Biochemistry 1986, 25, 6768-6778
6768
Moscarello, M. A., Gagnon, J., Wood, D. D., Anthony, J., & Epand, R. M. (1973) Biochemistry 12, 3402-3406. Murty, V . L. N., Sarosiek, J., Slomiany, A., & Slomiany, B. L. (1 984) Biochim. Biophys. Res. Commun. 121,521-524. Nguyen-Le, T., Nicot, C., Alfsen, A., & Barrat, M. D. (1976) Biochim. Biophys. Acta 427, 44-56. Nicot, C., Nguyen Le, T., Lepretre, M., & Alfsen, A. (1 973) Biochim. Biophys. Acta 332, 109-123. Schlesinger, M. J., & Magee, A. I. (1982) Biophys. J . 37, 126-127. Schmidt, M. F. G. (1983) Curr. Top. Microbiol. Immunol. 102, 101-129.
Sherman, G., & Folch-Pi, J. (1970) J . Neurochem. 17, 597-605. Stoffel, W., Hillen, H., Schroeder, W., & Deutzman, R. (1983) Hoppe-Seyler’s 2. Physiol. Chem. 364, 1455-1462. Stoffyn, P., & Folch, J. (1971) Biochem. Biophys. Res. Commun. 44, 157-161. Weinryb, I., & Steiner, R. F. (1970) Biochemistry 9, 135-146. Weller, A. (1961) Prog. React. Kinet. 1, 187-212. Wetlaufer, D. B. (1962) Adu. Protein Chem. 17, 303-310. Wu, H. C., Hou, C., Liu, J. J. C., & Yem, D. W. (1977) Proc. Natl. Acad. Sci. U.S.A. 74, 1388-1392.
N-Terminal Domain of the Bacteriophage h Repressor: Investigation of Secondary Structure and Tyrosine Hydrogen Bonding in Wild-Type and Mutant Sequences by Raman Spectroscopyt George J. Thomas, Jr.,* Betty Prescott, and James M . Benevides Department of Chemistry, Southeastern Massachusetts University, North Dartmouth, Massachusetts 02747 Michael A. Weiss Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 021 39 Received June 1 1 , 1986
Laser Raman spectroscopy has been employed to investigate structures of the X repressor N-terminal fragment, which recognizes operator D N A . Examination of repressor fragments containing deuterated amide groups and specifically labeled deuteriotyrosines has enabled the assignment of many of the conformation-sensitive Raman bands. By use of Fourier deconvolution and signal averaging techniques, the spectra of both wild-type and mutant sequences have been obtained as a function of the total protein concentration in aqueous solution over the range 5-100 mg/mL. This analysis has permitted monitoring of the monomerdimer association of the repressor fragment and determination of the effects of dimerization upon individual side-chain interactions and main-chain secondary structure. T h e spectra are interpreted to reveal the hydrogen-bonding environments of four tyrosines of the N-terminal fragment (Y22, Y60, Y85, and Y88). The fifth tyrosine (Y101) is known from N M R experiments to be exposed to solvent molecules. The results show that in the dimer Y22 and Y85 are each acceptors of a strong hydrogen bond from a positive donor group, while Y88 is the donor of a strong hydrogen bond to a negative acceptor and Y60, like Y101, is involved in both a donor role and an acceptor role. Y60, Y85, and Y88, which are all near the dimer interface, undergo a collective change in hydrogen-bonding environment with dissociation of the dimer. The net effect of this change is the conversion of one acceptor tyrosine, deduced to be Y88, to a combined donor and acceptor role. The Raman results also indicate a predominantly a-helical structure for the N-terminal fragment in aqueous solution, with 70 f 4% of the residues incorporated into helical domains. The amount of a-helix determined from the Raman spectrum is consistent with X-ray and prediction results and is altered neither by the mutations C85 Y85 and C88 Y88 nor by dissociation of the dimer. ABSTRACT:
-
-
x e bacteriophage X repressor is a DNA-binding protein that contains 236 amino acids and comprises two structural domains. The N-terminal domain binds specifically to the X operators, and the C-terminal domain contains dimer and higher order contacts (Pabo et al., 1979; Sauer et al., 1979). The crystal structure of an N-terminal fragment of X repressor (residues 1-92) has been determined at 3.2-A resolution (Pabo ‘This is paper XVIII in the series YStructuralStudies of Viruses by Laser Raman Spectroscopy”. This research was supported by NIH Grants AI11855 (G.J.T.), A116892 (R. T. Sauer), and GM30804 (M. Karplus). * Author to whom correspondence should be addressed.
0006-2960/86/0425-6768$01.50/0
& Lewis, 1982). The secondary structure of this fragment consists of five a-helices, as shown in Figure 1, and the tertiary structure is compatible with binding to double-stranded B DNA. A detailed model of the complex of X repressor and operator DNA of the B form has been proposed (Lewis et al, 1983). Major features of the model have been supported by studies on genetically altered repressors (Hecht et al., 1983; Nelson et al., 1983; Eliason et al., 1985; Nelson & Sauer, 1985). The active species in operator binding is a dimer (Chadwick et al., 1970). Because dimerization and DNA binding are coupled equilibria, dimerization contributes to the apparent operator affinity of the repressor. Both C-terminal and N -
O 1986 American Chemical Society
RAMAN SPECTROSCOPY OF
A
REPRESSOR
B 22
VOL. 2 5 , NO. 2 2 , 1986
6769
The positions of Y22, Y60, Y85, and Y88 in the crystal structure of the 1-92 fragment (Pabo & Lewis, 1982) are shown in Figure 1. Y22 projects from helix 1 into the interior of this N-terminal domain, where it is partially stacked against F51. Y60, at the turn between helices 3 and 4, is on the surface of the domain. Y85 and Y88 are in helix 5, which is involved in dimerization. The Raman spectrum permits the average hydrogen-bonding environment of these side chains to be determined. In addition, the contributions of individual side chains (as donor, acceptor, or both) can be inferred through the study of genetically altered repressors containing single amino acid substitutions.
EXPERIMENTAL PROCEDURES
FIGURE 1 : (Left) Cartoon showing the five helical domains (cylinders) of each subunit in the dimer of the X repressor N-terminal fragment,
residues 1-92. as determined from the crystal structure of Pabo and Lewis (1982). The 2-fold symmetry axis of the dimer is normal to the page and at the center of the interface between helices 5 and 5' (unlabeled) of the respective subunits. (Right) a-Carbon skeleton of the dimer in the same orientation as at left and including selected aromatic side chains (Pabo & Lewis, 1982). Y60 and Y85 of each subunit at the dimer interface are labeled, as are aromatics of helix 1 (tyrosine-22 and phenylalanines-51 and -76). Note that in this model YlOl is absent and, for clarity, the side chain of Y88 has not been labeled.
terminal domains appear to contain dimer contacts. However, the isolated C-terminal fragment forms dimers in solution (Kd < 10") much more readily than does the isolated N-terminal fragment (Kd = (Pabo et al., 1979; Weiss et al., 1986). In the crystals, the isolated N-terminal fragment forms the dimer shown in Figure 1 (Pabo & Lewis, 1982). Its functional importance has been demonstrated through the study of mutations in the dimer that disrupt operator binding (Hecht et al., 1983; Nelson et al., 1983; Weiss et al., 1986). In order to investigate directly the side-group interactions involved in dimerization of the N-terminal domain, we have obtained and interpreted the Raman spectra of wild-type and genetically altered repressor fragments. Data have been collected from both unlabeled and specifically deuterated fragments in H 2 0 and D,O solutions in order to assign individual Raman bands. We have observed the vibrational Raman bands at a series of protein concentrations within the range 5-100 mg/mL, enabling comparison of the structures of the DNA-binding domain in its monomeric and oligomeric states. The results of these Raman studies are compared and contrasted with the results of recent one- and two-dimensional IH N M R studies of aqueous solutions of the intact X repressor and N-terminal fragments (Weiss et al., 1983, 1984, 1986). Raman spectroscopy of proteins provides information on both the conformations of the peptide backbone and the interactions of selected side chains. The subject has been reviewed recently, and several applications to nucleic acid binding proteins,have been discussed (Thomas et al., 1983; Thomas, 1986). The Raman spectra are particularly informative concerning hydrogen-bonding interactions involving the phenolic O H group of tyrosine (Siamwiza et al., 1975). When a single tyrosine residue is present, the role of the p-hydroxyl group as hydrogen-bond donor, or acceptor, or both, may be deduced. Here, we apply this technique to the N-terminal domain of X repressor (residues 1-102). this domain contains five tyrosines: Y22, Y60, Y85, Y88, and Y101.
Preparation and Purification of X Repressor and N - Terminal Fragments. X repressor, genetically altered repressors, deuterium-labeled repressors, and their N-terminal fragments were purified as described (Johnson et al., 1980; Sauer et al., 1986). The purified proteins were exhaustively dialyzed against 0.2 M ammonium bicarbonate and lyophilized. Lyophilizates were redissolved in either H 2 0 or D,O buffers, depending upon whether the Raman spectra were to be obtained from fragments with normal or deuterated amide groups. In the latter case, the fragment was first incubated at 33 "C in D 2 0 at pD 7.5 for 72 h and then lyophilized and stored in an atmosphere free of HzO. The complete deuteration of peptide groups was confirmed by the absence of amide IH resonances in N M R spectra, as well as by the absence of amide I11 vibrations in Raman spectra. All samples gave a single sharp band on sodium dodecyl sulfate (SDS)-polyacrylamide gels. Sample Handling f o r Raman Spectroscopy. All proteins were dissolved initially at the highest concentration to be investigated, usually 100 mg/mL, in a buffer consisting of 0.2 M KCl, 1 mM NaN,, 0.1 mM ethylenediaminetetraacetic acid (EDTA), and 50 mM tris(hydroxymethy1)aminomethane (Tris) at pH 7.5 f 0.1. Aliquots of the concentrated solutions were diluted in pure HzO (or DzO, 99.8% deuterium, Aldrich) to maintain a constant ratio of protein to counterions in the more dilute solutions. pH (or pD) was maintained at 7.5 f 0.3, unless otherwise indicated. All sample handling was carried out at 4 OC. Protein solutions were sealed within capillary cells (Kimax 34507), which were thermostated at 10 OC in the sample illuminator while Raman spectra were recorded. For the wild-type N-terminal fragment (residues 1-102), spectra were obtained over the interval 300-1800 cm-' for protein concentrations of 100, 50,40, 25, 20, 12, and 5 mg/mL. For other fragments, the spectrum in this extended interval was obtained only for the more concentrated solutions, usually 50 and 100 mg/mL. Normally, six to eight scans of the 300-1800-~m-~ interval were required to obtain a satisfactory signal to noise ratio. For dilute solutions, up to 100 scans of the narrower interval, 780-880 cm-', were required for satisfactory monitoring of the intensity ratio ZsSo/Z8,0 of the tyrosine doublet at 850 and 830 cm-'. Additionally, for several of the proteins, the Raman amide I (1600-1750-cm-') and amide I11 (12001350-cm-') regions were obtained by averaging up to 50 scans. By use of signal averaging techniques, we have greatly reduced the lower limit of protein concentration required in the classical (Le. nonresonance) Raman scattering experiments to 5 mg/mL (0.5 weight % or approximately 0.5 mM in the N-Yerminal fragment). All spectra were excited with the 514.5-nm line of P Spectra Physics 171-18 argon laser and were recorded on a Spex Ramalog VI spectrometer under microcomputer control (Li
6770
B I OC H E M ISTR Y
et a]., 1981). Data were collected a t increments of 1 cm-’, with integration time of 1 s and spectral slit width of 8 cm-’. The laser power a t the sample was always less than 300 mW. Frequencies of the Raman lines in all spectra were reproducible to f l cm-’ and are believed accurate to within f 2 cm-’. Analysis of Raman Intensities of Tyrosine Residue. The work of Siamwiza et al. (1975) has established a quantitative relationship between the intensity ratio ( 1 8 5 0 / 1 8 3 0 ) of the pair of conformation-sensitive tyrosine Raman lines at ca. 850 and ca. 830 cm-l (“tyrosine doublet”) and the number of tyrosines involved as hydrogen-bonding donor and acceptor groups in proteins. The correlation developed by these workers was derived from the Raman peak heights (as opposed to band areas) measured over a flat base line drawn between the wings of the spectral doublet. For many proteins, including the repressor fragments, there are no interfering bands near 780 and 880 cm-’, thus allowing a flat base line to be conveniently drawn tangent to these two points and the tyrosine peak heights to be measured therefrom. W e have adopted this method for measuring I850/I830 from the spectra presented below. (The absence of interfering peaks is largely a consequence of the absence of tryptophan in the 102 residues of the X repressor N-terminal fragment.) In some of the figures that follow, we have partially deconvolved the tyrosine doublet using an iterative Fourier deconvolution scheme described previously (Thomas & Agard, 1984). Deconvolution enhances the separation of the overlapping 850- and 830-cm-’ band components of the tyrosine doublet while preserving the component band areas, thus facilitating the measurement of relative intensity changes in the doublet. However, the peak heights are obviously not conserved in the deconvolution procedure, and therefore, we have not employed the peak heights of the deconvolved doublet for structural conclusions by the method of Siamwiza et al. (1975). Structural interpretation of the I850/I830 ratio is based solely upon the peak heights measured in the manner described in the preceding paragraph from spectra that have not been resolution enhanced by deconvolution. RESULTS In this study, seven different protein fragments related to the N-terminal domain of X repressor have been investigated: (i) residues 1-102 of the wild-type sequence (this sequence in one-letter symbols starting from the amino terminus is
STKKKPLTQEQLEDARRLKAIYEKKKNELGLSQESVADKMFNGINALNAYGMGQSGVGALNAALLAKILKVSVEEFSPSIAREIYEMYEAVSMQPSLRSEYE, hereafter abbreviated as RF102); (ii) residues 1-102 containing 3,5-dideuterio-labeled tyrosines (RF102-3,5-d2); (iii) residues 1-1 02 containing 2,6-dideuterio-labeled tyrosines (RF102-2,6-d2); (iv) residues 1-102 containing the mutation of Y85 to cysteine (RF102-C85); (v) residues 1-102 containing the mutation of Y88 to cysteine (RF102-C88); (vi) residues 1-102 containing a carboxymethylated cysteine a t position 88 (RF102G38CM); (vii) residues 5-98 of the wild-type sequence. The lengths and amino acid compositions of these fragments have been characterized as described by Sauer et al. (1986) and Weiss et al. (1986). Spectra and Assignments of Raman Lines of X Repressor N-Terminal Fragment. Figure 2 shows the Raman spectra of H 2 0 and D 2 0 solutions of the wild-type X repressor fragment consisting of residues 1-102 (RF102). Labels indicate the positions of the important tyrosine doublet and of the amide I and amide I11 bands in the H 2 0 solution spectrum and amide I’ and amide 111’ bands in the D,O solution spectrum. A complete listing of the frequencies, intensities, and assignments
THOMAS ET A L .
Table I: Raman Frequencies and Assignments of X Repressor N-Terminal Fragment“ frequency relative assignfrequency relative assign(cm-’) intensitv ment (cm-’) intensity ment 375 1 a-helix 1057 2 K, S 415 Ib skeletal 1075 2 E 525 2 a-helix 1104 2 A 548 2s a-helix 1125 4 CC str 57 1 1s amide VI 1168 Ib CH, rck 620 1 F 1206 2 Y 642 2 Y 1245 3s amide I11 710 2b M, amide IV 1264 5 Y, amide 148 2 M, V, L, I I11 719 1s v 1300 6 amide 111 829 3 Y 1320 7 CH, tws 85 1 5 Y 1337 6 CH, tWS 1 E, D 5 S, G 1403 892 920 5s I, E 1421 2 CH, scr 8 a-helix 1447 9 CH, scr 933 2s F 5s CHI rck 1599 950 2b I 1616 3s Y 980 1000 5 F 1650 IO amide I 1029 2 F, G 1675 4s amide I, 1042 2 G, S N, Q “Intensities (0-10 scale) are based on Figure 2 peak heights. b and s denote broad band and shoulder. Str, rck, tws, and scr denote stretch, rock, twist, and scissor modes, respectively. Assignments (one-letter symbols) are based on model compounds (Thomas et al., 1983). Note that this sequence, like the Pfl subunit, lacks both W and C residues.
for the H 2 0 solution spectrum of RF102 is given in Table I. For RF102, we find that the two peaks of the tyrosine doublet are centered a t 851 and 829 cm-l, as noted in Table I. For other fragments, we find little or no variation ( f 2 cm-’ or less) in the peak positions, as noted in subsequent figures and tables. For convenience, we shall refer to the frequencies of the tyrosine doublet as “850” and “830” cm-’ throughout, even though small and possibly significant differences of f 1 or f 2 cm-I may exist from one fragment to another or for the same fragment a t different experimental conditions. More detailed discussions of the assignment of the tyrosine doublet and the basis for its intensity and frequency dependencies upon molecular environment have been given by Siamwiza et al. (1975). These authors have also established the validity of the doublet assignment for proteins and have discussed the absence of interfering Raman lines from other protein side chains in the region of the doublet. In order to confirm the tyrosine doublet assignment for RF102, we have obtained Raman spectra in the region 300-1800 cm-I of the following deuterio isomers of this fragment: (i) RF102-3,5-d2 in which all five tyrosines are deuterated at the ring positions ortho to the OH group and (ii) RF102-2,6-d2in which all tyrosines are deuterated a t the positions meta to the OH group. We anticipate that deuterium substitution of the phenolic ring will alter Raman lines of tyrosine only, including the tyrosine doublet. On the other hand, frequencies and intensities of Raman lines of the peptide backbone and of side groups other than tyrosines should not be affected by the phenolic ring deuterations. Raman spectra of H 2 0 solutions of N-terminal fragments containing the deuterium-labeled tyrosines are compared with the spectrum of RF102 in Figure 3. The points of inflection in the difference spectra of Figure 3 confirm that the effects of deuteration are consistent with the assignments of Siamwiza et al. (1975); i.e., only the tyrosine doublet a t 850 and 830 cm-’ and the additional tyrosine ring modes near 642, 1206, and 16 16 cm-’ are significantly shifted in frequency by phenolic ring deuterations. The additional small negative bands in the difference spectra, particularly those near 1000 and 1200-1 240
RAMAN SPECTROSCOPY OF
VOL. 2 5 , NO. 22, 1986
REPRESSOR
6771
H
> e I4 v)
z z I4
W
e
H H
I->
w v1
-
DZO SOLUTION
E
4
z W c z w
-
500
700
900
1100
1300
1500
1700
cm-1 2: Raman spectra in the region 300-1800 cm-' of the 102-residueN-terminal fragment of X repressor. (Top) H 2 0 solution at pH 7.5 and 10 O C corrected for background and buffer (0.2M KCI 1 mM NaN3 0.1 mM EDTA 50 mM Tris). Labels indicate the frequencies of the tyrosine doublet, the intensities of which are informative of tyrosine hydrogen bonding. The positions of conformation-sensitive amide I and 111 bands, which are seen to shift with deuteration (amide I' and HI'), are also indicated. (Bottom) As above, except with D 2 0 replacing H20. FIGURE
+
cm-', are assigned to vibrations of the deuteriotyrosines of RF102-3,5-d2and RF102-2,6-d2. The same frequencies are also observed in the Raman spectra of the simple amino acids tyrosine-3,.5-d2 and tyrosine-2,6-d2 (spectra not shown). We note the following further evidence that there are no interfering Raman lines of significant intensity in the region of the tyrosine doublet (780-880 cm-I). First, the doublet of RF102-3,5-d2 is well resolved as an intense line centered at 831 f 1 cm-' and a weaker companion centered at 859 f 1 cm-'. A deep minimum occurs at 844 cm-', indicating no underlying band near this frequency. (Note that 844 cm-' is also the center-of-gravity of the doublet.) Both experimental and deconvolved band shapes give no indication of hidden bands that might be associated with other residue vibrations in the interval 780-880 cm-' (Figure 4a). Second, the center-of-gravity of the doublet in the model compound tyrosine-3,5-d2 (amorphous solid sample) also occurs at 844 cm-', even though the two components of the doublet are shifted apart to 825 and 863 cm-' in the solid, with a concomitant disproportionation of their relative intensities (Figure 4b). This is precisely the behavior expected of a Fermi resonance doublet for different interaction energies. It indicates that the hydrogen-bonding state of the phenolic OH group in the solid model compound tyrosine-3,5-d2 differs from that of tyrosines in the aqueous protein RF102-3,5-d2. Third, the tyrosine
+
+
doublet of RF102-2,6-d2 similarly consists of well resolved components at 830 f 1 and 858 f 1 cm-', with no evidence of hidden Raman lines from other protein constituents (data not shown). Many of the other assignments listed in Table I are deduced from observations described in more detail in the subsequent sections. Secondary Structure of N-Terminal Fragment of X Repressor in Aqueous Solution. The 102-residue N-terminal fragment of wild-type repressor (RF102) exhibits Raman amide I and amide I11 frequencies and intensities (Table I ) characteristic of a protein that is rich in a-helical secondary structure (Lord, 1977). The strong and sharp amide I band centered near 1650 cm-I and the distribution of amide 111 intensity between 1280 and 1300 cm-' (Figure 2) are features reminiscent of the Raman spectra of filamentous bacterial viruses, the coat proteins of which are predominantly a-helical. In fact, the amide bands of RF102 closely resemble those of the subunits of filamentous viruses (Thomas et al., 1983; Thomas, 1985). Other qualitative indications of secondary structure are the following. A weak shoulder on the high-frequency side of the amide I peak, barely resolved at ca. 1675 cm-I, indicates nonhelical secondary structure for a minority of residues of the RF102 backbone. We assign also the weak 1245-cm-'
6772
BIOCHEMISTRY
THOMAS ET AL.
FRAGMENT WITH
-
-
N
FRAGMENT WITH TYROSINE-E. 6-62
DIFFERENCE: (a)
- OIFFERENCC: I
1
500
- (bl
-
(a1 (c)
I
1
700
I
I
900
I
I
1100
I
I
1300
I
I
I
1500
I
1700
cm-l FIGURE
3: Raman spectra in the region 300-1800 cm-I of three isomers of the N-terminal fragment of the X repressor and their corresponding
difference spectra: (a) repressor fragment containing unlabeled tyrosines; (b) repressor frzgment containing deuterium substitution of positions 3 and 5 of all tyrosine rings; (c) repressor fragment containing deuterium substitution of positions 2 and 6 of all tyrosine rings. Labels in (a) indicate the frequencies of the prominent Raman lines of tyrosine residues and the positions of amide I and 111 bands. The former are shifted by tyrosine ring deuterations [tic marks in (b) and (c)], while the latter are insensitive to tyrosine deuterations (vertical dashed lines). These properties are revealed in the difference spectra (lower two tracings). shoulder of the main amide I11 band to nonhelical structure. The 1675- and 1245-cm-' frequencies together indicate that the minor secondary structure component of RF102 is most probably a combination of irregular structural domains including turns (Lord, 1977). There is no indication in the spectrum of RF102 (Figure 2) of significant @-sheetstructure. The Raman amide I and amide 111 assignments and their structural interpretation are consistent with the amide I' and amide 111' bands observed for deuterated RF102 in D 2 0 solution (Figure 2, bottom spectrum). These D 2 0 solution data also demonstrate that the bands remaining near 1206, 1264, 1320, and 1337 cm-' in the fully deuterated fragment cannot be due to amide vibrations but most likely originate from the side groups indicated in Table I. In order to obtain a quantitative estimate of the percentages of residues of RF102 that are distributed between helical and nonhelical segments, we have deconvolved the amide I profile using the same procedure applied previously to several filamentous bacterial viruses (Thomas, 1985). The deconvolution approach is particularly effective for proteins containing predominantly a-helical secondary structure (Thomas & Agard, 1984). The resolution enhancement resulting from deconvolution permits reliable measurement of the integrated intensities of the helix and nonhelix components of the complex amide I band shape, and these areas are considered to be
Table 11: Raman Amide I Bands and Secondary Structures of X ReDressor N-Terminal Fragment (Residues 1-1 02) secondary deconvolved deconvolved observed frequency frequency band area structure (cm-I)' (cm-I ) (%Ib type'
*
1650 1647 70 4 a-helical 1675 1679 30 h 4 nonhelical "Raman data were obtained from the H20solution spectrum of wild-type repressor fragment at 50 mg/mL in pH 7.5 Tris buffer at 10 "C. Identical data were obtained for the 100 mg/rnL concentration (Figure 2). For deconvolution, a 22-cm-' Gaussian-Lorentzian function was employed (Figure 4). Similar results were obtained for the C85 and C88 mutant sequences. 'The deconvolution results provide quantitative estimates of the band areas at 1647 and 1679 cm-', which correspond respectively to the percentages of helical (70%) and nonhelical (30%) secondary structures. The predicted secondary structure for the 102-residue fragment is 78% helix, 15% sheet, 6% coil, and 1% turn by the method of Gamier et al. (1978) with decision constants DCH = -100 and DCS = -87.5. By the method of Chou and Fasman (1978), the prediction is 67% helix, 12% sheet, 12% coil, and 9% turn.
directly proportional to the respective secondary structure content (Lord, 1977; Williams, 1983). Other assumptions and limitations of this method have been discussed previously (Thomas & Agard, 1984). The results of the amide I deconvolution are shown in Figure 5 . In Table 11, for comprison with other proteins examined previously, we list the Raman
RAMAN SPECTROSCOPY OF
REPRESSOR
e00 820 840 a60 mo Observed (+) and deconvolved (-) spectra in the region 780-880 cm-l showing the tyrosine doublet of (a) aqueous RF1023,5-d2and (b) solid tyrosine-3,5-d2.Frequencies of the deconvolved components of the tyrosine doublet are labeled in each spectrum. The different frequency and intensity patterns in the two compounds are as expected for different interaction energies of the Fermi doublet. FIGURE 4:
4MIOE I REGION
4MlOE I11 REGION
> bm
3
Y
E
1235
1205
cn-1
1335
1550
1600
I650
1700
cm-l
Observed (+) and deconvolved (-) profiles of the amide I (right) and amide 111 (left) regions of the spectrum of the N-terminal fragment, residues 1-1 02. Labels indicate frequencies and assignments as discussed in the text and in Tables I and 11. FIGURE 5:
frequencies and normalized intensities of the amide I components resolved by deconvolution. The percent helical secondary structure of RF102 is 70 f 4%, which is similar to values obtained for coat proteins of filamentous viruses Xf and Pf3 (Thomas, 1985). The percent of amide I intensity not attributable to a-helix secondary structure (30%, contained in the deconvolved band at 1679 cm-') presumably arises from all nonhelical domains of RF102. A very small, virtually negligible, contribution to the 1679-cm-' intensity may also be expected from the C=O vibrations of the side chains of the five glutamine and three asparagine residues of RF102. Also included in Figure 5 is a deconvolution of the amide I11 profile of RF102, which reveals that the major components of amide I11 (near 1280-1300 cm-I) occur at the frequencies expected for a-helix, thus confirming the results of the amide I deconvolution. The minor amide I11 component is resolved
V O L . 2 5 , N O . 2 2 , 1986
6773
by deconvolution at 1244 cm-', which is the frequency expected for irregular backbone structure (Lord, 1977), including 0turns (Seaton, 1986). We note, however, that the approximation of intrinsically equal Raman scattering intensities from different secondary structures contributing to amide I is not necessarily valid also for the amide I11 intensities in proteins (Lord, 1977). Therefore, the relative intensities of the amide I11 components assigned in Figure 5 to helix (1280-1 300 cm-I) and nonhelix (1 244 cm-') domains are not necessarily directly proportional to the percentages of peptide residues in the corresponding backbone conformations, even though this assumption appears to be well founded for amide I (Williams, 1983). The deconvolutions shown in Figure 5 were carried out on experimental data collected from a solution of RF102 at 50 mg/mL in H 2 0 at pH 7.5. The same deconvolution results were obtained for data recorded from 100 mg/mL solutions of the protein. At these conditions RF102 does not exist as a monomer but is associated to form primarily dimers and possibly higher order oligomers as well. Therefore, the secondary structure monitored by the Raman measurements, and estimated quantitatively by deconvolution, is presumed to apply to RF102 in a state close to that of the biologically active repressor dimer. The present results are also in good quantitative agreement with the molecular structure determination by X-ray crystallography (Pabo & Lewis, 1982) and with secondary structure predictions based on sequence analyses (Table I1 footnote). We have also monitored the amide I and amide I11 bands of RF102 for more dilute solutions, viz., 40, 25, 20, 12, and 5 mg/mL. In each case, we observed no change in the contour or peak positions of amide I and amide I11 band envelopes, despite the fact that RF102 exists predominantly as a monomer at the lower protein concentrations (see below). Therefore, we conclude that the secondary structure of RF102 is invariant to dissociation of the dimer of the N-terminal fragment, within the precision (f4%) of the Raman measurements. The results of Figure 5 indicate the possibility that additional unresolved components contribute to the amide I and amide I11 bands of RF102. For example, the deconvolved amide I helix component, peaked at 1647 cm-I, is clearly not symmetrical in shape and gives evidence of a shoulder at 1653 cm-I. Similarly, the amide I11 deconvolution suggests as many as three helix components to the rather broad contour extending from 1286 to 1297 cm-'. At present, there exists neither a theoretical nor an empirical basis for interpreting these multiple components to helix amide I and amide I11 Raman lines. Nor is it clear whether such putative fine structure would originate from different helical domains or from different geometries within contiguous helical stretches. Further study of the deconvolved band shapes of model proteins is required to provide more detailed interpretation of these results. We note as well the partial separation by deconvolution of a very weak shoulder ca. 1235 cm-I to the amide I11 1244-cm-' band. The former is at the position expected for @-sheetstructure (Lord, 1977) and may indicate a very small percentage of @ structure in RF102. In addition to amide I and amide 111 peaks noted above, Figure 5 reveals prominent Raman lines at 1206, 1599, and 1616 cm-I, all of which can be assigned unambiguously to tyrosines and phenylalanines of RF102. Likewise, the Raman lines at 13 18 and 1345 cm-I can be assigned in a straightforward manner to aliphatic side-chain CH, twisting and wagging modes. (Thomas et al., 1983). The line resolved by deconvolution at 1264 cm-l is assigned primarily to tyrosines.
6774
THOMAS ET AL.
BIOCHEMISTRY
.
!
FRAGMENT 1-102
,$.
$00 ng/ml
0 0
hydrogen bonds, as, for example, when the OH group is exposed to solvent H 2 0 molecules or simultaneously t o other donor and acceptor groups. We designate this as the “E” state, which is commonly observed in proteins. (iii) If 1850/1830 = 0.30, then OH is the donor of a strong hydrogen bond to a negative acceptor group, for example, carboxylate and phosphate anions, etc. We designate this the “D” state. Tyrosines in the D state occur in subunits of several RNA plant viruses (Verduin et al., 1984; Prescott et al., 1985). For RF102 at 100 mg/mL and pH 7.5, Figure 2 shows that I g j O / I 8 3 0 has the value 1.51 f 0.07. This value reflects the average hydrogen-bonding environment of the five tyrosines of RF102. It is consistent with any of the following three distributions (but not with any other possible distribution) of tyrosines among the above-defined hydrogen-bonding states:
FRAGHENT 1-102 12 mg/ml
0
0 0
800
820
840
860
800
880
820
840
860
880
+ 2E + I D (1.56) WT-I1 = 3A + OE + 2D (1.62) WT-I11 = 1A + 4E + OD (1.50) WT-I = 2A
m-1
cnl-1
FIGURE 6: Tyrosine doublet intensities of the N-terminal fragment
a t concentrations of 100 (left) and 12 mg/mL (right). Solid curves indicate the experimental data from which the quantity 18s0/1830 was calculated. Open circles indicate the deconvolved band shapes from which the band area ratio A85O/A830 was calculated, as discussed in the text.
The calculated value of I 8 5 0 / I 8 3 0 for each distribution of eq 1 is given in parentheses. ( B ) Concentration Dependence of ~ ~ j O / I 8 3of 0 RF102. In order to determine which distribution of eq 1 is the most probable for RF102 and also to identify specifically which tyrosines of the probable distribution occur in A, E, and D states, we have examined 1850/1830 in wild-type and mutant sequences. Our first goal, however, has been to correlate the value of 1850/1830 with the dissociation of RF102 dimers (or higher order oligomers) into monomers. For this purpose, we have examined the spectra of RF102 down to sufficiently low protein concentrations that the equilibrium concentration of dimers or higher oligomers is small in comparison to monomer concentration, viz., to 5 mg/mL. Figure 6 shows the tyrosine doublet of RF102 at protein concentrations of 100 and 12 mg/mL. Additional data were collected for RF102 concentrations of 50, 40, 25, 20, and 5 mg/mL in the same buffer. The molar ratio of protein to buffer was maintained constant by appropriate dilution of aliquots of the 100 mg/mL stock solution in distilled H 2 0 . Both the observed and deconvolved band shapes are shown in
Consistent with all of the above assignments of the 12001350-cm-’ interval is the fact that the Raman spectrum of deuterated RFI 02 (Figure 2, bottom) manifests isotope shifts only for the 1244- and 1280-1300-cm-’ amide I11 features. Finally, we note the absence of appreciably intense Raman lines between 1550 and 1590 cm-’, consistent with the absence of tryptophan from RF102. Determination of Molecular Environments of Tyrosine Residues Y22, Y60, Y85, Y88, and YlOl in X Repressor N- Terminal Fragment. ( A ) Tyrosine Doublet Intensity Ratio (I850/1830) of M10.2. Siamwiza et al. (1975) have shown that the Raman intensity ratio (1850/1830) of the tyrosine doublet is diagnostic of three different hydrogen-bonding states of the p-hydroxyl group: (i) If 1850/1830 = 2.50, then the OH group is the acceptor of a strong hydrogen bond from a positive donor group. We denote this state as “A” in the discussion that follows. Tyrosines in the A state are observed in Pfl and fd virions (Thomas et al., 1983). (ii) If I 8 j O / I 8 3 0 = 1.25, then OH acts as both donor and acceptor of moderately strong
WILD TYPE
(1-102)
(1)
C85 MUTANT fl-102)
C88 MUTANT (1-102) 2.00-
-
,b _ _i _ -- -
1
1.80. 0
.r.
1.60-
1.40.
v)
c u 0)
c
Y
1.20. -
1.00.
o 20
40
60
rgla I
eo
io0
0
20
40
60
8ghI
EO 100
0
20
’
- .
40 60 8g/6l
C .
’
-
- =
BO 100
7: Plots of the tyrosine doublet intensity ratio as a function of total protein concentration for wild-type N-terminal fragment (a), C85 mutant (b), and C88 mutant (c). In each case, the solid line indicates the plot of 1850/1830 vs. concentration, and the dashed line indicates the plot of & o / A 8 3 0 vs. concentration. See text. FIGURE
RAMAN SPECTROSCOPY OF
VOL. 2 5 , NO. 22, 1986
REPRESSOR
Figure 6. Values of 1 8 5 0 / 1 8 3 0 measured from the unrefined spectra of all samples are plotted as a function of total protein concentration in Figure 7a (solid line). In the same figure we have included a plot (broken line) of the ratio of integrated intensities of the deconvoluted doublet components, Le., A 8 5 0 / A 8 3 @ As noted above, Z850/1830 is the parameter of interest for the structural correlations of Siamwiza et al. (1975). The companion plots of & O / A 8 3 0 , nevertheless, indicate consistency in the intensity measurements and confirm that no underlying bands are present. The data of Figures 6 and 7 indicate a strong dependence of 1 8 5 0 / 1 8 3 0 on RFlO2 concentration. Although 1 8 5 0 / 1 8 3 0 remains constant at 1.57 f 0.05 down to 50 mg/mL, it drops sharply thereafter, reaching the value of 1.20 f 0.05 at 5 mg/mL. If we assume for RF102 a dimer dissociation constant of 5 X 10” (Weiss et a]., 1984), the dimer would be only 10% dissociated at 100 mg/mL but 80% dissociated at 5 mg/mL. Therefore, the value Z850/1830= 1.57 at 100 mg/mL corresponds primarily to the dimer state, while the value 1 8 5 0 / 1 8 3 0 = 1.20 at 5 mg/mL corresponds primarily to the monomer state of RF102. The following are the only tyrosine distributions consistent with the monomer state (calculated 1 8 5 0 / 1 8 3 0 values in parentheses):
+ 3E + 1D (1.31) WT-11’ = OA + 5E + OD (1.25) WT-111’ = 2A + OE + 3D (1.18) WT-I’ = 1A
(2)
Crystallographic and N M R results show that Y22 is packed in the interior of the protein where it is unlikely to form hydrogen bonds with solvent molecules or equivalent donor and acceptor groups (Pabo & Lewis, 1982; Weiss et a]., 1986); Le., Y22 is likely to be either an A- or D-type tyrosine. The N M R data also show that YlOl exists in a random coil state that is exposed to solvent (E type). For these reasons, distributions 11, 11’, 111, and 111’ would require unexpected alterations of RF102 structure, and distributions I and I’ would represent the tyrosine configurations for dimer and monomer, respectively, which are most consistent with N M R and X-ray data. Therefore, we consider the most likely distribution of the five tyrosines to be 2A 2E + 1D for the dimer and 1A + 3E 1D for the monomer, with the requirement that YlOl be of E type in both dimer and monomer for consistency with N M R results. The further assignment of individual tyrosines to specific hydrogen-bonding states requires Raman data from mutant sequences, as next discussed. (C‘) Tyrosine Doublet of C85 Mutant Repressor Fragment. The 102-residue fragment containing cysteine instead of tyrosine at position 85 (hereafter RF102-C85) is expected to provide the value of 1850/1830 characteristic of Y22, Y60, Y88, and Y101. Comparison with results of the preceding section should then permit the hydrogen-bonding state of Y85 in RF102 to be deduced. The validity of this approach requires that the wild-type (RF102) and mutant (RF102-C85) sequences contain the same global structure or at least that the structures do not differ in such a way as to affect hydrogenbonding environments of other tyrosines. This requirement is apparently well met, since the Raman spectra of RF102 and RF102-C85 show no significant differences in their conformation-sensitive amide I and amide I1 bands (G. J., Thomas, Jr., and M. A. Weiss, unpublished results). Similarly, the N M R spectra of RF102 and RF102-C85 indicate no differences in tertiary or quaternary interactions (Weiss et al., 1986). The mutant fragment also binds operator as strongly as the wild-type fragment and forms a more stable dimer than does the wild-type fragment; Le., RF102-C85 exhibits an association
+
+
6775
constant at least 10-fold greater than that of RF102 (Sauer et al., 1986). Therefore, little or no dissociation of RF102-C85 dimer to monomer is expected to occur at the conditions of the present experiments, even for a total RF102-C85 concentration of only 5 mg/mL. Figure 7b shows the values of 1850/1830 and A 8 s O / A 8 3 0 measured over the range 5-100 mg/mL. The data indicate significant concentration dependence of the tyrosine doublet intensity. Since RF102-C85 is essentially all dimer at 5 mg/mL, the results of Figure 7b suggest that formation of larger oligomers may take place at higher protein concentrations, with attendant changes in tyrosine hydrogen-bonding environment. A gradual change is indicated by the shape of the 1 8 5 0 / 1 8 3 0 curve (solid line in Figure 7b), and a limiting state of oligomerization (or aggregation) may not have been reached at the highest concentration studied. We find 1 8 5 0 / 1 8 3 0 = 1.34 f 0.05 at 5 mg/mL (RF102-C85 dimer) and I 8 5 0 / 1 8 3 0 = 1.87 f 0.07 at 100 mg/mL (N-mer). The tyrosine distributions consistent with the RF102-C85 dimer state are
+ 2E + 1D (1.32) C85-I1 = 2A + OE + 2D (1.40) (285-1 = 1A
(3)
Since YlOl should remain as an E-type tyrosine in RF102-C85, the distribution (28.5-I1 of eq 3 can be rejected. Therefore, the RF102-C85 dimer is deduced to contain the distribution 1A 2E lD, which means that the net result of the C85 mutation has been the elimination of a tyrosine of the A type. This conclusion does not unambiguously identify Y85 of the wild-type sequence as the acceptor tyrosine, however. It is still possible that the roles of two or more tyrosines may have been interchanged to produce the observed net result. The present data do not provide further information on either the extent of RF102-C85 oligomerization at higher protein concentrations or the possible heterogeneity of N-mers that may be present. However, the results clearly show, by virtue of the large increase in 1 8 5 0 / 1 8 3 0 with total protein concentration, that there is a net transfer of tyrosines from states of lesser to greater hydrogen-bonding acceptor roles. The only distributions consistent with the N-mer state are 3A + OE + 1D and 2A + 2E OD. Again, the former may be excluded on the basis of the required E role for Y101. Therefore, no D-type tyrosine is indicated for RF102-C85 at 100 mg/mL. This further imposes the constraint that Y85 cannot assume the D role in the wild-type dimer RF102. At this point it is appropriate to consider the probable hydrogen-bonding roles of individual tyrosines of the wild-type dimer consistent with the observed Raman (and NMR) data: not E type (probably A or D) Y22: Y60: E or D type not D type (probably A or E) Y85: Y88: A, D, or E type Y101: E type In order to resolve unambiguously the hydrogen-bonding states of Y22, Y60, Y85, and Y88, we next consider the mutant containing C88 in place of Y88. ( D ) Tyrosine Doublet of C88 Mutant Repressor Fragment. The average hydrogen-bonding environment of Y22, Y60, Y85, and YlOl is revealed by the tyrosine doublet intensity ratio from the 102-residue fragment containing a cysteine mutation at position 88 (RF102-C88). This mutant, RF102-C88, contains an intermolecular disulfide bond between the C88 side chains of two subunits and is therefore constrained to exist as a “covalent dimer” even at low concentrations (Weiss et al., 1984). Not surprisingly, both 1 8 5 0 / 1 8 3 0 and As50/As30 exhibit no dependence on concentration over the
+
+
+
6776
BIOCHEMISTRY
THOMAS ET AL.
Table 111: Hydrogen-Bonding States of Tyrosines of X N-Terminal Fragment (Residues 1-102) sequence/structureb tyrosine hydrogen wild type/dimer Y22 Y60 Y85 Y88 YlOl wild type/monomer
Y22 Y60 Y85 Y88 YlOl
Repressor bonding state" A E A D E A E E D E
"The acceptor (A), donor (D), and combined acceptor/donor (E) states for hydrogen bonding of the phenolic OH group of tyrosine are defined in the text. See also Siamwiza et al. (1975). bThe amino acid sequence (one-letter symbols) for the 102-residue fragment of the wild-type X repressor is 'STKKKPLTQEQLEDARRLKAIYEKKK-
NELGLSQESVADKMGMGQSGVGALFNGINALNAYNAALLAKILKVSVEEFSPSIAREIYEMYEAVSMQPSLRSEYE'02. The C85 mutant contains one less A-type tyrosine, presumed to be Y85. The C88 mutant contains one less D-type tyrosine, presumed to be Y88. At high concentrations, the C85 (but not the C88) mutant appears to further aggregate with the conversion of Y88 from D type to E tY Pe.
range 5-100 mg/mL (Figure 7c). These results also provide an excellent control for the constancy of the tyrosine doublet intensity ratio when no change occurs in tyrosine hydrogenbonding states. Apparently, unlike RF102-C85, this sequence does not further oligomerize a t higher concentrations. (Alternatively, even if oligomerization does take place, no net change in the tyrosine hydrogen-bonding states results therefrom.) W e find for RF102-C88 dimer Is50/ls30 = 1.95 f 0.07, indicating the distributions of eq 4. It is interesting (288-1 = 2A
+ 2E + OD (1.88)
(4)
to note that the distribution of eq 4 was also obtained for the so-called N-mers of RF102-C85 (see Tyrosine Doublet of C85 Mutant Repressor Fragment). This suggests that the putative hydrogen-bonding interaction that mediates further association of RF102-C85 dimers, i.e., the additional acceptor role for one of the tyrosines of RF102-C85, cannot occur for RF102-C88. It is therefore likely that the phenolic oyxgen of Y88 may act as the additional hydrogen bond acceptor in the association of RF102-C85 beyond the dimer state. The Raman results obtained on RF102, RF102-C85, and RF102-C88 are thus seen to be internally consistent. They indicate that Y88 is of type D and, consequently in conjunction with the earlier results, Y22 and Y85 are of type A, while Y60 and YlOl are of type E in the repressor fragment dimer. These assignments are summarized in the upper section of Table 111. From these results follow also the assignments indicated in the lower section of Table I11 for the RF102 monomer. ( E ) Tyrosine Doublet of Fragment Containing a Carboxymethylated Cysteine Residue at Position 88. Reduction of the covalent dimer of RF102-C88 and selective carboxymethylation of C88 with iodoacetic acid results in the formation of an N-terminal fragment containing only four tyrosines (Y22, Y60, Y85, and Y101) and chemically blocked from dimerization (Weiss et al., 1986). W e have examined the Raman spectrum of this fragment and found ISS0/1S30 = 1.65 f 0.15, which corresponds to the distribution 1A 3E + OD that is expected (Table 111) for the monomer lacking Y88. The amide I, 1', 111, and 111' Raman bands of this derivative are also in accord with the data obtained from RF102, confirming the same secondary structure in monomer
+
and dimer forms of the N-terminal domain. (F)Tyrosine Doublet of Wild-Type Fragment Containing Residues 5-98. The Raman spectra of the wild-type fragment containing residues 5-98 provides additional confirmation of the assignments of Table 111. This sequence lacks Y101, and the observed value of 1.74 f 0.15 for ZsS0/ls30 indicates that the hydrogen-bonding states of Y22, Y60, Y85, and Y88 are consistent with the distribution 2A 1E 1D.
+
+
DISCUSSION Secondary Structure. W e have found that 70 f 4% of the 102 peptide residues of the repressor N-terminal fragment (RF102) are incorporated into regions of a-helical secondary structure when the protein exists as a dimer in aqueous solution. This is in reasonable accord with the value of 61% helix obtained by X-ray diffraction analysis of the 1-92 fragment (RF92) (Pabo & Lewis, 1982). In determining the value of 61% helicity in the crystal structure, we have followed the convention of Pabo and Lewis in excluding all of the ( I O ) residues a t the ends of the helices (Figure 1). Were these residues to be included, a value of 7 1% helix would be obtained for RF92, virtually identical with the helicity of aqueous RF102 as determined by Raman spectroscopy. Thus, the X-ray (crystal) and Raman (solution) structures appear to be in complete agreement. The overall secondary structures and presumably also the tertiary folding of the repressor fragments RF102 (aqueous dimer) and RF92 (crystal dimer) are therefore very similar to one another, irrespective of the different morphological states of the samples. The present Raman results are also in substantial agreement with the N M R findings on aqueous RF92 (Weiss et al., 1986). The secondary structure observed for the wild-type fragment is not perturbed by the mutations Y85 C85 and Y88 C88. The Raman data further indicate that the predominantly a-helical secondary structure of RF102 dimer is retained upon its dissociation into monomers. This is consistent with previous observations that these genetically altered repressors bind D N A a t least as well as wild-type repressor (Sauer et al., 1986). We note that predictions of secondary structure (Chou & Fasman, 1978; Garnier et al., 1978) are also in agreement with those observed experimentally for wild-type and mutant sequences. Tyrosine Hydrogen Bonding. Y22 is shown by the X-ray structure to be near the end of the first a-helical domain (residues 9-23) of the repressor fragment (Pabo & Lewis, 1982). N M R studies indicate that the Y22 side chain is buried in the hydrophobic interior of the protein (Weiss et al., 1986). The present results confirm that the phenolic OH group of Y22 is not exposed to solvent hydrogen bonding and demonstrate further that this group is the acceptor of a strong hydrogen bond from a positive donor group. The acceptor hydrogen-bonding role of Y22 is indicated by the Raman experiments for both the dimer and monomer states of RF102. This latter finding is consistent with the crystal dimer, which contains no quaternary interactions involving the first four helices (Figure l), and with 'H N M R studies, which indicate no change in tertiary structure accompanying dimerization. Examination of the model of Pabo and Lewis (1 982) suggests that the donor group of the S32 side chain could serve as a candidate for donating a hydrogen bond to the phenolic oxygen of Y22. Other candidates as positive donors are the termini of the side chains of K25 and K26. However, these are not favorably oriented in the crystal structure to permit hydrogen-bonding with Y22. Hydrogen bonding of the Y22 acceptor with either of the lysyl donors in the aqueous repressor would require alteration of the tertiary structure found in the crystal.
-
-
RAMAN SPECTROSCOPY OF
VOL. 25, N O . 2 2 , 1986
REPRESSOR
Y85 and Y88 are located in the X-ray structure at or near the dimer interface (Pabo & Lewis, 1982). Although NMR studies provide no detailed information about tyrosine hydrogen bonding, they do reveal that Y88 resonances are shifted to high field, consistent with the mutual stacking of Y88 and Y88’ in the dimer and with the proximity of Y85 (Figure 1). Not surprisingly, the Raman experiments demonstrate that the hydrogen-bonding roles of these tyrosines are most affected by the dimer/monomer dissociation (Table 111). Although Y60 is deduced here to retain an E-type role in both dimer and monomer forms of RF102 (Le., moderate hydrogen bonding of the O H group as both donor and acceptor), it is possible that its hydrogen-bonding partners may be quite different in the two states. The X-ray structure indicates that there are many side chains that could act as suitable candidates for both donor and acceptor hydrogen-bonding roles with Y60 (Pabo & Lewis, 1982). In the case of Y85 and Y88, one additional caution should be emphasized. The A role for Y85 and the D role for Y88 are predicated upon the assumption that each tyrosine retains its normal (wild-type) role when the other is mutated to cysteine. However, the model of Pabo and Lewis (1982) shows that the two tyrosine rings are situated almost on the same side of helix 5. We, therefore, should consider the possibility that the normal hydrogen-bonding partner of Y85 (a strong electropositive donor in the RF102 dimer) could become accessible to Y88 in the RF102-C85 mutant and vice versa for the RF102-C88 mutant. In such a case, one or more additional tyrosines would also be required to change hydrogen-bonding roles to preserve the correct value of the tyrosine doublet intensity ratio in each mutant. We view such a scenario as unlikely to take place without disruption of the native tertiary and quaternary structures. Nevertheless, we cannot categorically exclude this possibility on the basis of the available data. In fact,. an interchange of hydrogen-bonding partners between Y85 and Y88 in the wild-type repressor would also go undetected in the Raman spectra. CONCLUSIONS The present study shows that laser Raman spectroscopy in combination with site-specific mutagenesis can be employed to determine the hydrogen-bonding roles of individual tyrosines in globular proteins. The hydrogen-bonding environments of the five tyrosines of the 102-residue N-terminal fragment of the X repressor have been resolved for both the dimer and the monomer states. This investigation represents the first application of classical Raman scattering spectroscopy to a protein as dilute as 5 mg/mL in aqueous solution. The success of the method depends upon the use of signal averaging to enhance the spectral intensities and Fourier deconvolution to enhance the spectral resolution in the frequency domain. In addition to the information obtained about the molecular environments of tyrosine residues, the Raman spectra have provided detailed quantitative information on the repressor secondary structure. The total a-helical content of the repressor fragment (residues 1-102) was determined to be 70 f 4%. This value is unchanged by dissociation of the dimer in aqueous solution. It is not inconsistent with the 61% helical content deduced from the crystal structure of a smaller fragment (residues 1-92). In a recent paper (Prescott et al., 1986), we have described the use of Raman spectroscopy to determine conformational properties of the DNA operator site OL1,which is the strongest binding site in the phage X genome. Future studies will employ the Raman spectrum to detect and identify the nature of
6777
conformational changes and molecular interactions between X repressor and operator DNA. ACKNOWLEDGMENTS We thank Anna Jeitler-Nilsson, Kathy Hehir, and Donna Sargent for help with protein purification, Professor Carl Pabo (Johns Hopkins School of Medicine) and Mitchell Lewis for crystal coordinates, and Professors Robert T. Sauer (MIT), Martin Karplus (Harvard), and Richard C. Lord (MIT) for helpful discussions. We are also indebted to Robert Stearman, whose original work led to the availability of the mutant repressors, and to Eric Suchanek for contributions regarding experiment design. Registry No. L-Tyrosine, 60- 18-4. REFERENCES Chadwick, P., Pinota, V., Steinberg, R., Hopkins, N., & Ptashne, M. (1970) Cold Spring Harbor Symp. Quant. Biol. 35, 283-294. Chou, P. Y., & Fasman G. (1978) Ado. Enzymol. Relat. Areas Mol. Biol. 47, 45-148. Eliason, J. L., Weiss, M. A,, & Ptashne, M. (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 2339-2343. Gamier, J., Osguthorpe, J., & Robson, B. (1978) J. Mol. Biol. 120, 97-120. Hecht, M. H., & Sauer, R. T. (1985) J. Mol. Biol. 186, 53-63. Hecht, M. H., Nelson, H. C. M., & Sauer, R. T. (1983) Proc. Natl. Acad. Sci. U.S.A. 80, 2676-2680. Johnson, A. P., Pabo, C. O., & Sauer, R. T. (1980) Methods Enzymol. 65, 839-856. Lewis, M., Jeffrey, A,, Wang, J., Ladner, R., Ptashne, M., & Pabo, C. 0. (1983) Cold Spring Harbor Symp. Quant. Biol. 47, 653-673. Lewis, M., Wang, J., & Pabo, C. (1985) in Biological Macromolecules and Assemblies (Jurnak, F. A., & McPherson, A., Eds.) Vol. 2, pp 265-287, Wiley, NY. Li, Y., Thomas, G. J., Jr., Fuller, M., & King, J. (1981) Prog. Clin. Biol. Res. 64, 271-283. Lord, R. C. (1977) Appl. Spectrosc. 31, 187-194. Nelson, H. C. M., & Sauer, R. T. (1985) Cell (Cambridge, Mass.) 42, 549-558. Nelson, H. C. M., Hecht, M. H., & Sauer, R. T. (1983) Cold Spring Harbor Symp. Quant. Biol. 47, 441-449. Pabo, C. O., & Lewis, M. (1982) Nature (London) 298, 443-447. Pabo, C. O., & Sauer, R. T. (1984) Annu. Reu. Biochem. 53, 293-321. Pabo, C . O., Sauer, R. T., Sturtevant, J., & Ptashne, M. (1979) Proc. Natl. Acad. Sci. U.S.A. 76, 1608-1612. Prescott, B., Sitaraman, K., Argos, P., & Thomas, G. J., Jr. (1985) Biochemistry 24, 1226-1231. Prescott, B., Benevides, J. M., Weiss, M. A., & Thomas, G. J., Jr. (1986) Spectrochim. Acta, Part A 42A, 223-226. Sauer, R. T., Pabo, C. O., Meyer, B. J., Ptashne, M., & Backman, K. C. (1979) Nature (London) 279, 396-400. Sauer, R. T., Yocum, R. R., Doolittle, R. F., Lewis, M., & Pabo, C. 0. (1982) Nature (London) 298, 447-451. Sauer, R. T., Hehir, K., Stearman, R., Weiss, M. A., Jeitler-Nilsson, A., Suchanek, E., & Pabo, C. 0. (1986) Biochemistry (submitted for publication). Seaton, B. A. (1986) Spectrochim. Acta, Part A 42A, 227-232. Siamwiza, M. N., Lord, R. C., Chen, M. C., Takamatsu, T., Harada, I., Matsuura, H., & Shimanouchi, T. (1975) Biochemistry 14, 4870-4876.
Biochemistry 1986, 25, 6778-6784
6778
Thomas, G. J., Jr. (1985) Spectrochim. Acta, Part A 41A, 2 17-221. Thomas, G. J., Jr. (1986) in Biological Applications of Raman Spectroscopy (Spiro, T. G., Ed.) Vol. 1, Wiley, New York (in press). Thomas, G. J., Jr., & Agard, D. A. (1984) Biophys. J . 46, 763-768. Thomas, G . J., Jr., Prescott, B., & Day, L. A. (1983) J . Mol. Biol. 165, 321-356.
Verduin, B. J . M., Prescott, B., & Thomas, G . J., Jr. (1984) Biochemistry 23, 4301-4308. Weiss, M. A., Karplus, M., Patel, D. J., & Sauer, R. T. (1983) J . Biomol. Struct. Dyn. 1, 151-157. Weiss, M. A., Sauer, R. T., Patel, D. J., & Karplus, M. (1984) Biochemistry 23, 5090-5095. Weiss, M. A., Pabo, C. O., Karplus, M., & Sauer, R. T. (1986) Biochemistry (submitted for publication). Williams, R. W . (1983) J . Mol. Biol. 166, 581-603.
'H NMR Studies on Bovine Cyclophilin: Preliminary Structural Characterization of This Specific Cyclosporin A Binding Protein' D. C. Dalgarno,* M. W. Harding,s A. Lazarides,§ R. E. Handschumacher,§ and I. M. Armitage*,'$l Departments of Molecular Biophysics and Biochemistry, Diagnostic Radiology, and Pharmacology, Yale University School of Medicine, New Hacen, Connecticut 06510 Received April 14, 1986: Reuised Manuscript Received July 24, 1986
High-field 'H NMR spectroscopy has been used to study the conformation of the cytosolic cyclosporin A binding protein cyclophilin. For the drug-free form of cyclophilin, spectral editing methods in conjunction with a p H titration were used to identify all four His residues present in the protein, and two-dimensional COSY and R E L A Y spectroscopy was used to elucidate the scalar connectivities in the aromatic and upfield methyl regions of the spectrum. From these scalar connectivities, it was possible to distinguish between inter- and intraresidue dipolar interactions within the aromatic and upfield methyl regions of cyclophilin in the NOESY spectrum. T h e results of this analysis showed extensive interresidue crossrelaxation among and between these latter spectral regions indicative of the proximal relationships of several of these residues and the presence of a hydrophobic core within cyclophilin. ABSTRACT:
c y c l o p h i l i n is the specific cytosolic binding protein responsible for the concentration of the immunosuppressant cyclosporin A (CsA) by lymphoid and nonlymphoid mammalian cells (Merker & Handschumacher, 1984). Furthermore, cyclophilin binds a series of cyclosporin analogues in proportion to their immunosuppressive activity in a mixed lymphocyte reaction, consistent with the hypothesis that cyclophilin plays an important role in the immunosuppressive action of CsA (Handschumacher et al., 1984). Cyclophilin has been purified to homogeneity from bovine thymus and human spleen tissues. Two isoforms were identified from both of these sources that bind one molecule of CsA, have an apparent molecular weight of 17 kDa, and possess very similar amino acid compositions. The complete 163 amino acid sequence of bovine cyclophilin has recently been determined ( M , 17 737 Da) and found to contain a single Trp and four Cys residues. The first 72 ",-terminal residues of human cyclophilin were also determined and found to be identical with bovine cyclophilin, indicating a highly conserved structure. No significant sequence homologies with other 'This work was supported by grants from the National Institutes of Health (AM18778 and CA09200) and the American Cancer Society (CH67) and benefitted from instrumentation provided through the shared instrumentation programs of the National Institute of General Medical Science (Grant GM 3224351) and the Division of Resources of the NIH (Grant RR02379). *Address correspondence to this author at the Department of Molecular Biophysics and Biochemistry. *Department of Molecular Biophysics and Biochemistry. 5 Department of Pharmacology. 1' Department of Diagnostic Radiology.
protein sequences in the National Biomedical Research Foundation data base were detected (Harding et al., 1986). Additional physicochemical studies have shown that the CsA-binding activity of cyclophilin is sulfhydryl dependent and that the protein shows a 2-fold enhancement in the intrinsic fluorescence of its single Trp residue consequent to the binding of CsA (Handschumacher et a]., 1984). These characteristics suggest that the Trp residue and probably a single Cys residue are critical determinants in the CsA binding site. Our long-term objective is to use ' H nuclear magnetic resonance (NMR) methods to elucidate the molecular details of the interaction between CsA and cyclophilin. Ultimately, this necessitates identification in the 'H N M R spectrum of the amino acid residues involved in the CsA binding site and, subsequently, characterization of any protein conformational changes induced by drug binding. In this study, we present initial results in which both one- and two-dimensional ' H N M R methods have been used to identify several of the spin systems in the 'H N M R spectrum of the major isoform of bovine cyclophilin. These data were subsequently used to identify several spin systems coupled by interresidue crossrelaxation pathways in the NOESY spectrum. The identity and extent of the interresidue distance constraints derived from these data provide evidence for the the existence of a very hydrophobic core within the protein of suggested importance in the interaction with CsA. MATERIALS A N D METHODS Protein Preparation. The major isoform of calf thymus cyclophilin was purified to homogeneity as described previously
0006-2960/86/0425-6778$01.50/00 1986 American Chemical Society
