ACTIVATED CONFORMATIONS OF THE RAS-GENE-ENCODED P21 PROTEIN .2. COMPARISON OF THE COMPUTED AND HIGH-RESOLUTION X-RAY CRYSTALLOGRAPHIC STRUCTURES OF GLY-12 P21

The ras-oncogene-encoded p21 protein is a G-protein that has been show n to cause the malignant transformation of normal cells and has been i mplicated in causing human tumors. p21 is thought to be activated by t he binding of GTP in place of GDP to the protein. We have previously c onstructed the three-dimensional structure of the p21 protein bound to GDP from an available alpha-carbon tracing of this protein using a co mbination of molecular dynamics and energy minimization (Dykes, et al. , J. Biomol. Struct. Dynamics, 9:1025-1044). Until the recent publicat ion of the all-heavy-atom x-ray crystallographic molecular coordinates of p21 residues 1-166 bound to a non-hydrolyzable GTP derivative (Gpp Np), no all-atom structure of the p21 protein has been available in th e Brookhaven National Laboratories Protein Data Bank (PDB). In this co mmunication we compare our computed structure for the p21-GDP complex to this x-ray crystal structure. We find that the two structures agree quite closely with one another, the overall RMS deviation for the bac kbone being 1.47 angstrom and 2.71 angstrom for all of the atoms. We h ave identified the regions of the protein that are responsible for the most significant deviations between the two structures. i.e., residue s 32-40 and 61-74. We find that the main chain in the 32-40 segment de viates significantly from residue 32 to residue 36 and the side chain phenolic rings of residue 32 differ greatly between the two structures . The 61-74 region is the least-well defined region in the whole prote in crystallographically having, by far, the highest temperature factor (B-factor). The backbone and side chain conformations in the 61-74 se gment differ markedly, the overall RMS deviation being 3.1 angstrom fo r the backbone and 5.7 angstrom for all atoms. Both of these regions h ave been found in x-ray crystallographic studies of p21-GDP and p21-GT P complexes to undergo significant changes in conformation upon the bi nding of GTP in place of GDP to the protein. We have further compared our computed structure of the p21 protein with the x-ray crystal struc ture with regard to the conformations of individual segments, in parti cular, structurally conserved sequences (SCR), i.e., those sequences t hat have structural and sequence homology to corresponding sequences i n the related G-protein. bacterial elongation factor Tu (EF-Tu), and v ariable loop regions. Besides finding close agreement in backbone and side chain conformations in these segments, except for the two regions noted above, we find that there is a good correlation between the RMS deviation of a given segment and the average B-factor for that segmen t. This result suggests that, besides the differences in conformation in residues 32-40 and 61-74 caused by the presence of different ligand s bound to the protein, differences in structures between computed and x-ray structures may be caused by thermal fluctuations of segments in the x-ray structure. Overall, our method for computing the structure of the p21 protein from its alpha-carbons appears to be reliable and o f general use. A basic observation that emerges from this comparative study is that the SCRs of the protein appear to determine the low ener gy conformations. of variable loop regions. including the side chain c onformations in these regions. This observation is consistent with the hypothesis that folding proteins contain nucleation sequences that gu ide the folding process by adopting their native-like structures and c ompel adjacent sequences to adopt compatible conformations.