"The three-dimensional Patterson function of ribonuclease II"

Date:: 1956

Reference:: PP/CRI/I/1/9

Credit: "The three-dimensional Patterson function of ribonuclease II". Source: Wellcome Collection.

2/5

$BEATRICE S. MAGDOFF, F. H. C. CRICK AND V. LTJZZATI 157 varied from 1-0 to 1-3, for the second from 1-0 to 2-1. It was therefore important not to omit it. A record of the intensities, corrected for polarization and Lorentz factor, is available.* jErrors It is now known that this set of data is not as accurate as it might be, but it is believed that the inaccuracies are not great, and that the percentage error is at least as small as that of any other three- dimensional set of intensities of a protein crystal, nbonuclease or otherwise. The errors are due to three causes. First the crystals were mounted rather dry and thus their intensities were probably fluctuating somewhat, as has been described elsewhere (Magdoff & Crick, 1955a). Thus the recorded intensities probably do not all refer to exactly the same state. Secondly no elaborate checks were made to see that the instrument was always exactly set on the reflections, and this may have produced some errors at higher values of 1 jd. Thirdly, the capillary correction was in one case rather large, and as this correction can only be assessed approx imately it would have been worth while to work with a thinner-walled capillary and/or a thicker crystal. To get some idea of the inaccuracies, the hOl's (out to 20 = 30° using Cu Km radiation) were com pared with what is believed to be an accurate set (see Fig. 7 in Magdoff & Crick, 1955a). The value of A = E\AI\¡XI rises roughly linearly with ljd to about 20% at ljd - 0-28 Â -1 , so that the errors, though not as small as they might be, are certainly accept able. Their effect is likely to smear over the Patterson a little, but we do not believe that these errors will materially affect our conclusions. Computation The Patterson function was calculated, using IBM machines, by a new method (Magdoff, unpublished). There were about 5700 terms, not allowing for symme try. It was found necessary to include elaborate checks to prevent accidental errors, which were more common than was expected. The final Patterson function is believed to contain only minor errors, say less than 10 e. 2 Â 3 at any point. The values were calculated at intervals of a/40, è/40 and c/80. The calculation took one person about 4 weeks. Experimental results (a) The Wilson curve The plot of the average intensity, </>, against 1/d the ADI Auxiliary Publications project, Photoduplication Service, Library o£ Congress, Washington 25, D.C., Û.S.A. A copy may be secured by citing the Document number and by remitting $21-25 for photoprints, or 86-25 for 35 mm. micro film. Advance payment is required. Make checks or money orders payable to: Chief, Photoduplication Service, Library of Congress. is given in Fig. 1. It shows a sharp maximum in the 5 Â region, and the suggestion of a small maximum in the 10 Â region. The powder diagram of ribonuclease published by Arndt & Riley (1955) confirms that this latter peak really exists, although it is small by com- Fig. 1. Plot of </> against 1/d. The circles are computed values. The broken curve is 1560.ftys) exp (-4-56Id?), which values parison with the corresponding peak in many other proteins. This in itself suggests that ribonuclease is unlikely to consist mainly of parallel 'rods' spaced 10 Â apart. The high values of Fig. 1 near the origin are caused by the low electron density of the solvent contrasted with that of the hydrated protein. It is not easy to choose suitable constants to fit a curve of the form SP exp (—A/d 2 ) to Fig. 1, especially as the number of ordered atoms in the crystal is not known. Preliminary measurements to put the data on an absolute scale have been made by Dr Thomas C. Furnas, Jr., and his provisional value has been used throughout this paper. The dotted line of Fig. 1 shows the curve 1560/|(s) exp (—4-56/d 2 ), and these values were used to remove the origin peak. (b) The Patterson function The Patterson function was plotted by a novel method. The actual IBM sheets contained the figures set out in a rectangular array. Contours were first drawn on these sheets. They were naturally distorted from the correct shape because the basic array did not correspond to the monoclinic cell. Correct contours were then drawn on to the final sheets, having an underlying grid of the correct size and shape, by com paring the points where the contours intercepted their grids in the two eases. This method proved to be THE THREE-DIMENSIONAL PATTERSON FUNCTION OF RIBONUCLEASE II 158 quicker than first copying all the computed figures on to the scaled sheets and then drawing the contours. Each contour represents a level of about 100 e. 2 Â~ 3 . Fig. 2 shows the 21 sections of the three-dimensional Patterson function, from y = 0 to y = 206/40. The IBM sheets from which they were plotted have been deposited with the American Documentation Insti tute,* whence copies can be obtained. As might be expected from Fig. 1, the Patterson has a very marked 5 Â shell. This shell is not com-$

Content advisory

"The three-dimensional Patterson function of ribonuclease II"

Licence and re-use