I1 NH N-O 0 ~ OH3 CH3 CH3JN*CH3 SPIN-LABELED HEMOGLOBIN CRYSTALS* a*-b* X-ray precession photograph is consistent with the space group C2.

Transcription

1 SPIN-LABELED HEMOGLOBIN CRYSTALS* BY S. OHNISHIt J. C. A. BOEYENS,4 AND H. M. MCCONNELL STAUFFER LABORATORY FOR PHYSICAL CHEMISTRY, STANFORD, CALIFORNIA Communicated June 30, 1966 In the present paper we describe the results of a preliminary study of the paramagnetic resonance of spin-labeled crystals of horse oxyhemoglobin (HbO2). The central purpose of this work is to show that it is possible to obtain strong anisotropic paramagnetic resonance spectra from spin-labeled single crystals of proteins, and to show that structural information can be obtained from such spectra. Materials and Methods.-Defibrinated horse blood was obtained from the Colorado Serum Company, lot no. 763, and the hemoglobin obtained by the method of Benesch and Benesch.1 The hemoglobin was spin-labeled with the nitroxide maleimide described previously,2 and also with the following nitroxide maleimide, 0 I1 NH N-O 0 ~ - 11 which is more resistant to hydrolysis. The preparation and properties of I will be described elsewhere.3 The spin-labeled HbO2 was crystallized according to the procedure given by Green, Ingram, and Perutz.4 The spin-labeled crystals are isomorphous with native HbO2. We find the same crystal morphology, and our a*-b* X-ray precession photograph is consistent with the space group C2. The unit cell dimensions a and b for the spin-labeled crystal are the same as those reported by Bragg and Perutz5 for the unlabeled HbO2 crystals. Paramagnetic resonance spectra were taken with a Varian Associates X-band spectrometer; the crystals had to be kept immersed in an (NH4)2SO4-(NH4)2HP04 solution to prevent dessication. During the course of our experiments it was observed that the hemoglobin crystals became darker, presumably due to the formation of methemoglobin. The observed resonance spectra were not changed, however, so our results presumably apply as well to methemoglobin. Results.-The paramagnetic resonance spectra of sterically protected nitroxide radicals containing the group R R' CH3JN*CH3 OH3 CH3 can be accounted for in terms of the spin Hamiltonian,6 809

2 810 CHEMISTRY: OHNISHI, BOEYENS, AND McCONNELL PROC. N. A. S. SC = IBS g.ho + hs-t*i + gsniiho. (1) From studies of a number of radicals containing the group II in liquid and single crystal solid solution,6 it has been concluded that the odd electron is confined primarily to a 2pxr atomic orbital on the nitrogen atom,7 that the principal axes of the hyperfine tensor T and g-factor tensor g are those sketched in Figure 1 (z II jxorbital axis, x N-O bond direction), that T, >>T T,,, > 0, and that gxx > gyy > g2,. Under these conditions the orientation of the nitroxide group relative to the crystallographic axes can be deduced from the paramagnetic resonance spectra. It is very likely that for the spin labels under consideration here (where R and R' are linked so as to form 5- or 6-membered saturated rings) the nitrogen atom, oxygen atom, and the two tertiary carbon atoms are all coplanar, since the hyperfine tensor indicates the odd-electron to be largely confined to a 2pir atomic orbital, corresponding to sp2 hybridization around the nitrogen atom. The paramagnetic resonance spectra of crystals of HbO2 spin-labeled with nitroxide-maleimide I can be accounted for rather simply as follows. Our previous study8 of the nitroxide-maleimide labeling of HbO2 has shown that most of the spins are "strongly immobilized" due to attachment of the label to -SH groups on the,3 chains (p3-93) together with a subsequent reaction, presumably involving catalyzed opening of the ring and hydrogen bonding of the carboxyl acid group of the opened ring. These "strongly immobilized" spins give powder-type paramagnetic resonance spectra in solution,8 since the correlation time for tumbling of the relatively rigid tertiary protein structure is long compared to the anisotropic terms in (1), and furthermore, the label cannot "wiggle" appreciably in its attached configurations. A smaller proportion of spins (- 20%) are only "weakly immobilized" in solution, presumably arising from covalent bonding of the label to the flexible lysine side chains. Since the "strongly immobilized" spin-labels are thought to have a definite orientation relative to the HbO2 molecule, on crystallization one expects, and finds, a simple anisotropic paramagnetic resonance spectrum. For example, Figure 2 shows the spectrum when the applied field is parallel to the crystallographic b axis (Ho b), which is also the twofold molecular axis.9 The splitting CH CH3 FIG. 2.-Paramagnetic resonancespectrum of nitroxide-maleimide I spin-labeled horse oxy- FIG. 1.-Spin Hamiltonian principal hemoglobin, with the applied axis system for sterically protected magnetic field parallel to the nitroxide radicals. crystallographic b axis.

3 VOL. 56, 1966 CHEMISTRY: OHNISHI, BOEYENS, AND McCONNELL 811 of the nitrogen hyperfine triplet is 91.3 Me, very close to that observed for -- a number of sterically protected nitroxide radicals when the applied field is parallel to the 7r-orbital. For other orientations of the magnetic field relative to the crystallographic axes, the strongly anisotropic spectra (corresponding to strongly immobilized spin-labels in solution) are accounted for in terms of the following parameters of the spin Hamiltonian. Tzz = Me Tar = Tyy = zX = L vv = i gzz = Figure 3 shows a plot of resonance line position versus crystal orientation when the applied field lies in the (001) plane. The heavy curve is a plot of theoretical splittings using the above parameters and a principal axis orientation such that z b and x II a. Labels having this unique orientation automatically fulfill the condition of being related to one another by the twofold molecular axis b, and necessarily give identical spectra for all field directions. In addition to the strong signals plotted in Figure 3, there were observed in field rotations in the (001) plane other somewhat weaker signals shown by the dashed lines in Figure 3. These correspond to another set of oriented radicals related to one another by the twofold molecular axis b but having principal axes x' and x' displaced from x( = a) by ca. +15 and -15. This second label orientation may arise from the two types of-hemoglobin molecule known to be present in horse blood,10 or from any other effect that could make (3-subunits nonequivalent to one another, such as the labeling itself. These two principal axis orientations might also arise from isomeric forms of the attached spin-label I. It is hoped to resolve this point in future work. The extra broad line in Figure 3 arises from a relatively weak powder component in the. crystal spectra. It is likely that these spins V / a -J correspond to the weakly immobilized spins seen in the solution spectra. 3O@ / 0 Experiments were also carried out on HbO2 / h1 crystals spin-labeled with the nitroxide-malei- 60 mide described previously.2 Here the signalto-noise ratio in the observed spectra was 90_ b rather poor, but it was clear that here again / the 7r-orbital (z-axis) of the label attached to 120'O - SH on A3-93 is nearly parallel to the molecular twofold axis. '50- We were able to account for the observed DO0 G orientation of the principal axes of the label by I80, constructing Pauling-Corey-Koltum atomic models of the hemoglobin,8-chain regions F, FIG. 3.-Angular variation of hyper- H, and FG, using the three-dimensional paper fine structure in the paramagnetic resonance spectra of horse oxyhemoglobin model described by Schroeder and Jones"1 as with the applied field in the (001) plane.

4 812 CHEMISTRY: OHNISHI, BOEYENS, AND McCONNELL PROC. N. A. S. FIG. 4.-Schematic two-dimensional representation of "strongly immobilized" (a) and"weakly immobilized" (b) modes of binding of nitroxide-maleimide spin-labels to,3-93-sh of hemoglobin. The heavy black dot represents the sulfur atom, and the arrow the unpaired spin. In the first case (a) the label goes into the hydrophobic pocket, then forms a covalent bond to sulfur, is immobilized, and cannot get out. In the second case (b) the label bonds to sulfur first, is only weakly immobilized, and cannot get into the hydrophobic pocket. a guide. It was assumed that after addition of p3-93 -SH to the carbon-carbon double bond of the maleimide ring, the carbon-nitrogen bond of the maleimide ring most removed from the sulfur atom is hydrolyzed. With this covalent bonding, the saturated nitroxide ring can be tucked snugly into a hydrophobic pocket between the hemoglobin segments F, G, H, and FG. When the hemoglobin molecule is viewed along the a axis, then we may say, very crudely, that the top of the pocket is formed by the helical region F, the back of the pocket by the heme group, the right-hand wall of the pocket by the helical region H, and the left-hand and bottom walls by regions G and FG. A significant structural feature assumed in the model is the hydrogen bond between tyrosine p3-145 (H23) and the a-carbonyl of residue A-98 (FG5) described by Perutz et al. 12 The pocket constrains the ring so that the magnetic i x axis is nearly parallel to a, as observed, and also constrains the ring so that the 7r-orbital is parallel to b, as observed. The evident close fit in this pocket accounts for the otherwise unexpected result that both the 5- and 6-4nembered nitroxide rings have essentially identical 7r-orbital orientations. The model also shows the previously suggested8 hydrogen bonding of the hydrolyzed maleimide ring to the imidazole ring of histidine From the model it is clear that removal of two residues'3 from the C-terminal end of the : chains with carboxypeptidase A destroys one wall of the pocket, thus accounting for the observed effect8 of this enzyme in transforming the stronglyimmobilized spin label at (-93-SH to a weakly immobilized spin-label. The tightness of the fit and the close proximity to the heme group is compatible with the previous observations that the attachment of the nitroxide-maleimide label to deoxyhemoglobin does not yield an immobilized spin, if it is here assumed the pocket is too small. Finally, the irreversible nature of the spin immobilization in HbO2 with respect to subsequent deoxygenation is clear from the model, since once the saturated ring is in the pocket and the attachment to sulfur subsequently takes place, the ring cannot turn around so as to become weakly immobilized on the surface of the protein (see Fig. 4). Labeling of (-93 -SH in deoxyhemoglobin is irreversible in a similar sense, since the weakly immobilized spin remains weakly immobilized on subsequent oxygenation.8 Here again, after the chemical attachment the model clearly shows that the nitroxide ring cannot turn around so as to go into the strongly immobilizing pocket. Summary.-This work has demonstrated that strong anisotropic paramagnetic resonance spectra can be obtained from nitroxide-maleimide spin-labeled hemoglobin crystals. The spectra show the known twofold symmetry of the hemoglobin molecule. The orientation of the labeling molecule relative to the hemoglobin molecule deduced from the paramagnetic resonance spectra can be accounted for in terms of a rough atomic model of oxyhemoglobin, based on the work of Perutz and co-workers. The model also accounts for previously observed "allosteric" effects of oxygen on the immobilization of nitroxide-maleimide spin-labels.8

5 VOL. 56, 1966 CHEMISTRY: OHNISHI, BOEYENS, AND McCONNELL 813 * Sponsored by the National Science Foundation under grants GP 3430 and GB t Present address: Department of Chemistry, Faculty of Science, Kyoto University, Kyoto, Japan. t On leave from the National Physical Research Laboratory of the Council for Scientific and Industrial Research, Pretoria, South Africa. I Benesch, R. E., and R. Benesch, Biochemistry, 1, 735 (1962). 2 Griffith, 0. H., and H. M. McConnell, these PROCEEDINGS, 55, 8 (1966). 3 Hamilton, C. L., and H. M. McConnell, in preparation. 4 Green, D. W., V. M. Ingram, and M. F. Perutz, Proc. Roy. Soc. (London), A225, 287 (1954). 5 Bragg, W. L., and M. F. Perutz, Acta Cryst., 5, 323 (1952). Griffith, 0. H., D. W. Cornell, and H. M. McConnell, J. Chem. Phys., 43, 2909 (1965). 7Stone, T. J., T. Buckman, P. L. Nordio, and H. M. McConnell, these PROCEEDINGS, 54, 1010 (1965). 8 Boeyens, J. C. A., and H. M. McConnell, these PROCEEDINGS, 56, 22 (1966). 9 Cullis, A. F., H. Muirhead, M. F. Perutz, and M. G. Rossmann, Proc. Roy. Soc. (London), A265, 161 (1962). 10 Perutz, M. F., L. K. Steinrauf, A. Stockell, and A. D. Baugham, J. Mol. Biol., 1, 402 (1959). "1 Schroeder, W. A., and R. T. Jones, Fortschr. Chem. Org. Naturstoffe, 23, 111 (1965). 12 Perutz, M. F., J. C. Kendrew, and H. C. Watson, J. Mol. Biol., 13, 669 (1965). 13 Antonini, E., J. Wyman, R. Zito, A. Rossi-Fanelli, and A. Caputo, J. Biol. Chem., 236, PC60 (1961).