Deficient uracil base excision repair leads to persistent dUMP in HIV proviruses during infection of monocytes and macrophages

Replication-competent HIV-1 in human alveolar macrophages and monocytes despite nucleotide pools with elevated dUTP

BACKGROUND

Although CD4⁺ memory T cells are considered the primary latent reservoir for HIV-1, replication competent HIV has been detected in tissue macrophages in both animal and human studies. During in vitro HIV infection, the depleted nucleotide pool and high dUTP levels in monocyte derived macrophages (MDM) leads to proviruses with high levels of dUMP, which has been implicated in viral restriction or reduced transcription depending on the uracil base excision repair (UBER) competence of the macrophage. Incorporated dUMP has also been detected in viral DNA from circulating monocytes (MC) and alveolar macrophages (AM) of HIV infected patients on antiretroviral therapy (ART), establishing the biological relevance of this phenotype but not the replicative capacity of dUMP-containing proviruses.

RESULTS

As compared to in vitro differentiated MDM, AM from normal donors had sixfold lower levels of dTTP and a sixfold increased dUTP/dTTP, indicating a highly restrictive dNTP pool for reverse transcription. Expression of uracil DNA glycosylase (UNG) was eightfold lower in AM compared to the already low levels in MDM. Accordingly, ~ 80% of HIV proviruses contained dUMP, which persisted for at least 14-days due to low UNG excision activity. Unlike MDM, AM expression levels of UNG and SAM and HD domain containing deoxynucleoside triphosphate triphosphohydrolase 1 (SAMHD1) increased over 14 days post-HIV infection, while dUTP nucleotidohydrolase (DUT) expression decreased. These AM-specific effects suggest a restriction response centered on excising uracil from viral DNA copies and increasing relative dUTP levels. Despite the restrictive nucleotide pools, we detected rare replication competent HIV in AM, peripheral MC, and CD4⁺ T cells from ART-treated donors.

CONCLUSIONS

These findings indicate that the potential integration block of incorporated dUMP is not realized during in vivo infection of AM and MC due to the near absence of UBER activity. In addition, the increased expression of UNG and SAMHD1 in AM post-infection is too slow to prevent integration. Accordingly, dUMP persists in integrated viruses, which based on in vitro studies, can lead to transcriptional silencing. This possible silencing outcome of persistent dUMP could promote viral latency until the repressive effects of viral dUMP are reversed.

Repair of APOBEC3G-Mutated Retroviral DNA In Vivo Is Facilitated by the Host Enzyme Uracil DNA Glycosylase 2

Apolipoprotein B mRNA editing enzyme catalytic subunit 3 (APOBEC3) proteins are critical for the control of infection by retroviruses. These proteins deaminate cytidines in negative-strand DNA during reverse transcription, leading to G-to-A changes in coding strands. Uracil DNA glycosylase (UNG) is a host enzyme that excises uracils in genomic DNA, which the base excision repair machinery then repairs. Whether UNG removes uracils found in retroviral DNA after APOBEC3-mediated mutation is not clear, and whether this occurs in vivo has not been demonstrated. To determine if UNG plays a role in the repair of retroviral DNA, we used APOBEC3G (A3G) transgenic mice which we showed previously had extensive deamination of murine leukemia virus (MLV) proviruses. The A3G transgene was crossed onto an Ung and mouse Apobec3 knockout background (UNG-/-APO-/-), and the mice were infected with MLV. We found that virus infection levels were decreased in A3G UNG-/-APO-/- compared with A3G APO-/- mice. Deep sequencing of the proviruses showed that there were significantly higher levels of G-to-A mutations in proviral DNA from A3G transgenic UNG-/-APO-/- than A3G transgenic APO-/- mice, suggesting that UNG plays a role in the repair of uracil-containing proviruses. In in vitro studies, we found that cytoplasmic viral DNA deaminated by APOBEC3G was uracilated. In the absence of UNG, the uracil-containing proviruses integrated at higher levels into the genome than those made in the presence of UNG. Thus, UNG also functions in the nucleus prior to integration by nicking uracil-containing viral DNA, thereby blocking integration. These data show that UNG plays a critical role in the repair of the damage inflicted by APOBEC3 deamination of reverse-transcribed DNA. IMPORTANCE While APOBEC3-mediated mutation of retroviruses is well-established, what role the host base excision repair enzymes play in correcting these mutations is not clear. This question is especially difficult to address in vivo. Here, we use a transgenic mouse developed by our lab that expresses human APOBEC3G and also lacks the endogenous uracil DNA glycosylase (Ung) gene and show that UNG removes uracils introduced by this cytidine deaminase in MLV reverse transcripts, thereby reducing G-to-A mutations in proviruses. Furthermore, our data suggest that UNG removes uracils at two stages in infection-first, in unintegrated nuclear viral reverse-transcribed DNA, resulting in its degradation; and second, in integrated proviruses, resulting in their repair. These data suggest that retroviruses damaged by host cytidine deaminases take advantage of the host DNA repair system to overcome this damage.