Eur. J. Biochem. 62, 103-107 (1976)
A Study of the Lysyl Residues in the Basic Pancreatic Trypsin Inhibitor using 'H Nuclear Magnetic Resonance at 360 MHz Larry R. BROWN, Antonio DE MARCO, Gerhard WAGNER, and Kurt WUTHRICH Institute of Molecular Biology and Biophysics, Swiss Federal Institute of Technology, Zurich (Received August 18/November 5, 1975)
Fourier transform 'H nuclear magnetic resonance (NMR) experiments at 360 MHz using convolution difference techniques to improve the spectral resolution were employed to investigate the resonances of the lysyl residues in bovine pancreatic trypsin inhibitor. The observations in both native protein and in chemically modified protein containing Ne-dimethyllysine show that three of the four lysines extend predominantly freely into the solvent, whereas lysine-41 is involved in an intramolecular interaction with tyrosine-10. Since in the single crystal structure tyrosine-10 is involved in an intermolecular interaction with arginine-42 of the neighboring protein molecule, the NMR data thus reveal a local conformation difference for bovine pancreatic trypsin inhibitor in solution and in the crystalline form which appears to result primarily from intermolecular interaction in the crystal lattice.
Use of proton magnetic resonance (lH NMR) to probe the structure and function of proteins has had many important successes, but has been limited to a certain extent by difficulty in resolving and assigning resonances arising from specific protons of the protein. In particular, relatively little information has SO far been extracted from the spectral region from 1 - Sppm, which contains the bulk of the protein resonances. Recent reports have however indicated that with use of advanced NMR techniques, the heavily overlapped aromatic resonances characteristic of protein spectra can to a certain extent be analyzed [l -41. The present paper describes investigations of the lysine 8-methylene resonances located in the crowded 2- 4 ppm region of the 'H NMR spectrum of the basic pancreatic trypsin inhibitor. This inhibitor is a particularly favourable protein for such studies due to its low molecular weight (6500), known crystal structure [5] and highly stable solution conformation [6,7]. Indeed, recent work on the complex aromatic region of the 'H NMR spectrum has shown that many resonances can be resolved and assigned [2,8,9]. A combination of high magnetic field (84.6 kG) and convolution difference techniques [3] has now permitted resolution of a considerable number of resonances in the spectral Ahhruviations. NMR, nuclear magnetic resonance; pprn, parts per million; the inhibitor, basic pancreatic trypsin inhibitor.
region from 1- 5 ppm. Thus the pH-titration of the four lysyl residues in the inhibitor could be followed by observation of the E-CH, resonances. The lysines15, 21 and 46 show normal, unperturbed titration behaviour, whereas lysine-41 has altered characteristics apparently arising from specific interaction with tyrosine-10. In the Discussion we consider the implications of these observations relative to the molecular conformation of the inhibitor.
MATERIALS AND METHODS Basic pancreatic trypsin inhibitor (TrasylolE, Bayer Leverkusen, Germany) was obtained from the Farbenfabriken Bayer AG. Methylation was by previously described methods [lo, 111. Amino acid analyses of methylated inhibitor were performed by Dr K . Wilson (Biochemisches Institut der Universitat Zurich) after hydrolysis in 6 M HCI at 110 "C for 24 h. Protein titrations were performed in 'HzO using 1 M KO'H to avoid sodium ion effects at high pH. Quoted pH values are pH meter readings uncorrected for isotope effects. N M R spectra were recorded on a Bruker HXS360-MHz instrument. Chemical shifts, 6, are quoted relative to internal sodium 2,2,3,3-tetradeutero-3-
' H NMR of the Lysines in the Pancreatic Trypsin Inhibitor
104
I
I
I
I
I
Fig. 2. Plot of' the chemical shifi vs p H meter reading for the iwo Iysine E-CH, resonances A (0) and B (m) in the inhibitor (see Fig. I ) . The solid lines correspond to the function pH = pK log [(a, - S)/(S - S,)] [lo, 131 using the S,, Sband pK values shown in Tablc 1. The broken line indicates thc corresponding titration curve for free lysine [lo].
+
Table 1. MMR characteristics of the four Iysyl residues und .f tvrosine-I0 in basic pancreatic trypsin inlzibirnr and its N'-dimeth.slIysine derivative 6 , and hb are the chemical shifts in the fully protonated and fully deprotonated forms, respectively. pK,,, is the apparent pKdetermined from plots of log ( 6 , - 6)/(S - 6,) vs pH Resonance
6,
PKapp
6,
-
PPm Lysine E-CH,
I
4.0
3.0
t
2 .O
6 (wm)
1 .o
0,
Fig. 1 . Specrral region from 0 to 4.5 ppm of ihe convolution dijyerence Fourier transform ' H N M R spectra at 360 M H z o f a 0.01 M solution of inhibitor in 'H20at the dijyerent p H meter readings indicated. A and B are the resonances assigned to emethylene protons of lysine (see text)
Free amino acid [lo] Peak A in inhibitor (3 lysines) Peak B in inhibitor (1 lysine)
10.9
3.03
2.61
0.42
10.6
3.04
2.64
0.40
10.8
2.96 2.56 _ _ Lysine N e - ( C H 3 ) , in Ne-dimethyllysine inhibitor
0.40 _
Lysine-41
~
~
11.2 10.3 10.3 10.3
2.61 2.86 2.92 2.93
1.90 2.15 2.18 2.18
0.71 0.71 0.74 0.75
Inhibitor 9.7 N'-Dimethyllysine inhibitor 10.2
7.07 7.10
6.70 6.69
0.37 0.41
Lysines-15,21 and 46
~
3,5 Protons of tyrosine-I0
trimethylsilyl-propionate. Spectral resolution was improved using convolution difference techniques [3,121. For titratable resonances, we have plotted log (6, - S)/(S - 6,) vs pH and taken the pH for which log (6, - S)/(S - S,) = 0 as the pK [10,11]. This will be a meaningful measure of pK (based on concentration) for groups which have simple one-proton titration behaviour and hence show straight lines with slopes near one for such plots. For groups with more complex titration behaviour, this method gives an apparent pK (pK,,,) corresponding to the pH at which the observed resonance has titrated to a half way position between the fully protonated (6,) and fully deprotonated (6,) resonance positions. A linear least-squares program was used to draw the straight lines in these plots.
RESULTS The 'H NMR spectra between 0 and 4.5 ppm of the inhibitor (Fig. 1) are well resolved and thus should be amenable to detailed interpretation. For example, we are able to observe two triplet resonances A and B which show a splitting of 7.5 Hz and shift to higher field as pH is increased. The chemical shift, titration curves, apparent pK and total shift upon titration (Fig. 2, Table 1) all indicate that these two triplet resonances can be assigned to lysine emethylene protons, since no other amino acid could match all these characteristics. The intensity ratio of the two triplet resonances, A:B z 3 : 1, indicates that with
~
L. R. Brown. A. De Marco. G. Wagner. and K . Wiithrich
105
P cm -0 -1
3
Fig. 3. Plof qflog /(IS,, - S ) / ( b - IS,)/ vspH. (I) Lysine E-CH, resonances A (0)and B (A) in tbe inhibitor. (11) 3.5-proton resonances of ryrosine-10 in the inhibitor (a) and N'-dimethyllysine inhibitor (A). (111) "-Methyl resonances of N'-dimethyllysine inhibitor. Lysine-41 (A), the lysines-15, 21 and 46 (+, 0.0) have not been individually assigned
regard to N M R characteristics, three of the four lysine &-methylenegroups in the inhibitor are nearly equivalent and one is altered. Plots of log (6, - 6)/(S - 6,) vs pH for peaks A and B are shown in Fig. 3-1. The three lysyl residues contributing to peak A show identical, simple titration behaviour and we therefore consider that the pK,,, of 10.58 for these lysines represents the pK in the inhibitor of surface lysines free of local perturbations. The single lysine contributing to peak B shows a straight line with slope near one only above pH 10 (Fig. 3-1). The deviation observed for this lysine below pH 10 suggests interaction with another titratable group having a pK near pH 10. Since the other three lysines show simple titration behaviour, interaction with tyrosyl residues in the inhibitor is the most likely cause of this perturbation. Inspection of a molecular model shows that only tyrosine-10 and lysine-41 are likely to be sufficiently close in the solution conformation for such an interaction to occur. The 3,5-proton and 2,6-proton resonances of tyrosine10 have been previously assigned by study of the chemically modified inhibitor [8] and confirmed by titration of analogues lacking tyrosine-10 [13]. Both the titration curves of the 3,5 and 2,6 resonances (Fig. 4) and the plot of log (6, - 6)/(6 - 6,) vs pH (Fig. 3-11) show that tyrosine-10 deviates from simple titration behaviour. Comparing Fig. 3-1 and 3-11, it is apparent that the deviations from a slope near one, and hence from simple titration behaviour, occur below pH 10 for lysine-B and above pH 10 for tyrosine-10. Since the lysine-41 titration curve would be perturbed primarily in the pH region where tyrosine-10 titrates and vice versa, the observed deviations are consistent with interaction between lysine-41 with a pK of 10.8 and tyrosine-10 with a pK of 9.8. To confirm the apparent interaction between lysine-41 and tyrosine-10, we have examined the NMR
7'5
r
1
6.5
7
I
I
I
I
1
8
9
10
11
12
PH
Fig. 4. Comparison of rhe plots of chemical .shift v s p H nwfer reading ,for the ring proton resonances of ryrosine-I0 in the inhihitor (A,mI and N'-dimethyllysine inhibitor ( A , 0). Upper curve: 2,6 protons. Lower curve : 3.5 protons. The solid lines were obtained as in Fig. 2
spectra of inhibitor in which the lysine &-aminogroups have been methylated [lo, 111. Amino acid analysis showed 87 7;modification of lysine and no appreciable change in the yield of other amino acids. Titration of methylated inhibitor in 6 M guanidinium chloride after performic acid oxidation of the disulfide bonds confirmed that the lysine residues were dimethylated at the c-amino group, and gave no indication for either Ne-mono-methyllysine or a-aminomethylation of arginine-1 . NMR spectra of the aromatic region of methylated inhibitor at various pH values show that, except for tyrosine-10, all resonances are virtually identical to the native inhibitor. This is good evidence that, except possibly near tyrosine-10, methylation of the lysines causes no appreciable change in the conformation of the inhibitor. For tyrosine-10, the titration curves of the 3,5 and 2,6 protons (Fig. 4) and the plot of log (6, - S)/(6 - S,) vs pH (Fig. 3-11) are both altered compared to the native protein.
106
As observed previously [lo, 111 the methyl groups of NE-dimethyllysine can be easily observed by 'H NMR, even without convolution difference. The plots of log (6, - 6)/(6 - 6,) vs pH (Fig. 3-111) and apparent pK values (Table 1) for the N"-dimethyllysine resonances in methylated inhibitor indicate that three lysyl residues are very similar, whereas the fourth one shows perturbed behaviour. Thus, as in the native inhibitor, there is evidence for interaction of tyrosine10 with one of the four lysyl residues. As previously noted, the crystal structure [5] suggests that this interaction is with NE-dimethyllysine-41.
DISCUSSION The present results lead to two principal conclusions. First, the spectral detail resolved in the 'H NMR spectra of Fig. 1 indicates that, with advanced instrumentation, resonances lines between 1 and 4.5 ppm may become of practical interest as natural probes of protein conformation and function. The present use of the E-CH, protons of lysine as a conformational probe in basic pancreatic trypsin inhibitor is a first step in the analysis of the crowded spectral region of the resonances of the aliphatic amino acid side chains. We suggest, on the basis of present experience and previous studies of methylated lysyl residues [lo, 111, that the E-CH, resonances of lysine should prove quite generally useful as a probe for studies of proteins. Second, structural information on this inhibitor is obtained which is pertinent to the view that the single crystal conformation of globular proteins is essentially maintained in the native solution conformation. These structural implications are discussed in more detail in the following paragraphs. The overall similarity of conformation of the inhibitor in single crystals and in solution has previously been substantiated [7,9] and hence the above assignment of the lysine interacting with tyrosine-10 as residue 41 appears well founded. On the other hand, in the crystal structure tyrosine-10 is in close proximity to arginine-42 of the nearest-neighbor inhibitor molecule, so that a close intramolecular interaction with lysine-41 is excluded. Accordingly, in the single crystals all of the four lysyl residues, including lysine-15 in the active site, extend freely into the hydration water and the positions of the E-methylene groups are rather poorly defined [14] (and J. Deisenhofer, private communication). Inspection of the three-dimensional electron density contours shows, however, that in the absence of the crystal lattice and with only strictly localized spatial rearrangement of the side chain, lysine-41 could occupy a position near tyrosine-10 (J. Deisenhofer, private communication). Hence the crystal structure of the inhibitor is consistent with lysine-41- tyrosine-10 interaction in solution. In addi-
'H NMR of the Lysines in the Pancreatic Trypsin Inhibitor
tion to the pH titration (Fig. 1-41, direct evidence for close proximity of lysine-41 and tyrosine-10 is suggested by the observation that for the lysine E-CH, resonances in the inhibitor and the lysine N'-methyl resonances in N&-dimethyllysineinhibitor, the resonance assigned to lysine-41 shows an upfield shift (Table 1). Calculations using the single crystal atomic coordinates of lysine-41 and tyrosine-10 [5] and the Johnson-Bovey tables of ring current effects [15] indicate that upfield shifts of the lysine-41 resonances are to be expected from the ring current field of tyrosine-10, but further predict that the lysine41 NE-methyl resonance should be less shifted than the E-methylene resonance. In contrast to this prediction we observe a four-fold greater upfield shift for the N'-methyl resonance, which suggests that in the solution conformation the lysine-41 side chain extends toward tyrosine-10. The NMR data therefore reveal a local conformation difference for the inhibitor in solution and in the crystalline form which appears to result primarily from intermolecular interactions in the crystal lattice. In general, interactions between surface ionizable groups in proteins may be more prevalent than suggested by crystal structure data since protein crystals are commonly prepared in high-ionic-strength solutions. In the present case the exact nature of the interaction observed between lysine-41 and tyrosine-10 is difficult to ascertain, but could be simply electrostatic interaction between nearby ionizable groups or possibly a hydrogen bond. Certain characteristics suggest that a hydrogen bond may exist. As discussed previously, the structural rearrangement necessary to allow a hydrogen bond appears entirely possible. The presence of a hydrogen bond would also be expected to raise the pK of lysine (proton donor) and lower the pK of tyrosine (proton acceptor) relative to the pK values in the absence of such interaction. We note in the native inhibitor that tyrosine-I0 has a considerably lower pK than the other tyrosine residues [8,9] and that for lysine-41 the pK is slightly higher than for the other three apparently normal lysines (Table 1). The slopes of the plots of log (6, - 6)/(S - 6,) vs pH for tyrosine-10 and NE-dimethyllysine-41 in the methylated inhibitor (Fig. 3-11, 111) indicate each experiences a strong interaction with another titratable group. Although the values of pK,,, for both groups show an increase relative to the native inhibitor, that of lysine-41 is particularly large since dimethylation normally decreases pK,,, [lo, 111. Thus pK,,, for lysine-41 relative to other lysines is +0.20 in native inhibitor and almost + 1.O in methylated inhibitor. This is consistent with the observation that titration of a tertiary amine involved in a hydrogen bond requires concurrent breakage of the hydrogen bond, whereas this is not so for a primary amine.
L. R. Brown, A. Dc Marco. G. Wagner, and K. Wiithrich
107
I
I
H
On the other hand, the observation of an exchangeaveraged position for the N~-dimethyllysine-41resonance as a function of pH implies rapid exchange (> lo3 s-l) of a possible hydrogen bond. To conclude, use of advanced instrumentation has permitted observation of lysine E-methylene proton resonances lying at chemical shift values where the large number of overlapping resonances has previously made detailed analysis difficult. We have consequently been able to define the characteristics of unperturbed, surface lysine residues in basic pancreatic trypsin inhibitor and to demonstrate a specific interaction on the protein surface between lysine-41 and tyrosine10. Since lysine-15 is an integral part of the active site of the inhibitor [ 5 ] , future applications of the lysine E-CH, NMR probe in this system might include studies of intermolecular interactions in the complexes formed between the inhibitor and proteinases. Further, the present demonstration that lysine E-CH, resonances are potential probes for specifying interactions between lysine and neighboring groups indicates the inherent usefulness of this technique for studying the many other proteins which contain lysyl residues with important conformational or functional properties. We would like to thank Dr R. Schmidt-Kastner, Fdrbenfabriken Bayer AG Tor a generous gift of the inhibitor (Trasylol%), Dr J. Deiscnhofer and Dr R. Huber for access to thc X-ray data and helpful discussions. Financial support by the Schweizrischer
H
Nutzonulfonds (project 3.1510.73), the Roche Research Foundation (fellowship to L.R.B.), and the Italian N.R.C. (fellowship to A. De M.) is gratefully acknowledged.
REFERENCES 1. Karplus, S., Snyder, G. H. & Sykes, B. D. (1973) Biochemistry, 12, 1323- 1329. 2. Wuthrich, K. & Wagner, G. (1975) FEBS Lett. 50, 265-268, 3. Campbell, 1. D., Dobson, C. M., Williams, R. J. P. & Xavier, A.'V. (1973) J . Mugn. Resonance, 11, 172-181. 4. Moore, G. R. & Williams, R. J. P. (1975) FEBS Lerr. 53, 334-338. 5. Huber, R., Kukla, D., Riihmann, A. & Steigemann, W. (1971) Cold Spring Harbor Symp. Quaiit. Biol. 36, 141-151. 6. Vincent, J. P., Chicheportiche, R. & Lazdunski, M. (1971) Eur. J . Biochem. 23,401-411. 7. Masson, A. & Wiithrich, K. (1973) FEES Lett. 31, 114-118. 8. Snyder, G. H., Karplus, S., Rowan, R. & Sykes, B. D. (1975) Biochemistry, 14, 3165- 3177. 9. Wagner, G. & Wiithrich, K . (1974) 6th In/. Conf on Mugnetic Resonance in Biological Sjstems, Kandersteg, Switzerland. 10. Bradbury, J. H. & Brown, L. R. (1973) Eur. J. Riochem. 40, 565 - 576. 11. Brown, L. R. & Bradbury, J . H. (1975) Eur. J . Biochem. 54, 219-227. 12. Ernst, R. R. (1966) Adv. Magn. Resonance, 2, 1 - 135. 13. Wiithrich, K., Wagner, G. & Tscheschc, H. (1975) Pro(:. X X I I I Colloq. Protides qf the Biologicul Fluids, Pergamon Press, Oxford, in press. 14. Deisenhofer, J. & Steigemann, W. (1974) in Proteinuse Inhibitors, Bayer-Symposium V (Fritz, H., Tschesche, H., Greene, L. J. & Truscheit, E., eds) Springer, New York. 15. Johnson,C.E. & Bovey,F.A. (1958) J. Chrm. Phys. 29, 1012- 1014.
L. R. Brown, A. De Marco. G. Wagner, and K. Wiithrich, Institut fur Molekularbiologie und Biophysik, E.T.H. Zurich-Honggerberg. CH-8049 Zurich, Switzerland
A Study of the Lysyl Residues in the Basic Pancreatic Trypsin Inhibitor using 'H Nuclear Magnetic Resonance at 360 MHz Larry R. BROWN, Antonio DE MARCO, Gerhard WAGNER, and Kurt WUTHRICH Institute of Molecular Biology and Biophysics, Swiss Federal Institute of Technology, Zurich (Received August 18/November 5, 1975)
Fourier transform 'H nuclear magnetic resonance (NMR) experiments at 360 MHz using convolution difference techniques to improve the spectral resolution were employed to investigate the resonances of the lysyl residues in bovine pancreatic trypsin inhibitor. The observations in both native protein and in chemically modified protein containing Ne-dimethyllysine show that three of the four lysines extend predominantly freely into the solvent, whereas lysine-41 is involved in an intramolecular interaction with tyrosine-10. Since in the single crystal structure tyrosine-10 is involved in an intermolecular interaction with arginine-42 of the neighboring protein molecule, the NMR data thus reveal a local conformation difference for bovine pancreatic trypsin inhibitor in solution and in the crystalline form which appears to result primarily from intermolecular interaction in the crystal lattice.
Use of proton magnetic resonance (lH NMR) to probe the structure and function of proteins has had many important successes, but has been limited to a certain extent by difficulty in resolving and assigning resonances arising from specific protons of the protein. In particular, relatively little information has SO far been extracted from the spectral region from 1 - Sppm, which contains the bulk of the protein resonances. Recent reports have however indicated that with use of advanced NMR techniques, the heavily overlapped aromatic resonances characteristic of protein spectra can to a certain extent be analyzed [l -41. The present paper describes investigations of the lysine 8-methylene resonances located in the crowded 2- 4 ppm region of the 'H NMR spectrum of the basic pancreatic trypsin inhibitor. This inhibitor is a particularly favourable protein for such studies due to its low molecular weight (6500), known crystal structure [5] and highly stable solution conformation [6,7]. Indeed, recent work on the complex aromatic region of the 'H NMR spectrum has shown that many resonances can be resolved and assigned [2,8,9]. A combination of high magnetic field (84.6 kG) and convolution difference techniques [3] has now permitted resolution of a considerable number of resonances in the spectral Ahhruviations. NMR, nuclear magnetic resonance; pprn, parts per million; the inhibitor, basic pancreatic trypsin inhibitor.
region from 1- 5 ppm. Thus the pH-titration of the four lysyl residues in the inhibitor could be followed by observation of the E-CH, resonances. The lysines15, 21 and 46 show normal, unperturbed titration behaviour, whereas lysine-41 has altered characteristics apparently arising from specific interaction with tyrosine-10. In the Discussion we consider the implications of these observations relative to the molecular conformation of the inhibitor.
MATERIALS AND METHODS Basic pancreatic trypsin inhibitor (TrasylolE, Bayer Leverkusen, Germany) was obtained from the Farbenfabriken Bayer AG. Methylation was by previously described methods [lo, 111. Amino acid analyses of methylated inhibitor were performed by Dr K . Wilson (Biochemisches Institut der Universitat Zurich) after hydrolysis in 6 M HCI at 110 "C for 24 h. Protein titrations were performed in 'HzO using 1 M KO'H to avoid sodium ion effects at high pH. Quoted pH values are pH meter readings uncorrected for isotope effects. N M R spectra were recorded on a Bruker HXS360-MHz instrument. Chemical shifts, 6, are quoted relative to internal sodium 2,2,3,3-tetradeutero-3-
' H NMR of the Lysines in the Pancreatic Trypsin Inhibitor
104
I
I
I
I
I
Fig. 2. Plot of' the chemical shifi vs p H meter reading for the iwo Iysine E-CH, resonances A (0) and B (m) in the inhibitor (see Fig. I ) . The solid lines correspond to the function pH = pK log [(a, - S)/(S - S,)] [lo, 131 using the S,, Sband pK values shown in Tablc 1. The broken line indicates thc corresponding titration curve for free lysine [lo].
+
Table 1. MMR characteristics of the four Iysyl residues und .f tvrosine-I0 in basic pancreatic trypsin inlzibirnr and its N'-dimeth.slIysine derivative 6 , and hb are the chemical shifts in the fully protonated and fully deprotonated forms, respectively. pK,,, is the apparent pKdetermined from plots of log ( 6 , - 6)/(S - 6,) vs pH Resonance
6,
PKapp
6,
-
PPm Lysine E-CH,
I
4.0
3.0
t
2 .O
6 (wm)
1 .o
0,
Fig. 1 . Specrral region from 0 to 4.5 ppm of ihe convolution dijyerence Fourier transform ' H N M R spectra at 360 M H z o f a 0.01 M solution of inhibitor in 'H20at the dijyerent p H meter readings indicated. A and B are the resonances assigned to emethylene protons of lysine (see text)
Free amino acid [lo] Peak A in inhibitor (3 lysines) Peak B in inhibitor (1 lysine)
10.9
3.03
2.61
0.42
10.6
3.04
2.64
0.40
10.8
2.96 2.56 _ _ Lysine N e - ( C H 3 ) , in Ne-dimethyllysine inhibitor
0.40 _
Lysine-41
~
~
11.2 10.3 10.3 10.3
2.61 2.86 2.92 2.93
1.90 2.15 2.18 2.18
0.71 0.71 0.74 0.75
Inhibitor 9.7 N'-Dimethyllysine inhibitor 10.2
7.07 7.10
6.70 6.69
0.37 0.41
Lysines-15,21 and 46
~
3,5 Protons of tyrosine-I0
trimethylsilyl-propionate. Spectral resolution was improved using convolution difference techniques [3,121. For titratable resonances, we have plotted log (6, - S)/(S - 6,) vs pH and taken the pH for which log (6, - S)/(S - S,) = 0 as the pK [10,11]. This will be a meaningful measure of pK (based on concentration) for groups which have simple one-proton titration behaviour and hence show straight lines with slopes near one for such plots. For groups with more complex titration behaviour, this method gives an apparent pK (pK,,,) corresponding to the pH at which the observed resonance has titrated to a half way position between the fully protonated (6,) and fully deprotonated (6,) resonance positions. A linear least-squares program was used to draw the straight lines in these plots.
RESULTS The 'H NMR spectra between 0 and 4.5 ppm of the inhibitor (Fig. 1) are well resolved and thus should be amenable to detailed interpretation. For example, we are able to observe two triplet resonances A and B which show a splitting of 7.5 Hz and shift to higher field as pH is increased. The chemical shift, titration curves, apparent pK and total shift upon titration (Fig. 2, Table 1) all indicate that these two triplet resonances can be assigned to lysine emethylene protons, since no other amino acid could match all these characteristics. The intensity ratio of the two triplet resonances, A:B z 3 : 1, indicates that with
~
L. R. Brown. A. De Marco. G. Wagner. and K . Wiithrich
105
P cm -0 -1
3
Fig. 3. Plof qflog /(IS,, - S ) / ( b - IS,)/ vspH. (I) Lysine E-CH, resonances A (0)and B (A) in tbe inhibitor. (11) 3.5-proton resonances of ryrosine-10 in the inhibitor (a) and N'-dimethyllysine inhibitor (A). (111) "-Methyl resonances of N'-dimethyllysine inhibitor. Lysine-41 (A), the lysines-15, 21 and 46 (+, 0.0) have not been individually assigned
regard to N M R characteristics, three of the four lysine &-methylenegroups in the inhibitor are nearly equivalent and one is altered. Plots of log (6, - 6)/(S - 6,) vs pH for peaks A and B are shown in Fig. 3-1. The three lysyl residues contributing to peak A show identical, simple titration behaviour and we therefore consider that the pK,,, of 10.58 for these lysines represents the pK in the inhibitor of surface lysines free of local perturbations. The single lysine contributing to peak B shows a straight line with slope near one only above pH 10 (Fig. 3-1). The deviation observed for this lysine below pH 10 suggests interaction with another titratable group having a pK near pH 10. Since the other three lysines show simple titration behaviour, interaction with tyrosyl residues in the inhibitor is the most likely cause of this perturbation. Inspection of a molecular model shows that only tyrosine-10 and lysine-41 are likely to be sufficiently close in the solution conformation for such an interaction to occur. The 3,5-proton and 2,6-proton resonances of tyrosine10 have been previously assigned by study of the chemically modified inhibitor [8] and confirmed by titration of analogues lacking tyrosine-10 [13]. Both the titration curves of the 3,5 and 2,6 resonances (Fig. 4) and the plot of log (6, - 6)/(6 - 6,) vs pH (Fig. 3-11) show that tyrosine-10 deviates from simple titration behaviour. Comparing Fig. 3-1 and 3-11, it is apparent that the deviations from a slope near one, and hence from simple titration behaviour, occur below pH 10 for lysine-B and above pH 10 for tyrosine-10. Since the lysine-41 titration curve would be perturbed primarily in the pH region where tyrosine-10 titrates and vice versa, the observed deviations are consistent with interaction between lysine-41 with a pK of 10.8 and tyrosine-10 with a pK of 9.8. To confirm the apparent interaction between lysine-41 and tyrosine-10, we have examined the NMR
7'5
r
1
6.5
7
I
I
I
I
1
8
9
10
11
12
PH
Fig. 4. Comparison of rhe plots of chemical .shift v s p H nwfer reading ,for the ring proton resonances of ryrosine-I0 in the inhihitor (A,mI and N'-dimethyllysine inhibitor ( A , 0). Upper curve: 2,6 protons. Lower curve : 3.5 protons. The solid lines were obtained as in Fig. 2
spectra of inhibitor in which the lysine &-aminogroups have been methylated [lo, 111. Amino acid analysis showed 87 7;modification of lysine and no appreciable change in the yield of other amino acids. Titration of methylated inhibitor in 6 M guanidinium chloride after performic acid oxidation of the disulfide bonds confirmed that the lysine residues were dimethylated at the c-amino group, and gave no indication for either Ne-mono-methyllysine or a-aminomethylation of arginine-1 . NMR spectra of the aromatic region of methylated inhibitor at various pH values show that, except for tyrosine-10, all resonances are virtually identical to the native inhibitor. This is good evidence that, except possibly near tyrosine-10, methylation of the lysines causes no appreciable change in the conformation of the inhibitor. For tyrosine-10, the titration curves of the 3,5 and 2,6 protons (Fig. 4) and the plot of log (6, - S)/(6 - S,) vs pH (Fig. 3-11) are both altered compared to the native protein.
106
As observed previously [lo, 111 the methyl groups of NE-dimethyllysine can be easily observed by 'H NMR, even without convolution difference. The plots of log (6, - 6)/(6 - 6,) vs pH (Fig. 3-111) and apparent pK values (Table 1) for the N"-dimethyllysine resonances in methylated inhibitor indicate that three lysyl residues are very similar, whereas the fourth one shows perturbed behaviour. Thus, as in the native inhibitor, there is evidence for interaction of tyrosine10 with one of the four lysyl residues. As previously noted, the crystal structure [5] suggests that this interaction is with NE-dimethyllysine-41.
DISCUSSION The present results lead to two principal conclusions. First, the spectral detail resolved in the 'H NMR spectra of Fig. 1 indicates that, with advanced instrumentation, resonances lines between 1 and 4.5 ppm may become of practical interest as natural probes of protein conformation and function. The present use of the E-CH, protons of lysine as a conformational probe in basic pancreatic trypsin inhibitor is a first step in the analysis of the crowded spectral region of the resonances of the aliphatic amino acid side chains. We suggest, on the basis of present experience and previous studies of methylated lysyl residues [lo, 111, that the E-CH, resonances of lysine should prove quite generally useful as a probe for studies of proteins. Second, structural information on this inhibitor is obtained which is pertinent to the view that the single crystal conformation of globular proteins is essentially maintained in the native solution conformation. These structural implications are discussed in more detail in the following paragraphs. The overall similarity of conformation of the inhibitor in single crystals and in solution has previously been substantiated [7,9] and hence the above assignment of the lysine interacting with tyrosine-10 as residue 41 appears well founded. On the other hand, in the crystal structure tyrosine-10 is in close proximity to arginine-42 of the nearest-neighbor inhibitor molecule, so that a close intramolecular interaction with lysine-41 is excluded. Accordingly, in the single crystals all of the four lysyl residues, including lysine-15 in the active site, extend freely into the hydration water and the positions of the E-methylene groups are rather poorly defined [14] (and J. Deisenhofer, private communication). Inspection of the three-dimensional electron density contours shows, however, that in the absence of the crystal lattice and with only strictly localized spatial rearrangement of the side chain, lysine-41 could occupy a position near tyrosine-10 (J. Deisenhofer, private communication). Hence the crystal structure of the inhibitor is consistent with lysine-41- tyrosine-10 interaction in solution. In addi-
'H NMR of the Lysines in the Pancreatic Trypsin Inhibitor
tion to the pH titration (Fig. 1-41, direct evidence for close proximity of lysine-41 and tyrosine-10 is suggested by the observation that for the lysine E-CH, resonances in the inhibitor and the lysine N'-methyl resonances in N&-dimethyllysineinhibitor, the resonance assigned to lysine-41 shows an upfield shift (Table 1). Calculations using the single crystal atomic coordinates of lysine-41 and tyrosine-10 [5] and the Johnson-Bovey tables of ring current effects [15] indicate that upfield shifts of the lysine-41 resonances are to be expected from the ring current field of tyrosine-10, but further predict that the lysine41 NE-methyl resonance should be less shifted than the E-methylene resonance. In contrast to this prediction we observe a four-fold greater upfield shift for the N'-methyl resonance, which suggests that in the solution conformation the lysine-41 side chain extends toward tyrosine-10. The NMR data therefore reveal a local conformation difference for the inhibitor in solution and in the crystalline form which appears to result primarily from intermolecular interactions in the crystal lattice. In general, interactions between surface ionizable groups in proteins may be more prevalent than suggested by crystal structure data since protein crystals are commonly prepared in high-ionic-strength solutions. In the present case the exact nature of the interaction observed between lysine-41 and tyrosine-10 is difficult to ascertain, but could be simply electrostatic interaction between nearby ionizable groups or possibly a hydrogen bond. Certain characteristics suggest that a hydrogen bond may exist. As discussed previously, the structural rearrangement necessary to allow a hydrogen bond appears entirely possible. The presence of a hydrogen bond would also be expected to raise the pK of lysine (proton donor) and lower the pK of tyrosine (proton acceptor) relative to the pK values in the absence of such interaction. We note in the native inhibitor that tyrosine-I0 has a considerably lower pK than the other tyrosine residues [8,9] and that for lysine-41 the pK is slightly higher than for the other three apparently normal lysines (Table 1). The slopes of the plots of log (6, - 6)/(S - 6,) vs pH for tyrosine-10 and NE-dimethyllysine-41 in the methylated inhibitor (Fig. 3-11, 111) indicate each experiences a strong interaction with another titratable group. Although the values of pK,,, for both groups show an increase relative to the native inhibitor, that of lysine-41 is particularly large since dimethylation normally decreases pK,,, [lo, 111. Thus pK,,, for lysine-41 relative to other lysines is +0.20 in native inhibitor and almost + 1.O in methylated inhibitor. This is consistent with the observation that titration of a tertiary amine involved in a hydrogen bond requires concurrent breakage of the hydrogen bond, whereas this is not so for a primary amine.
L. R. Brown, A. Dc Marco. G. Wagner, and K. Wiithrich
107
I
I
H
On the other hand, the observation of an exchangeaveraged position for the N~-dimethyllysine-41resonance as a function of pH implies rapid exchange (> lo3 s-l) of a possible hydrogen bond. To conclude, use of advanced instrumentation has permitted observation of lysine E-methylene proton resonances lying at chemical shift values where the large number of overlapping resonances has previously made detailed analysis difficult. We have consequently been able to define the characteristics of unperturbed, surface lysine residues in basic pancreatic trypsin inhibitor and to demonstrate a specific interaction on the protein surface between lysine-41 and tyrosine10. Since lysine-15 is an integral part of the active site of the inhibitor [ 5 ] , future applications of the lysine E-CH, NMR probe in this system might include studies of intermolecular interactions in the complexes formed between the inhibitor and proteinases. Further, the present demonstration that lysine E-CH, resonances are potential probes for specifying interactions between lysine and neighboring groups indicates the inherent usefulness of this technique for studying the many other proteins which contain lysyl residues with important conformational or functional properties. We would like to thank Dr R. Schmidt-Kastner, Fdrbenfabriken Bayer AG Tor a generous gift of the inhibitor (Trasylol%), Dr J. Deiscnhofer and Dr R. Huber for access to thc X-ray data and helpful discussions. Financial support by the Schweizrischer
H
Nutzonulfonds (project 3.1510.73), the Roche Research Foundation (fellowship to L.R.B.), and the Italian N.R.C. (fellowship to A. De M.) is gratefully acknowledged.
REFERENCES 1. Karplus, S., Snyder, G. H. & Sykes, B. D. (1973) Biochemistry, 12, 1323- 1329. 2. Wuthrich, K. & Wagner, G. (1975) FEBS Lett. 50, 265-268, 3. Campbell, 1. D., Dobson, C. M., Williams, R. J. P. & Xavier, A.'V. (1973) J . Mugn. Resonance, 11, 172-181. 4. Moore, G. R. & Williams, R. J. P. (1975) FEBS Lerr. 53, 334-338. 5. Huber, R., Kukla, D., Riihmann, A. & Steigemann, W. (1971) Cold Spring Harbor Symp. Quaiit. Biol. 36, 141-151. 6. Vincent, J. P., Chicheportiche, R. & Lazdunski, M. (1971) Eur. J . Biochem. 23,401-411. 7. Masson, A. & Wiithrich, K. (1973) FEES Lett. 31, 114-118. 8. Snyder, G. H., Karplus, S., Rowan, R. & Sykes, B. D. (1975) Biochemistry, 14, 3165- 3177. 9. Wagner, G. & Wiithrich, K . (1974) 6th In/. Conf on Mugnetic Resonance in Biological Sjstems, Kandersteg, Switzerland. 10. Bradbury, J. H. & Brown, L. R. (1973) Eur. J. Riochem. 40, 565 - 576. 11. Brown, L. R. & Bradbury, J . H. (1975) Eur. J . Biochem. 54, 219-227. 12. Ernst, R. R. (1966) Adv. Magn. Resonance, 2, 1 - 135. 13. Wiithrich, K., Wagner, G. & Tscheschc, H. (1975) Pro(:. X X I I I Colloq. Protides qf the Biologicul Fluids, Pergamon Press, Oxford, in press. 14. Deisenhofer, J. & Steigemann, W. (1974) in Proteinuse Inhibitors, Bayer-Symposium V (Fritz, H., Tschesche, H., Greene, L. J. & Truscheit, E., eds) Springer, New York. 15. Johnson,C.E. & Bovey,F.A. (1958) J. Chrm. Phys. 29, 1012- 1014.
L. R. Brown, A. De Marco. G. Wagner, and K. Wiithrich, Institut fur Molekularbiologie und Biophysik, E.T.H. Zurich-Honggerberg. CH-8049 Zurich, Switzerland
