No abstract

All retroviral nucleocapsid (NC) proteins contain one or two copies of an invariant 3Cys‐1His array (CCHC = C‐X2‐C‐X4‐H‐X4‐C; C = Cys, H = His, X = variable amino acid) that are essential for RNA genome packaging and infectivity and have been proposed to function as zinc‐binding domains. Although the arrays are capable of binding zinc in vitro, the physiological relevance of zinc coordination has not been firmly established. We have obtained zinc‐edge extended X‐ray absorption fine structure (EXAFS) spectra for intact retroviruses in order to determine if virus‐bound zinc, which is present in quantities nearly stoichiometric with the CCHC arrays (Bess, J.W., Jr., Powell, P.J., Issaq, H.J., Schumack, L.J., Grimes, M.K., Henderson, L.E., & Arthur, L.O., 1992, J. Virol. 66, 840–847), exists in a unique coordination environment. The viral EXAFS spectra obtained are remarkably similar to the spectrum of a model CCHC zinc finger peptide with known 3Cys‐1His zinc coordination structure. This finding, combined with other biochemical results, indicates that the majority of the viral zinc is coordinated to the NC CCHC arrays in mature retroviruses. Based on these findings, we have extended our NMR studies of the HIV‐1 NC protein and have determined its three‐dimensional solution‐state structure. The CCHC arrays of HIV‐1 NC exist as independently folded, noninteracting domains on a flexible polypeptide chain, with conservatively substituted aromatic residues forming hydrophobic patches on the zinc finger surfaces. These residues are essential for RNA genome recognition, and fluorescence measurements indicate that at least one residue (Trp37) participates directly in binding to nucleic acids in vitro. The NC is only the third HIV‐1 protein to be structurally characterized, and the combined EXAFS, structural, and nucleic acid‐binding results provide a basis for the rational design of new NC‐targeted antiviral agents and vaccines for the control of AIDS.

A new method is described for determining molecular structures from NMR data. The approach utilizes 2D NOESY back-calculations to generate simulated spectra for structures obtained from distance geometry (DG) computations. Comparison of experimental and back-calculated spectra, including analysis of cross-peak buildup and auto-peak decay with increasing mixing time, provides a quantitative measure of the consistence between the experimental data and generated structures and allows for use of tighter interproton distance constraints. For the first time, the "goodness" of the generated structures is evaluated on the basis of their consistence with the actual experimental data rather than on the basis of consistence with other generated structures. This method is applied to the structure determination of an 18-residue peptide with an amino acid sequence comprising the first zinc fingerlike domain from the gag protein p55 of HIV. This is the first structure determination to atomic resolution for a retroviral zinc fingerlike complex. The peptide [Zn(p55F1)] exhibits a novel folding pattern that includes type I and type II NH-S tight turns and is stabilized both by coordination of the three Cys and one His residues to zinc and by extensive internal hydrogen bonding. The backbone folding is significantly different from that of a "classical" DNA-binding zinc finger. Residues C(1)-F(2)-N(3)-C(4)-G(5)-K(6) fold in a manner virtually identical with the folding observed by X-ray crystallography for related residues in the iron domain of rubredoxin; superposition of all main-chain and Cys side-chain atoms of residues C(1)-K(6) of Zn(p55F1) onto residues C(6)-Y(11) and C(39)-V(44) of rubredoxin gives RMSDs of 0.46 and 0.35 A, respectively. The side chains of conservatively substituted Phe and Ile residues implicated in genomic RNA recognition form a hydrophobic patch on the peptide surface.

The early NMR research on nucleic acids was of a qualitative nature and was restricted to partial characterization of short oligonucleotides in aqueous solution. Major advances in magnet design, spectrometer electronics, pulse techniques, data analysis and computational capabilities coupled with the availability of pure and abundant supply of long oligonucleotides have extended these studies towards the determination of the 3-D structure of nucleic acids in solution.

The three-dimensional structure of d(CGCGTTTTCGCG) in solution has been determined from proton NMR data by using distance geometry methods. The rate of dipolar cross-relaxation between protons close together in space is used to calculate distances between proton pairs within 5 A of each other; these distances are used as input to a distance geometry algorithm that embeds this distance matrix in three-dimensional space. The resulting refined structures that best agree with the input distances are all very similar to each other and show that the DNA sequence forms a hairpin in solution; the bases of the loop region are stacked, and the stem region forms a right-handed helix. The advantages and limitations of the technique, as well as the computer requirements of the algorithm, are discussed.