Vulnerability to APOBEC3G linked to the pathogenicity of deltaretroviruses

Abstract

Human retroviruses are derived from simian ones through cross-species transmission. These retroviruses are associated with little pathogenicity in their natural hosts, but in humans, HIV causes AIDS, and human T-cell leukemia virus type 1 (HTLV-1) induces adult T-cell leukemia-lymphoma (ATL). We analyzed the proviral sequences of HTLV-1, HTLV-2, and simian T-cell leukemia virus type 1 (STLV-1) from Japanese macaques (Macaca fuscata) and found that APOBEC3G (A3G) frequently generates G-to-A mutations in the HTLV-1 provirus, whereas such mutations are rare in the HTLV-2 and STLV-1 proviruses. Therefore, we investigated the mechanism of how HTLV-2 is resistant to human A3G (hA3G). HTLV-1, HTLV-2, and STLV-1 encode the so-called antisense proteins, HTLV-1 bZIP factor (HBZ), Antisense protein of HTLV-2 (APH-2), and STLV-1 bZIP factor (SBZ), respectively. APH-2 efficiently inhibits the deaminase activity of both hA3G and simian A3G (sA3G). HBZ and SBZ strongly suppress sA3G activity but only weakly inhibit hA3G, suggesting that HTLV-1 is incompletely adapted to humans. Unexpectedly, hA3G augments the activation of the transforming growth factor (TGF)-β/Smad pathway by HBZ, and this activation is associated with ATL cell proliferation by up-regulating BATF3/IRF4 and MYC. In contrast, the combination of APH-2 and hA3G, or the combination of SBZ and sA3G, does not enhance the TGF-β/Smad pathway. Thus, HTLV-1 is vulnerable to hA3G but utilizes it to promote the proliferation of infected cells via the activation of the TGF-β/Smad pathway. Antisense factors in each virus, differently adapted to control host cellular functions through A3G, seem to dictate the pathogenesis.

Keywords: APOBEC3G; HBZ; HTLV-1; TGF-β; deltaretrovirus.

Fig. 1.

A3G induces G-to-A mutations in HTLV-1, but not in HTLV-2 and STLV-1. (A) Frequency of mutations detected in the proviruses of 65 HTLV-1 ACs by deep sequencing. (B) These HTLV-1 mutations occurred primarily at human APOBEC3G (hA3G) preferred target sequences. (C) Frequency of nonsense mutations observed in 65 HTLV-1 ACs. (D) Heatmaps showing the frequency of nonsense mutations observed in the tax, rex, and HBZ genes in 65 HTLV-1 ACs. (E) Frequency of mutations detected in the proviruses of 11 HTLV-2a-infected individuals by deep sequencing. (F) These HTLV-2a mutations did not typically occur at hA3G preferred target sequences (GG→AG). (G) Frequency of nonsense mutations observed in 11 HTLV-2a-infected individuals. (H) Frequency of mutations detected in the proviruses of 16 STLV-1-infected Japanese macaques by deep sequencing. (I) Most of these STLV-1 mutations do not typically occur at sA3G target sequences. (J) Frequency of nonsense mutations observed in 16 STLV-1-infected Japanese macaques. Mutations detected in more than 10% of total cases are shown (A and C in HTLV-1 ACs, E and G in HTLV-2a-infected individuals, and H and J in STLV-1-infected Japanese macaques). (K) Frequency of G-to-A mutations detected in HTLV-1-infected cells expressing hA3G or sA3G in vitro. (L) Frequency of G-to-A mutations detected in HTLV-2-infected cells expressing hA3G in vitro. (M) Frequency of G-to-A mutations detected in STLV-1-infected cells expressing hA3G or sA3G in vitro. Experiments were performed twice (K–M).

Fig. 2.

Antisense proteins interact with A3G. (A) Schematic representation of the A3G-mediated deaminase activity assay (Left) and quantifications of A3G deaminase activity in the presence of increasing concentrations of HBZ, APH-2, and SBZ (normalized mean ± SD of triplicate experiments; one-way ANOVA with Tukey correction; ***P < 0.001) (Right). (B) Coimmunoprecipitation experiment showing the interaction between human or sA3G and the antisense proteins of HTLV-1, HTLV-2, and STLV-1 in transfected HEK293T cells. IP, immunoprecipitation; IB, immunoblot. (C) Coimmunoprecipitation experiment showing the interaction of endogenous hA3G with endogenous HBZ in HTLV-1-infected T cell lines (MT-2 and MT-4), and an ATL cell line (ATL-55T+). IP, immunoprecipitation; IB, immunoblot. (D) Complex of hA3G with HBZ in primary ATL cells (n = 2; ATL#1 and ATL#2) and 3-dimensional image of a fresh ATL cell (ATL#2) shown by confocal z-stacking, detected by the Duolink Proximity Ligation Assay (Left) and rate of nuclear localization of HBZ-hA3G complex (Right). (E) Immunoblot showing the incorporation of both A3G and antisense proteins into HTLV-1, HTLV-2, and STLV-1 VP. All experiments were performed at least twice.

Fig. 3.

hA3G cooperates with HBZ to activate the TGF-β/SMAD pathway. (A) Expression of hA3G in healthy donors (n = 9) and patients with aggressive ATL (acute, and lymphoma type ATL) (n = 20) by RT-qPCR (triplicate experiments; one-way ANOVA with Tukey correction; ***P < 0.001). (B) Gene signatures that are significantly enriched in hA3G-knockdown ATL cells (ATL-55T+), based on GSEA analysis with RNA-seq. (C) Luciferase activity of 3TP-Lux under the control of a TGF-β responsive element in cells expressing hA3G or sA3G (normalized mean ± SD of triplicate experiments; two-tailed unpaired Student’s t test; **P < 0.01; ***P < 0.001). (D) Luciferase activity of 3TP-Lux under the control of a TGF-β responsive element in cells coexpressing HBZ with hA3G (Left), APH-2 with hA3G (Middle), or SBZ with sA3G (Right) (normalized mean ± SD of triplicate experiments; two-tailed unpaired Student’s t test; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). (E) Luciferase activity of 3TP-Lux under the control of a TGF-β responsive element in cells expressing deletion mutants of the N-terminal region of hA3G (normalized mean ± SD of triplicate experiments; two-tailed unpaired Student’s t test; ***P < 0.001). IB, immunoblot. (F) Luciferase activity of 3TP-Lux under the control of a TGF-β responsive element in cells expressing alanine scanning mutants of the 12 to 18th amino acids of hA3G (normalized mean ± SD of triplicate experiments; two-tailed unpaired Student’s t test; ****P < 0.0001). IB, immunoblot.

Fig. 4.

Interaction between A3G and Smad proteins. (A) Coimmunoprecipitation experiment showing the interaction of endogenous hA3G with endogenous Smad2/3 in HTLV-1-negative T cell lines (Hut78, H9), an HTLV-1-infected T cell line (MT-2), and an ATL cell line (ATL55T+). IP, immunoprecipitation; IB, immunoblot. (B) Expression of Smad2, Smad3, and Smad7 in healthy donors (n = 9) and patients with aggressive ATL (n = 20) by RT-qPCR (triplicate experiments; one-way ANOVA with Tukey correction; *P < 0.05). (C) Immunofluorescence microscopy images showing the hA3G protein and Smad3 protein in the presence or absence of TGF-β (Left) and percentage of localization (Right) in transfected HeLa cells. N: nucleus, C: cytoplasm. (D) Complex of hA3G with Smad3 in primary ATL cells with or without TGF-β treatment, detected by the Duolink PLA (Left) and the rate of nuclear localization (Right; two-tailed unpaired Student’s t test; **P < 0.01). (E) Immunoblot showing that hA3G expression induces the phosphorylation of Smad3 in a dose-dependent manner in transfected HepG2 cells. (F) Immunoblotting reveals no dose-dependent phosphorylation of Smad3 induced by HBZ expression in transfected HepG2 cells. Experiments were performed at least twice (A and C–F).

Fig. 5.

Blockage of TGF-β/Smad signaling suppresses ATL cell proliferation. (A and B) Cell growth of MT-1 (A) Hut78 (B) cells treated with SB431542 (Left), and cell growth rates of assorted HTLV-1-infected T cell lines (n = 9) (A) and HTVL-1 noninfected T cell lines (n = 6) (B) treated with SB431542 at day 4 are shown (Right) (normalized mean ± SD of triplicate experiments). (C and D) Cell growth of MT-1 (C) Hut78 (D) cells treated with SIS3-HCl cell (Left), and cell growth rates of assorted HTLV-1-infected T cell lines (n = 9) (C) and HTVL-1 noninfected T cell lines (n = 6) (D) treated with SIS3-HCl at day 4 are shown (Right) (normalized mean ± SD of triplicate experiments). (E) Hallmark gene sets down-regulated in MT-1 and ED cells upon treatment with SB431542, analyzed by RNA-seq. (F) Significantly enriched gene signatures for Myc-related gene sets in SB431542 treated ED cells and MT-1 cells, determined by GSEA analysis with RNA-seq. (G) BATF3 and IRF4 expression in ED and MT-1 cells treated with SB431542 or SIS3-HCl, analyzed by RT-qPCR (one-way ANOVA with Tukey correction; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). (H) Correlation of TGF-β protein with BATF3 protein or IRF4 protein in fresh ATL cells (n = 26), detected by multiplex mass-cytometry analysis (CyToF) (Pearson correlation analysis). (I) Venn diagram of differently expressed genes (DEG) by ChIP-seq in ED cells, MT-1 cells, and Hut78 cells. (J) Enrichment of phosphorylated Smad3 in BATF3 and IRF4 genes demonstrated by ChIP-seq analysis in an ATL cell line (ED) and a non-ATL T cell line (Hut78).

Fig. 6.

In vivo effect of blockage of TGF-β/Smad signaling on ATL. (A) Schematic representation of the xenograft model of ATL. (B) ATL tumor growth in mice treated with control (n = 7), SB431542 (n = 8), or SIS3-HCl (n = 7). (C and D) ATL tumor weights (normalized mean ± SD) (C) and volumes (D) in mice treated with control (n = 7), SB431542 (n = 8), or SIS3-HCl (n = 7) are shown (one-way ANOVA with Tukey correction; **P < 0.01; ***P < 0.001; ****P < 0.0001).