Retroviral DNA integration into the genome of the host cell is essential for retroviral replication. Three distinct steps have been delineated in this reaction (Fig. ). In the first step, processing, nucleotides (usually two) are removed from the 3′ ends of the viral DNA. In the second step, joining, these newly created 3′ ends are joined to staggered phosphates in complementary strands of host cell DNA. In the resulting integration intermediate, 5′ ends of the viral DNA are separated by single-strand gaps from the host DNA. These first two steps are catalyzed by the retroviral enzyme integrase (IN) (13
). In the third and last step of integration, the single-strand DNA gaps are repaired, creating a stably integrated provirus flanked by short direct repeats. The length of the repeats is characteristic of the species of retrovirus, i.e., 4 bp for Moloney murine leukemia virus, 5 bp for human immunodeficiency virus, and 6 bp for avian sarcoma virus (ASV) (3
), and is a property determined by each specific IN protein.
Three steps of retroviral DNA integration.
We have recently proposed that the cellular nonhomologous end-joining (NHEJ) repair complex plays an important role in this last step of retroviral DNA integration (6
). The NHEJ proteins are known to be required for repair of double-strand breaks induced by ionizing radiation and certain DNA-damaging drugs and for V(D)J recombination during the generation of immunoglobulin producing cells (20
). Mutations in any of the genes that specify the protein components of the NHEJ complex cause defects in double-strand break repair. They include XRCC4, which encodes a 38-kDa protein that associates with DNA ligase IV and stimulates its ligase activity; XRCC5 and XRCC6, which encode the 86- and 70-kDa DNA-binding subunits of Ku protein, respectively; and XRCC7, which specifies the 460-kDa catalytic subunits of the phosphatidylinositol 3-kinase-related, DNA-dependent serine/threonine protein kinase (PK) DNA-PKCS
). NHEJ-deficient cells are hypersensitive to DNA damage (5
), and we have shown that retroviral infection induces apoptosis of DNA-PK-deficient scid
). As the scid
cell death was induced by an integration-competent virus but not a virus carrying an inactive IN, we proposed that retroviral DNA integration is detected by the cell as DNA damage which, in the absence of NHEJ proteins, induces programmed cell death (6
). One plausible hypothesis is that the integration intermediate produced by IN (Fig. ) is the relevant damage signal and that NHEJ proteins normally participate in its repair (8
). If this were the case, then the first two, IN-mediated, steps of integration should proceed normally in NHEJ-deficient cells.
Despite the proposed requirement for NHEJ in retroviral DNA integration, some NHEJ-deficient cells (10 to 20% of normal) can be stably transduced (6
). As these transductants were selected only for expression of reporter genes (the neomycin resistance-encoding gene or lacZ
), it seemed possible that the incorporation of reporter sequences in such cells could have taken place via a reaction that caused a loss of viral (or host) DNA sequences. For example, one possible model is that NHEJ proteins are needed to recruit the enzymes required for repair of gaps or other discontinuities in the integration intermediate (Fig. ); in the absence of such repair, these lesions may be attacked by a cellular endonuclease, producing double-strand breaks. Such breaks might then be repaired by a less efficient, single-strand annealing pathway (2
), in which an endonuclease exposes 3′ single strands that can be joined at short regions of homology. If this were the case, provirus-host DNA junctions in NHEJ-deficient cells might show loss of viral or host DNA sequences. Alternatively, other cellular proteins may compensate for the NHEJ deficiency and produce normal junctions, albeit with reduced efficiency. To distinguish between these possibilities, we determined the nucleotide sequences of provirus-host DNA junctions from singly infected NHEJ-deficient cells that were selected for expression of the retrovirus-transduced neomycin resistance-encoding gene.
Selection of NHEJ-deficient cells containing a single provirus.
In the case of adherent cell lines (CHO-K1, XR-1, xrs-6, and ST.SCID), cells were plated at 105
per 60-mm-diameter dish and infected for 2 h at a multiplicity of infection of 0.01 infectious particles/cell with an ASV vector bearing an amphotropic murine envelope protein and carrying a neomycin resistance-encoding reporter gene (6
). To select transduced cells, G418 (1 mg/ml) was added 24 h after infection. After approximately 7 days, single clones were picked, transferred to fresh plates, and grown to confluence for extraction of genomic DNA. In the case of pre-B cell lines, which grow in suspension, individual infected clones were selected by plating cells in semisolid methylcellulose-containing medium. As expected (6
) the number of stably transduced, NHEJ-deficient cells was only ~10 to 20% of the number with the control, NHEJ-proficient lines infected at the same multiplicity of infection (results not shown).
Cloning and sequencing of the junctions between host and retroviral DNAs.
DNA was isolated from the individual cell clones (17
), and inverse PCR was used to amplify the junctions between host and proviral DNAs (19
). DNA (500 ng) was digested overnight with Alu
I or Sau
3AI (New England Biolabs) restriction enzymes in total volumes of 80 μl. Restriction fragments were purified by phenol-chloroform extraction and circularized by ligation of their ends at a low DNA concentration by using T4 DNA ligase (Gibco BRL) in a final volume of 50 μl. All PCRs were performed with Biolase DNA polymerase (Denville Scientific Inc., Metuchen, N.J.) and the primers listed in Table , with different pairs for amplification of left and right junctions. Primers 2 and 3 were used for the first PCR, and primers 1 and 4 were used for the second PCR amplification of fragments that included the left junction sequences; primers 6 and 7 were used for the first PCR, and primers 5 and 8 were used for the second PCR amplification of fragments that included the right junction sequences. Twenty-five picomoles of each primer and 50 ng of ligated chromosomal DNA (5 μl of ligase reaction mixture) template were used for each PCR in a total volume of 30 μl. Amplification conditions were 94°C for 20 s, 58°C for 20 s, and 72°C for 30 s, with 30 rounds of amplification in each of the two PCRs. Five-microliter samples of the product obtained in the first step of amplification were used as a template in the second step. Following the amplification, 15-μl reaction mixture aliquots of each PCR were subjected to electrophoresis in a 2% agarose gel, and DNA bands were cut out of the gel and purified with a QIAGEN gel extraction kit. Each DNA sample was eluted with 30 μl of buffer (QIAGEN gel extraction kit) and 5-μl aliquots were used for PCR with the second set of primers. Amplified DNA bands were obtained with all of the singly transduced cell clones selected, indicating that none contained long terminal repeat deletions encompassing the PCR primer sequences. The resultant products were purified using a QIAGEN purification kit and sequenced with an automated sequencing system and the same primers used for PCR amplification of the junctions. Sets of primers for amplification of the unoccupied integration site were selected from the sequences of host DNA, approximately 60 bp from the end of the provirus sequence. PCR was performed to amplify these sequences using conditions as described above, and the products were purified using the QIAGEN purification kit and sequenced.
Oligonucleotides used in this study
In these studies, single integration sites were analyzed for seven NHEJ-proficient control clones (Table ), nine DNA-PKCS
clones, seven Ku86−
clones, and three XRCC4−
clones (Table ). The results showed that in every clone analyzed the proviral DNA ends terminated with the expected 5′TG
… CA3′ sequences. In addition, all seven control clones and 14 of 19 NHEJ-deficient clones showed a 6-bp duplication of host DNA flanking retrovirus as expected for ASV integration (12
); the rest of the clones were flanked by 5-bp repeats. Comparison of the frequencies of 6-bp junctions among control and NHEJ-deficient cells (Fisher's exact test) failed to show a statistically significant difference between the two groups in this data set (P
= 0.289). However, as the duplication length is generally assumed to be constant in normal infection (3
), this variation in proviruses from NHEJ-deficient cells was unexpected.
Provirus-host DNA junction sequencesa in NHEJ-proficient, control clones
Provirus-host DNA junction sequencesa in NHEJ-deficient clones
Sequencing of the unoccupied integration sites.
It seemed possible that the 5-bp duplications might have resulted from a deletion during the repair step (Fig. ). To test this hypothesis, the original (unoccupied) integration sites for all clones with 5-bp duplications were sequenced. As illustrated for one such clone in Fig. , we found that the occupied and unoccupied integration sites of all of these clones had the same nucleotide sequence. Thus, the shorter duplications did not arise via deletions at the host target sites. As three of the five clones had ambiguous junction sequencing, we cannot exclude the possibility that these arose from a normal 6-bp staggered joining in which one viral DNA end had lost three rather than two nucleotides in the processing reaction. However, this seems unlikely as the expected viral end sequences were observed in all junctions analyzed.
Sequences of the occupied and unoccupied integration sites in the XR-1 clone with a 5-bp flanking repeat (see Table ).
Another possible explanation for the presence of 5-bp repeats in some NHEJ-deficient cells is that the fidelity of the integration reaction is somewhat compromised in these cells. Based on our current understanding of the biochemistry of retroviral DNA integration, the end of the proviral DNA is determined by the processing step, and the length of the host cell duplication by the staggered cut introduced in the host cell DNA during the joining step (3
) (Fig. ). Variations in length of this duplication have been reported in a few cases with murine leukemia proviruses (4
). In addition, two out of five sequenced ASV proviruses were recently reported to be flanked by 5-bp duplications (16
). In the latter case, the authors speculated that this lack of fidelity might be attributed to 3- and 4-bp sequence similarities in host and viral DNA at the integration sites (16
). However, no such sequence similarities are present in the proviruses we analyzed (Table ). We note that variation in the length of flanking duplications appears to be very frequent in in vitro reconstituted experiments with purified IN protein. For example, in three independent studies with ASV IN, the sizes of the duplications at the sites of concerted DNA integration were 4, 5, or 7 bp, in 33 to 50% of the integrants analyzed (1
). Thus, as we suggested earlier (14
) it may be that host proteins, which are missing in the in vitro reactions, can affect the fidelity of the joining reaction. If so, such proteins may also be absent or nonfunctional in NHEJ-deficient cells.
Apart from the small variation in target site duplication, the joining step of retroviral DNA integration appears to be normal in NHEJ-deficient cells. In addition, in all cases that could be interpreted unambiguously viral DNA ends were processed correctly by IN in the NHEJ-deficient transductants. These results are consistent with the proposal that NHEJ proteins participate in the last step of integration, repair of the intermediate produced by IN. Furthermore, the absence of any gross abnormalities at the provirus-host DNA junctions in the NHEJ-deficient cells is inconsistent with a role for the single-strand annealing pathway for double-strand break repair, or any other repair pathway that might cause deletions or nucleotide misincorporations. We have recently obtained evidence that the residual integration in NHEJ-deficient cells depends on the activity of the phosphatidylinositol 3-kinase-related protein kinase ATM (for ataxia-telangiectasia mutated) (7
). ATM is known to signal cell cycle arrest in response to DNA damage and to contribute to DNA repair. Either or both of these activities could affect the efficiency of retrovirus-mediated transduction in NHEJ-deficient cells. Our results indicate that the mechanism which facilitates repair of the integration intermediate in these cells does so without introducing any gross alterations at the viral-host DNA junctions. Further studies will be directed at elucidating the mechanism by which NHEJ or ATM mediates such repair.