Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 50
X-RAY ANALYSIS OF HAEMOGLOB1N ANN F. CULLIS, H. M. DINTZIS AND M. F. PERUTZ* 1. PHYSICAL PRINCIPLES We are trying to determine the structure of haemoglobin by direct x-ray analysis, using isomorphous substitution with heavy metals. So far this has given us a picture of the electron density distribution in the haemoglobi~ ' molecule in projection on a plane, at a resolution of about 3 A. This picture can be proved to be correct, but on ac- count of the great thickness of the molecule and the con- sequent overlapping of its components in proj ection we have been unable to recognize in it any polypeptide chains or haem groups. The only definite information obtained so far concerns the shape of the molecule, and the positions of the sulfhydryl groups. Before describing the results we should like to outline some of the physical principles involved in the C-ray anal- ysis. These are best illustrated by analogy with optical diffraction, using a device invented by Braggi and per- fected by Hanson, Lipson and Taylor.' As figure 1 shows, this consists essentially of a light source, a pinhole, two plane-convex lenses and a microscope. A picture of a simple molecule, e.g. hexamethylbenzene, is stamped out on a mask of black paper, each atom being represented by a circular hole. The mask is placed between the two plane-convex lenses and its Fraunhofer diffraction pattern observed through the microscope. The pattern, illustrated in figure 2a, consists of bright patches of positive or neg- at~ve amplitude separated by nodes, where the amplitude passes through zero on changing sign. We must now consider the relation between the dif- \\ o---l l A —B ., D 1 it, FIG. 1. Optical device used for ob- serving diffraction patterns of simple molecules (Repro- duced from Han- son, Lipson and Taylor, Proc. Roy. Soc., Series A, 21S: 371, 1953 ) . fraction by a single molecule and that of a crystal in which many molecules are arranged in a regular array. As a first step we may examine the diffraction by two mole- cules of hexamethylbenzene placed side by side (fig. 3a). Interference between them causes the diffraction pattern of figure 2a to be divided by a series of vertical dark fringes whose spacing is inversely proportional to the distance between the molecules. Note that the d ffraction pattern of the pair of molecules exhibits bright patches only at positions where the pattern of the single molecules also shows a bright patch. * This paper was presented by Dr. Perutz. 50
OCR for page 51
X-RAY ANALYSIS CULLIS, DINTZIS AND PERUTZ 51 · - ~ ~ ;W, ·. ,,,.,,- ., ,,. , i,, ,˘..;, ,., ..~ . '"." '. ~'.'""'""""'' _ ' ~ ~..... _............................................ ,............................................ - . ~ I.... - ....'.. _ "~ ~ _ '"''""''""'""~ ~' '"~, ~. ............................................................. (~6 - ~: ~ ; ~,~ e : : ~ FIG. 2. Diffraction from single molecule of hexamethylbenzene without and w ith an extra atom added at the centre. The diffraction patterns shown in this figure, and in figures 3 and 4, were kindly provided by Professor H. Lipson at the University of Manchester. ...,.~................ ii~ .,,.,. ... i it_ ........................... ........................ .:.::::::: :.:::::::::: :: :::::::::: :::::: ::::: ::: :::: ::: .... :: : : . . :.:: .: By... ~ _L ........ - . ~_Z - : _ _]k - . ~ . ..: ~ ::.. }:: :: ~ i: :: ::.... ~ .::~ id: it: ~ :::: ~ ~ ~ :::: :: ~ ~ ~ ~ ~ : ~ ~ :~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ i: - - - : ~ ~ ~ :~ : ~:~ ~ ~ ~ ~ ::::: :: ~ ~ ~ ~ :::::: ~ ~ :~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Hi::: : :::::::: ::.: . :::: :: :: :: ;:;; ;;;;;;;;;;;;;;;;;. ;; ;;; ~ ::::::::::::::::::::~:::::: ~:: ~ ::::^ ... 2.'.'.2"""""'""""'"""''"""'."'."'.'.""".""""""""" "' '' ''"'I''''"'"' :' ' ~.'.'.~.'. ~t, f . ~.~.~.~.. ~ ~. ~.~.~. ~... - i:: ::: ::::::: ~ I ~ - L:: Id: :.: ::: ..... Em. a ..... - :: :: FIG. 3. Diffraction by two molecules of hexamethylbenzene. In other words the effect of the additional molecule is merely to modify the diffraction pattern by cutting it into strips. If four molecules are placed at the corners of a parallelogram the diffrac- tion pattern is divided by a second set of dark fringes cutting across the first one (fig. 4a). As a result the pattern is now divided into spots lying at the corners of a parallelogram outlined by the molecules. Note again that bright spots appear only in positions corresponding to bright patches in the diffraction pattern of the single molecule. If the four molecules are replaced by a large number in a regular t~vo-dimensional array, as in the pattern of phthalocyanin illustrated in figure 5, the spots become sharper and form a completely regular
OCR for page 52
52 PART I. STRUCTURE OF HEMOGLOBIN FIG. 4.- Diffraction by four molecules of hexamethylbenzene as arranged in the crystal. pattern. The distance between the spots is inversely proportional to the dis- tc~nce between the molecules, but their brightness depends only on the bright- r~ess of the corresponding region in the diffraction pattern of the single phthal- ocvanin molecule. J FIG. 5. Left: pattern representing the b projection of the phthalocyanin structure; right: optical diffraction given by it. (Reproduced from C. W. Bunn, Chemical Crys- tallography, Oxford University Press, 1945).
OCR for page 53
X-RAY ANALYSIS—CULLIS, DINI'ZIS \ND PERUTZ 53 In x-ray analysis one is faced with the task of reconstituting the unknown picture of a molecule from the x-ray diffraction pattern of a crystal. This can be done because diffraction is reversible. For instance, if a mask were prepared corresponding to the diffraction pattern of the hexamethylbenzene molecule (figure 2a), the patches of positive amplitude being left open while those of negative amplitude are covered with a phase plate giving half a wavelength retardation, and if this mask were now placed between the plane- conve~x lenses of the diffraction machine, then the superposition of the dif- fracted waves would produce an image of the hexamethylbenzene molecule at the focus of the microscope. Similarly the array of phthalocyanin mole- cules can be reconstituted by punching a mask with a regular array of holes of varying sizes, corresponding to the spots of varying brightness in figure 5, and leaving open the holes of positive amplitude, while covering those of negative amplitude with a phase plate. Thus the image of an object can be reconstituted from its diffraction pat- tern provided the sign of each diffracted spot is known. In an x-ray diffraction picture, however, such as that of haemoglobin show-e later in figure 8, only the intensities of the spots can be measured, while their signs are unknown. The difficulty of x-ray analysis consists in the determination of the unknown signs or phases. One well-known method of sign determination depends on the use of two kinds of crystal, identical in all respects, except that in one crystal each molecule contains a heavy atom in a position occupied by a light atom in the other. The effect of such an ison~orphous substitution upon the diffraction pattern may again be explained with the help of an optical analogue. Suppose for the sake of argument that an additional atom could be put at the centre of the hexamethylbenzene molecule, as in figure 2b. Due to its central position, this atom makes a positive contribution over the whole area of the diffraction pattern, thus increasing the amplitude of the positive patches and diminishing that of the negative ones, as is evident from a com- parison of figures 2a and 2b. By changing their brightness the additional atom thus enables us to recognize the signs of the different patches, and to gather the information needed for the reconstitution of the image from the diffraction pattern. The effect of the additional atoms on the diffraction pat- terns by two and by four molecules is shown in figures 3b and 4b. In each case it enables us to~ find the signs of the different patches or spots, simply by observing whether their brightness goes up or down. It will also be noted that the positions of spots of positive or negative amplitude in the diffraction pattern from two or more parallel molecules coincide with those of the corresponding positive or negative patches in the diffraction pattern of the single molecule. This is the physical basis underlying the x-ray analysis of organic compounds by isomorphous substitution. In actual x-ray analysis the reconstitution of the image from its diffraction pattern is done by calculation rather than optically, but the principle is the
OCR for page 54
54 PART I. STRUCTURE OF HEMOGLOBIN 1 1 1 1 1 1 1 1 o o 5 A FIG. 6.—Fourier projection of sodium penicillin (Reproduced from Crowfoot e' al., The Chemistry of Penicillin, pp. 310-367, Princeton University Press, 1949). same. The result of the image reconstruction is expressed in the form of a contour map showing the molecule in projection on a plane. The contours are drawn at levels of equal electron density. The maps are called Fourier projections, because Fourier series are used in their computation. As an example figure 6 shows a Fourier projection of penicillin.3 Some atoms are clearly resolved and appear as individual peaks, awhile others overlap in pro- jection, especially in the p-lactam ring. A projection of a structure on a plane, such as figure 6, is calculated from only part of the diffraction pattern of a crystal, namely the x-ray reflexions from lattice planes lying normal to the plane of projection. Overlapping of atoms can be avoided and complete resolution achieved by extending the x-ray analysis to the entire diffraction pattern from the crystal, which may be visualized as a three-dimensional network of spots of varying brightness. At this stage, however, the analysis of a complex enantiomorphous structure becomes exceedingly difficult, because the amplitudes of the general spots are no longer restricted to plus or minus. Instead there is associated with each
OCR for page 55
X-RAY ANALYSIS CULLIS, DINTZIS AND PERUTZ 55 FIG. 7. Electron density distribution along a series of parallel sections through the penicillin molecule (Reproduced from Crowfoot et al., The Chemistry of Penicillin, pp. 310-367, Princeton University- Press, 1949 ) . spot a phase angle which may have any value between 0° and 360°, and reconstitution of the image is possible only by determining the precise values of all the phase angles. This is the situation at present confronting us in the >;-ray analysis of proteins. There is a way of determining these phase angles by direct x-ray analysis, but it is a difficult one, especially because of the exacting demands it makes upon the ingenuity of the chemists assisting the C-ray crystallographer. In- stead of a single isomorphous pair, it is necessary to have a series of at least three, and if possible more, isomorphous compounds, each having a heavy atom attached to a different site on the protein molecule. By measuring the amount by which the intensity of each diffraction spot changes in each of these isomorphous substitutions, the value of the phase angle can be deter- mined, and the electron density of the molecule can then be calculated in three dimensions. As an example of the advantages to be gained by three-dimensional crystal structure analysis, as compared to two-dimensional prod ections, figure 7 shows the results of such an analysis in penicillin. The electron density has been calculated along a series of parallel sections through the penicillin mole-
OCR for page 56
56 PART I. SlRUCTURE OF HEMOGLOBIN `:ule, rather like the sections the histologist cuts through the organ of an animal. For purposes of illustrations the sections are all superimposed and drawn together on one plane. They show the atoms of penicillin lying at different levels above the plane of projection clearly resolved, thus allowing the investigator to build a three-dimensional model of the molecule. The building of such a model is the ultimate aim of the protein crystallog- rapher. In principle it is possible to attain it, provided a series of isomorphous heavy atom compounds can be made. In practice it will be difficult, at any rate with a protein as big as haemoglobin, to obtain a resolution comparable with that of figure 7. The best that can be hoped for is a picture allowing one to recognize the configuration and lay-out of the polypeptide chains, together with the positions of the haem groups and of the more prominent side-chains If the amino-acid sequence of haemoglobin w ere to become known from chemical analysis, it should then be possible to build a fairly accurate model of the haemoglobin molecule. 2. APPLICATION TO HORSE METH.\EMOGLOBIN Our work has been done on horse, rather than human, haemoglobin, be- cause it leas the more favourable crystal form, all the molecules in the crystal being arranged in parallel. We have used methaemoglobin on account of its stability. Crystals of horse methaemoglobin are monoclinic, with two mole- cules in ~ face-centred unit cell, each lying on an axis of dyed symmetry. Ire such a structure there is only one projection, along the dyed axis, in which the phase angles of the diffracted rays are restricted to 0 or ~ (plus or minus). The first attempt at solving this projection of haemoglobin by iso- morphous substitution with heavy atoms was made by Green, Ingram and Perutz.4 They prepared compounds in which two of the sulphydryl groups of haemoglobin were combined with parachloromercuribenzoate or with silver ions. By comparing the intensities of the diffracted rays from the substituted and the unsubstituted haemoglobin, they w ere able to determine the signs of 87 out of 94 reflexions from lattice planes of spacings greater than 6.5 A (fig. 8~. From this and other, supplementary, information Bragg and Perutz' it. ~~ ..: I, I,. . .~.~...~. ~ A... ~~ ~.~,,,~,,,,' ~ .. , ,. ~'~.x'.: I.... ~ . ..... .~ . - . . I FIG. 8.—X-ray photograph of hOI re- flections of horse methaemoglobin extend- ing to a spacing of 6.5 A.
OCR for page 57
X-RAY ANALYSIS CULLIS, DINTZIS AND PERUTZ 57 were then able to calculate the first electron density projections of haemo- glol~in. Two kinds of projection were obtained. One was calculated from the intensity changes produced in the diffraction pattern when salt solution re- placed water as the suspension medium of the crystals and shows a contour map of the volume of hydrated protein into which salt cannot penetrate (fig. 9~. The contour levels represent thickness of the hydrated protein in projection. This picture has a very low resolution, because only the lowest order reflexions are affected by salt, but it gives a fairly accurate picture of the external shape of the molecule. This is seen to be an ellipsoid of dimen- sions 55 x 55 .x 70 9~. There is a depression at the centre corresponding to a dimple or a pair of dimples at the surface of the molecule. Act;. 9. (left) Volume of hydrated protein molecule into which salt cannot penetrate, . . as seen In projection along the dyed axis. The dyed is marked at the centre of the molecule. The broken part of the contours is extrapolated. (Reproduced from Bragg and Perutz, Proc. Roy. Soc., Series A, 225: 315, 1954). FIG. 10. (right) Electron density distribution in a single haemoglobin molecule sus- pended in salt-free water, seen in projection along the dyed axis. The contours near the right and left hand edges of the picture may be distorted due to overlapping of neighbouring molecules in the crystal. Note the difference in scale between this figure and figure 6. The other picture shows the electron density distribution within ~ haemo- globin molecule suspended in saltfree water (fig. 10~. It contains a flat area at the centre, corresponding to the central depression of figure 9, and a com- plex system of peaks and depressions which have so far defied any attempt at interpretation. The enigmatic appearance of the projection is due to two causes. First, the overlapping of a large thickness of matter in projection, and secondly, insufficient resolution. At the next stage of the analysis we have therefore attempted both to improve the resolution and to prepare some more heavy atom complexes in preparation for a three-dimensional analysis of the structure. Native horse haemoglobin has four available -SH groups arranged in two pairs.6 These combine with four silver ions or with two moles of parachloro- mercuribenzoate or other mercurials. If the -SH groups of haemoglobin are
OCR for page 58
58 PART I. STRUCTURE OF HEMOGLOBIN blocked by iodoacetamide, another mercury-birding site is uncovered form- irlg a complex with mercuriacetate. So far three complexes have proved useful for x-rav analysis: J I. Methaemoglobin + 2 moles of paramercuribenzoate 2. NIethaemoglobin + 2 moles of Hg~CHCOO- (dimercuriacetate) 3. Blocked methaemoglobin + 2 moles of Hg(CH3COO-~2 (mer- curidiacetate ~ ~ The first and most crucial task in the actual analysis consists in finding the positions of the heavy atoms, for on their correct location depends the subsequent calculation of the signs or phase angles of the x-ray reflexions. In our work on haemoglobin, the heavy atoms were found by a method which is largely automatic and free from assumptions about the nature of the pro- tein molecule or the number of heavy atoms attached to it. For each x-ray re- flexion the intensity charge produced by the heavy atom is measured. All the changes together are then put through a mathematical process of image reconstitution in which all the signs are taken as positive, and the result is plotted ire the form of a contour map (fig. 11~. This map shows a high peak at the origin together with a system of smaller peaks distributed around it. If the substitution with the heavy metal is truly isomorphous, and if it has attached itself to one unique kind of site, then the map shows either one peak or a small number of well-defined peaks which are high compared to the general fluctuations of the background. The line connecting the peaks to the origin correspond to the vector joining the different heavy atoms on the protein molecule. In figure 10 there is just one such peak, corresponding to the vector between the two mercury atoms in PCMB-haemoglobin. Fig- ure 12 shows the positions of the mercury atoms in the three different deriv- atives as determined by the method j ust outlined. One group of mercury , . / .. ~ . . , . . ~ c , · · . , ., ~ . · , / · . . / ., , , , . , FIG. 11. Difference Patterson pro- jection of me rcu ribenzo ate ha emoglob- in, showing the vector between the two mercury atoms attached to the haemo- globin molecule.
OCR for page 59
X-RAY ANALYSIS CULLIS, DINTZIS \ND PERUTZ 59 / \/ , c / ~~ / ~ / A / ~ /C O I I t I I 19 1 1 1~ 1 1 1~ 1 1 14 1 ~ A 7 ~ H9(OOCCH;2 \/ .-HclCHCOO- / \ °-~sCbH.coo / FIG. 12.—Positions of mercury atoms / in different haemoglobin derivatives. atoms clusters around the two pairs of sulphydryl groups, while the others. representing the mercuriacetate on blocked haemoglobin, are combined with a site whose chemical nature is still unknown. The positions of the mercury atoms bound to the sulphydryl groups show that these groups are spaced 30 A apart in the haemoglobin molecule. The positions of the heavy atoms having been found, the way was now opened for the calculation of signs. In the first round Green, Ingram and Perutz had confined themselves to 94 reflexions, extending to a spacing of 6.5 A. In the second round eve have tried to find the signs of the 470 re- flexions in figure 13, extending to a spacing of 2.8A. We have so far suc- ceeded in determining the signs of 422 of these. Our new electron density projection is thus calculated from 4.5 times as many reflexions as the original one of Bragg and Perutz and shows corres- pondingly greater detail (fig. 14~. For the sake of clarity the contours of the high density regions or peaks, and those of the low density regions or de- pressions have been drawn on separate diagrams. They show the density distribution within the area of the unit cell covered by one haemoglobin molecule suspended in ~ M ammonium sulphate solution. The contours neat the periphery of the molecule are not exact, being partly modified by the overlapping of neighbouring; molecules in the crystal lattice. The map con- ...... ~ ......... ,i .... ~0. · t ~ ; ~ ... .... ~ ~ ~ . my. ~ : 2' ,' ' :':': A: :~ : : : : : : :~: k :~ : c::: : : ,,,,,, ,,,.,,,,,,,.,, ,. , . , . . ~ . . .,,. ,. ,. ,, ~ . ,, I, . FIG. 13. X-ray photograph of hOI reflexions of horse methaemoglobin ex- tending to a spacing of 2.8 A.
OCR for page 60
60 PART I. STRUCTURE OF HEMOGLOBIN (a) <~—1 in. 9, I i, 9,,,, I,,,, I,,, I I ~ A FIG. 14. Elects on density distribution in the haemoglobin molecule seen in pro jec- tion along the dyed axis at a resolution of 2.8 A. The black dots in (a) represent the >;ulphydryl groups. (a) shows the contours of high electron density or peaks, (b) shows the contours of low density or depressions. Contours are drawn at intervals of 2.5 electrons/^', the base line being the density of 2 M ammonium sulphate solution, marked B. The lowest contour in figure 14a corresponds to 5 electrons/ above the level of 2 M ammonium sulphate solution. firms the existence of Ha flat region at the centre of the molecule where the density is hardly higher than in the surrounding salt solution. It shows, in addition, ~ striking system of peaks and depressions which appear as enig- matic as ever. Presumably the peaks represent projected positions of chains crossing over, or portions of chains running normal to the plane of projection, but clearly the projection contains too little information as yet to give any clue to the structure of the molecule. This was disappointing, but not un- expected, in view of the great thickness of matter projected on one plane. It is evident that the riddle cannot be solved without a solution of the crystal structure in three dimensions, giving us the density distribution in the molecule along a series of sections. This will have to be the next step. Acknowledgment: The authors wish to thank Professor H. Lipson of the University of Manchester for his kindness in preparing the optical diffraction patterns shown in figures 2, 3 and 4. REFEREN CES 1. Bragg, W. L.: A new type of x-ray microscope, Nature, 143: 678, 1939. 2. Hanson, A. W., Lipson, H., and Taylor, C. A.: The application of the principles of physical optics to crystal-structure determinations, Proc. Roy. Soc. (London), Series A, 218: 371, 1953. 3. Crowfoot, D., Bunn, C. W., Rogers-Low, B. W., and Turner-Jones, A., in "The Chemistry of Penicillin" pp. 310-367, Princeton University Press, 1949. 4. Green, D. W., Ingram, V. M., and Perutz, M. F.: The structure of haemoglobin IV. Sign determination by the isomorphous replacement method, Proc. Roy. Soc., (London), Series A, 225: 287, 1954. 5. Bragrg, Sir Lawrence and Perutz, M. F.: The structure of haemoglobin VI. Fourier projections on the 010 Plane, Proc. Roy. Soc. (London), Series A, 225: 325, 1954. 6. Ingram, V. M.: Sulphydryl groups in haemoglobin, Biochem. J. 59: 653, 1955.
Representative terms from entire chapter: