Antiretroviral APOBEC3 cytidine deaminases alter HIV-1 provirus integration site profiles

Abstract

APOBEC3 (A3) proteins are host-encoded deoxycytidine deaminases that provide an innate immune barrier to retroviral infection, notably against HIV-1. Low levels of deamination are believed to contribute to the genetic evolution of HIV-1, while intense catalytic activity of these proteins can induce catastrophic hypermutation in proviral DNA leading to near-total HIV-1 restriction. So far, little is known about how A3 cytosine deaminases might impact HIV-1 proviral DNA integration sites in human chromosomal DNA. Using a deep sequencing approach, we analyze the influence of catalytic active and inactive APOBEC3F and APOBEC3G on HIV-1 integration site selections. Here we show that DNA editing is detected at the extremities of the long terminal repeat regions of the virus. Both catalytic active and non-catalytic A3 mutants decrease insertions into gene coding sequences and increase integration sites into SINE elements, oncogenes and transcription-silencing non-B DNA features. Our data implicates A3 as a host factor influencing HIV-1 integration site selection and also promotes what appears to be a more latent expression profile.

Fig. 1. A3-mediated restriction of HIV-1 integration.

A Western blot analysis of virus producer cells. HIV-1 pseudotyped virus was produced in 293 T cells by co-transfection of plasmids coding for NL4-3-ΔEnv/ΔVif/eGFP, VSV-G, and either empty pcDNA 3 plasmid (HIV-1 no A3), or each of the A3 expression plasmids. The Control lane is the transfection of the pcDNA in absence of virus. Cell lysates were subjected to SDS-PAGE and Western blot analysis. B CEM-SS cells were infected with the amounts of virus as indicated after normalization to capsid (p24CA) protein, as determined by ELISA. Infection was measured as the percentage of eGFP+ cells by flow cytometry. Data are presented as mean values ± SD. C Integrated provirus in the CEM-SS cells from B was quantified using Alu-based PCR combined with nested qPCR. Data are presented as mean values ± SD. D 293 T cells were transfected for the expression of each of the A3 proteins, HIV-1 (no A3), or pcDNA (cells transfected with pcDNA plasmid). Viral and cell lysates were mixed together and subjected to co-immunoprecipitation using an anti-FLAG antibody and analyzed by western blotting. Data shown are representative of three independent experiments. Source data are provided as a Source Data file.

Fig. 2. A3F and A3G expression alters integration site targeting of genomic features.

A Frequency of integration sites within or at different distance intervals (1–499, 500–4999, 5000–49,999, or >49,999 bp) away from various common genomic features in CEM-SS T cells infected with HIV-1 generated in the presence of A3F-WT, A3F [E251A], A3G-WT, or A3G [E259A], or the absence of A3F or A3G (‘no A3’ control). Inset numbers refers to the percentage of total integration sites falling directly within the feature. The statistical comparison is with respect to the No A3 control. B Heatmaps depicting the fold-enrichment (blue shading) and depletion (red shading) of integration sites at various distance intervals compared to the ‘no A3’ control virus. C Pairwise distance matrix was used to determine the overall similarity between the integration site profiles of CEM-SS cells infected with either the no A3 control virus or A3F-WT, A3F [E251A], A3G-WT, or A3G [E259A] virus. The fold-enrichment and depletion values in each distance bin of each common DNA feature were used in the comparison. The heatmap shows the distance matrix calculated by Euclidean distance as the measurement method. Stronger relationships are indicated by the darker blue color and weaker relationships by darker red color. D, E Percentage of total integration sites located in genes for CEM-SS cells infected with A3F-WT, A3F [E251A], A3G-WT, or A3G [E259A] virus generated from cells expressing increasing concentrations of A3 protein. F, G Percentage of total integration sites located in SINEs for CEM-SS cells infected with A3F-WT, A3F [E251A], A3G-WT, or A3G [E259A] virus generated from cells expressing increasing concentrations of A3 protein. Shaded triangles represent the different distance bins with the darkest shading representing distances further away from the feature. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; Fisher’s exact test, two-sided. Source data are provided as a Source Data file.

Fig. 3. A3F and A3G alters integration site targeting of non-B DNA.

A Frequency of integration sites within or in different 50 bp distance intervals (1–500 bp) away from various non-B DNA features in CEM-SS T cells infected with HIV-1 generated in the presence of A3F-WT, A3F [E251A], A3G-WT, or A3G [E259A], or the absence of A3G or A3F (‘no A3’ control). The inset numbers refer to the percentage of total integration sites falling within 500 bp of the feature. The statistical comparison is with respect to ‘no A3’. B Heatmaps depicting the fold-enrichment (blue shading) and depletion (red shading) of integration sites at various distance intervals compared to the ‘no A3’ control virus. Black boxes highlight regions of notable enrichment. C Pairwise distance matrix was used to determine the overall similarity between the different integration site profiles. The fold-enrichment and depletion values in each distance bin for each non-B DNA feature were used in the comparison. The heatmap shows the distance matrix calculated by Euclidean distance as the measurement method. Stronger relationships are indicated by the darker blue color and weaker relationships by darker red color. Shaded triangles represent the different distance bins with the darkest shading representing distances further away from the feature. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; Fisher’s exact test, two-sided.

Fig. 4. A3 residues W94 and W127 play a role in integration site targeting.

A Frequency of integration sites within or at different distance intervals away from various common genomic features in CEM-SS T cells infected with HIV-1 generated in the presence of A3G-WT, A3G [W94A], A3G [W127A], or the absence of A3F or A3G (‘no A3’ control). The inset numbers refer to the percentage of total integration sites falling directly within the feature. The statistical comparison is with respect to A3G-WT. B Heatmaps depicting the fold-enrichment (blue shading) and depletion (red shading) of integration sites at various distance intervals compared to the ‘no A3’ control virus. C Pairwise distance matrix was used to determine the overall similarity between the integration site profiles of CEM-SS cells infected with either the no A3 control virus or A3F-WT, A3F [E251A], A3G-WT, A3G [E259A], A3G [W94A], or A3G [W127A] virus. The fold-enrichment and depletion values in each distance bin of each common DNA feature were used in the comparison. The heatmap shows the distance matrix calculated by Euclidean distance as the measurement method. Stronger relationships are indicated by the darker blue color and weaker relationships by darker red. D Frequency of integration sites within or at different distance intervals away from various non-B DNA features. The inset numbers refer to the percentage of total integration sites falling within 500 bp of the feature. The statistical comparison is with respect to A3G-WT. E Heatmaps depicting the fold-enrichment (blue shading) and depletion (red shading) of integration sites at various distance intervals compared to the ‘no A3’ control virus. Black boxes highlight regions of notable enrichment. F Pairwise distance matrix was used to determine the overall similarity between the integration site profiles. Shaded triangles represent the different distance bins with the darkest shading representing distances further away from the feature. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; Fisher’s exact test, two-sided. Source data are provided as a Source Data file.

Fig. 5. A3F and A3G reduce the number of integration hotspots and clustering of sites.

A Analysis of integration site hotspots. A hotspot was defined as a 1 kb window in the genome hosting 4 or more unique integration sites. Integration hotspots are shown as a proportion of total integration sites from CEM-SS cells infected with HIV-1 produced in the presence of no A3F or A3G (“No A3”) control (blue bars), or from cells expressing A3F-WT (black bars), A3F [E251A] (brown bars), A3G-WT (red bars), A3G [E259A] (green bars), A3G [W94A] (purple bars), or A3G [W127A] (yellow bars). B Integration site clustering was assessed by comparing the spacing between integration sites genome-wide to the same number of uniformly distributed (random) sites. Distances between sites are collected in seven length (L) ‘bins,’ with the shortest intersite lengths to the left and the longest to the right. A matched random control dataset was generated in silico (see methods for details). Fisher’s exact test, two-sided.

Fig. 6. G-to-A mutations in the 3’ LTR alters integration site targeting in vitro.

A LOGO representations of the terminal 56 nucleotides (A, green; C, blue; G, orange; T, red) of the 3’ LTRs of all integrated HIV-1 proviruses generated in the presence of either A3F, A3F [E251A], A3G, or A3G [E259A]. B Circa plots showing the integration sites of A3F- (left) or A3G- (right) containing viruses in the genome of infected CEM-SS cells. The outer ring represents the different chromosomes. The chromosomal locations of integration sites of proviruses containing GG (black), GA (red), AG (blue), and AA (brown) at positions 14 and 15 nucleotides from the LTR end are represented as colored ticks. The number and percentage of total sites are shown inside the circa plots. C Heatmaps depicting the fold-enrichment (blue shading) and depletion (red shading) of integration sites at various distance intervals from common genomic features (left) and non-B DNA features (right). Integration sites from A3F-LTR-GA (top) and A3G-LTR-AG (bottom) proviruses are shown. Fold changes are with respect to A3F-LTR-GG and A3G-LTR-GG, respectively. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; Fisher’s exact test, two-sided. Infinite number (inf) represents 1 or more integrations were observed when 0 integrations were expected by chance. Not a number (nan) represents 0 integrations were observed and 0 were expected by chance.

Fig. 7. G-to-A mutations in the 3’ LTR alters integration site targeting in vivo.

A LOGO representations of the terminal 27 nucleotides (A, green; C, blue; G, orange; T, red) of the 3’ LTRs of integrated HIV-1 proviruses in HIV-1 infected individuals. B Circa plot showing the integration sites of proviruses in the genome of infected individuals. The outer ring represents the different chromosomes. The chromosomal locations of integration sites of proviruses containing GG (black), GA (red), AG (blue), and AA (brown) at positions 14 and 15 nucleotides from the LTR end are represented as colored ticks. Numbers adjacent to ticks show the number of sites in that region that could not be distinguished by multiple ticks. The number and percentage of total sites are shown inside the circa plots. C Heatmaps depicting the fold-enrichment (blue shading) and depletion (red shading) of integration sites at various distance intervals from common genomic features (left) and non-B DNA features (right). Integration sites from proviral LTR-AG is shown and the fold changes are with respect to proviral LTR-GG integration sites. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; Fisher’s exact test, two-sided. Infinite number (inf) represents 1 or more integrations were observed when 0 integrations were expected by chance. Not a number (nan) represents 0 integrations were observed, and 0 were expected by chance.

Fig. 8. A3F and A3G promote latent infections in vitro.

A Schematic of the HIV-1 GKO reporter vector showing the csGFP gene (green box) under transcriptional control of the HIV-1 5’ LTR promoter and the mKO2 gene (orange box) under control of the constitutive EF1alpha promoter. B CEM-SS cells were infected with HIV_GKO or HIV_GKO-ΔU3LTR in the presence or absence of A3F, A3F[E251A], A3G, or A3G[E259A] for 48 h. Using flow cytometry, live cells were gated on using Zombie NIR^TM staining and the percentage of double-positive (csGFP+, mKO2+) (blue bars) and single-positive cells ((csGFP−, mKO2+) (red bars) or (csGFP+, mKO2−) (gray bars)) is shown. Statistical analysis was performed using two-way ANOVA with Dunnett’s multiple comparisons test (degrees of freedom = 36). C The average proportion of latently infected cells (csGFP−, mKO2+) from panel B is shown. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparisons test (degrees of freedom = 12). Data shown represents the mean values (±S.E.M.) from three independent experiments. Source data are provided as a Source Data file.