Integration of the reverse-transcribed viral DNA into the host genome is an essential step in the lifecycle of retroviruses. Retrovirus integrase (IN) catalyzes insertions of both ends of the linear viral DNA into a host chromosome 1. IN from HIV-1 and closely related retroviruses share the three-domain organization, consisting of a catalytic core domain flanked by N- and C-terminal domains essential for the concerted integration reaction. Although structures of the tetrameric IN-DNA complexes have been reported for IN from prototype foamy virus (PFV) featuring an additional DNA-binding domain and longer interdomain linkers 2–5, the architecture of a canonical three-domain IN bound to DNA remained elusive. Here we report a crystal structure of the three-domain IN from Rous sarcoma virus (RSV) in complex with viral and target DNAs. The structure shows an octameric assembly of IN, in which a pair of IN dimers engage viral DNA ends for catalysis while another pair of non-catalytic IN dimers bridge between the two viral DNA molecules and help capture target DNA. The individual domains of the eight IN molecules play varying roles to hold the complex together, making an extensive network of protein-DNA and protein-protein contacts that show both conserved and distinct features compared to those observed for PFV IN. Our work highlights diversity of retrovirus intasome assembly and provides insights into the mechanisms of integration by HIV-1 and related retroviruses.

The human immunodeficiency virus type 1 (HIV-1) life cycle involves the reverse transcription of the viral RNA genome into cDNA and subsequent integration of the viral DNA by integrase (IN) into the host chromosomes. Although the amino acid sequences of INs differ significantly between viruses, INs share three conserved structural domains and their associated functions (13). The N-terminal domain (ϳ50 residues) contains a zinc-binding region (7), promotes multimerization (46), and is necessary for 3Ј-OH processing and strand transfer. The catalytic core domain (CCD) (ϳ162 residues) contains the highly conserved acidic D, D-35-E motif (27) that is involved in coordinating Mg 2ϩ for 3Ј-OH processing and strand transfer activities (4, 13). The catalytic core is also involved in target binding for strand transfer (3,20,26,40). The C-terminal domain (ϳ35 residues) binds to the viral DNA ϳ6 to 9 bp from the long terminal repeat (LTR) end (15); the C-terminus and CCD are also involved in IN multimerization (1, 24, 29a).IN, along with the reverse transcriptase and protease, is an antiretroviral target (2,35,38). Highly active antiretroviral therapy, consisting of various combinations of reverse transcriptase and protease inhibitors, has significantly decreased HIV-1 replication in humans. The emergence of multidrugresistant HIV-1 mutants and undesirable side effects associated with certain drug combinations necessitates continuing efforts to develop novel and effective combinational therapies. The addition of inhibitors of HIV-1 IN function would enhance highly active antiretroviral therapy. Raltegravir (MK-0518), an analog of the strand transfer inhibitor L-870,810 used in this report, is in phase III human clinical trials (18,33).Oligonucleotide-based assays in vitro have identified a large number of compounds that inhibit HIV-1 IN activities in vitro (25), the majority of which are ineffective at preventing HIV-1 replication in cell culture. The "strand transfer inhibitors" were identified as being effective against recombinant IN, suppressed HIV-1 replication in cell culture and in vivo, and were so named because of their selectivity in both cases towards inhibiting strand transfer over 16,21,22,38). The first generation of strand transfer inhibitors possessed a 1,3-diketo acid (DKA) pharmacophore, which served as a template in the development of the naphthyridine carboxamide inhibitors. They are structurally analogous to and function identically to DKA inhibitors but exhibit improved metabolic and pharmacokinetic properties and are represented here by compounds L-870,810 and L-870,812 (21, 23). DKA-mediated inhibition is accomplished by the contact of the DKA moiety with the divalent metal ion in the CCD of IN (19,38), and efficient inhibitor binding occurs only with IN bound to the viral DNA substrate (14,22,38

Summary A macromolecular nucleoprotein complex in retrovirus-infected cells, termed the preintegration complex, is responsible for the concerted integration of linear viral DNA genome into host chromosomes. Isolation of sufficient quantities of the cytoplasmic preintegration complexes for biochemical and biophysical analyses is difficult. We investigated the architecture of human immunodeficiency virus type-1 (HIV-1) nucleoprotein complexes involved in the concerted integration pathway in vitro. HIV-1 integrase (IN) non-covalently juxtaposes two viral DNA termini forming the synaptic complex, a transient intermediate in the integration pathway and shares properties associated with the preintegration complex. IN slowly processes two nucleotides from the 3′ OH ends and performs the concerted insertion of two viral DNA ends into target DNA. IN remains associated with the concerted integration product, termed the strand transfer complex. The synaptic complex and strand transfer complex can be isolated on native agarose gels for biochemical and biophysical analyses. In this report, in-gel fluorescence resonance energy transfer measurements demonstrated that the energy transfer efficiencies between the juxtaposed Cy3 and Cy5 5′-end labeled viral DNA ends in the synaptic complex (0.68 ± 0.09) was significantly different than observed in the strand transfer complex (0.07 ± 0.02). The calculated distances were 46 ± 3 Å and 83 ± 5 Å, respectively. DNaseI footprint analysis of the complexes revealed that IN protects U5 and U3 DNA sequences up to ~32 bp from the end, suggesting two IN dimers were bound per terminus. Enhanced DNaseI cleavages were observed at nucleotide positions 6 and 9 from the terminus on U3 but not on U5 suggesting independent assembly events. Protein-protein cross-linking of IN within these complexes revealed the presence of dimers, tetramers, and a larger-size multimer (>120 Kd). Our results suggest a new model where two IN dimers individually assemble on U3 and U5 ends prior to the non-covalent juxtaposition of two viral DNA ends, producing the synaptic complex.

An autosomal dominant congenital cataract in human is associated with mutation of Arg-116 to Cys (R116C) in alpha A-crystallin. To investigate the molecular basis of cataract formation, rat alpha A-crystallin cDNA was cloned into pET-23d(+), and the site-directed mutants S142C (similar to wild-type human alpha A) and R116C/S142C or R116C (similar to human R116C variant) were generated. These were expressed in E. coli and the recombinant alpha A-crystallins purified by Sephacryl size-exclusion chromatography. The chaperone-like function of mutant R116C determined at 37 degrees C with insulin and alcohol dehydrogenase as target proteins was about 40% lower than those of wild-type and mutant S142C. Based on size-exclusion chromatography data, the oligomeric size of the R116C mutant was about 2000 kDa at 25 degrees C, 1400 kDa at 37 degrees C, and 900 kDa at 45 degrees C. In comparison, alpha A-wild-type and alpha A-S142C ranged from 477 to 581 kDa. Heat stability studies corroborated the effect of temperature on the dynamic quaternary structure of the R116C mutant. Circular dichroism spectra showed secondary and tertiary structural changes, and ANS fluorescence spectra showed loss of surface hydrophobicity in the R116C mutant. These findings suggest that the molecular basis for the congenital cataract with the alpha A-R116C mutation is due to the generation of a highly oligomerized alpha A-crystallin having a modified structure and decreased chaperone-like function.

Background: Structure of the three-domain Rous sarcoma virus integrase with viral DNA is lacking. Results: Soluble (Ͼ1.5 mg/ml) and stable synaptic complex using Rous sarcoma virus integrase, DNA, and HIV-1 strand transfer inhibitors was produced. Conclusion: Two integrase dimers assemble onto viral DNA to produce a synaptic complex. Significance: This is the first report producing a high concentration of soluble and kinetically stabilized RSV synaptic complex.