Replication of HIV-1 requires the covalent integration of the viral cDNA into the host chromosomal DNA directed by the virus-encoded integrase protein. Here we explore the importance of a protein surface loop near the integrase active site using protein engineering and X-ray crystallography. We have redetermined the structure of the integrase catalytic domain (residues 50-212) using an independent phase set at 1.7 Å resolution. The structure extends helix R4 on its N-terminal side (residues 149-154), thus defining the position of the three conserved active site residues. Evident in this and in previous structures is a conformationally flexible loop composed of residues 141-148. To probe the role of flexibility in this loop, we replaced Gly 140 and Gly 149, residues that appear to act as conformational hinges, with Ala residues. X-ray structures of the catalytic domain mutants G149A and G140A/G149A show further rigidity of R4 and the adjoining loop. Activity assays in vitro revealed that these mutants are impaired in catalysis. The DNA binding affinity, however, is minimally affected by these mutants as assayed by UV cross-linking. We propose that the conformational flexibility of this active site loop is important for a postbinding catalytic step.Integration of the viral cDNA into a host chromosome is required for viral replication. Prior to integration, viral cDNA is cleaved on each strand near the 3′ end by integrase (terminal cleavage), probably to remove nontemplated extra bases occasionally added by reverse transcriptase (1, 2). Integrase then catalyzes the attachment of the recessed 3′ ends to the target DNA (strand transfer) (3-5). In vitro, purified integrase protein can carry out the terminal cleavage (6, 7) and strand transfer reactions (8-11). Purified integrase can also carry out an apparent reversal of strand transfer, termed "disintegration" (12).The structure of full-length integrase has not yet been identified. However, the structure of each of its three domains has been determined in isolation. The amino-terminal domain (amino acids 1-50) is composed of three R-helices with an embedded zinc binding site (13,14). The central domain [amino acids 50-212 (50-212 1 )] is composed of mixed R-helix and -sheet (15), forming a compact fold seen previously in several polynucleotidyl phosphotransferases (16)(17)(18)(19). This domain by itself can carry out covalent chemistry on permissive disintegration substrates, an indication of its role in catalysis (20,21). A conserved amino acid sequence motif, D,D-35-E, is found in the central domain of integrases and some related prokaryotic transposases. The carboxyl-terminal domain (amino acids 213-288) is composed of a five-stranded -barrel resembling an SH3 domain, and contributes to DNA binding (22,23). All three domains form dimers independently, and the full-length integrase is likely to act as a higher-order multimer (reviewed in ref 24).We have used X-ray crystallography and protein engineering to investigate the active site of HIV-1 integrase....