A n X -ray study of horse m ethaem oglobin. I By Joy Boyes-Watson, Edna Davidson and M. F. Pertjtz Cavendish Laboratory and Molteno Institute, University of Cambridge (<Communicated by Sir Lawrence Bragg, F.R.S.—Received 3 February 1947) [Plates 5 and 6] The paper describes a detailed study of horse methaemoglobin by single crystal X-ray diffraction methods. The results give information on the arrangement of the molecules in the crystal, their shape and dimensions, and certain features of their internal structure.
84 J. Boyes-Watson, E. Davidson and M. F. Perutz motion a vast volume of research into the functions and properties of proteins. Numerous biological, chemical and physical methods have been applied to their study and have led to the gradual emergence of a coherent picture of protein con stitution. The striking advances which have been made in the methods of amino- acid analysis have led to the determination of almost complete balance sheets of the amino-acid composition of certain proteins (Chibnall 1945), and are beginning to yield information on the number of separate polypeptide chains of which protein molecules are composed, on the sequence of amino-acid residues within these chains, and on the types of chemical linkages between them (Synge 1943; Sanger 1945). Knowledge of protein structure, on the other hand, has so far been confined to protein fibres, where Astbury has shown polypeptide chains to follow certain structural patterns whose geometry is independent of the detailed amino-acid composition (see Astbury 1940, 1942). Among the crystalline proteins no such structural classes have as yet emerged. It is known that the molecules of crystalline proteins do not consist of single polypeptide chains coiled up in random fashion, but are probably composed of one or several chains of definite length and con figuration, linked together to form particles of definite shape and dimensions, with a characteristic pattern of active side chains facing the solvent; though we can infer that such definite features exist we do not know what they are, and we possess only partial and indirect information regarding the molecular shape and dimensions.
X-ray study of methaemoglobin. I 85 the presentation of what is admittedly a very limited advance towards the solution of this problem, the logical sequence of the argument tends to become lost among a mass of essential detail. For this reason an outline of the main points of the argu ment is given below.
86 J. Boyes-Watson, E Davidson and M. F. Perutz combined with changes in the layer spacing. Patterson projections indicate that the number of liquid layers per unit cell cannot be greater than two and is more likely to be one.
X-ray study of methaemoglobin. I 87 the li quid of crystallization, in which case the concentration of solute ions throughout the liquid of crystallization would be much lower than in the suspension medium, as shown by its lower density. Alternatively, it may be assumed that the ‘ bound water ’ is directly attached to the protein molecules and that the concentration of solute ions in the remaining liquid (the ‘ free ’ liquid of crystallization available to diffusing ions) is the same as in the suspension medium. Since the ‘ bound ’ water would occupy a considerable amount of space, the second assumption, if valid, would also reduce the size of the regions through which solute ions can diffuse, and thus alter the dimensions of the liquid regions whose phase contribution it is desired to calculate. Fortunately, there is an obvious reversal of phase of the 001 reflexion at a certain salt concentration from which it can be shown to follow that the electron density in the ‘free ’ liquid of crystallization must be the same as in the suspension medium, thus deciding in favour of the second assumption. At this point the way is clear for calculating the phase contributions of the heavy ions, and the signs of some of the 00Z reflexions can be found, though not with complete certainty in every case.
88 J. Boyes-Watson, E. Davidson and M. F. Perutz 2. Preparation of crystals Horse methaemoglobin solutions were prepared by the method of Keilin& Hartree (1935). The protein was crystallized in simple diffusion cells by dialysis of 10 ml. 1 % methaemoglobin solution against 30 to 40 ml. concentrated salt solution. Diffusion took place through cellophane membranes. Initially, great difficulties were en countered in preparing crops of sufficiently perfect crystals reproducibly. Attempts were made, therefore, to discover the factors influencing crystal size, habit and twinning. In series of experiments, temperature, rate of diffusion, protein and salt concentration, presence of colloidal impurities or dust particles and pH of the precipitating solution were tested as possible factors influencing growth, but none except pH and a factor which, for lack of a more precise definition, may be called the state of the protein had a decisive influence. The optimum pH for crystallization was found to be 7-2 to 7-3. At this- hydrogen-ion concentration half-saturation with ammonium sulphate is sufficient to effect complete precipitation of the protein. As a rule, solutions made up of 2 parts 4m-(NH4)2S04+ 1 part 2m-(NH4)2HP04 were used for precipitating the protein. Alternatively, a solution containing 3 m-K2HP04 +1 m-HC1 was employed. Dr E. F. Hartree devised another method for growing the crystals which was to prove of decisive importance in the X-ray analysis. Making use of the low solubility of methaemoglobin in the absence of salt he electrodialyzed a concentrated solution of methaemoglobin against tap water and produced an excellent crop of large single crystals after some 24 hr. The very concentrated solutions of methaemoglobin needed for this method can be obtained by oxidizing a mush of oxyhaemoglobin crystals with a few drops of saturated potassium ferricyanide solution. The linear dimensions of single crystals in a good crop vary from 0T to 1*5 mm.
X-ray study of methaemoglobin. I 89 from the surface towards the centre of the crystals. If the change is followed with a microspectroscope, the absorption spectrum is seen to change from that of acid methaemoglobin to that of its addition compound azide-methaemoglobin. Similarly, the change from acid to alkaline methaemoglobin can be followed in the interior of the crystals by watching the changing absorption spectrum.
90 J. Boyes-Watson, E. Davidson and M. F. Perutz 4. Unit-cell dimensions and space group (a) Experimental technique For the purpose of taking X-ray pictures, wet haemoglobin crystals are mounted as follows. Capillaries of either borosilicate glass or quartz are prepared, with a wall thickness of 0-01 to 0-02 mm., a bore of about 1 mm. and a length of about 30 mm. A single protein crystal is isolated on a slide with the help of a dissecting needle and drawn into the capillary with a drop of mother liquor. A firm cotton thread, made by twisting cotton-wool between the fingers, is now introduced half-way into the capillary and the liquid drawn off very gently through the thread, leaving the crystal behind. The thread is now withdrawn and the liquid re-introduced into both suspension medium picein Figure 1. Wet protein crystal mounted for X-ray diffraction work.
92 J. Boyes-Watson, E. Davidson and M. F. Perutz reflexions occur in a series of concentric annular regions whose origin is explained in figure 2. Each ring on the photograph represents part of a reciprocal lattice plane intersecting the sphere of reflexion. Reciprocal lattice planes passing through the origin give rise to rings which pass through the centre of the photograph. On plate 5 that central ring has the index 1 = 0, and the reflexions on it can be indexed simply by counting from the centre. Systematic absences are often obvious from mere inspection of the pictures.