Repair of oxidative DNA base damage in the host genome influences the HIV integration site sequence preference

Abstract

Host base excision repair (BER) proteins that repair oxidative damage enhance HIV infection. These proteins include the oxidative DNA damage glycosylases 8-oxo-guanine DNA glycosylase (OGG1) and mutY homolog (MYH) as well as DNA polymerase beta (Polβ). While deletion of oxidative BER genes leads to decreased HIV infection and integration efficiency, the mechanism remains unknown. One hypothesis is that BER proteins repair the DNA gapped integration intermediate. An alternative hypothesis considers that the most common oxidative DNA base damages occur on guanines. The subtle consensus sequence preference at HIV integration sites includes multiple G:C base pairs surrounding the points of joining. These observations suggest a role for oxidative BER during integration targeting at the nucleotide level. We examined the hypothesis that BER repairs a gapped integration intermediate by measuring HIV infection efficiency in Polβ null cell lines complemented with active site point mutants of Polβ. A DNA synthesis defective mutant, but not a 5'dRP lyase mutant, rescued HIV infection efficiency to wild type levels; this suggested Polβ DNA synthesis activity is not necessary while 5'dRP lyase activity is required for efficient HIV infection. An alternate hypothesis that BER events in the host genome influence HIV integration site selection was examined by sequencing integration sites in OGG1 and MYH null cells. In the absence of these 8-oxo-guanine specific glycosylases the chromatin elements of HIV integration site selection remain the same as in wild type cells. However, the HIV integration site sequence preference at G:C base pairs is altered at several positions in OGG1 and MYH null cells. Inefficient HIV infection in the absence of oxidative BER proteins does not appear related to repair of the gapped integration intermediate; instead oxidative damage repair may participate in HIV integration site preference at the sequence level.

Figure 1. Retroviral integration.

(A) Viral cDNA is depicted by a thin line and host target DNA is indicated by a thick line. Base pairs in the host target DNA are numbered. The HIV LTR ends are covalently joined to the target DNA 5 base pairs apart. The intervening host DNA denatures yielding an integration intermediate with two 5 base pair gaps. (B) The sequence preference observed at HIV integration sites. The numbering is identical to (A) and the points of joining are indicated by “IN”. Base pairs in green are favored and base pairs in red are disfavored at HIV integration sites.

Figure 2. Polβ polymerase activity is not required for efficient HIV infection.

Murine embryonic fibroblasts with a deletion of the Polβ cDNA (PolB−/−) were complemented with (A) an empty vector (Neo), (B) the wild type cDNA (PolB), (C) a polymerase defective point mutant gene (PolB (pol-)), (D) a lyase deficient mutant cDNA (PolB (lyase-)), or (E) an enzymatically dead mutant (PolB (pol-lyase-)). (F) These cell lines and a matched wild type cell line were infected with an HIV based vector expressing GFP following integration. Cells were analyzed by flow cytometry for GFP expression, an indicator of successful infection efficiency. Results are from three independent experiments of duplicates and are expressed relative to wild type cells. Error bars indicate the standard deviations.

Figure 3. Oxidative damage during HIV infection.

(A) OGG1 and MYH both recognize 8-oxo-dG damage (shown as G_o). OGG1 repairs 8-oxo-dG to G. Replication of 8-oxo-dG results in an 8-oxo-dG:A mispair, which is recognized by MYH. The MYH glycosylase initiates repair of the mispaired A to C yielding an 8-oxo-dG:C base pair. The product of the MYH repair reaction must still be repaired by OGG1. (B) Wild type cells were infected with an HIV based vector expressing GFP. Cells were treated with increasing concentrations of H₂O₂ immediately prior to infection. Viability in the absence (open diamonds) or presence of HIV (closed diamonds) was measured by trypan blue exclusion. Error bars indicate the standard deviations from three independent experiments.

Figure 4. Effects of 8-oxo-dG specific glycosylases or hydrogen peroxide on HIV integration site sequence preference.

Wild type, OGG1 null, MYH null, and wild type cells treated with 10 µM or 30 µM H₂O₂, were infected with an HIV based retroviral vector at 0.8 MOI. 10 µM H₂O₂ is less than the 50% lethal dose for the wild type cells and 30 µM H₂O₂ is greater. After 7 days, genomic DNA was purified. HIV integration sites were subcloned, sequenced, and mapped to the mouse genome. The random frequency of G or C in the mouse genome is 0.205 and A or T is 0.295. The differences in observed base frequencies relative to the random frequencies are shown. Base numbering relative to the HIV points of joining is as in Figure 1. Boxes indicate the known HIV integration base preferences for wild type cells. Green, red, and gray highlights indicate a statistically significant difference of >0.10 from random frequency with a p value of <0.005. Green highlights are positive differences and red highlights are negative previously published palindromic prefered bases. Deletion of the OGG1 gene leads to a loss of HIV sequence preference at positions −2 and −1. Deletion of the MYH gene also leads to loss of the HIV sequence preference at positions −2 and 7. While treatment of wild type cells with 10 µM H₂O₂ did not dramatically alter the sequence preference at integration sites, treatment with 30 µM H₂O₂ led to the disfavor of C at position −3 and disfavor of G at position 8, highlighted in red.

Figure 5. HIV integration near genomic elements.

HIV integration sites were mapped to genomic elements in OGG1 null cells, MYH null cells, untreated wild type cells, and wild type cells treated with 10 µM or 30 µM H₂O₂. (A) HIV has a preference for integration to transcription units. (B) HIV shows no preference for integration to promoters. (C) There is no preference for HIV integration within 5 kb of CpG islands.

Figure 6. G:C content surrounding HIV integration sites.

The percentage of HIV integration sites compared to the percentage of G:C base pairs within a 5 kb window is shown for wild type cells compared to (A) OGG1 null cells, (B) MYH null cells, (C) wild type cells treated with 10 µM H₂O₂, and (D) wild type cells treated with 30 µM H₂O₂. The frequency of G:C base pairs in the mouse genome is 0.41. There is no significant difference between HIV integration sites in untreated wild type cells and OGG1 null (p = 0.96), MYH null (p = 0.99), 10 µM H₂O₂ treated (p = 0.67), or 30 µM H₂O₂ treated cells (p = 0.88).