MOLECULAR BIOLOGY OF RETROVIRUSES

Our laboratory studies molecular aspects of retroviral replication. In recent years, we have focused on elucidating the mechanism of retroviral DNA integration, a process of both practical and scientific importance. The integration of viral DNA into the host cell genome is essential for replication of the retroviruses and, therefore, an important target for the development of antiviral drugs to treat AIDS. Because of the ability to integrate their genetic material, retroviruses have potential utility as vectors in gene therapy. The integration reaction is also a valuable model for other types of DNA rearrangements and transpositions that normally take place in eukaryotic cells. Finally, because host functions are co-opted to facilitate specific steps in the process, study of retroviral integration provides unique opportunities to uncover critical aspects of cellular biology.

Retroviruses are distinct among animal viruses in their possession of a specific recombinase, the integrase, which splices the ends of viral DNA into host cell genomic DNA. We are able to investigate the chemistry of this reaction because it can be reconstructed in the test tube with purified integrase and model DNA substrates. The recent solution of the three dimensional structure of isolated domains of integrase proteins has also provided significant insight into their function. However, many fundamental questions remain concerning protein organization and mechanistic features of the reaction. To address such questions, we study integrase proteins of the avian sarcoma and leukosis virus (ASV, ALV) and the human immunodeficiency virus type 1 (HIV-1). A second, equally active area of research in our laboratory centers on integration-related events that take place inside of infected cells. In such cells, integrase functions as a component of a large nucleoprotein complex that is formed in the cytoplasm shortly after infection. This complex includes viral DNA, other viral proteins, and probably host proteins as well. We are especially interested in understanding how this complex gains access to host DNA in the nucleus and which viral and host proteins participate in the process.

STRUCTURE AND MECHANISM OF THE INTEGRASE CATALYTIC CORE DOMAIN. KATZ, GRAVUER, MERKEL, DI CANDELERO, in collaboration with WLODAWER,^d LUBKOWSKI,^d ALEXANDRATOS,^d YANG,^d POMMIER,^e HARRISON^f

Three highly conserved carboxylate residues (the D,D(35)E constellation) comprise the active site of retroviral integrases (Figure 1). Our collaborators in the Wlodawer laboratory have solved the crystal structure of the central, catalytic domain of ASV integrase (amino acids 50-207). These studies revealed that the three conserved carboxylate residues are adjacent in the structure and are able to coordinate a single Mg²⁺ or Mn²⁺ ion (the required cofactors), or two Zn²⁺, Ca²⁺, or Cd²⁺ ions. To investigate the role of the carboxylate residues in metal binding and enzymatic activity, we have introduced several substitutions into the active site of the catalytic domain using site-directed mutagenesis. As Asp-64 is involved in binding both metal ions in the Zn²⁺, Ca²⁺, and Cd²⁺-complexed structures, it should be quite critical to activity. As expected, we found that substitution of Asp-64 with Asn (D64N) results in loss of catalytic activity of the isolated core. Our collaborators then solved the crystal structure of a catalytic domain that included the D64N substitution. They observed that the Asn-64 side chain was rotated in the molecule and not in position to bind metal. A similar configuration of the native Asp-64 side chain was observed in crystals of the wild-type catalytic domain formed at low pH. Thus, it appears that the mechanism of inactivation by both low pH and Asn-64 substitution involves a shift in the side chain at this position that is incompatible with metal binding (1). This was the first description of the structure of an integrase-inactivating substitution. This year, we also reported the first crystal structure of an integrase core domain (ASV) complexed with an inhibitor molecule (2). Analysis of this complex suggested that repositioning of a flexible loop near the active site contributes to inactivation by the inhibitor. Future work will examine the crystal structure of integrase proteins containing other active site substitutions. We also continue to test compounds from collaborators at the National Cancer Institute and Thomas Jefferson University for inhibition of both HIV-1 and ASV integrases. The most potent of these dually active compounds will then be used for co-crystallization studies.

Although the structure of each isolated domain of the HIV-1 integrase has been determined, it is not yet known how these domains are positioned relative to each other in the full-length protein or in a protein-DNA complex. Thus, we are continuing efforts to determine the structure of full-length and various multi-domain fragments of both ASV and HIV-1 integrase alone or in complex with a variety of DNA substrates and inhibitors. We have recently identified three monoclonal antibodies (Nos. 17, 4, and 33) from our panel directed against HIV1 integrase, that bind each of the three major domains and are capable of solubilizing the full-length protein at low salt concentrations. Fab fragments isolated from these monoclonal antibodies are similarly active, and may facilitate crystallization of full-length integrase for structural studies. We are now mapping the binding sites of the Fab fragments using limited proteolysis-mass spectroscopy analysis. As each of these antibodies inhibits integrase activity, this information may suggest new strategies for the development of inhibitors to treat AIDS.

The reaction catalyzed by integrase comprises two distinct steps (Figure 2): site-specific cleavage of two nucleotides from both 3' ends of viral DNA (processing), and joining of the recessed viral end to random sites in target DNA (joining). Real-time kinetic studies have allowed us to address the following two fundamental questions: 1) How can integrase distinguish between its viral and target DNA substrates? and 2) What is the role of metal ions in these interactions? In these studies, DNA fragments were immobilized on a biosensor chip, and the binding and dissociation of HIV-1 integrase was detected by surface plasmon resonance.

The results showed that in the absence of metal cofactors HIV-1 integrase forms equally tight complexes with duplex oligonucleotides that represent the viral DNA ends and those that represent a target DNA substrate. The stability of these complexes appears to be highly dependent on structural features of the integrase protein. Divalent metal cofactors were found to exert two distinct effects on the integraseDNA interaction. Mn²⁺ inhibits integrase binding to a model target DNA with the apparent dissociation constant (Kd) increasing approximately 3-fold in the presence of this cation. On the other hand, Mn²⁺ (or Mg²⁺) stimulates the binding of integrase to a model viral DNA end decreasing the apparent Kd of this complex approximately 6-fold. Metal-mediated stimulation of the binding of integrase to the viral DNA is dependent on DNA sequence and primary structure. Thus, our results indicate that preferential recognition of viral DNA by HIV1 integrase is achieved only in the presence of the metal cofactor, and that it requires a wild-type viral DNA end. Our kinetic analyses have also suggested that multiple forms of the HIV-1 integrase protein can interact with DNA. Further investigation will be directed at determining the composition of DNA-binding forms of the protein and investigating the role of each of the three domains in the assembly of the complexes.

ANALYSIS OF THE STRUCTURE OF VIRAL DNA BOUND TO INTEGRASE. KATZ, DI CANDELERO

Interaction between ASV integrase and model viral DNA termini has been investigated with the thymidine-specific chemical probe, potassium permanganate (KMnO₄), which can be used to detect protein-induced DNA distortions such as base unpairing and/or unstacking. We found that in the presence of ASV integrase accessibility of this probe to the three terminal base pairs in DNA was increased, indicating significant distortion of the DNA ends. A positive correlation was observed between the extent of integrase-mediated distortion and processing activity. For example, introduction of DNA mispairs near the termini increased integrase activity as well as probe accessibility. Conversely, mutations that stabilized the viral DNA ends resulted in reduced integrase activity and probe accessibility. From these and other results, we conclude that the integrase mechanism involves unpairing and distortion of viral DNA termini. Because the DNA sequence requirements for this activity are relaxed, it is possible that distortion also occurs during sequence-independent binding of target DNA. The DNA-distorting activity was also observed with the isolated catalytic core of ASV integrase, indicating that the critical DNA binding determinants reside in this domain. We are currently attempting to identify the specific integrase residues responsible for the activity.

We have shown previously that addition of cellular DNA-binding proteins of the high mobility group (HMG) class are able to stimulate the concerted integration of both ends of a mini-viral ASV DNA substrate into a target plasmid DNA by ASV integrase in vitro. Subsequently a specific HMG protein, HMG-I(Y), was found by others to be a component of pre-integration nucleoprotein complexes isolated from HIV-1-infected cells. Furthermore, HMG-I(Y) was found to be required for efficient integration of HIV DNA into an exogenously supplied target DNA. We have recently reconstituted an HMG-stimulated, in vitro concerted integration reaction with HIV1 integrase; the products exhibit all of the expected hallmarks of viral integration (3). The stimulation by HMG requires the DNA binding domain of this HMG protein. These and other results with the ASV and HIV in vitro systems indicate that the HMG proteins probably function by altering the architecture of the viral DNA substrate rather than through direct interaction with the integrases. Future studies will examine the details of HMG interaction with integration substrates.

We have been investigating the mechanisms by which the ASV pre-integration complex gains accesses to host DNA in the nucleus. We previously identified a nuclear localization signal (NLS) within ASV integrase that might participate in the active transport of this complex from the cytoplasm into the nucleus (Figure 2). The boundaries of the NLS (Figure 1, top) were initially mapped by fusion of integrase sequences to a reporter protein (5). More recently, we have shown that synthetic peptides corresponding to the NLS are also capable of directing nuclear import of a reporter protein in an in vitro nuclear import assay using semi-permeabilized cells. With this assay, we have verified that transport does, indeed, occur through nuclear pores, and that it requires ATP. Moreover, our competition experiments and depletion studies suggest that the ASV integrase NLS utilizes a cellular import pathway that is distinct from the karyopherin alpha/beta pathway, a pathway reported to be responsible for nuclear import of HIV-1 integrase. In addition, import mediated by the ASV integrase NLS appears to be independent of GTP hydrolysis. These results suggest that the ASV integrase NLS utilizes a novel nuclear import pathway and our current efforts are focused on identification of the cellular proteins that mediate this import.

To examine the biological relevance of this nuclear import activity, we introduced NLS-inactivating mutations into a replication-competent ASV DNA clone. These mutations caused a partial or complete block of viral replication in cycling cells (5). Such results are consistent with a role for the integrase NLS in nuclear import of viral DNA. To test this hypothesis directly, we have been examining nuclear import of viral DNA under conditions in which the cell cycle is blocked and, thus, nuclear import requires active transport through the nuclear pore. Preliminary results indicate that significant levels of viral DNA can accumulate in the nucleus in the absence of cell division. NLS mutants can now be analyzed in a similar manner.

The integrase-catalyzed reactions produce a gapped intermediate in which the viral DNA 5'-ends are unjoined (Figure 2, bottom). The mechanism by which the gaps are repaired and the 5'-ends are joined is unknown. It was generally assumed that cellular enzymes catalyze these final repair steps, but it has also been suggested that viral enzymes alone, e.g. reverse transcriptase together with integrase, could perform this function. In mammalian cells, repair of non-homologous, double-stranded DNA ends is mediated by a pathway that includes the DNA-dependent protein kinase (DNA-PK), which is a complex comprised of a large catalytic subunit (DNA-PK_CS) and a DNA-binding heterodimer (Ku70/Ku86). DNA-PK_CS is mutated in the severe combined immunodeficient (scid) mouse. This mutation inactivates the DNA-PK pathway and causes scid mouse cells to be defective in DNA repair as well as immunoglobulin gene rearrangement.

To determine if the DNA-PK pathway is required for retroviral integration, DNA-PK_CS-deficient scid cell lines, derived some years ago by our Fox Chase colleagues Dr. Bosma and Gail Bosma, were infected with a recombinant retroviral vector. The results were striking--infection of these scid cell lines induced cell death, but no loss of viability was seen with a matched, control cell line. If these scid cells were infected with an integrase deficient virus, no killing was observed, suggesting that abortive integration is the trigger for cell death. The observed death was due to apoptosis, as determined by assays for specific markers of this process in the infected scid cells. Moreover, we observed ca. 10-fold reduction in integration with primary scid fibroblasts when compared to normal primary fibroblasts. To determine whether cell death can be induced by different retroviruses, scid cell lines were infected with a pseudotyped HIV1-based vector and with Moloney murine leukemia virus. Again, we observed cell killing in scid, but not normal cells, coupled with a reduced integration in scid cell lines. Finally, we found that infection of HeLa cells with a retroviral vector induced a 2- to 3-fold increase in DNA-PK activity in infected cells. Taken together, our data suggest that retroviral integration is completed with the help of the host cell's DNA-PK repair pathway, and that retroviral enzymes alone do not suffice (4).

The finding that a cellular DNA repair protein is required for HIV-1 integration may have practical ramifications. Cellular repair protein(s) may represent a new set of targets for inhibition of HIV replication. Additionally, virus-mediated killing of scid cells can provide a relatively simple assay for inhibitors that block early steps in retrovirus integration, without toxic side effects. Elucidation of the molecular mechanism underlying the DNA-PK requirement in retroviral integration will be a major goal of future experiments. We also plan to examine if other cellular DNA repair proteins are required for retroviral integration.

1.   LUBKOWSKI, J., YANG, F., ALEXANDRATOS, J., MERKEL, G., KATZ, R.A., GRAVUER, K., SKALKA, A.M., WLODAWER, A. Structural basis for inactivating mutations and pHdependent activity of ASV integrase. J. Biol. Chem.273:32685-32689, 1998.

2.   LUBKOWSKI, J., YANG, F., ALEXANDRATOS, J., WLODAWER, A., ZHAO, H., BURKE, T.R., Jr., NEAMATI, N., POMMIER, Y., MERKEL, G., SKALKA, A.M. Structure of the catalytic domain of avian sarcoma virus integrase with a bound HIV1 integrase-targeted inhibitor. Proc. Natl. Acad. Sci. USA 95:4831-4836, 1998.

3.   HINDMARSH, P., RIDKY, T., REEVES, R., ANDRAKE, M., SKALKA, A.M., LEIS, J. HMG protein family members stimulate HIV1 and ASV concerted DNA integration in vitro. J. Virol.73:2994-3003, 1999.

4.   DANIEL, R., KATZ, R.A., SKALKA, A.M. A role for DNA-PK in retroviral DNA integration. Science 284:644-647, 1999.

ASANTE-APPIAH, E., SKALKA, A.M. Structural determinants of metal-induced conformational changes in HIV1 integrase. J. Biol. Chem. 273:35078-35087, 1998.

ASANTE-APPIAH, E., SKALKA, A.M. HIV1 integrase: Structural Organization, Conformational Changes, and Catalysis. In Advances in Virus Research, Academic Press, NY, Vol. 52, 1999.

BOUCK, J., FU, X.D., SKALKA, A.M., KATZ, R.A. Role of the constitutive splicing factors USAF65 and SAP49 in suboptimal RNA splicing of novel retroviral mutants. J. Biol. Chem. 273:15169-15176, 1998.

BOUCK, J., LITWIN, S., SKALKA, A.M., KATZ, R.A. In vivo selection for intronic splicing signals from a randomized pool. Nucl. Acids Res. 26:4516-4523, 1998.

KATZ, R.A.,GRAVUER, K., SKALKA, A.M. A preferred target DNA structure for retroviral integrase in vitro. J. Biol. Chem. 273:24190-24195, 1998.

SAUTER, M., ANDRAKE, M.D., KUKOLJ, G., SKALKA, A.M. The ASV integrase NLS uses a GTPase-independent pathway for nuclear import. J. Cell Science (in press).

SKALKA, A.M. (Editor) Retroviral DNA integration. In Advances in Virus Research, Vol. 52. Academic Press, NY (in press).

YI, J., ASANTE-APPIAH, E., SKALKA, A.M. Divalent cations stimulate preferential recognition of a viral DNA end by HIV1 integrase. Biochemistry (in press).

5.   KUKOLJ, G., KATZ, R.A., SKALKA, A.M. Characterization of the nuclear localization signal in the avian sarcoma virus integrase. Gene 223:157-163, 1998.

^a   E. Asante-Appiah: Present address--Merck Frosst Center for Therapeutic Research, Dept. of Biochemistry and Molecular Biology, Pointe-Claire-Dorval, Quebec H9R 4P8, Canada

^b   M. Sauter: Present address--Medical College of Wisconsin, Dept. of Microbiology and Molecular Genetics, Milwaukee, WI 53226

^c   J. Bouck: Present address--Baylor College of Medicine, Dept. of Molecular and Human Genetics, Houston, TX 77030

^d   A. Wlodawer, J. Lubkowski, J. Alexandratos, F. Yang: NCIFrederick Cancer Research and Development Center, Frederick, MD 21701

^e   Y. Pommier: National Institutes of Health, Bethesda, MD 20892

^f   R. Harrison: Thomas Jefferson University, Philadelphia, PA 19107

^g   P. Hindmarsh, J. Leis: Case Western Reserve University, Cleveland, OH 44106

^h   R. Reeves: Washington State University, Pullman, WA 99164

Illustrations or unpublished data in these reports should not be used without permission of the author.

Fox Chase Cancer Center

Scientific Report 1998