Biochemical differentiation of APOBEC3F and APOBEC3G proteins associated with HIV-1 life cycle

Abstract

APOBEC3G and APOBEC3F are cytidine deaminase with duplicative cytidine deaminase motifs that restrict HIV-1 replication by catalyzing C-to-U transitions on nascent viral cDNA. Despite 60% protein sequence similarity, APOBEC3F and APOBEC3G have a different target consensus sequence for editing, and importantly, APOBEC3G has 10-fold higher anti-HIV activity than APOBEC3F. Thus, APOBEC3F and APOBEC3G may have distinctive characteristics that account for their functional differences. Here, we have biochemically characterized human APOBEC3F and APOBEC3G protein complexes as a function of the HIV-1 life cycle. APOBEC3G was previously shown to form RNase-sensitive, enzymatically inactive, high molecular mass complexes in immortalized cells, which are converted into enzymatically active, low molecular mass complexes by RNase digestion. We found that APOBEC3F also formed high molecular mass complexes in these cells, but these complexes were resistant to RNase treatment. Further, the N-terminal half determined RNase sensitivity and was necessary for the high molecular mass complex assembly of APOBEC3G but not APOBEC3F. Unlike APOBEC3F, APOBEC3G strongly interacted with cellular proteins via disulfide bonds. Inside virions, both APOBEC3F and APOBEC3G were found in viral cores, but APOBEC3G was associated with low molecular mass, whereas APOBEC3F was still retained in high molecular mass complexes. After cell entry, both APOBEC3F and APOBEC3G were localized in low molecular mass complexes associated with viral reverse transcriptional machinery. These results demonstrate that APOBEC3F and APOBEC3G complexes undergo dynamic conversion during HIV-1 infection and also reveal biochemical differences that likely determine their different anti-HIV-1 activity.

FIGURE 1. A3 HMM and LMM complexes in 293T cells

A, isolation of A3A, A3F, and A3G HMM and LMM complexes by equilibrium density centrifugation. Cytosolic fractions were prepared from 293 cells expressing human A3A, A3F, and A3G. After RNase A or mock treatment, they were subject to 4 – 40% sucrose velocity gradient ultracentrifugation. Twelve fractions were collected from each tube. The proteins from each fraction were precipitated by trichloro-acetic acid and analyzed by Western blot. B, molecular mass of A3F and A3G complexes in high or low density fractions. Samples in these fractions were further subject to FPLC analysis. In total, 60 fractions from each analysis were collected for Western blot, and detected proteins with marked molecular mass were presented. C, A3G cytidine deaminase assay. Samples from high density, low density, and total cell lysate with or without RNase A treatment of 293 cells stably expressing A3G were subjected to an in vitro cytidine deamination assay as described under “Experimental Procedures.” Control (Ctrl) was the total cell lysate of the parent 293 cell. The error bars represent the standard deviations in at least three independent experiments.

FIGURE 2. Mapping A3G RNase-sensitive region

A, a schematic representation of A3G, A3F, and their recombinant forms A3F+G and A3G+F. The cytidine deaminase motif was squared, and the boundary for generation of A3F and A3G recombinant proteins was positioned. B, RNase sensitivity of A3F+G and A3G+F HMM complexes. A3F, A3G, A3F+G, and A3G+F protein complexes were prepared as in Fig. 1A and subject to Western blot.

FIGURE 3. Mapping the key determinant for A3F and A3G HMM complex assembly

A, schematic representation of mutations introduced to the cytidine deaminase motif in A3F and A3G. Four mutants were generated for each of A3F and A3G, which were designated as A3F_HXE1, A3F_PCXXC1, A3F_HXE2, A3F_PCXXC2, A3G_HXE1, A3G_PCXXC1, A3G_HXE2, and A3G_PCXXC2. B, anti-HIV-1 activities of these A3F and A3G mutants. pNL-LucΔvif was co-transfected with these mutant constructs into 293T cells, and viral infectivity was analyzed in GHOST cells. Ctrl, control plasmid. C and D, A3G and its mutants (C) or A3F and its mutants (D) were expressed in 293T cells, and their protein complexes were isolated and analyzed similarly as in Fig. 1A with the exception of RNase A treatment.

FIGURE 4. Interaction of A3G with cellular proteins by disulfide bond

A, detection and mapping of labile disulfide bond in A3G proteins. A3F-GST, A3G-GST, A3F+G-GST, and A3G+F-GST fusion proteins were expressed in 293T cells, and the proteins were purified by GSH-Sepharose beads. The samples were then treated with 8 mM NEM and analyzed by nonreducing SDS-PAGE followed by Western blot with anti-V5 antibody. B, absence of A3G-A3G interaction via disulfide bond. A3F-GST or A3G-GST was co-expressed with A3F or A3G, respectively, and a similar experiment was performed as in A. After nonreducing SDS-PAGE resolution, the proteins were analyzed by Western blot with anti-V5 antibody for A3-GST fusions or anti-FLAG antibody for A3 only. Those supershift bands were marked with *.

FIGURE 5. A3FandA3GproteincomplexinHIV-1virion

The top section provides a schematic description for three protocols to isolate virion-associated A3G or A3F complexes. Virus pellets were enriched by spinning through 20% sucrose cushion of the culture medium from 293T cells transfected with pNL4–3Δvif and A3G or A3F expression vector and used to determine A3G or A3F protein complexes in purified virion where virus pellets were directly loaded on the top of a16–65% sucrose gradient (A), purified viral core where virus pellets were loaded on the top of a 1% Triton X-100 layer followed with a 16–65% sucrose gradient (B), or complete viral lysates where virus pellets were first incubated with 1% Triton X-100 at 37 °C for 3.5 h and then loaded on a 16–65% sucrose gradient (C). These samples were then subjected to ultracentrifugation, and twelve fractions were collected. After trichloroacetic acid precipitation, the proteins were analyzed by Western blot.

FIGURE 6. A3F and A3G protein complex in target cells

Isolation of A3G (A) and A3F (B) protein complexes in HIV-1 infected cells. HIV-1 virions pseudotyped by VSV-G were produced by co-transfection of pNL4 –3Δvif, A3G or A3F, and VSV-G expression vectors. Viruses were used to infect GHOST cells. Four hours later, the cells were washed extensively, and cytosolic fraction was prepared. After RNase A or mock treatment, the samples were spun through a 4 – 40% sucrose gradient and analyzed by Western blot for A3G or A3F and viral proteins p32^IN, p17^MA, and p24^CA to localize viral reverse transcription machinery.