the Cornel! Biotechnology Center. M A . was an N I H postdoctoral fellow, 1987-1988. * To whom correspondence should be addressed. f Present address: Biophysics Research Division, Institute of Science and Tcchnology, University of Michigan, Ann Arbor, MI 48109.

0006-2960/90/0429-82 1 1S02.50/0

in the molecules containing the nonnative trans isomer of Pro-93. Previous results (Schmid & Baldwin, 1978) support the conclusion that both Ui and U:' contain nonnative isomer(s) of the four proline peptide bonds of which two, 93 and 1 14, are cis in the native protein and the other two, 42 and 117, are trans. Presumably, the 20% of the protein that folds in the fast phase, Up contains no nonnative isomers that inhibit the refolding. RNase A also contains four disulfide bonds. The disulfide bond (Cfl-S-S-Cp) has two stable conformations. The CB-S bonds are mutually perpendicular. Therefore, both left- and right-handed conformations of the CB-S-S-Cfl moiety are possible. The internal rotational energy barrier about the S-S bond is estimated to be as high as 15.5 kcal mol-' based on studies with model compounds (Kessler & Rundel, 1968; Fraser et al., 1971), and might be higher inside a globular protein. Once the protein is unfolded, the four disulfide bonds are free to isomerize into the nonnative conformation. It has been proposed that these nonnative isomers of the disulfide bond might give rise to slow-folding species (Mui et al., 1985). The disulfide bonds themselves form a series of intertwined loops. Nall et al. (1978) have proposed that these loops may become tangled upon unfolding, thus giving rise to a slowfolding form(s) of the protein. Given the large number of bonds that can isomerize into nonnative forms, the question becomes not why does 80% of RNase A refold so slowly, but why does 20% fold so quickly. There are nine separate arrangements (four prolines, four disulfides, and the pattern of disulfide loops), all of which could presumably adopt nonnative isomers in the unfolded state. The 1 Abbreviations: AMX, a three-spin system such as the (Y and 0 protons of Ser, Cys, Asp, Asn, His, Phe, Tyr, and Trp (these amino acids cannot be distinguished in a DQF-COSY spectrum); DQF-COSY, double quantum filtered correlated spectroscopy; NOE, nuclear Overhauser effect; NOESY, nuclear Overhauser effect spectroscopy; N M R , nuclear magnetic resonance; ppm, part per million; DSS, 2,2-dimethyl-2-silapentanesulfonate; GdnaHCI, guanidine hydrochloride; RNase A, bovine pancreatic ribonuclease A; UI, Ut, and U:', single fast and two slow phases found when RNase A refolds.

0 1990 American Chemical Society

82 12 Biochemistry, Vol. 29, No. 36, 1990 percentage of molecules that retain the native conformation of all nine bonds should be vanishingly small. To date, the isomers that give rise to the slow-folding forms have not been unambiguously identified, and the question remains unanswered. Two theories are invoked in the literature to explain why so much of the protein folds in the rapid phase. One states that the protein refolds even in the presence of nonnative isomers. I n staphylococcal nuclease (Fox et al., 1986; Evans et al., 1987, 1989) and in calbindin DgK(Chazin et al., 1989), both the cis and trans isomers of proline are found in the fully folded protein. Furthermore, the cis and trans isomers of unfolded staphylococcal nuclease refold at roughly the same rate (Evans et al., 1989). The result shows that proteins can fold properly even if the molecule contains a nonnative conformational isomer. Alternatively, there may be sufficient native structure in the unfolded protein to retain the native conformational isomer of the relevant bonds. Lin and Brandts (1983b, 1984, 1987b) have assayed the conformational states of three of the four prolines of denatured RNase A using isomer-specific proteases (ISP). Their results show that all three prolines are predominantly i n the native form (see Discussion for more details). Their evidence indicates that short-range interactions stabilize the native isomers in the unfolded protein and reduce the population of slow-folding species. Unfortunately, it has been difficult to measure the percentages of different isomers in the unfolded protein, and hence different conclusions based on different approaches have been reached. A direct way to measure the proportions of the native and nonnative isomers of the appropriate bonds is needed to resolve this question. Estimates of the amount of trans Pro-93 and Pro- 1 14 are presented in this paper based on the 1D N M R spectrum of heat-unfolded RNase A. The results indicate that, in the heat unfolded protein, the nonnative trans proline is the dominant form for both Pro-93 and Pro-I 14. Various models for the refolding of RNase A are discussed in light of the new results. MATERIALS A N D METHODS Bovine pancreatic ribonuclease A (RNase A) type IIa (Sigma, St. Louis, MO) was further purified by the method of Taborsky ( 1 959) and was a generous gift from D. M. Rothwarf. The exchangeable protons were removed by lyophilizing the sample 3 times from 99.6% D 2 0 (Aldrich, Milwaukee, WI). The sample concentration was approximately 5 mM. N o buffer was used. The pH was adjusted to the acidic range using concentrated HCI before the first lyophilization, and then checked after use. Sample pH varied from 2 to 3 with no significant effect on the spectra other than variations i n the melting temperature. The spectra were referenced to the most upfield resonance which in turn had been calibrated against DSS (Adler & Scheraga, 1988). Measurements were made either on a General Electric GN500 (Figures 1 and 2) or on a Varian VXL-400 (Figure 3) N M R spectrometer. The 90" pulse width was 15 ~s for the G E and 20 ps for the Varian. All experiments, except the DQF-COSY experiment, were carried out on both instruments. For the 2D measurements, approximately 5 12 t , values were measured, and 2048 (Varian) or 4096 (General Electric) t 2 data points were collected. Sine bell shifted by 60" was used in both directions for the DQF-COSY spectrum (Figure 3). A 60" squared sine bell was used in t , and a 90' squared sine bell in t , to process the NOESY spectrum (Figure 2). The integrated intensities of the C"H resonances were normalized to the peak areas of the 43 aromatic ring protons

Adler and Scheraga

0.0

7.5

1.0

6:5

6.0 PPm

5:5

5:O

'H NMR spectrum of heat-unfolded RNase at 50 "C, pH 2. The peaks are labeled as follows: (a) the combined resonances of the 15 phenylalanine ring protons and 4 C6H histidines; (b) the 12 Tyr C*H;(c) the 12 Tyr C'H; (d) the C"H of Asn- 1 I3 that precedes trans Pro-1 14; (e) the C"H of Tyr-92 that precedes trans Pro-93. The inset shows peaks d and e more clearly. FIGURE 1:

that lie between 6.7 and 7.5 ppm. These protons fall into 3 resolved groups, 19 protons from the 15 Phe ring and 4 C6H His, 12 from the C6H Tyr, and 12 from the C'H Tyr. Each of the three groups of resonances was integrated separately. The carrier frequency was shifted to halfway in between the resonances of the aromatic ring protons and the CaH of Tyr-92 and Asn-113, approximately 6 ppm, to minimize pulse width distortion between the resonances. A 10-s delay between pulses was used in order to avoid partial saturation. A total of 256 free induction decays were collected. Measurements were made over a range of temperatures from 50 to 75 "C. The relative intensities did not vary significantly with temperature. Distortions in the base line were removed by fitting selected points between the resonances to a third-order polynomial (Abramowitz & Stegun, 1968). At least two points for the base-line fit were selected from the upfield and downfield sides of both the Asn and Tyr CaH resonances. The RNase A was checked for covalent modification after the experiment by H P L C using a Mono S high-performane cation-exchange column at p H 7 (Pharmacia, Upsala, Sweden). There was no detectable degradation of the protein (