© Nature Publishing Group1953

No. 4356 April 25, 1953 NATURE 737

This figure is purelydiagrammatic. The tworibbons symbolize thetwo phosphate-sugarchains, and the hori­zontal rods the pairs ofbases holding the chainstogether. The verticalline marks the fibre axis

equipment, and to Dr. G. E. R. Deacon and thecaptain and officers of R.R.S. Discovery II for theirpart in making the observations.1 Young, F. B., Gerrard, H;, and Jevons, W., Phil. Mag., 40, 149

(1920).• Longuet-Higgins, M. S., Mon. Not. Roy. Astra. Soc., Geophys. Supp.,

6, 285 (1949).• Von Arx, W. S., Woods Hole Papers in Phys. Oceanog. Meteor., 11

(3) (1950).'Ekman, Y. W.. Arkiv. Mat. Astron. Fysik. (Stockholm), 2 (11) (1905).

MOLECULAR STRUCTURE OFNUCLEIC ACIDS

A Structure for Deoxyribose Nucleic Acid

WE wish to suggest a structure for the saltof deoxyribose nucleic acid (D.N.A,). This

structure has novel features which are of considerablebiological interest.

A structure for nucleic acid has already beenproposed by Pauling and Corey'. They kindly madetheir manuscript available to us in advance ofpublication. Their model consists of three inter­twined chains, with the phosphates near the fibreaxis, and the bases on the outside. In our opinion,this structure is unsatisfactory for two reasons:(I) We believe that the material which gives theX-ray diagrams is the salt, not the free acid. Withoutthe acidic hydrogen atoms it is not clear what forceswould hold the structure together, especially as thenegatively charged phosphates near the axis willrepel each other. (2) Some of the van del' Waalsdistances appear to be too small.

Another three-chain structure has also been sug­gested by Fraser (in the press). In his model thephosphates are on the outside and the bases on theinside, linked together by hydrogen bonds. Thisstructure as described is rather ill-defined, and for

this reason we shall not commenton it.

We wish to put forward aradically different structure forthe salt of deoxyribose nucleicacid. This structure has twohelical chains each coiled roundthe same axis (see diagram). Wehave made the usual chemicalassumptions, namely, that eachchain consists of phosphate di­ester groups joining ~-D-deoxy­

ribofuranose residues with 3',5'linkages. The two chains (butnot their bases) are related by adyad perpendicular to the fibreaxis. Both chains follow right­handed helices, but owing tothe dyad the sequences of theatoms in the two chains runin opposite directions. Eachchain loosely resembles Fur"berg's' model No.1; that is,t he bases are on the inside ofthe helix and the phosphates onthe outside. The configurationof the sugar and the atomsnear it is close to Furberg's'standard configuration', thesugar being roughly perpendi­cular to the attached base. There

is a residue on each chain every 3·4 A. in the z-diree­tion. We have assumed an angle of 36° betweenadjacent residues in the same chain, so that thestructure repeats after 10 residues on each chain, thatis, after 34 A. The distance of a phosphorus atomfrom the fibre axis is 10 A. As the phosphates are onthe outside, cations have easy access to them.

The structure is an open one, and its water contentis rather high. At lower water contents we wouldexpect the bases to tilt so that the structure couldbecome more compact.

The novel feature of the structure is the mannerin which the two chains are held together by thepurine and pyrimidine bases. The planes of the basesare perpendicular to the fibre axis. They are joinedtogether in pairs, a single base from one chain beinghydrogen-bonded to a single base from the otherchain, so that the two lie side by side with identicalz-co-ordinates. One of the pair must be a purine andthe other a pyrimidine for bonding to occur. Thehydrogen bonds are made as follows: purine position1 to pyrimidine position 1; purine position 6 topyrimidine position 6.

If it is assumed that the bases only occur in thestructure in the most plausible tautomeric forms(that is, with the keto rather than the enol con­figurations) it is found that only specific pairs ofbases can bond together. These pairs are: adenine(purine) with thymine (pyrimidine), and guanine(purine) with cytosine (pyrimidine).

In ~ther words, if an adenine forms one member ofa pair, on either chain, then on these assumptionsthe other member must be thymine; similarly forguanine and cytosine. The sequence of bases on asingle cl1ain does not appear to be restricted in anyway. However, if only specific pairs of bases can beformed, it follows that if the sequence of bases onono chain is given, then the sequence on the otherchain is automatically determined.

It has been found experimentally3,. that the ratioof the amounts of adenine to thymine, and the ratioof guanine to cytosine, are always very close to unityfor deoxyribose nucleic acid.

It is probably impossible to build this structurewith a ribose sugar in place of the deoxyribose, asthe extra oxygen atom would make too close a vandel' Waals contact.

The previously published X-ray data5•6 on deoxy­ribose nucleic acid are insufficient for a rigorous testof our structure. So far as we can tell, it is roughlycompatible with the experimental data, but it mustbe regarded as unproved until it has been checkedagainst more exact results. Some of these are givenin the following communications. We were not awareof the details of the results presented there when wedevised our structure, which rests mainly though notentirely on published experimental data and stereo­chemical arguments.

It has not es~aped our notice that the specificpairing we have postulated irrimediately suggests apossible copying mechanism for the genetic material.

Full details of the structure, including the con­ditions assumed in building it, together with a setof co-ordinates for the atoms, will be publishedelsewhere.

We are much indebted to Dr. Jerry Donohue forconstant advice and criticism, especially on inter­atomic distances. We have also been stimulated bya knowledge of the general nature of the unpublishedexperimental results and ideas of Dr. M. H. F.Wilkins, Dr. H. E. Franklin and their co-workers at

© Nature Publishing Group1953

738 NATURE April 25, 1953 VOL. 171

King's College, London. One of us (J. D. W.) has beenaided by a fellowship from the National FOlUldationfor Infantile Paralysis.

J. D. 'WATSON

F. H. C. CRICKMedical Research Council Unit for the

Study of the Molecular Structure ofBiological Systems,

Cavendish Laboratory, Cambridge.April 2.

1 Pauling, I,., and Corey, R. B., Nature, 171, 346 (1953); Proc. U.S.Nat. Acad. Sci., 39, 84 (1953).

2 Jo'urberg, S., Acta Chem. Scand., 6, 634 (1952) .• Chargaff, E., for references see Zamenhof, S., Brawerman, G., and

Chargaff, E., Biochim. et Biophys. Acta, 9, 402 (1952) .• Wyatt. G. R., J. Gen. Phys'iol., 36, 201 (1952)., Astbury, W. '1.'., Syrnp. Soc. Exp. BioI. I, Nucleic Acid, 66 (Camb.

Dniv. Press, 1947).• Wilkins, M. H. :1<'., and Randall, J. '1.'., Biochim. et Biophys. Acta.

10. 192 (1953).

Molecular Structl,lre of DeoxypentoseNucleic Acids

'WHILE the biological properties of deoxypentosenucleic acid suggest a molecular structure con­taining great complexity, X-ray diffraction studiesdescribed here (cf. Astburyl) show the basic molecularconfiguration has great simplicity. The purpose ofthis communication is to describe, in a preliminaryway, some of the experimental evidence for the poly­nucleotide chain configuration being helical, andexisting in this form when in the natural state. Afuller account of the work will be published shortly.

The structure of deoxypentose nucleic acid is thesame in all species (although the nitrogen base ratiosalter considerably) in nucleoprotein, extracted or incells, and in purified nucleate. The same linear groupof polynucleotide chains may pack together parallelin different ways to give crystallinel- 3, semi-crystallineor paracrystalline material. In all cases the X-raydiffraction photograph consists of two regions, onedetermined largely by the regular spacing of nucleo­tides along the chain, and the other by the longerspacings of the chain configuration. The sequence ofdifferent nitrogen bases along the chain is not madevisible.

Oriented paracrystalline deoxypentose nucleic acid('structure B' in the following communication byFranklin and Gosling) gives a fibre diagram as shownin Fig. I (cf. ref. 4). Astbury suggested that thestrong 3 ·4-A. reflexion corresponded to the inter­nucleotide repeat along the fibre axis. The "-' 34 A.layer lines, however, are not due to a repeat of apolynucleotide composition, but to the chain con­figuration repeat, which causes strong diffraction asthe nucleotide chains have higher density than theinterstitial water. The absence of reflexions on ornear the meridian immediately suggests a helicalstructure with axis parallel to fibre length.

Diffraction by Helices

It may be shown" (also Stokes, unpublished) thatthe intensity distribution in the diffraction patternof a series of points equally spaced along a helix isgiven by the squares of Bessel functions. A uniformcontinuous helix gives a series of layer lines of spacingcorresponding to the helix pitch, the intensity dis­tribution along the nth layer line being proportionalto the square of J n, the nth order Bessel function.A straight line may be drawn approximately through

Fig. I. Fibre diagram of deoxypentosc nucleic acid from B. coli.Fibre axis vertical

the innermost maxima of each Bessel flUlction andthe origin. The angle this line makes with the equatoris roughly equal to the angle between an element ofthe helix and the helix axis. If a unit repeats n timesalong the helix there will be a meridional reflexion(J 0 2) on the nth layer line. The helical configurationproduces side-bands on this fundamental frequency,the effect5 being to reproduce the intensity distributionabout the origin around the new origin, on the nthlayer line, corrosponding to C in Fig. 2.

We will now briefly analyse in physical terms someof the effects of the shape and size of the repeat unitor nucleotide on the diffraction pattern. First, if thenucleotide consists of a unit having circular symmetryabout an axis parallel to the helix axis, the wholediffraction pattern is modified by the form factor ofthe nucleotide. Second, if the nucleotide consists ofa series of points on a radius at right-angles to thehelix axis, the phases of radiation scattered by thehelices of different diameter passing through eachpoint are the same. Summation of the correspondingBessel functions gives reinforcement for the inner-

./ Co ...........

~ ~, , .

,/ .

,, .,

A A,A A, ,

~ :- A A :----: ~ A A -+ B .:..-

---.:. ~~ B B

~ / "- B ~0

Fig. 2. Diffraction pattern of system of helices corresponding tostructure of deoxypentose nucleic acid. The squares of Besselfunctions arc plotted about 0 on the equator and on the first,second, third and fifth layer lines for half of the nucleotide massat 20 A. diameter and remainder distributed along a radius, themass at a l(iven radius being proportional to the radiu.. AboutC on the tenth layer line similar functions are plotted for an outer

diameter of 12 A.