CRYSTALLOGRAPHY.PROTEINS, NUCLEIC ACIDS, AND VIRUSES.Introduction.-The study of these complex structures has greatlyadvanced since it was last fully reported,l owing to the convergence of anumber of lines of attack both chemical and physical. The analyticalchemical methods of Sanger and others have succeeded in revealing thesequence of amino-acid residues in a number of complex peptides and ininsulin as well as indicating the position of -S-S- bridges. Thus it may besaid that the structure of all pure proteins is, in the purely chemical sense,in principle, determinable. This information, though essential, is not,however, sufficient to explain the physical, colloidal, and biochemicalproperties of such complicated bodies. It is necessary also to form someidea of how the long and relatively flexible polypeptide chains are coiledand folded in both fibres and discrete molecules.It is now apparent that large aliphatic heteropolymeric molecules canbe considered as arranged in a succession of levels of ordering of which thechemical structural formula giving the topology of the covalent bonds isonly the first or primary structure. The secondary structure refers to themutual stereochemical relations of sequences of residues along each chain.These are determined in the first place by conditions of steric hindrance andfurther among configurations sterically possible by the presence of hydrogenbonds.These can stabilise various types of coiled or folded chain struc-tures according to the arrangements of bonds between residues of the samechain (intracatenary hydrogen bonds) or of neighbouring chains (inter-catenary hydrogen bonds).The knowledge of primary and secondary structure serves in general todescribe adequately those of synthetic polypeptides, but for naturallyoccurring fibres and for globular or crystalline proteins it is insufficient.Itis necessary in addition to know the tertiary structure, namely the lengths ofthe coils and the way in which they are grouped or folded together, forexample as a coiled coil as has been proposed for keratin (p. 386) or inreversed parallel coils in one model of insulin (p. 384).Even higher orders of structure can now be discussed, namely in thearrangements of small protein molecules to form larger, relatively stable,aggregates.The unit molecule of insulin of molecular weight 6000 is, itseems, rarely found in solution. In crystals it appears in groups of six, infibres in groups of t ~ e n t y , ~ and in solution in groups of two to eight.The protein coating of virus particles seems to be composed of such high-order aggregates .So far the chief advances have been made in determining the secondarystructure of proteins and nucleic acids, though information on the tertiarystructure is beginning to accumulate. The crystallographic attack has been1 Ann. Reports, 1951, 48, 238-248, 362-382.a Cf. Ann. Reports, 1953, 50, 268-280.8 D. F. Waugh, D. F. Wilhelmson, S. L. Commerford, and M. L. Sackler, J. Amer.Chem. SOC., 1953, 75, 2592BERNAL AND CARLISLE.381pressed from both within and without, that is, by starting with molecularmodels and verifying their correctness with X-rays or starting from X-rayanalysis and interpreting the results by means of molecular models. Nocomplete analysis of the type carried out by Mrs. Hodgkin on vitamin B,,(p. 403) has yet been carried out for a protein but it is not unlikely thatthis may be done soon.the idea of a hydrogen-bonded spiral first put forward by Huggins * and refined by Pauling et aZ.6has proved most fruitful. Especially valuable was Pauling’s conceptionof a non-integral spiral allowing the greatest freedom for packing andhydrogen bonding. To verify this by X-rays involved the development byCochran, Crick, and Vand of the theory of the scattering by spiral structuresexpressible in terms of Bessel functions and leading to very rapid inductionsof plausible structures from simple inspection of fibre photographs. Com-bined with the optical method perfected by Taylor, Hinde, and Lipson 7of producing Fourier transforms similar to X-ray photographs from molecularmodels, this method has led to the elucidation of many extremely complexstructures such as collagen (p.388), nucleic acids (p. 395), and tobacco-mosaic virus (p. 400).The value of these methods has been greatest in the case of fibroussubstances, especially synthetic fibres which lack the complication of vanedside groups. The interplay between the study of corresponding syntheticand natural proteins has been most fruitful.Thus poly(methy1 glutamate)provided the key to the a-keratin structures (p. 385), poly-L-alanine tothat of silk (p. 387), and polyproline to that of collagen (p. 388). The useof polarised infrared absorption has enabled conclusions to be drawn as tothe orientation of hydrogen bonds in synthetic polymers which would bedifficult to determine for natural fibres owing to the influence of side groups.The structures of three main types of protein fibres distinguished by Ast-bury have been determined in main outiine. The a-type (keratin, myosin,epidermim, fibrin) seems to conform, with variations, to the Pauling a-helix(p. 385) in which all hydrogen bonds are within the chain. The p-type(silk) consist of extended chains linked by hydrogen bonds in pleated sheets(p.387). The most intractable form, that of collagen, now seems to berevealed (p. 389) as a twinning of three partially extending chains mutuallyheld together by hydrogen bonds and thus intermediate between CL and p.The most spectacular advance of fibre analysis, already briefly reported,ghas been in the structure of the polynucleotides. Here a number of workersin Cambridge and London, notably Crick, Watson, Wilkins, and Franklin(p. 397), have co-operated to determine a structure for deoxyribose nucleicacid. This structure involves a double spiral of linked phosphate and sugargroups stabilised by the piling of sheets of hydrogen-bonded purine andpyrimidine pairs lying at right-angles to the axis. This structure is likelyto prove of considerable biochemical and even of genetic importance.In the structure of crystalline proteins the degree of complexity has4 M.L. Huggins, Chem. Rev., 1943, 32, 195.6 L. Pauling, R. B. Corey, and H. R. Branson, Proc. Nut. Acad. Sci., 1951, 37, 206. * W. Cochran, F. H. C . Crick, and V. Vand, Acta Cryst., 1962, 5, 581.7 C. A. Taylor, R. N. Hinde, and H. Lipson, zbid., 1951, 4, 458.8 W. T. Astbury, J . Inst. Soc. Leather Trades’ Chemists, 1940, 24, 69. * Ann. Reports, 1954, 51, 270-279.In the prediction of the structure of protein382 CRYSTALLOGRAPHY.made both approaches from models or from diffraction data extremelydificult. Indeed, despite much observation the most that could be deduceddirectly from unaltered proteins was something of the size, shape, andmutual relations of protein molecules in crystals.A decisive advance onthe crucial problem of the determination of the phases of X-ray diffractionwas made when Perutz et aZ.1° (see p. 392) used heavy atoms attached tothe protein molecule for this purpose. Since then work has gone rapidlyforward on a number of protein crystals with promising results, though thesheer labour of observation and calculation makes it unlikely that completestructures will be determinable for some years.The papers of four discussions l1 on the structure of proteins are avail-able : the first, initiated by Astbury, centres on the theme of the newly borna-helklla The second lib is that of the Institut International de ChimieSolvay, held at Brussels in April, 1953.This is an excellent review of theposition concerning chemical and structural information of the globularproteins up to 1953. The third is the “ Nature and Structure of Collagen,’’held at King’s College, London, 1953, and the fourth is that of the Symposiaof the Society of Experimental Biology,lld No. IX on “ Fibrous Proteinsand their Biological Significance.” The emphasis throughout the latter ison the significance of chemical, X-ray diffraction, and electron-microscopicalstudies on the interpretation of structure. Current knowledge in thesestudies is reviewed as it concerns collagen, keratin, and muscle, and theirbiological , biophysical, and biogenetic implications.An excellent survey has been given l2 by Kendrew of the position in thestudy of the crystalline proteins up to 1954, with particular reference tostructural hypothesis; Springall’s book l3 is an attempt to portray ourpresent knowledge on proteins to honours students reading chemistry.Thelatter is an excellent survey of the subject covering the chemical and physico-chemical aspects of protein structure. Another outstanding contributionin this field is “ The Proteins ” l4 (two volumes, four parts) ; it containsvaluable surveys of many aspects of proteins and nucleoproteins by peopleactively engaged in research in their fields. Another contribution l5 con-stituting a valuable survey is “ The Nucleic Acids,” in which the attempt ismade to collect all the information at present available on this subject intoa single comprehensive work.General and Theoretical.-In addition to the proposed configurations forpolypeptide chains, the u- and y-helices,s Huggins l6 claimed the existenceof a 306,~ helix, and Low and Baybutt l7 suggested a 4*4,, helix, with an10 D.W. Green, V. F. Ingram, and M. F. Perutz, Proc. Roy. Soc., 1954, A , 225, 287.11 (a) Proc. Roy. Soc., 1953, B , 141, 1 : (b) Proceedings of April Meetings, 1953, ofthe Institut Internatiqyal de Chimie Solvay, ed. R. Stoops, Brussels; (c) ‘‘ NFFure andStructure of Collagen, ed. J. T. Randall, Butterworths, London, 1953 : ( d ) FibrousProteins and their Biological Significance,” Symposia of the Society of ExperimentalBiology No. 9, Cambridge Univ. Press, 1955.12 J. C. Kendrew, Frogy. Bio9hysics Biophys.Chem., 1954, 4.13 H. D. Springall, The Structural Chemistry of Proteins,” Butterworths, London,1954.14 “The Proteins,” ed. H. Neurath and K. Bailey, Accademic Press, NewYork, 1955.15 “ The Nucleic Acids,” ed. E. Chargaff and J. N. Davidson, in “ Chemistry andBiology,” Vols. I and 11, Academic Press, New York, 1955.16 M. 1;. Huggins, J. Amer. Chem. SOL, 1962, 74, 3963.17 B. W. Low and R. B. Raybutt, ibid., p. 5806BERNAL AND CARLISLE. 383axial translation of 1-15 A per residue parallel to the axis.has made an assessment of six hydrogen-bonded helix configurations, namelythe a- and y-helices and the 4.416, 4-3,,, 3.010, and 2.2, helices. The last isthe 2,b originally put forward by Huggins* but the flattened ribbon has atwist which gives the chain a super-helix effect.Quantitative estimates ofbond energies show that proposed structures are decreasingly stable in theorder 3.6,,, 4.416, 2*2,, 3.0,,, 5.1,,, and 403,~. A combination of a favourablehydrogen-bond system and unknown external influences may allow poly-peptide chains to fold in one or another of the configurations which at firstglance seem to be rather less reasonable than the 3-6,, (a)-helix.In a subsequent paper Donohue 1M calculated the radial distributionfunctions for the 3.6,,, 413, 3.010, and 4 ~ 4 , ~ helices for the @-carbon atoms inboth position 1 and 2, and also that for the 2.2, helix with the @-carbon atomin position 1. The 41, helix was originally suggested by Bragg, Kendrew,and Perutz19 which closely resembles that of the 3*6,, helix.At smallradial distributions (0-2.0 A) there is little to distinguish one curve fromany of the others, but for spacings greater than 2.7 A it is possible to distin-guish the 2.2, and 3.OlO helices from the others which are quite similar toone another-the latter all show a marked peak in the vicinity of 5 A, withlittle to choose between those with the @-carbon atoms in positions 1 or 2.The stereochemical uncertainties in such calculations can only lead to verygeneral considerations concerning protein structure.Arndt and Riley20 have examined the X-ray scattering of thirty amor-phous proteins and two synthetic polypeptides, and suggest that there is asingle polypeptide chain common to all as a predominant constituent.Furthermore, they claim that it is possible to distinguish from such scatteringdata, for instance, in which position the @-carbon atom is to be found inserum albumin.Small-angle X-ray investigations 21 have been carried outon bovine serum albumin, human mercaptalbumin, and mercaptalbumin-mercury dimer, which are shown to have radii of gyration of 29.8, 31-0, and37-2 A respectively. I t is suggested that the molecules have internaland external hydrations of 0.37 and 0.11 g. of water per g. of proteinrespectively.Complete theoretical treatments of the X-ray diffraction to be expectedare available for the single strand a-helix and for " coiled coil " arrange-ments of a-helices in two and three strand ropes.22 The form factors havebeen calculated for the 18-residue 5-turn y-helix, without the inclusion oftemperature, Lorentz, or polarisation fact01-s.~~ A description of the theoryof X-ray fibre diagrams suitable for biologists is given by Stokes.24 Harker 25discusses the inapplicability of Wilson's statistical procedures 26 to thediffraction data of protein crystals, and Luzzati 27 has further indicated theDonohueJ.Donohue, PYOC. Nut. Acad. Sci., (a) 1953, 39, 470; ( 6 ) 1954, 40, 377.W. L. Bragg, J. C . Kendrew, and M. F. Perutz, Proc. Roy. SOL, 1951, A , 203, 321.2o U. W. Arndt and D. P. Riley, Phil. Trans., 1955, 24'4, A , 409.z1 J. W. Anderegg, W. W. Beemen, C . Shulman, and P. Kaesberg, J . Amer. Chew.22 F. H. C . Crick, Acta Cryst., 1953, 6, 685.23 L.Pauling, R. B. Corey, H. L. Yakel, jun., and R. E. Marsh, ibid., 1955, 8, 853.24 A. R. Stokes, Progr. Biophysics Biophys. Chem., 1955, 5.25 D. Harker, Actu Cryst., 1953, 6, 731.26 A. J. C. Wilson, Nature, 1942, 150, 152.27 V. Luzzati, Acta Cryst., 1955, 8, 795.SOL, 1955, 77, 2927384 CRYSTALLOGRAPHY.limitations inherent in Wilson’s statistical method and has derived newtheoretical laws more applicable to protein crystals.Low and Grenville-Wells 28 have constructed nornographs relating theatomic co-ordinates of atoms on a helix to the vertical translation of the coilper residue. Lindley 29 has shown that there are simple ways of modifyingthe rigidity of the or-helix by incorporating a sense of change in the chain.This can result in side-chains participating in main chain bonding, henceleading to the masking of groups.Lindley and RoHett 30 have, on the basisof the known amino-acid sequence of its four chains, put forward a proposedstructure for insulin based on a-helices. They conclude that it is possible tobuild a satisfactory model of a globular protein on the a-helix, and the factthat such a possibility can exist is all the more important in view of thedoubts expressed about the presence of the a-helix in globular proteins,Techniques-Single-crystal X-ray work on proteins requires the carefulindexing and the measuring of intensities of thousands of reflections. Thisis a tedious operation and Furnas and Harker31 have made the first stepstowards designing a Geiger-counter diff ractometer that permits of theaccurate location of reflections and the measurements of their integratedintensities.The heart of the apparatus is a “ Eulerian cradle ” permittingrotation of the crystal about each of the Eulerian axes. This eliminates theneed to raise the counter out of the equatorial plane and thus enables thereciprocal lattice to be surveyed by a “ zone ” or “ cone ” scheme.One difficulty in determining the molecular weights of proteins by theX-ray method is that of obtaining accurate and reliable density measure-ments of the crystals. Law and Richards32 have described a gradientcolumn technique for the measurement of densities and a microbalancewhich combines the procedures of weighing the crystal and simultaneouslyenabling the operator, by taking suitable photomicrographs, to determinethe dimensions of the crystal and hence its volume by optical means.Increasing use is being made of the Lipson-Taylor diffractometer origin-ally suggested by Bragg in the study of the diffraction intensities of helicalstructures. A recent modification of this instrument has been successfullyused34 in the study of synthetic polypeptides, and Hooper, Seeds, andStokes 35 describe a photographic process for the preparation of masks forthe Lipson-Taylor diffractometer in their studies of the larger molecules likedeaxyribonucleic acid and collagen.has shown two methods ofproducing fibre diagrams of a given structure by means of t h e opticaldiffractometer by superposing a number of diffraction patterns of theprojections of the structure.An expression is obtained for the minimumnumber of projections required for a faithful representation of the fibrediagram.New types of space-filling atomic model for use in building organic28 B. W. Low and H. J. GrenviHe-Wells, Proc. Nal. Acad. Sci. , 1953, 39, 785.29 H. Lindley, Biochim. Biophys. Ada, 1955, 38, 194.SO H. Lindley and J. S. Rdlett, abid., p. 183.32 B. D. Low and F. M. Richards, J . Amer. Chem. SOL, 1952, 74, 1660; Nature,33 W. L. Bragg and A. R. Stokes, Nature, 1945, 156, 332; see also ref. 7.34 A. Elliott and P. Robertson, Actu Cryst., 1955, 8, 736.35 C. W. Hooper, W. E. Seeds, and A. R. Stokes, Nature, 1955, 175, 679.313 A. R. Stokes, Actu Cryst., 1955, 8, 27.T. C.Furnas and D. Harker, Rev. Sci. Instv., 1955, 26, 449.1952, 170, 412BERNAL AND CARLISLE. 385structure models and polypeptide chains are de~cribed.~' These are nowessential aids in organic and protein-structure X-ray analysis.Artificial and Natural Fibrous Proteins.-In this section is given a briefaccount of the advances in knowledge of the structure of the three maintypes of fibrous protein : (1) the a, or helical keratin, type; (2) the p, orextended silk, type; and (3) the twined, or collagen, type. In each casethe evidence provided by the relatively simple synthetic polymers is firstpresented followed by that for natural products.There is good evidence from the use of X-rays andpolarised infrared rays that synthetic polypeptide chains with neutral residuestake up the a-helix configuration.New measurements are available 38for better oriented films of poly-(y-methyl L-glutamate) with cell dimensionsa = 11-95 A, b = 20-70 A, c = 43.2 A (fibre axis), which are closely related tothe hexagonal cell previously described.39 This work shows that the generalpattern of non-meridional reflections is in at least qualitative agreement withthat calculated by Cochran and In line with these conclusions isthat of the work carried out on poly-(c-benzyloxycarbonyl-L-lysine) and acomplex of a form of poly-(y-methyl ~-glutamate).*l The former crystalliseswith one molecule in a simple hexagonal cell and gives an X-ray diffractionpattern that is almost identical with the latter. There appears to be someuncertainty about the correct choice of unit cell for poly-(7-methyl L-glutamate) , but its cylindrical Patterson function 42 shows qualitativeagreement with a theoretical cylindrical Patterson projection for the 3:6-residue a-helix.The interpretation of the polarised infrared absorption bands for syntheticpolypeptides can still only be applied in a qualitative manner owing to lackof information concerning the precise nature of the atomic motions associatedwith certain absorption bands.43 Particularly is this so for the molecularmotion concerned with the CEO band which is known to be complex.Nevertheless, the method has been useful in distinguishing between the a- andthe @-form of polypeptides, particularly where orientation is difficult, bynoting the differences in the characteristic frequencies of the N-H deform-ation mode.44 Elliott 45 has used the characteristic frequencies of the C=Oband to distinguish between optically active a-polypeptides and the meso-forms.A recent study of poly-L-alanine46 shows that the fibres exist inthe a- and the @-form. The a form is similar to poly-(y-methyl L-glutamate),but there appear to be 47 residues in 13 turns of the helix. A detailedreport 47 of the investigations on this synthetic polypeptide has shown that37 G. S. HartIey and C. Robinson, Trans. Faraday SOC., 1952, 48, 847; C. Robinsonand E. J. Ambrose, ibid., p. 854; R. B. Corey and L. Pauling, Rev. Sci. Instr., 1953,84, 621.38 C. H. Bamford, L. Brown, A. Elliott, W. E. Hanby, and I.F. Trotter, Proc.Roy. Sot., 1953, A , 141, 49.39 C. H. Bamford, W. E. Hanby, and F. Happey, PYOC. Roy. Soc., 1951, A, 205, 30.40 W. Cochran and F. H. C. Crick, Nature, 1952, 169, 234.41 H. L. Yakel, jun., Acta Cryst., 1953, 6, 724.42 H. L. Yakel, jun., and P. N. Schatz, ibid., 1955, 8, 22.4s R. D. B. Fraser and W. C. Price, Nutwe, 1952, 170, 490.44 E. J. Ambrose and A. Elliott, Proc. Roy. SOL, 1951, A , 205, 47.45 A. Elliott, ibid., 1953, A , 221, 104.46 C. €1. Bamford, L. Brown, A. Elliott, W. E. Hanby, and I. F. Trotter, Nutwe,4? L. Brown and I. F. Trotter, Trans. Furuday Soc., 1956, in the press.HeZicaZ type (E).1954, 173, 27.REP.-VOL. LII 386 CRYSTALLOGRAPHY.some distortion in methyl side-groups is necessary in the helical structure toaccount for the presence of meridional reflections not supposed to exist onthe basis of the a-helix with equivalent residues.No significant major discoveries concerning the keratin-myosin-epi-dermin-fibrin group have been made since the last Rep0rt.l It was alwaysdifficult to explain the 5.1 A meridional reflection for a-keratin in termsof a single-chain a-helix although the presence of a 1.49 %i reflection recordedby MacArthur 48 supports Pauling and Corey's suggestion that a-keratincan be interpreted in terms of their helix.Both Pauling and Corey49 andCrick M, suggested that the polypeptide chains of a-keratin might assumecompound helical configurations. By this Pauling and Corey mean thatthe straight a-helix is now to be replaced by an a-helix that is itself a super-helix repeating after about 35 turns of the a-helix (126 residues in 190A),and in assuming this configuration it is possible for six such super-helices tosurround a straight-chain helix in such a manner as to give rise to a 7-strandcable of the " coiled coil " type, having a diameter of 30 A.The sense ofthe outer strands of the 7-strand cable must be opposite to that of thecentral strand in order that the compound a-helices may pack around thecentral cable. The 5.1 A meridional arc is explained by this proposedstructure. Crick 51 has shown that when a-helices of the same sense packtogether they will probably do so about 20" out of parallel. Two simplemodels, the 2- and 3-strand rope, are described and used to illustrate thediffraction theory already developed for such structures.6, 22Astbury and Haggitt 52 have shown that before the a-p transformationa-keratin shows a stretching of the 5.1 A and 1.5 spacings which takes placereversibly over a range of 2%. This experiment demonstrates that the twospacings must be intimately related although the authors point out thatthere is no suggestion that these reflections are not orders of the samediffraction series.The low infrared dichroism of the natural fibrous proteins is accountedfor 53 by the presence of a relatively large component not possessing di-chroism. Fraser 54 has studied side-chain orientation in fibrous proteinsand Malcolm 55 has shown that only 15% of muscle can be oriented parallelto the muscle axis, the remainder being folded in secondary folds oramorphous, giving zero dichroism.The proteins from flagella of Proteus vulgaris and BaciZZus subtilis showthe normal a-type diagram together with that of a cross-B-type.11ds56 Inaccordance with other proteins of the keratin group, the meridional 5.1 Aand 1-(5 The fibres are about 120 A thick, witha microperiod of 410 %i which corresponds very closely to the 411 A fibreperiod observed for skeletal muscle by Philpott and>zent-Gyorgyi 584 8 I .MacArthur, Nature, 1943, 152, 38.4s L. Pauling and R. B. Corey, ibid., 1953, 171, 59.F. H. C. Crick, ibid., 1952, 170, 882.f 1 Idem, Acta Cryst., 1952, 5, 381.f2 W. T. Astbury and J . W. Haggitt, Biochim. Biophys. Acta, 1953, 10, 483.53 K.D. Parker, ibid., 1955, 17, 148.j4 R. D. B. Fraser, Nature, 1955, 176, 358.f b B. R. Malcolm, in ref. l l ( d ) .5 6 W. T. Astbury and N. N. Saha, Nature, 1953, 171, 280.5 7 R. S. Bear, J. Amer. Chew. SOC., 1943, 65, 1784.reflections are observed.D. E. Philpott and A. G. Szent-Gyofgyi, Biochini. Biophvs. Acfa, 1954, 15, 165BERNAL AND CARLISLE. 387have recently shown that one of the components of myosin, light mero-myosin, obtained by tryptic digestion shows cross-striations of about420 A. Such close correlation between these periods cannot pass unnoticed.Suggestions are made by Astbury and his co-workers 56 for the geometry ofthe arrangement of subfibrils in flagella and the explanation of the helicalmovements of the flagella in motion.Elliott and Malcolm 59 have been investigating water-soluble silks whichare obtained by dissolving the fibroin in aqueous inorganic salts, followed bydia1ysis.M Infrared and X-ray evidence, in the case of water-soluble Bombyxsilk, points to the existence of a folded configuration for the polypeptidechains.More definite conclusions on chain configuration appear to befound in the water-soluble Anaphe nzoloneyi silk which to a large extent isa copolymer of glycine and alanine. By comparison with poly-L-alanine 46whose configuration is shown to be that of the a-helix and with the copolymerL-alanine-glycine (2 : 1) it is suggested 59 that Anaphe moloneyi silk castfrom trifluoroacetic acid takes up the cc-helix configuration. Freeze-driedAnaphe silk, however, seems to resemble water-soluble Bombyx silk, whereinfrared and X-ray data are not conclusive enough to point in favour of theu-helix.Kratky, Sekora, and Pilz 62 give evidence for the existence of a-helices insolutions of natural silk gel from small angle X-ray scattering.Pleated sheet structures (p).In this category there are two main types,the extended synthetic polypeptides and the natural silks. The X-raypattern of P-poly-L-alanine 469 47 can be indexed on an orthorhombic unit cellcontaining two chains which are hydrogen bonded into sheets in a mannersimilar to that observed 63 in nylon 66. There are interesting comparisonsbetween the fibre diagrams of this polypeptide and silk structures. Marsh,Corey, and Pauling 64 point out that the unit cell dimensions indicated byBamford et al.for p-poly-L-alanine 46 are in good agreement with the pseudo-unit cell of tussore silk based on antiparallel-chain pleated sheets.The early X-ray work on fibroin has been s ~ m m a r i s e d . ~ ~ ~ 66 Warwickerhas based his suggested structure for Bombyx mori silk on an orthorhombicunit cell containing four residues. This is a pleated sheet structure resem-bling that first put forward by Pauling and Corey 67 to account in generalfor the P-type proteins. Marsh, Pauling, and Corey 68 have since putforward alternatives which are still not fully established. They have,however, indicated discrepancies 68 that arise as a result of Warwicker’schoice of atomic co-ordinates and they propose instead a structure that hasa pseudo-monoclinic unit cell, with space group P2,.The four chains packin the cell so that the extended polypeptide chains are held together bylateral N-H . . . 0 hydrogen bonds to form antiparallel-chain pleated sheets,5g A. Elliott and B. R. Malcolm, Trans. Faraday Soc., 1956, in the press.6o P. P. von Weimarn, “ Colloid Chemistry,” Chemical Catalog Co., New York,1932, p. 363; D. Coleman and F. 0. Howitt, Proc. Roy. Soc., 1947, A , 190, 145; E. J.Ambrose, C. H. Bamford, A. Elliott, and W. E. Hanby, Nature, 1951, 168, 264.62 0. Kratky, A. Sekora, and I. Pilz, 2. Naturforsch., 1954, 9b, 803.63 C. W. Bunn and E. V. Garner, Proc. Roy. SOC., 1947, A , 189, 37.64 R. E. Marsh, R. B. Corey, and L. Pauling, Acta Cryst., 1955, 8, 710.6 5 J.0. Warwicker, ibid., 1954, 7, 567.66 R. E. Marsh, R. B. Corey, and L. Pauling, Biochim. Biophys. Acta, 1955, 16, 1.L. Pauling and R. B. Corey, Proc. Nut. Acad. Sci., 1951, 37, 251.68 R. E. Marsh, L. Pauling, and R. B. Corey, Acta Cryst., 1955, 8, 62388 CRYSTALLOGRAPHY.the sheets in turn being staggered with respect to each other. The sequenceglycyl-alanyl(or sery1)-glycyl-alanyl(or seryl) predominates throughout thestructure, so that adjacent sheets pack parallel together at distances ofabout 3.5 A and 5.7 A.In light of this work, Marsh, Pauling, and Corey 68 consider it is un-necessary to suggest an amorphous phase for silk. Yet according to Am-brose and Elliott 69 and Elliott, Hanby, and Malcolm '* there is strongspectroscopic evidence for the existence of such a phase.Further con-firmation that silk fibroin is an extended polypeptide chain comes from thework of Andreeva and Iveronova 71 on Bombyx mori silk. An investigationon tussore silk reveals that the pseudo-unit cell is orthorhombic with aspace group P2,2,2, containing 8 amino-acid residues. This space groupdiffers from that of the Bonabyx mori silk pseudo-unit cell which is P2,. Thisdifference is due to the method of packing of adjacent sheets in Bombyxmori silk which are separated alternately by 3.5 and 5.7 A, whereas in tussoresilk all sheets are at equally spaced intervals 5.3 A apart. Within eachpleated sheet adjacent polypeptide chains are held in antiparallel sense byhydrogen bonds.The principal difference accordingly between the pseudo-structures of the two sheets is in the method of packing pleated sheets. * InBombyx silk the alternation of intersheet distances of 3.5 and 5.7 A corre-sponds to back-to-back and front-to-front packing of the sheets-thisappears to be necessary for the location of larger residues like tyrosine. Intussore silk the regular spacing of pleated sheets implies the lack of residueswith large atomic grouping and is in keeping with chemical evidence thatabout 85% of the silk contains glycine, alanine, and serine, in the cruderatio 2 : 1 : 1.Interhelical hydrogen- bonded collagenous ,types. Although collagen 72was long known to give a diffraction pattern different from the a- andp-type proteins it is only recently that a scheme of classification has beenfound for its constituent polypeptide chains. This has arisen largely as aresult of the X-ray study uf the synthetic polypeptides, poly-L-proline andpolyglycine 11, which indirectly have led to a suitable structure for collagen.Polyproline 73 as prepared by Berger, Kurtz, and Katchalski 74 contains anaverage of 20 residues in the chain and the X-ray powder pattern can beindexed on a hexagonal lattice with indications that there is a three-foldscrew axis parallel to c.There are three proline residues in the unit cell andthe structure consists of polypeptide chains situated at trigonal positions,each having three equivalent residues related by an exact three-fold screwaxis, the residue repeat along the c axis being 3.12 A.It is possible to builda three-fold helix with a repeat of 3.12 using planar amide groups if somedistortion is allowed at the a-carbon atom and it is found that the helix isleft-handed if the absolute configuration at the carbon atom is assumed.An observed perpendicular dichroism for the infrared G O stretching fre-quency is in keeping with such a model.The original X-ray diffraction effect of polyglycine obtained by Meyer69 E. J. Ambrose and A. Elliott, Proc. Roy. Soc., 1951, A , 206, 206.'O A. Elliott, W. E. Hanby, and B. R. Malcolm, Brit. J. AppZ. Phys., 1954, 4, 377.71 N. S. Andreeva and V. J. Iveronova, Doklady Akad. Nauk S.S.S.R., 1954,99,991.7a W. T. Astbury and W. R. Atkin, Natuve, 1933, 152, 348.73 P.M. Cowan and S . McGavin, Nature, 1955, 116. 501.7 4 A. Berger, J. Kurtz, and E. Katchalski, J . Anzer. Chem. SOC., 1954, 76, 5552BERNAL AND CARLISLE. 389and Go 75 has been reinvestigated by Bamford et Using infrared andX-ray methods they have identified two forms of the synthetic polypeptide.Form I, which shows spacings of 4.4,3.45, and 1-16 A, is identified as a (3-typestructure in agreement with Astbury's earlier findings on this syntheticpolypeptide.77 Form 11, precipitated from water by lithium bromidesolution, shows strong reflections at 4-15 and 3.10 A. No satisfactoryexplanation could be put forward for these spacings except that they clearlycould not be explained on the basis of an a-helix type structure. Crick andRich 78 connected the existence of this 3-10 A reflection in polyglycine I1with the residual repetition of 3-12A observed in poly-L-proline and haveconsequently suggested a close similarity in structure for these two syntheticpolypeptides.The polyglycine chains are packed in hexagonal array, eachchain being hydrogen-bonded to each of its six neighbours, forming an infinitesequence of hydrogen bonds running from one cell to the next. The planarpeptide groups in each chain are inclined at about 35" to the fibre axis.The structure contains three residues in 9.3 A and can be described asa = 4.8 A, c = 9.3 Owing to the fact that thereis no asymmetric carbon atom the mirror-image structure P3, is equallypossible. As there are no bulky side groups, interchain hydrogen bondingis dictated by the interaction between neighbouring chains.This is thefirst example reported of interchain hydrogen bonding between helical typepolypeptide chains and it is a warning that no rigid demands are made, inthe case of the natural and globular proteins that all hydrogen-bonds areinternally satisfied in the folded configuration of their respective chains.A cylindrical Patterson function 79 has been calculated €or collagen.There is some difficulty in interpreting the function in terms of the a-helix.In the attempt to maintain planar amide groups and to account for infrareddichroism Randall et al. suggested a sheet-type structure for collagen whoseconfiguration was formally similar to that for a 2,b chain earlier suggestedby Bamford, Hanby, and Happey.81 This structure, like those previouslysuggested,4Ssa failed for the simple reason that it could not account for thestrong meridional reflections at 2.8-3.1 A.Cohen and Bear 83 and Cowan,McGavin, and North 8Q have provided strong arguments based on X-rayevidence that collagen must be a helical type structure. A structure pro-posed by Ramachandran and Kartha,*5 being a revision of an earlier 3-chainmodel based on a 3-chain coiled coil does account for the meridionalspacing of collagen. Rich and Crick 86 found that Ramachandran andKartha's structure was stereochemically unsatisfactory and was incom-75 K. H. Meyer and Y. Go, Helc. Chim. Acta, 1934, 17, 1488.76 C. H. Bamford, L. Brown, E. M. Cant, A. Elliott, W.E. Hanby, and B. R. Malcolm,77 W. T. Astbury, Natuve, 1949, 163, 722.'@ H. L. Yakel, jun., and P. N. Schatz, A d a Cryst., 1955, 8, 22.with space group P3,.Nature, 1955, 176, 396.F. H. C. Crick and A. Rich, Nature, 1955, 176, 780.J. T. Randall, R. D. B. Fraser, S. Jackson, A. V. W. Martin, and A. C. T. North,C. H. Bamford, W. E. Hanby, and F. Happey, Proc. Roy. Soc., 1961, A , 205, 308a W. T. Astbury and F. 0. Bell, Nature, 1940,145, 421 ; L. Pauling and R. B. Corey,1 3 ~ C. Cohen and R. S. Bear, J . Amer. Ckem. SOC., 1953, 75, 2783.84 P. M. Cowan, C. S. McGavin, and A. C. T. North, Nature, 1955, 176, 1062.85 G. N. Ramachandran and G. Kartha, Nature, 1954, 174, 269; 1955,178, 59.Nature, 1952, 169, 1029.Proc. Nat. Acad. Sci., 1951, 37, 272.A.Rich and F. H. C. Crick, Nutuve, 1955, 175, 915390 CRYSTALLOGRAPHY.patible with recent chemical evidence 87 which shows that the amino-acidsequence glycine-proline-hydroxyproline occurs very frequently. Startingfrom the structure of polyglycine I1 78 they showed that if three adjacentchains were twisted as in the Ramachandran-Kartha structure, ie., themajor helix is right-handed while the minor ones are left-handed, two stereo-chemically completely satisfactory structures could be obtained dependenton the two different ways it is possible to select three chains in polyglycine 11.Both structures have one systematic set of hydrogen bonds lying approxim-ately perpendicular to the fibre axis, in agreement with infrared data 88 andthe effect of super-coiling permits room for residues like proline and hydroxy-proline. In one of the structures the hydroxyproline residues are internallyhydrogen-bonded to a G O group of one of the three chains, while in the otherthe hydroxyproline is able to form a hydrogen bond to an adjacent set ofthree chains which appears to be in keeping with the suggestion made byGustavson 89 that hydroxyproline is largely responsible for stabilising thecollagen structure. Although this model has not as yet been established asa correct structure it does broadly satisfy the requirements of the wide-angle X-ray data of collagenous type fibres.The swelling of collagen has been investigated by X-ray diffraction byRougvie and Bear.go Water adsorption gives rise to the straightening ofchains for dry collagen resulting in parallel alignment.Further adsorptionresults in lateral separation of the fibrillar groups with further straighteningof the chains. At low and high adsorption values the sorbed water contri-butes to the increase of the apparent molecular volume in the expectedmanner. Low-angle X-ray diffraction studies on the ultrastructure ofbone 91 of perch and pike indicate rod-shaped particles approximately65 A wide and 200 A long, with the long axis aligned parallel to the fibreaxis of collagen. A close relation was shown to exist between the form ofcrystallisation of the apatite and the structure of collagen. It is interestingto point out that Cowan et aLg2 from improved wide-angle X-ray photo-graphs of collagen have definite evidence of a layer line spacing of 200 A.Vitrosin, prepared from vitreous humour of cattle has been charac-terised from wide- and small-angle X-ray diffraction, electron diffraction,and chemical analysis as a member of the collagen group.It is a fibrousprotein, 100-150 A in diameter, containing an axial periodicity of 640 A.The hydroxyproline and glycine contents are 11.7 and 19% respectively,indicating the ratio of these two residues to be higher than for most verte-brate collagens. X-Ray investigations of the changes with age in thestructure of the human intervertebrate disc have been carried Ageingof the disc, in general, results in fibrillation and precipitation of the collagen.On the basis of X-ray investigations Astbury and Bellg6 placed elastin in87 W.A. Schroeder, L. M. Kay, J. Le Gette, L. Homer, and F. C. Green, J . Amer.Chem. SOC., 1954,76,3556; T. D.Kroner, W.Tabroff, and J. J. McGarr.ibid., 1955,77,3356.88 G. B. B. M. Sutherland, K. N. Tanner, and D. L. Wood, J . Chem. Plzys., 1954,22, 1621 ; G. N. Ramachandran, ibid., 1955, 22, 600.wo M. Rougvie and P. S. Bear, J. Amer. Leather Chemists' Assoc., 1953, 48, 735.w 1 D. Carstrom and J. B. Finean, Biochim. Biophys. Acta, 1954, 13, 183.v2 P. M. Cowan, A. C. T. North, and J. T. Randall, Nature, 1954, 174. 1143.g3 J . Gross, A. G. Matolsty, and C. Cohen, J . Biophys. Biochenz. Cvtol., 1955, 1, 21.5.g4 F. Happey, T. P. Macras, and K. Naylor, ref. l l c , p. 66.v 5 W. T. Rstbury and F. 0.Bell, Tabah Biologic@, 1939, 17, 90.K. H. Gustavson, Nature, 1955, 175, 90BERNAL AND CARLISLE. 891the collagen group, and now recent electron-micrograph studies, combinedwith biochemical and histological studiesF6 have shown the apparenttransformation of collagen fibrils into elastin. Randall and his collaborators’pertinent observations 97 on native and precipitated collagen are worthnoting. Attempts are made (1) to correlate electron-micrograph studies ofcollagen fibrils with low-angle X-ray scattering, (2) to examine the im-portant factors in fibrinogenesis of collagen from solution, both experi-mentally and theoretically, and (3) to define the conditions of formation oflong-spacing fibrils and segmented fibrils more closely. It is found as resultof these studies (1) that the 640 A periodicity arises from a single proteinconstituent, (2) that neither fibrous nor segmented long-spacing materialhas been observed in connective tissues in vivo, and (3) that there is insuffi-cient evidence to decide whether or not structures of collagenous type arebuilt up of units of short length. Their most recent observations show thatthe structure factors calculated from density functions along the fibril axisof different tendons derived from electron micrographs can be successfullycompared with the corresponding structure factors derived from low-angleX-ray photographs of dry fibres.Globular Proteins.-The chief crystalline globular proteins now underX-ray investigations are various hzemoglobins (molecular weight 66,700) andmyoglobins (14,000), insulin (12,000), ribonuclease (13,400) and lysozyme(14,000).For the purpose of this Report it is convenient to deal withhzemoglobins, which have been most intensively investigated, and myo-globins together, although it is fully recognised that there is no necessarystructural connection between these two proteins. Of the remaining three,ribonuclease has so far been examined in most detail.These three groups have been studied by the heavy-atom technique, butonly in the case of hzmoglobins and myglobins has this led to informationon the internal structure of the molecules. For the rest all that is known istheir shape and size, and some tentative hypotheses as to the possiblealignments of their polypeptide chains.Hwmoglobins and myoglobins.The names of these are put in the pluralas many crystalline varieties derived from horse, sheep, whale, seal, ox,tortoise, penguin, carp, and man have been studied. Though differing indetail, these show a common basic structure.Six papers entitled “ The Structure of Hzmoglobin ” Qg describe the firststeps in the attempt to solve the structure of the haemoglobin moleculeusing direct methods. It has been possible, by following lattice changes andconsequent changes in the modulus of the molecular transform at the latticepoints in crystals at different shrinkage stages using absolute measurements,to give reliable dimensions for the hemoglobin molecule, treated as a spheroidof 53 X 53 X 71 A and 45 x 45 x 65A for the hydrated and the dry96 D.Burton, D. A. Hall, M. K. Keech, R. Reed, H. Saxl, R. E. Tunbridge, andM. J . Wood, Nature, 1955, 167, 1966.s7 J . T. Randall, F. Booth, R. E. Burge, S. F. Jackson, and F. C. Kelly, ref. lld,p. Ig27; R. E. Burge and J . T. Randall, Proc. Roy. Soc., 1955, A , 223, 1.(Sir) L. Bragg and M. F. Perutz, Proc. Roy. Soc., 1952, A , 213, 425; (Sir) L.Bragg, E. R. Howells, and M. F. Perutz, ibid., 1954, A, 222, 33 ; (Sir) L. Bragg andM. F. Perutz, ibid., 1954, A , 225, 264; D. W. Green, V. M. Ingram, and M. F. Perutz,ibid., p. 287; E. I<. Howells and M. F. Perutz, ibid., p. 308; (Sir) L. Bragg and M. F.Perutz, zbid., p. 315392 CRYSTALLOGRAPHY.molecule, respectively. By observing these lattice changes an attempt hasbeen made to obtain the absolute signs of reflections for h and I up toA M = 0.24 for that part of the crystal transform which is real.The absolutesigns of layers with h > 2 are left in doubt, but the number of alternativecombinations has been reduced from 296 to 213.Confirmation of the sign determination by application of transformprinciples has come from a study of haemoglobin crystals containing p -mercuribenzoate and silver ions severally, which are isomorphous withnormal monoclinic methzemoglobin. The changes in F(h0I) were used todetermine the x and z parameters of the pair of heavy atoms attached toeach haemoglobin molecule. This was carried out for the normal wet latticeand for one of the acid expanded lattices. The positions of the heavy atomsproved to be slightly different in each case, and this allowed just over two-thirds of the signs to be found with certainty. All the sign relationsestablished by the transform method were confirmed and uncertaintiescleared up.In this way the signs of 87 out of 94 reflections were foundwith certainty. A further check on the signs was made by the study ofglyoxaline-methzemoglobin where there is a close correlation in the alignmentof the molecules in these crystals which are orthorhombic with the orientationof the molecules in the monoclinic form.This work has now resulted in the calculation of an electron-density mapof a single row of molecules on the (010) projection. This is shown inFig. 1 for hzemoglobin molecules suspended in salt-free water.It has beencalculated with terms whose interplanar spacings are larger than 7.0 A ;hence one is looking at a poor resolution of a molecule some 50 A thick.Nevertheless, the picture must be substantially correct and it is also thefirst picture of a protein molecule derived without chemical or physicalassumptions. The broken lines running across the positive regions of themap indicate the outline of the molecule along the a direction.Although the internal structure of the molecule is still obscure the mapis in general agreement for a tilted spheroid 71 x 54A and much greaterresolution is needed to confirm or refute the arguments put forward for thepresence of parallel polypeptide chains in the molecule.X-Ray studies on crystalline myoglobin are being actively pursued byKendrew 99 at Cambridge.Intense activity at the moment is being centredon the isomorphous replacement technique, so successful in the case ofhamoglobin. The crystal symmetry and Patterson projection of a mono-clinic, pseudo-orthorhombic form of horse myoglobin has been described byKendrew and Trotter.lW The unit cell is very closely related to the mono-clinic form of the protein previously described. lol Interesting speciesspecificities of myoglobin in relation to antigen-antibody reactions have beendescribed by Kendrew, Parrish, Marrack, and Ostens.lo2 It is of interestthat crystallographic and immunological techniques have been used inconjunction for the first time in an attempt to gain a deeper insight into thedifficult field of immunochemistry.fie J.C . Kendrew, personal communidation.loo J . C . Kendrew and I. F. Trotter, Acta Cryst., 1954, 7, 347.Io1 J. C. Kendrew, R o c . Roy. SOL, 1950, A , 201, 62.lo2 J. C. Kendrew, R. G. Parrish, J. R. Marrack, and E. S. Ostens, Nalzwe, 1954,174, 946BERNAL AND CARLISLE. 393Other publications have included the demonstration of an improvedtechnique for studying the discontinuous lattice changes in haemoglobincrystals,lW the study of the form birefringence of the ha3moglobin moleculeand the possible application of such a method to the determination of the10 20 30 40 50 a =109*2I t r r r l i i i i l r r ~ r l i i r i l ~ i r ~ lFIG. 1. Fouvier projection of u YOU of hawtoglobin molecules in salt-free water.The zerocontour corresponds to the density found where the whole depth of the unit cell is fclledwith water. The other contours are drawn at intervals of I eZectron/A4 above or belowthat level. (Reproduced, with permission, from Bragg and Perutz, Proc. Roy. Soc.,1954, A , 226, 315.)form of other protein molecules,lw the study of polarisation dichroism, formbirefringence, and molecular orientation in crystalline haemoglobins lo5 andan X-ray study of reduced human haemoglobin,lo6 showing its similaritiesto horse methzmoglobin.The X-ray work on globular proteins may now be divided into (a) thatwhich deals with the direct analysis of the structure, the isoxnorphousheavy-atom replacement technique being used with the minimum of chemicalassumption, and (b) attempts to interpret the Patterson functions of thesecrystals. Critical analyses of the latter have been of use in rejecting hypo-theses of very simple models for globular proteins.has suggested from a study of the three-dimensional Patterson functions ofhaemoglobin that the molecule cannot be regarded as a simple alignment ofFor instance, Crickl** H.E. Huxley and J. C. Kendrew, Acta Cryst., 1953, 6, 76.lo4 W. L. Bragg and A. B. Pippard, ibid., p. 865.lo6 M. F. Perutz, ibid., p. 859.lo6 M. F. Perutz, I. F. Trotter, E. R. Howells, and D. W. Green, zbid., 1955, 8, 241.F. H. C. Crick, Pi1.D. Thesis, Cambridge, 1953; Acta Cryst., 1952, 5, 381;1953, 6, 600394 CRYSTALLOGRAPHY.polypeptide chains folded backwards and forwards on themselves, but it isonly by postulating irregularities such as parallel alignment of chains overshort lengths that an explanation can be given for the low vector density ofthe Patterson functions.Furthermore, from a study of the strength of the10-A reflections it is possible that hzmoglobin could consist of a-helices notnecessarily parallel to one another. Caution is expressed that this must notbe taken to mean that the molecule is largely composed of a-helices. Argu-ments that hzmoglobin is made up of globulite molecules are put forward byWrinch lo* who has further concluded from a study of the three-dimensionalvector maps of hzmoglobin that there is no evidence for the existence ofparallel polypeptide chains in this molecule.lo9 Arguments against theexistence of parallel polypeptide chains in myoglobin have been putforward.lloInsuligt. The anomalous situation exists that for this protein moleculethe sequence of amino-acids in the sub-unit of 6000 comprising 2 chains iscompletely known.lll Yet no model so far built involving the four chainscoiled in the a-helix configuration 112 has satisfactorily accounted for theX-ray crystal data. A preliminary account has been published of theorientation of polypeptide chains in orthorhombic air-dried crystals of acidinsulin sulphate. 113 A complete three-dimensional Patterson synthesis hasbeen calculated by using 96 reflections, which shows marked ridges of vectordensity in approximately hexagonal packing at about 10-13 A from oneanother running parallel to the a axis.I t is suggested that these ridges ofhigh vector density might correspond with possible chain directions of themolecules in the crystal.The amino-acid composition of this enzyme has beenrecently determined by Hirs, Stein, and Moore 114 who find that the moleculeis composed of 128 amino-acid residues, and preliminary studies by Afinsen,Redfield, Choate, Page, and Carroll 115 have shown that the molecule consistsof one chemical chain.reports the preparation of ribonuclease crystals containingdyes with and without heavy atoms, and the cell dimensions and space groupsof at least four different protein dye complex crystals. Crick and Magdoff 117have described a crystalline form of ribonuclease containing iodophenolblue, and details are given 118 of a new type of shrinkage phenomenon inthese crystals involving appreciable alterations in the intensities of theX-ray diffraction by the crystal. Attention is drawn to the obtaining ofaccurate measurements of cell dimensions in the light of solvent effects onRibonuclease.Harkerlo* D.Wrinch, J . Chem. Phys., 1952, 20, 1332; Acta Cryst., 1952, 5, 694.lo* Idem, ibid., 1953, 6, 562, 638.110 Idem, ibid., 1952, 5, 694.ll1 F. Sanger and H. Tuppy, Biochem. J . , 1951, 49, 463, 481; F. Sanger andE. 0. P. Thompson, ibid., 1953, 53, 353, 366.112 D. P. Riley and U. W. Arndt, Nature, 1953, 172, 245 i C. Robinson, $bid., p. 27;H. Lindley and J. S . Rollett, Biochim. Biophys. Acta, 1955, 18, 183.113 B.W. Low, Nature, 1952, 189, 955.114 C . H. W. Hirs, W. H. Stein, and S. Moore, J . B i d . Chem., 1954, 211, 941.lt5 C. B. Anfinsen, R. R. Redfield, W. L. Choate, J. Page, and W. R. Carroll, ibid.,116 D. Harker, Progress Report of Protein Structure Project, 1-3-1954 to 28-2-117 F. H. C . Crick and B. Magdoff, ibid., 1955, 8, 468.llB Idem, ibid., p. 461.1954, 207, 201.1955; Acta Cryst., 1954, 7, 654BERXAL AND CARLISLE. 395the crystal. A third paper 119 describes the three-dimensional Pattersonvectors of the monoclinic form of ribonuclease they call 11. This Pattersonvector distribution is calculated with reflections having spacings greaterthan 3 A and the authors point out that there is no apparent evidence forany polypeptide chain direction in the crystal.have calculated a second three-dimensional Pattersonvector distribution of the monoclinic form of ribonuclease based on a new setof X-ray results, and find that the Patterson vector maps are in substantialagreement with a previous three-dimensional Patterson calculation of thecrystal for which a smaller number of terms was used.This has been con-sidered to supply more evidence to justify the original suggestion that thepolypeptide chains of the molecule are lying in or near the c axis direction ofthe crystal, in agreement with infrared dichroic measurements 121 and insupport of the earlier findings by Carlisle, Scouloudi, and Spier 122 that it isexceedingly difficult to interpret the X-ray data of this crystal in terms of ahexagonal packing of ct-helices.Wrinch has put forward evidence forglobular units, with low-density interiors, as forming the basis of the ribo-nuclease molecule.Lysozyme. Corey, Donohue, and Trueblood 124 have calculated thethree-dimensional Patterson function of air-dried tetragonal lysozymechloride crystals. Only 198 observed reflections from the air-dried crystalbeing used, it is not surprising that the authors saw no distinctive character-istics in the vector distribution that would support the presence of eitherIX- or y-helices in the molecule.In his attempts to find proteins of low molecular weight suitable fordetailed study Crick 125 has given the cell dimensions for lysozyme nitrate.King 126 has recently completed a measurement of the wet and the dry celldimensions of lysozyme nitrate and iodide which are isomorphous in bothstates.The wet crystals contain 4 molecules per unit cell and the driedcrystals 2 molecules per unit cell; Carlisle has found I2O that lysozymebromide behaves similarly.Nucleic Acids.-The unravelling of the structure of deoxyribonucleicacid has been a triumph of combined chemical and crystallographic attack.(A good general account 127 is given in Chargaff and Davidson’s book,‘‘ TheNucleic Acids.”) The 3’ : 5’-diester linkage proposed by Brown and Toddfor a method of joining nucleotides through phosphates has been amply con-firmed,128 but X-ray analysis was required to bring out the stereochemistryof this arrangement. Deoxyribonucleic acid can be made in the form oforiented fibres.These were studied by Astbury f29 who, noting the strongmeridional reflection at 3.4A, concluded that the structure consisted of aCarlisle et al.ll0 F. H. C. Crick and B. Magdoff, Acta Cryst., 1956, 9, in the press.lZo C. H. Carlisle, M. Ehrenberg, G. S. D. King, RI. Levy, and H. Scouloudi,121 A. Elliott, Proc. Roy. Soc., 1952, A , 211, 490.122 C. H. Carlisle, H. Scouloudi, and M. Spier, Proc. Roy. Soc., 1953, B, 141, 85.123 D. Wrinch, Biol. Bull., 1953, 105, 354.12* R. 13. Corey, J , Donohue, and K. N. Trueblood, Acta Cryst., 1952, 5, 701.125 F. H. C. Crick, Acta Cryst., 1953, 6, 221.lzG G. S. D. King, personal coniinunication.lZ7 Ref. 15, p. 461lea Cf. - 4 n n . Reports, 1952, 49, 246; 1954, 51, 275.lz9 W.T. Astbury, Sywip. SOL. Exp. Biol., 1947, 1, 66pcrsonal com munication396 CRYSTALLOGRAPHY.pile of purine and pyrimidine groups at right angles to the fibre axis whichwould also serve to explain the strong negative birefringence. Next Fur-berg,130 studying the crystal structure of the nucleoside cytidine, showed thatthe plane of the sugar molecule was roughly at right angles to that of thepyrimidine base. He accordingly put forward two alternative structures(Fig. 2) for nucleic acid, both involving a pile of purine pyrimidine basesQ nf4- 3-4AI 3.4A3 , 4 i i I0 C,N,OModel I @ p 0&3A ModelIIFIG. 2. Two models of thymonucleic acid based on nucleotides of the “ standard ” conjigura-The planes of the $urine and pyrimidine rings are perpendicular to the p2ane of(Reproduced, with permission, from Furberg, Acta Chenz.Scand., 1952,tion.?he paper.6, 634.)in one of which the phosphate sugar groups were peripheral, in the othercentral. The second alternative was further developed by Pauling andCorey,131 who introduced the concept of a helical structure, but this led tocontradictions. The first led to a structure of unexpected complexity butof great beauty and ultimate significance. Its establishment was the resultof the combination of careful X-ray study of good preparations of nucleicacid and the impetus of the ideas of helical structures already developed forproteins. It arose out of consultations between workers at King’s College(London) and at the Cavendish Laboratory (Cambridge) and has led in turnto a deeper appreciation of the part helical systems play in biologicalstructures (see section on Plant Viruses).Franklin and Gosling 132 foundthat sodium deoxyribonucleate fibres can occur in two reversible statesA and B, and that in both states the phosphate groups are accessible towater and hence must lie on the outside of any roposed structure. StatesA and B show layer line spacings of 28 and 34 x respectively : the formerexists at about 75% relative humidity and possesses a high degree ofcrystallinity ; the other exists at higher humidities and its X-ray diffraction130 S. Furberg, Acta Claeni. Scatid., 1952, 6, 634.131 L. Pauling and R. B. Corey, Pvoc. Nut. Acnd. Sci., 1953, 39, 84; Natzrre, 1953,133 R.E. Franklin and R. G. Gosling, Acta Cryst., 1953, 8, 073.171, 346397 BERNAL AND CARLISLE.pattern shows a lower degree of crystallinity but is even inore characteristicof the type of diffraction pattern given by a helical structureeS6Watson and Crick, 133 considering this ex perimen tal evidence, arrived a tthe conclusion that deoxyribonucleic acid was not composed of piles ofsingle nucleotides as previously supposed but of pairs of nucleotides (Fig. 3)0 1 2 3 4 5 i t ' " " ' FIG. 3. The $airing of adenine and guanine in deoxyribonucleic acid. Hydrogen bondsThe arrow vepresents(Reproduced, with permission, from Crick and Watson,are shown dotted.the crystaZZograph.ic diad.Proc. Roy. SOL, 1954, A , 223, SO.)One carbon atom of each sugar is shown.formed by hydrogen bonding.Each pair was composed of a pyrimidine anda purine residue, and consequently were either thymine-adenine or cytosine-guanine. This idea suggested in turn that a double instead of a single helicalphosphate sugar chain (Fig. 4) was coiled around the same axis, repeatingFIG. 4. This Figure i s purely diagrammatic. The two ribbons symbolise the two phosphate-sugar chains, and the vertical rods the $airs of bases holding the chains together.The horizontal line marks the fibre axis. (Reproduced, with permission, fromWatson and Crick, Nature, 1953, 171, 737.)10 nucleotides a t 34 intervals. This model provided an explanatione X-ray and titration observations of sodium deoxyribonucleate fibres133 J.D. Watson and F. El. C. Crick, Nature, 1953, 171. 737; Proc. Roy. Soc., 1954,A, 223, 80398 CRYSTALLOGRAPHY.of different h~1nidities.l~~ It also had other implications for biochemistryand genetics. The two chains postulated follow right-handed helices butrun in opposite directions; consequently if one chain has the sequence :adenine-cytosine-thymine-adenine, then the corresponding sequence on theother chain must be thymine-guanine-adenine-thymine. This is in keepingwith the chemical evidence 134 that the ratio of purine to pyrimidine basesare usually close to unity. Furthermore, the structure is compatible lZ99 135with X-ray and optical evidence for structure B. Further detailed workdGfining the geometry of deoxyribonucleic acid has been published.1367 13iA cylindrically symmetrical Patterson function and a three-dimensionalPatterson function have been calculated for the sodium salt of form A. Itis shown that the P-P vectors in these Patterson functions indicate a slightlymodified form of the model proposed by Watson and Crick,133 viz., that thephosphorus atoms form two coaxial helical strands of 9 A radius and 28.1 Apitch, with 11 phosphorus atoms spaced equally along each turn of eachstrand, and the separation of the helical strands in the direction of theircommon axis is 14 A. This model suggests that ionic links between phos-phate groups are primarily responsible for maintaining order in the three-dimensional crystal. I n structure A, the sodium ions hold the chainstogether and, as the relative humidity increases, the additional water insome way weakens the directional property of the phosphate link andstructure B then appears.Water absorption leads ultimately to gel fonn-ation and solution. These X-ray data give no direct evidence for the locationof the sugar and base rings in the structure but it is clear that they mustpoint inwards towards the helical axis and that the diameter of the helix ofstructure B should be about 20 A with a pitch of 34 A. A radial distributioncurve for the air-dried sodium salt 138 has been reasonably interpreted interms of the two-strand helix model.I n their studies of nucleoprotein, Wilkins and Randall 139 have pointedout that there is a similarity between the X-ray photographs of sperm headsand those of the fibres of pure sodium thymonucleate. Watson andCrick 140 suggest that there is room between the pair of polynucleot ide chainsfor a polypeptide chain in the p-configuration to wind around the same axis,and this argument is based on their interpretation of the relevant publishedX-ray pictures.129$ 139 Riley and Arndt,l*l in investigating the X-rayscattering of the air-dry compacts of nucleoprotein specimens from herringsperm and calf thymus together with the corresponding separated proteinsand nucleic acids, conclude that the nucleoproteins are simple additionproducts rather than the type of complex suggested by Watson and Crick.130 E.Chargaff, C. F. Crampton, and R. Lipschitz, Nature, 1953, 172, 289; G. R.Wyatt, “ The Chemistry and Physiology of the Nucleus,” Academic Press, New York,1952.135 R.E. Franklin and R. G. Gosling, Nature, 1953, 171, 740; M. H. F. Wilkins,A. R. Stokes, and H. R. Wilson, ibid., p. 738; M . H. F. Wilkins, R. G. Gosling, andW. E. Seeds, ibid., 1951,167, 759.136 R. E. Franklin and R. G. Gosling, Acta Cryst., 1953, 6, 67; Nature, 1953, 172,156; Acta Cryst., 1955, 8, 151.13’ M. H. F. Wilkins, W. E. Seeds, A. R. Stokes, and H. R. Wilson, Nature, 1953,172, 759.13* U. W. Arndt and D. P. Riley, Nature, 1953, 172, 803.130 M. H. F. Wilkins and J . T. Randall, Biochim. Biophvs. Acta, 1953, 10, 193.l40 J . D. Watson and F. H. C . Crick, Nature, 1953, 171,-964.141 D. P. Riley and U. W. Arndt, ibid., 1953, 172, 249BERNAL AND CARLISLE. 399Recently Feughelman et ~ 1 .l ~ ~ have shown that in the synthetic deoxyribosenucleoptrotein, made by combining nucleic acid and protamine, the X-rayphotograph can be reasonably well interpreted in terms of an extendedpolypeptide chain wound helically around the nucleic acid helix. Thisevidence obtained from oriented wet fibres is absent when the fibres areexamined in the dry unoriented state.Feughelman et aZ.142 have put forward a modified structure for deoxy-ribonucleic acid, in that it is a tighter two-chain helix conforming to dimen-sions suggested by Franklin and G0sling.13~ This reduces the generaldiameter of the molecule, so moving each pair of purine-pyrimidine basestowards the helix axis.Rich and Watson 143 have pointed out that the X-ray patterns of ribo-nucleic acid from different sources appear to be similar, implying that theunderlying three-dimensional configurations of these nucleic acids are possiblysimilar in spite of large differences in their respective purine : pyrimidineratios.Drawn fibres show negative birefringence, like deoxyribonucleic acid,suggesting a similar orientation of nucleotide groups with respect to the fibreaxis. As the water content increases the fibres swell, losing birefringenceand becoming amorphous. The fibres show " necked " regions on extensionand become positively birefringent in contrast to the rest of the structure;similar effects were previously noted in deoxyribonuclease by Wilkins,Gosling, and Seeds. 135Plant Viruses.-Following the success of the chemically feasible structurefor deoxyribonucleic acid based on a helical structure, Watson turnedto an interpretation of the wide-angle diffraction photographs of tobacco-mosaic virus which had earlier been studied by Bernal and F a n k ~ c h e n .~ ~ He suggested, on the basis of the theory of X-ray diffraction by helicalstructures,6 that the virus particle was one giant helical molecule, composedof a large number of equivalent protein sub-units helically arranged arounda core of ribonucleic acid. The proposed helix is repeated after 3 turns in68 It is considered that n isof the order of 10. By using this value and a molecular weight 146 of 5 x lo7,it was found that the weight of a protein building unit was 35,000, a figureabout twice that found by chemical methods.147 Schramm and Braunitzer 14*believe there is N-terminal proline.A detailed analysis by Franklin 149 of the X-ray pattern of tobacco-mosaic virus, by using a cylindrical Patterson function, has revealed thatWatson's preliminary considerations about the virus particle's being ahelical structure are essentially correct, but should be modified in detail.Forinstance, A value of 12 was suggested but 16 now seemsand contains (3n + 1) sub-units per repeat.is larger than 10.142 M. Feughelman, R. Langridge, W. E. Seeds, A. R. Stokes, H. R. Wilson, C. W.Hooper, M. H. I?. Wilkins, R. K. Barclay, and L. D. Hamilton, Nature, 1955, 175, 834.143 A. Rich and J. D. Watson, ibid., 1954, 173, 995; Proc.Nut. Acad. Sci., 1954,40, 759.144 J. D. Watson, Biochim. Biophys. Acta, 1954, 13, 10.145 J. D. Bernal and I. Fankuchen, J. Gen. Physiol., 1941, 25, 111.146 R. C. Williams, R. C. Backus, and R. L. Steere, J . Anzer. Chem. Soc., 1951, 73,14' J. L. Harris and C. A. Knight, Nature, 1952, 170, 613; C. A. Knight, Adv.148 G. Schramm and G. Braunitzer, 2. Natzwforsch, 1963, 8b, 61.149 I<. E. Franklin, Nutwe, 1955, 175, 379.2062.Vivus Research, 1954, 2, 153400 CRYSTALLOGRAPHY.more probable,l50 giving 49 building units in 3 turns of the helix of pitch23 A. There is evidence from both the cylindrical Patterson function anda detailed study of the X-ray pattern that there is an important structuraldiscontinuity between one turn of the helix and the next and this appears tobe associated with an external grooving of the particle.This groovingresults in the virus particle's having a larger surface area than that associatedwith a cylinder of 150 A diameter.145 Since the mean radius of the tobacco-mosaic virus particle is only about 75 A and, of this, the ribonucleic acidcore must occupy the inner 15-20 151 it is clear that a substantial partof the virus protein must lie in the proximity of the surface. The cylindricalPatterson function shows a double row of peaks about 11 A apart, consistentwith the idea that the protein units have a double layer of polypeptide chainsrunning perpendicular to the axis of the particle. Here there is agreementwith infrared measurements 152 which suggest that the protein may be inthe form of %-helices lying perpendicular to the axis of the virus.The helical groove in a mild U2 strain of tobacco-mosaic virus has beenshown 153 to be approximately 30 A deep and it is suggested that the groovein fact may be a common feature in all strains, being more marked in somethan in others.The existence of the groove implies some form of helicalridge around the particle and it is suggested that this takes the form of ahelical array of protuberances, one for each protein sub-unit. The form ofthe groove and ridge is such as to permit a high degree of interlockingbetween neighbouring particles. In further investigations it has beenshown that the X-ray diagrams of three strains of tobacco-mosaic virus anda cucumber virus are closely similar in their main features but differ signi-ficantly from one another in points of detail.A noticeable feature of theseX-ray diagrams is that the intensity maxima do not lie exactly on a set ofequally spaced layer lines. They are displaced to small distances on eitherside of the mean layer line position for layer lines of the type I = (3n + 1)and (3n + 2). The extent of the effect varies with the strain of the virusand this means that there must be a slight variation from the (3n + 1)protein sub-units in 3 turns of the helix.155 In the mild U2 strain there are,for instance, 31.05 &- 0.01 protein units in 3 turns of the helix if n = 10.In the other strains examined, such as U1 and the Rothamsted strain thereare approximately 31.02 units in 3 turns of the helix.Investigations aimed at determining the radial structure of tobacco-mosaic virus have been carried out by Caspar 156 who has studied the radialdistribution curves of the equatorial intensity maxima for the virus and thevirus soaked in lead acetate.It is found that the lead is located at sites25 A and 84 A from the axis of the particle, there being two lead atoms perprotein sub-unit. It is not clear yet what specific residues bind the leadatoms to the protein. Crystallographically, the importance of the work isthat, in noting the difference in intensities of the equatorial intensity maxima150 R. E. Franklin, unpublished work.151 G. Schramm, G. Schumacker, and W. Zillig, Nature, 1955,175,549; R.G. Hart,152 R. D. B. Fraser, Nature, 1952, 170, 491.153 R. E. Franklin and A. Klug, Biochim. Bioplzys. Actu, in the press.15* €3. E. Franklin, ibid., in the press.lS5 R. E. Franklin and A. Klug, Acta Cryst., 1955, 8, 777.156 D. Caspar, Ph.D. Thesis, Yale, 1955.PYOC. Nut. Acad. Sci., 1955, 41, 261BERNAL AND CARLISLE. 401for tobacco-mosaic virus and for the lead-bound virus, it has been possibleto give the first ten maxima their correct signs. A calculation of the radialdensity shows regions of high density at 24 and 40 A from the axis with aneffective radius of 84 A for the virus particle in solution.Watson,157 in suggesting that the ribonucleic acid forms a central coresome 35 A in diameter, drew his ideas from the study of turnip yellow-mosaicvirus,158 where X-ray and chemical evidence suggests that the ribonucleicacid component is situated within the virus particle.Beautiful confirm-ation of this comes from the electron-micrograph studies by Hart on theone hand and by Schramm et al. on the other 151* 159 who show very clearlyin their studies on degraded particles of tobacco-mosaic virus that the ribo-nucleic acid exists as a fibrous central core, some 40A thick. Equallyinteresting are their observations of a hole down the centre of the virusparticle when the ribonucleic acid has been removed.X-Ray studies on an abnormal protein associated with tobacco-mosaicvirus have been carried out by Rich, Dunitz, and Newark 160 and Franklinand Commoner.lG1 The material studied by Rich et al.is non-infectious,nucleic-acid free, and virtually identical with the protein component of thevirus, and at its isoelectric point forms rod-like structures, which are some5’3, shorter than the usual value of 68A associated with tobacco-mosaicvirus. Franklin and Commoner’s studies are connected with an abnormal(B8) non-virus protein polymerised in rods which differs little in chemicalconstitution from that examined by Rich et al. The particle has a repeatunit of 65A, but the protein sub-units are possibly not arranged helicallyaround the particle axis but rather stacked one above the other to form astructure which is grossly similar to that of the tobacco-mosaic virus protein.In further studies Franklin 162 shows that there is a marked structural re-semblance between a repolymerised protein freed from ribonucleic acid,known as Schramm’s A protein,163 and tobacco-mosaic virus.The gel has alower positive birefringence than the virus of the same concentration, whilethe birefringence of dry oriented polymerised A protein is weakly negative.This shows that the ribonucleic acid makes a positive contribution to thebirefringence of the virus and therefore has a structure which is unlike thatof deoxyribose nucleic acid 135 which is strongly negative. The protein gelshows an axial repeat of 69 A as in tobacco-mosaic virus and 62 in thedry state. It seems that when the nucleic acid core is replaced by waterthe structural arrangement of the protein in the virus particle remainsstable, but when this water is removed by drying the particle shrinks andbecomes partially disordered.Small-angle X-ray scattering measurements have been made on southernbean-mosaic virus, tobacco-necrosis virus and tomato-bushy stunt virus.164These nearly spherical viruses are found to have diameters of 286A for157 J. D. Watson, Biochim. Biophys. Acta, 1955, 13, 10.15* J. D. Bernal and C . H. Carlisle, Nutwe, 1948, 162, 139; R. Markham, Discuss.15D G. Schramm, G. Schumacker, and W. Zillig, 2. Naturforsch., 1955, lob, 481.160 A. Rich, J. D. Dunitz, and P. Newark, Nature, 1955, 175, 1074.lel R. E. Franklin and B. Commoner, ibid., p. 1076.162 R. E. Franklin, Biochim. Biophys. Ada, 1951, 18, 313.163 G. Schramm, 2. Naturforsch., 1947, 2b, 112, 249.164 B. R. Leonard, J. W. Anderegg, S. Shulman, P. Kaesberg, and W. W. Beeman,Faraduy SOC., 1951,11, 221.Biochim. Biophys. A d a , 1953, 12, 499402 CRYSTALLOGRAPHY.southern bean-mosaic virus, 280 for tobacco-mosaic virus, and 309 A fortomato-bushy stunt virus. A similar investigation has been carried out onsolutions of turnip-yellow mosaic virus. 165 Measurements of the virus andthe associated nucleic acid-free protein indicate that both particles are nearlyspherical and about 140tf in radius. It is suggested that all these virusparticles have large internal hydrations.* B c P. Schmidt, P. Kaesberg, and W.W. Beeman,Biochim., Biophys. A d a . , 19.54,14, 1 BERNAT, AND CARLISLE. 403Vitamin B1,.-Although vitamin B,, is not mainly a polypeptide thiswork is relevant both as a landmark in X-ray crystallography and as pro-viding information on the packing of residues in complex molecules. Whatis probably the most important X-ray investigation being carried out at themoment is that by Dr. Crowfoot Hodgkin and her schoo1.166 Almost thecomplete structure of a degradation product of the vitamin, a hexacarboxylicacid,I67 has now been determined. The 73 atoms in the molecule (Fig. 5)have been located from three-dimensional structure analyses and althoughit is not possible at this early stage to pick out double bonds it is neverthelesspossible to suggest a likely chemical formula for the degradation product,which provides a solution of the larger part of the chemical structure ofvitamin B,, now in progress. Main points arising so far are : (1) the cobaltatom at the centre of the molecule is attached to chlorine on one side andcyanide group on the other and surrounded by a nucleus of 63 atoms some-what characteri3tic of type I11 porphyrins. (2) The main nucleus consistsof 4 five-membered rings, two of which are directly linked, forming an almostplanar unit. The outer ring of atoms lie alternately above and below theplane of the minor ring ; they are reduced and carry side-chains which appearto be alternately acetic acids, lying on the same side of the ring as thechlorine atom, and propionic acids on the other. Not all the substituentshave been unequivocally located but sufficient is known for the structureanalysis to be carried to a satisfactory conclusion. There is confirmation ofthe same nucleus being present in vitamin BIZ. It appears that an exchangereaction occurs during the formation of the acid, the nucleotide in thevitamin being at the site of the cyanide group in the degradation product.With this reorientation it has been possible to identify the same groupingsin the vitamin, except for the lactam ring in the acid, which is representedby a free acetamide residue. A tentative complete structure correspondingto the empirical formula C,,H,oOl,N,,PCo has been put forward.168J. D. BERNALC. H. CARLISLE.D. C. Hodgkin, J. Peckworth, J . H. Robertson, K. N. Trueblood, R. J. Prosen,and J. G. White, Nature, 1955, 176, 326; Cf. J. C . Speakman, Ann. Reports, 1954, 51,399.16' J. R. Cannon, A. W. Johnson, and A. R. Todd, ibid., 1954, 174, 1168.168 D. C. Hodgkin, A. W. Johnson, and (Sir) A. R. Todd, Chem. SOC. Special Publ.,1956, No. 3, p. 109