Amide proton titration shifts in bull seminal inhibitor IIA by two-dimensional correlated 1H nuclear magnetic resonance (COSY)

J. Mol. Biol. (1984) 179, 283-288

LETTER TO THE EDITOR

Amide Proton Titration Shifts in Bull Seminal Inhibitor IIA by Two-dimensional Correlated ‘H Nuclear Magnetic Resonance (COW) Manifestation

of Conformational

Equilibria

Involving Carboxylate Groups

Amide proton titration shifts in H,O solution of bull seminal inhibitor IIA were measured over the pH range from 3 to 6 using two-dimensional correlated spectroscopy. These data enabled characterization of the pK, values for the majority of the carboxylate groups in the protein. Two glutamate side-chains were found to form hydrogen bonds with their own backbone amide proton. Different temperature variations of the populations of these local, cyclic structure elements are indicated for the individual sites.

In hull seminal inhibitor IIA the ‘H nuclear magnetic resonance spectrum was previously assigned (8trop et al., 1983) and the secondary structure in solution

was determined (Williamson et al., 1984). This letter describes the amide proton titration shifts in this protein and discusses structural implications of these data. From studies with model peptides it is known that only a very limited number of amide protons in a polypeptide chain show sizeable intrinsic titration shifts arising from through-bond interactions with ionizeable groups (Bundi 8: Wiithrich, 1977,1979). Thus, in the amino acid sequence of BUS1 IIAt (Meloun & Cechova, 1979; Strop et al., 1983) the C-terminal Cys57 would be expected to show an upfield shift with increasing pH of about -0.4 p.p.m., Cys7 should experience an upfield shift of about -0.06 p.p.m. due to the titration of Asp6, and the amide protons of the two aspartic acids in positions 6 and 12 should show upfield shifts of about -0.03 p.p.m. (residue 13 is proline, so that no observation is possible in this position).

In addition,

downfield

titration

shifts can arise from conformation-

dependent through space hydrogen bonding interactions between amide protons and carboxylate groups (Bundi & Wiithrich, 1979; Arseniev et al., 1981). All the amide protons in BUS1 IIA which show large titration shifts exchange too rapidly with the solvent to be observed by n.m.r. in 2H20 solutions of the protein, and the one-dimensional ‘H n.m.r. spectra recorded in H,O are too crowded for all amide proton lines to be unambiguously resolved and assigned. Therefore we recorded COSY spectra (Aue et al., 1976; Nagayama et al., 1980; Bax & Freeman, 1981) in a H,O solution of BUSI IIA at different pH values in the range from 3.2 to 5.4. Experimental details are given in the legends to Figures 1 and 2. Measurements were obtained at two temperatures, 18°C and 45”C, for t Abbreviations magnetic

nuclear

used: BUS1 IIA, bull seminal resonance; COSY, two-dimensional

0022-2836/84/300283%06

$03.00/O

inhibitor IIA; p.p.m., parts correlated spectroscopy. 283

0

1984 Academic

per million;

Press Inc. (London)

n.m.r.,

Ltd.

L

s’

Y16

$626

5 ?

C25 I

/

9

8

i

w2 ( p.p.m.) Fro. I. Contour plot of the spectral region (0, = 3.7 to 5.3 p.p.m., o2 = 7.0 to 9.6 p.p.m.) of two 500 MHz iH COSY spectra recorded at 45°C and at different pH values ((a) pH 4.0; (b) pH 5.2) in a 0.007 M-solution of BUS1 IIA in a mixed solvent of 90% H,O and 10% *H,O. which contained 0.1 M-NaCI. The spectra were obtained with a Bruker WM 500 spectrometer, using the standard COSY pulse sequence and phase cycles (Nagayama et al., 1980; Strop et al., 1983). The solvent resonance was suppressed by selective irradiation at all times except during t, and t, (Wider et al., 1983). Prior to Fourier transformation the time domain data matrix was multiplied along ti by sin (x(t+t,)/t,), with t,/t. = l/32, and along t, by sin*(x(t+t,)/t,), with t,/t. = l/64. The spectra are shown in the absolme value representation. The digital resolution is 5.8 Hz/point, the spectra were recorded in about 13 h. The spectral region presented here contains the “fingerprint” of the protein, i.e. the amide proton-C” proton cross peaks. Resonance assignments (Strop et al., 1983) are indicated by the one-letter symbol for the amino acid residue and the position in the amino acid sequence. Underlined symbols identify residues with large amide proton titration shifts. In (a) only the cross peaks of the residues discussed in the text are indicated; in (b) all observed peaks are identified. (Because of the water irradiation the cross peaks of Thrl8, Tyr32 and His53 are bleached out and those of Asn28 and Thr31 have very low intensity.)

LETTER

TO THE

EDITOR

286

5 li ci d 3 4

5

I

9

8 w2 (p.p.m.1

FIN. 2. Same as Fig. 1, except that the temperature the cross peaks of Asp6, Tyrl6 3 peaks identified by an asterisk correspond water irradiation or to ageing of the protein

temperature was 18°C ((a) pH 3.8; (b) pH4.5). At this and Gly26 are bleached out by the water irradiation. The to “impurities”. They might be due to artifacts from the solution (Strop et al., 1983).

which sequence-specific resonance assignments at pH 4.9 were previously obtained (Strop et al., 1983). Examples of the “fingerprint regions” in the spectra used are shown in Figures 1 and 2. The COSY fingerprint of a protein includes all amide proton-C” proton cross peaks (Wagner & Wiithrich, 1982), and in the spectra with pH values nearest to 4.9 the sequence-specific cross peak assignments are indicated (Figs 1 and 2). In the other spectra only the cross peaks with sizeable

2X6

S. EBINA

ANI)

K. WtfTHRI(‘H

tit,ration shifts (underlined) and those of Asp6, AspIS, Glu30 and the C-terminal Cys57 are identified. Between about pH 3-6, where the influence of acid denaturation becomes noticeable on numerous resonance lines (Strop & Wiithrich, 19X3), and pH 5.5 only five cross peaks in the COSY fingerprint of BUS1 IIA show amide proton titration shifts larger than 045 p.p.m. These peaks could readily be identified and their chemical shifts measured in the spectra recorded at different pH values. In Figure 3 the experimental points were fitted to one-proton titration curves, and 9.c )-

8.e

8.6

9.0

8.8

8.6 I

E 4 8.5 a 00 8.3

8.2

8.0

7.8

8.3

_--

.-a-

/ !

C57 ( C57, bockbone-----* ~. ~_

COOH)

8.1 3

4

5

6

PH

FIG. 3. Plots of the amide proton chemical shifts, 6, versus pH for 5 residues in BUS1 IIA. The continuous lines correspond to non-linear least-squares fits of the experimental points at 18°C (0) and 45°C (w) to one-proton titration curves (Bundi & Wiithrich, 1979). The pK, values at 18°C and 45°C are also indicated. For Ala8 a titration shift of 0.07 p.p.m. between pH 3.6 and 5.5 at 18°C could be fitted to a one-proton titration curve with pK, = 4.2 (not shown). At 45°C the corresponding shift was only about 0.02 p.p.m. (Figs 1 and 2). Amino acid residues which were found to cause the observed amide proton titration shifts are listed in parentheses (see the text).

LETTER

TO THE

EDITOR

287

the resulting pK, values are indicated in the Figure. The results of the fitting procedure support the conclusion that for the titrating amide groups the chemical shifts are not noticeably affected by protein denaturation down to pH 3.2 at 18°C and to pH 3.4 at 45°C. Inspection of Figures 1 to 3 shows that only one of the expected intrinsic titration shifts in BUS1 IIA was observed, i.e. for Cys7, which yields pK, values for Asp6 of 4-O at 18°C and 4.1 at 45°C. The intrinsic shifts for Asp6 and Asp12 are probably too small to be reliably measured with the digital resolution of the presently used COSY spectra, and at 18°C (Fig. 2) the cross peak of Asp6 is furthermore bleached out by the water irradiation (Wider et al., 1983). For Cys57 the onset of the titration seems to be near pH 3.5, and since a titration shift of 2 -0.4 p.p.m. is expected we conclude that the pK, value for the C terminus is < 3.0. The downfield titration shifts for Ala8, Glu9, Lysl4 and Glu20 must arise from hydrogen bonding interactions with carboxylate groups (Bundi & Wiithrich, 1979). The residues which cause these titration shifts are identified in Figure 3. These identifications relied on inspection of the three-dimensional structure of BUSI IIA (M. P. Williamson, T. Have1 & K. Wiithrich, unpublished results) and consideration of the pK, values for the different titrating amide protons, which showed that all other intramolecular interactions with carboxylates could for steric reasons be excluded. The tendency of glutamate residues to form cyclic structures with intra-residue hydrogen bonds is well documented (Bundi & Wiithrich, 1979; Mayer et al., 1979), and the pK, values for E9 and E20 (Fig. 3) are typical for solvent-accessible glutamic acid residues. The titration observed for Ala8 further indicates that in a small proportion of the molecules the side-chain of Glu9 interacts with the amide proton of Ala8. With regard to the determination of the tertiary structure of BUS1 IIA the results of Figure 3 contribute three local structural constraints, i.e. the intraresidual ring closures for Glu9 and Glu20 and the hydrogen bond from the sidechain carboxylate of Asp12 to the amide proton of Lysl4. For the local conformation of the dipeptide segment Ala8-Glu9 there is the indication that interactions between the side-chain carboxylate group and the amide proton of Ala8 a’re geometrically feasible, and the absence of a titration shift for Cys57 implies that’ the C terminus is not freely accessible to the solvent. For Glu30, where the amide proton is involved in the hydrogen bonding network of the antiparallel B-sheet of BUS1 IIA (Williamson et al., 1984), there is no evidence for side-chain interactions with backbone protons (Figs 1 and 2). For further physico-chemical studies of BUS1 IIA, in particular on the acid denaturation and the pH dependence of the amide proton exchange rates (Strop & Wiithrich, 1983; S. Ebina, G. Wagner & K. Wiithrich, unpublished work), it is of interest that the pK, values of all except one of the ionizeable groups which titrate between pH 3.0 and 8.0 are now known. The exception is Glu30, as discussed above. Alternative methods for measuring the pK, values given in Figure 3 would be by pH titration in the 13C n.m.r. spectra or in the aliphatic region of ‘H COSY spectra recorded in *H,O. Both techniques tend to be more laborious than the presently described experiments.

IXX

S. EI(I.NA

ASL)

K. N’V’THKL(‘H

Perhaps the most exciting observation in Figure 3 is that the extent of the titration shifts caused by XH ~ ~carboxglate hydrogen bonds and its temperature dependence is different for the individual groups. This indicates that amide proton titration can be used to monit,or t,he populations of local structural features under conditions that are non-denaturing for the overall globular protein conformation. For example, for Glu20 the titration shift is 0.35 p.p.m. at 18°C and 0.12 p.p,m. at 45°C; for Glu9 the corresponding numbers are 0.51 p.p.m. at 18Y’ and 0.43 p.p.m. at 45”C, and for Lysl4 0.30 p.p.m. at 18°C and 0.36p.p.m. at 45°C. These different titration shifts manifest primarily different populations of the hydrogenbonded local structures (Bundi & Wiithrich, 1979). For Glu20 the hydrogenbonded cyclic form would thus be approximately three times more abundant at 18°C than at 45”C, for Glu9 there is only a slight decrease of about 20yo at the higher temperature, and for Lysl4 there is actually an increase of about< 20% at 45°C. While proper analysis of these observations will need more work, it would appear at a first glance that the hydrogen-bonded forms of Glu9 and Glu20 are predominantly stabilized by hydrophilic interactions, whereas the interacticJn of Lysl4 NH with Asp12 appears to be stabilized t,o a larger extent by hydrophobic interactions (Kauzmann. 1959). Financial support by the Schweizerischer Nationalfonds (project 3.284.82 and international postdoctoral fellowship to S.E.) is gratefully acknowledged. We thank Drs D. cechov8 and P. &rop for a gift of BUS1 IIA, Drs T. Havel, G. Wagner and M. P. Williamson for helpful discussions, and Mrs E. Huber for the careful preparation of the manuscript. Institut fiir Molekularbiologie und Biophysik Eidgeniissische Technische Hochschule-Hijnggerberg CH-8093 Ziirich. Switzerland.

SATOSHI EBINA KURT W~~THRICII

Received 1 May 1984 REFERENCES Arseniev, A. S., Pashkov, V. S., Pluzhnikov. K. A., Rochat. H. & Bystrov, 1’. F. (1981). Eur. J. Biochem. 118, 453462. Aue, W. P., Bartholdi, E. & Ernst, R. R. (1976). J. Chem. Phys. 64, 2229-2246. Bax, A. & Freeman, R. (1981). J. Magn. Reson. 44, 542-561. Bundi, A. & Wiithrich, K. (1977). PEBS Letters, 77, 11-14. Bundi, A. & Wiithrich, K. (1979). Biopolymers, 18, 285-298. Kauzmann, W. (1959). Advan. Protein Chem. 14, l-63. Mayer, R., Lancelot, G. & Spach, G. (1979). Biopolymers, 18, 1293-1296. Meloun, B. & CechovB, D. (1979). Coil. Czechoslov. Chem. Commun. 44, 2710-2720. Nagayama, K., Anil Kumar, Wiithrich, K. & Ernst, R. R. (1980). J. Magn. Reson. 40, 321334. Strop, P. & Wiithrich, K. (1983). J. Mol. Biol. 166, 631-640. Strop, P., Wider, G. & Wiithrich, K. (1983). J. Mol. BioE. 166, 641-667. Wagner, G. & Wiithrich, K. (1982). J. Mol. Biol. 155, 347-366. Wider, G., Hosur, R. V. & Wiithrich, K. (1983). J. Magn. Reson. 52, 130-135. Williamson, M. P., Marion, D. & Wiithrich, K. (1984). ,I. Mol. Biol. 173, 341-359. Edited by H. E. Huxley