NO. 4408 November 12, 1955 NATURE 916 THE STRUCTURE OF COLLAGEN By DR. ALEXANDER RICH* and DR F. H. C. CRICK Medial Ramarch Coumil Unit for tha Stud Cavendish La to of the Modular Strutturn d Blofoglal Systems, ratory, CombrIdge ERY recently, Bsmachsndran end Karths' have mede an importent contribution by proposing a c&d-coil tin&m for c~Uegen. We belier this j&e to be bssically correct but the actual structure euegcsted by them to be wrong. Their structure consists of three palypeptide ~hsh~, as& having 8pprOrimk&td3' 8 three-fold screw axis. h addition, the &sins slowly wind around each other to form e c&led coil, thue mproducing the obeerved non-integer screw aria*. The major helix is right- handed, the minor one 1efManded. Bach &sin is held to ib neighbcura by tt~ sets of e-tic hydrogen bon&. The allowed sequence of reoiduee is -G-R-P-Q-R-P-, etc., where C implies glycine only, R implies any residue except proline or hydroxyproline, and P implies any &due, but ususlly proline or hydroxyproline. Iire believe this structure to be wrong for two reasons. (1) It is stereochemically unsati&ctory. In pnrticulw, there is 13 very short Csc, contact of 3.3 A. (normally 34-4-O A.) and an extremely short G-0 contact of 26 A. (normslly 32-34 A.). In addition, the hydrogen bond angles are on the out- rcido limit of the values usuelly found. (2) It is not compatible with recent work* on the amino-acid sequence, whioh shows t.hat -glY--Pro--hyp* is 8 common sequence in collegen. On the other hand, Dr. Perrline Cowan snd her co-worka* have infanned us that 8 preliminsry optic44 diffmction p&tern of thii structure agrees qualitatively With the obemved Wide-angle X-my pettern of adagem. Thy have &O pointed out to WI that the canflgnretion of the. bsckbone of the polypeptide chain is similsr to th& found by them for p~lypmliae~. Theee fect.6 suggeut that the &NC- ture, though incorrect, ie on the right geneml lines. Ourownworkoncollagenhasspnmgf?omour rfxent structure for polyglycine II *. We have t&m a ccmpact group of three ndjacent ch&s out of the polyglycine lattice (space group P 3,) and twisted them, 8s in the PRmeahendrarr-Kerthe Bt~cture, to form e &m&w coiled coil. Such e group csu be selected in two d&rent weys, since the metry is brigrmsl, not hexsgonal. We have adled the two resulting arraugementa structure I and structure If. Both have a right-handed major helix and e left- handed minor helix. srmnged to fit the ohserved non-integer screw. Both have only one set of system- atic hydrogen bonds linking neighbouring chains. In nt~ructure I the NHgroups point anticlockwise when viewed from the carboxyl ends of the c~18im5'; in RtNCtIUFi n ~b&WilH3. These two structures am thus similsr to th8t of Ramachandran and Kartha in being three-chain COiiCd-COil &~t?tUree : but they differ in having Ody one set of svstematic hydrogen bonds instead of two. ~Moreovec, they are both stereochemically completely .%tisfactory. o Psrmsnent addrem: Phymlal ???o??oo? 8ectioa. NatIona htltuta of afent4.l Edth. Bathada, Earr&ruL Both 6tNCtlX'C I and structure D will accommodate the sequence .--gly-pro-hypm- but they do so in 8 di%rent manner. In structure II the glycine position is nom the tube, end every third residue must be glycine. The proline and hydroxy- proline pcsitions, on the other hand, w far mxn~ved from the t&s. Either proline or hydxmypmline can go into either position-there see&4 to be no pm- fered in neither case can the OH of hydroxy- proline make a hydrogen bond wi4h e CO group of an adjacent chsiu within the same set of three ch&ns. ~tNctUI'f3 1 Will dt3C 8CCOI.XbUkOd8tC thC 8bOVE3 sequence; but when the hydroxyprolii is in the expected position (that is, previous to glycine) its OH group m form 8 hydrogen bond to one of the adjacent ch&ns within the me set. Careful structure building hee ahown that not all the possible pcsitions for hydroxyproline can he occupied if the &ucture is to fit together comfortably. This is compatible with the amino-acid snslys~ of collsgen, which show thst the amount of hydroxyproline present in bovine collagen' would 611 about one-third of these sites. The polypeptide backbones, being held together by oniy one set of hydrogen bonds, have 8 certain amount of flexibilitv. In the undeformed'structure. only glycine can 63 8ccommod8ted in the glycine sitee. However. if the structure is Morn& some- what, them sit& can accommodste other residues, though only to 8 iiited extent. This m8y explaiu ccwtain minor feetures of the amino-acid sequence d&8. We have made an exhaustive study of all poeaible strllctm (using tcpologicul enu?ner8 tion, simihu to t&et of Brsgg, Kendrew end Per&a) of this general type which are compatible with the observed mrew da-that is, with thme parallel chin lit&d by at least one systematic set of hydrogen bonds-and we find that : (1) no Btructure with &JO system&o set8 of hydrogen bonds is stereochemically pcssiile; (2) no other structure with one systematic set of hydrogen bonds is stereocbemically sstisf'ory except the two described above derived from poly- glycine. Neither were w ehIe to add occ&onal backbone hydrogen bonds to structures I and II in 8 convincing manner. All the above findings are independent of a&u- mente about the aide-chain aranpments. We therefore conclude that t3tNCtAUH I end II, though making few hydrogen bonda tptamaMy, are the be& that can be eohieved clang thy line+ Note thet, 86 in. polyglycine II*, it is not unpasslble th8t th;ehCon of one of the three ohaim may be Pmlim'inary work on the optical Mns&me of theee tw etruoturee (tmrried out with Dr. A. Elliott on his optical trsnsform machine) show h&h strut- turee to give Nugh E?tgmmmt with the x-rqr pattern, though structure II shows a diserep~oy on the fourth leyer-line. Both struotums M &o corn- putible with the in&~&a observ&ions. Further work will therefore he required to de&de hetweea them. We BIW at the moment inchned to bvour 916 NATURE November 12, 1955 vOL. l76 +thould `likel to th8llk Dr. A Elliott for l.& 8~91~t~pc8 in obt8iniig the optial transforms. roct. 10 1 m 0. N.. md X&h. 0.. Ndun. 1% 693 iigs), ' cob8o. a. and Bear a 8. J. A-. ulrr. 800. 78. 27f3 (lQ&) Cow8o.k Bl., NO&. A. 0. T. and Bwd8ll. J. `i... In "Tlw N& 8odstnuttMeofcollagw" htt8rwu~Lnodu lOtJ)* w In "Flbrona Proteiru and dbe1r Blolqdul S&cnncc" &c lk&. BloL syrup.. 9 w8lub. Univ. Plah 1955). Dew. IL s.; . fitruoture I, a it m8kea &%&ive and eysMm8Uo um of the hydroqqwoline raaidueo. "+avam;*t presented extenrtive evidemx mqgaamg . unusu8I 8minoaoid etebiliza f&e oollagen &uoture. wefeelth8tthisiaulorelikelytotak6placeby form8tion of hydrogen bon& within 8 group of three ch8iM th8n between cli&mnt group3 of cheins? In addition, &nIoture I expl8imJ in 8 n8tur8l way the amino-a&d sequence date. !l!hua while the peptide hypro-gly is common, gly-hypro ie not found. Sim- ihuly, @y-pro is common, where& pro-gly is r8re. In etructure I this preference ia expLained in terma of fhehydroxyproiine-hydrogenbondbetweentheoheias. >+Uthough we 08nnot et the moment m8ke 8 ihl8l decision between the two l#euam% we think it very probable that one of them ie correot (or, lees likely, both). The general agreement with the X-ray ' COM, P. N.L, ind xd38v4 8., N&w, 11). ii01 (1956). ' W F. H. Q. aml Blab. A., Ndn, 178,180 (19m. pattern, the close reeemblence to polyproEne snd to ' Born*. J. H.. aad Kenton. B. H.. B&&m. J.. U 9K8 119~. polyglycine II, ard the ability to erpl8in the major features of the amino-acid eequence dete all suggwt that COh@Xl h 8 8t?U~Ul'9 Of thifl type ; th8t ie, 8 three-chain, coiled-coil etruoture, with one set of _ _ sptematic hydrogen bonds. " t3~tmon, K. Il.. Nutuw. 175, 70 (1956). `Pmauml oommwlutloo. Alau aommuolc8tad 8t the Snl later. o8tlowl Cunmn of Blocbwnlrtn. Bmmcls. 1955. PtmS M. F.. Proc. Rag. k.,