A homozygous mutation in the kinase domain of ZAP-70, a T cell receptor-associated protein tyrosine kinase, produced a distinctive form of human severe combined immunodeficiency. Manifestations of this disorder included profound immunodeficiency, absence of peripheral CD8+ T cells, and abundant peripheral CD4+ T cells that were refractory to T cell receptor-mediated activation. These findings demonstrate that ZAP-70 is essential for human T cell function and suggest that CD4+ and CD8+ T cells depend on different intracellular signaling pathways to support their development or survival.

The selective encapsidation of retroviral RNA requires sequences in the Gag protein, as well as a cis-acting RNA packaging signal (psi site) near the 5' end of the genomic transcript. Gag protein of human immunodeficiency virus type 1 (HIV-1) has recently been found to bind specifically to the HIV-1 psi element in vitro. Here we report studies aimed at mapping features within the genetically defined psi locus that are required for binding of HIV-1 Gag or of its processed nucleocapsid derivative. The full-length HIV-1 Gag (p55) and nucleocapsid (p15) sequences were expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli. In a gel shift assay containing excess competitor tRNA, affinity-purified GST-p15 and GST-p55 proteins bound to a 206-nucleotide psi RNA element spanning the major splice donor and gag start codons but did not bind to antisense psi transcripts. Quantitative filter-binding assays revealed that both GST-p55 and GST-p15 bound to this RNA sequence with identical affinities (apparent Kd congruent to 5 x 10(-8) M), indicating that all major determinants of psi binding affinity reside within the nucleocapsid portion of Gag. Chemical and RNase accessibility mapping, coupled with computerized sequence analysis, suggested a model for psi RNA structure comprising four independent stem-loops. Filter-binding studies revealed that RNAs corresponding to three of these hypothetical stem-loops can each function as a independent Gag binding site and that each is bound with approximately fourfold-lower apparent affinity than the full-length psi locus. Interaction of Gag with these regions is likely to play a major role in directing HIV-1 RNA encapsidation in vivo.

Retroviral genomes must dimerize to be fully infectious. Dimerization is directed by a unique RNA hairpin structure with a palindrome in its loop: hairpins of two strands first associate transiently through their loops, and then refold to a more stable, linear duplex. The structure of the initial, kissing-loop dimer from HIV-1, solved using 2D NMR, is bent and metastable, its interface being formed not only by standard basepairing between palindromes, but also by a distinctive pattern of interstrand stacking among bases at the stem-loop junctions. This creates mechanical distortions that partially melt both stems, which may facilitate spontaneous refolding of this RNA complex into linear form.

Retrovirus particles each contain two copies of the viral genome in the form of a noncovalently linked RNA dimer. Earlier studies have mapped a cis-acting region near the 5 end of the human immunodeficiency virus type 1 (HIV-1) genome, termed the locus, which appears essential for initiation of genomic dimerization, as well as for interactions with the HIV-1 Gag protein that are thought to target the RNA into nascent virions. This HIV-1 locus is proposed to be organized in four independent RNA stem-loops; at least three (SL1, SL3, and SL4) contain binding sites for Gag, and one of these (SL1) is implicated in dimer initiation through a kissing-loop mechanism. In this study, we have created HIV-1 proviruses containing mutations that affect in vitro Gag binding, RNA dimerization, or both, and we have characterized the effects of these mutations on viral assembly and infectivity by using a single-step infectious assay. We find that various mutations which eliminate the Gag binding sites in SL1 or SL3 produce marked defects in genomic RNA packaging and viral infectivity. In each case, the reduced genomic content of the mutant virions is associated with an increased content of spliced viral transcripts, suggesting that both SL1 and SL3 contribute to the discrimination between spliced and unspliced RNAs. The structures, but not the specific sequences, of the SL1 and SL3 stems appear critical for RNA packaging. Disruption of the stem or deletion of SL1 also results in abnormal genomic dimerization, as assessed by nondenaturing gel electrophoresis of virion-derived RNA. Virions carrying less extensive mutations in the SL1 loop that are known to prevent in vitro dimerization have impaired infectivity despite normal virion RNA content. This suggests that RNA dimerization is not a prerequisite for genomic packaging but instead serves an independent function in the retroviral infectious cycle.

Sequences from the 5 end of type 1 human immunodeficiency virus RNA dimerize spontaneously in vitro in a reaction thought to mimic the initial step of genomic dimerization in vivo. Dimer initiation has been proposed to occur through a "kissing-loop" interaction involving a specific RNA stem-loop element designated SL1: the RNA strands first interact by base pairing through a six-base GC-rich palindrome in the loop of SL1, whose stems then isomerize to form a longer interstrand duplex. We now report a mutational analysis aimed at defining the features of SL1 RNA sequence and secondary structure required for in vitro dimer formation. Our results confirm that mutations which destroy complementarity in the SL1 loop abolish homodimer formation, but that certain complementary loop mutants can heterodimerize. However, complementarity was not sufficient to ensure dimerization, even between GC-rich loops, implying that specific loop sequences may be needed to maintain a conformation that is competent for initial dimer contact; the central GC pair of the loop palindrome appeared critical in this regard, as did two or three A residues which normally flank the palindrome. Neither the four-base bulge normally found in the SL1 stem nor the specific sequence of the stem itself was essential for the interaction; however, the stem structure was required, because interstrand complementarity alone did not support dimer formation. Electron microscopic analysis indicated that the RNA dimers formed in vitro morphologically resembled those isolated previously from retroviral particles. These results fully support the kissing-loop model and may provide a framework for systematically manipulating genomic dimerization in type 1 human immunodeficiency virus virions.