Identifying the assembly intermediate in which Gag first associates with unspliced HIV-1 RNA suggests a novel model for HIV-1 RNA packaging

Abstract

During immature capsid assembly, HIV-1 genome packaging is initiated when Gag first associates with unspliced HIV-1 RNA by a poorly understood process. Previously, we defined a pathway of sequential intracellular HIV-1 capsid assembly intermediates; here we sought to identify the intermediate in which HIV-1 Gag first associates with unspliced HIV-1 RNA. In provirus-expressing cells, unspliced HIV-1 RNA was not found in the soluble fraction of the cytosol, but instead was largely in complexes ≥30S. We did not detect unspliced HIV-1 RNA associated with Gag in the first assembly intermediate, which consists of soluble Gag. Instead, the earliest assembly intermediate in which we detected Gag associated with unspliced HIV-1 RNA was the second assembly intermediate (~80S intermediate), which is derived from a host RNA granule containing two cellular facilitators of assembly, ABCE1 and the RNA granule protein DDX6. At steady-state, this RNA-granule-derived ~80S complex was the smallest assembly intermediate that contained Gag associated with unspliced viral RNA, regardless of whether lysates contained intact or disrupted ribosomes, or expressed WT or assembly-defective Gag. A similar complex was identified in HIV-1-infected T cells. RNA-granule-derived assembly intermediates were detected in situ as sites of Gag colocalization with ABCE1 and DDX6; moreover these granules were far more numerous and smaller than well-studied RNA granules termed P bodies. Finally, we identified two steps that lead to association of assembling Gag with unspliced HIV-1 RNA. Independent of viral-RNA-binding, Gag associates with a broad class of RNA granules that largely lacks unspliced viral RNA (step 1). If a viral-RNA-binding domain is present, Gag further localizes to a subset of these granules that contains unspliced viral RNA (step 2). Thus, our data raise the possibility that HIV-1 packaging is initiated not by soluble Gag, but by Gag targeted to a subset of host RNA granules containing unspliced HIV-1 RNA.

Fig 1. Gag constructs: Diagrams and phenotyopes.

(A) Diagram of the different cis and trans expression systems used (Sets I–IV). Only WT constructs are shown here, with mutant constructs diagrammed in later figures. Set I consists of WT (and mutant) HIV-1 proviruses (pro-, delta env). Set II consists of codon-optimized WT (and mutant) Gag constructs tagged with GFP (Gag GFP) that are co-transfected with a modified genomic construct (V1B). V1B provides the genome for packaging in trans and expresses an assembly-defective truncated Gag. Set III consists of a WT provirus from Set I in which the nef gene was replaced with a puromycin resistance gene (PURO-R). H9 T cells were infected with this construct and maintained under puromycin selection to generate a chronically infected H9 T cell line expressing WT Gag from a provirus. Set IV consists of V1B constructs (see Set II) which were engineered to express WT or mutant Gag in cis (rather than the assembly-defective truncated Gag expressed by the Set II V1B construct). Set IV constructs were transfected into HeLa-MCP-GFP cells, which express a GFP-tagged MS2 capsid protein (MCP) that contains a nuclear localization signal. Since V1B genomic constructs also contain MS2 binding sites, MCP-GFP binds to V1B, resulting in GFP tagging of V1B viral RNA [12]. (B) Summary of expected Gag phenotypes. Diagram indicates whether WT Gag or each Gag mutant is expected to associate with intracellular unspliced HIV-1 RNA, leading to packaging initiation, and whether VLPs are known to be produced. Released VLPs that contain or lack viral RNA are indicated by +(+) and +(-), respectively. (C) Confirmation of expected Gag phenotypes. COS-1 cells were transfected with Set I constructs from A. Cell lysates and VLPs were harvested for analysis. Top row: Equivalent aliquots of cell lysates were analyzed by WB for Gag, as were VLPs harvested from the corresponding cell supernatants. VLPs and cell lysates were also analyzed by RT-qPCR for copies of unsplied HIV-1 RNA (graph). Similar results were obtained in 293T cells but are not presented. Bottom row: Cell lysates expressing proviruses encoding WT or mutant Gag (Set I constructs in panel A) were subjected to IP with αGag (G) or nonimmune (N) antibody followed by Gag WB (left). IP eluates were also analyzed by RT-qPCR for copies of unspliced HIV-1 RNA, with NI values subtracted (graph). Error bars show SEM from duplicate samples. Positions of kD markers are shown to the right of blots. Data are representative of three independent replicate experiments.

Fig 2. Non-translating unspliced HIV-1 RNA is primarily in diverse complexes ≥30S in the absence of assembling Gag.

(A-E) Lysate from COS-1 cells transfected with the assembly-incompetent MACA provirus (Set I constructs in Fig 1A) was divided into two pools that were either untreated (ribosomes intact) or treated with PuroHS (ribosomes dissociated). Both pools were analyzed in parallel by velocity sedimentation followed by RT-qPCR of paired gradient fractions using the appropriate qPCR primer sets to determine copy number of the indicated RNA (28S rRNA, 7SL RNA, unspliced HIV-1 RNA, HIV-1 Tat mRNA, or GAPDH mRNA). Quantity of the indicated RNA in gradient fractions is expressed as a distribution (% of RNA in all fractions) to allow comparison between different species of RNA. Blot in panel C shows migration of MACA protein, which is also graphed as a gray dotted line in C. Position of kD marker is shown to the right of the blot. Brackets at top show expected S value migrations, and horizontal bars show expected migrations of various ribonucleoprotein complexes. Error bars show SEM from duplicate samples. Data are from a single experiment that is representative of three independent replicate experiments.

Fig 3. The ~80S assembly intermediate contains Gag G2A associated with unspliced HIV-1 RNA while the ~10S Gag G2A assembly intermediate does not.

(A) COS-1 cells transfected with indicated MACA and G2A proviruses (Set I constructs in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced HIV-1 RNA copies was determined for cell lysates. (B) Lysates from A were analyzed by velocity sedimentation, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for each fraction and normalized to the inputs shown in A. (C) Gradient fractions from B were subjected to IP with human polyclonal HIV immune globulin (αGag), and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for IP eluates from each gradient fraction and normalized to the inputs shown in A. Similar results were obtained upon IP with monoclonal antibody to p24 but are not presented. (D) COS-1 cells transfected with the indicated Gag G2A provirus (Set 1 constructs in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for cell lysate. (E) Lysate from D was also analyzed by velocity sedimentation, and viral RNA copy number per 1000 cells was determined for each gradient fraction. (F) Gradient fractions from E were subjected to IP with αDDX6, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for IP eluates from each gradient fraction. Positions of kD markers are shown to the right of blots. Brackets at top show S value markers, and dotted lines demarcate assembly intermediates based on their migrations in the corresponding Gag WB. Error bars show SEM from duplicate samples. Data in each column are from a single experiment that is representative of three independent replicate experiments.

Fig 4. The ~80S assembly intermediate formed by WT Gag and two Gag mutants is the first intermediate to contain unspliced viral RNA.

(A) COS-1 cells transfected with the indicated WT and Gag G2A constructs (Set II constructs in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced viral RNA copies per 1000 cells was determined for cell lysates. (B) Lysates from A were analyzed by velocity sedimentation, and the number of unspliced viral RNA copies per 1000 cells was determined for each gradient fraction and normalized to inputs in A. (C) Gradient fractions from B were subjected to IP with αGFP, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for IP eluates from each gradient fraction and normalized to inputs in A. (D) COS-1 cells transfected with the indicated Gag W184A/M185A construct (Set II construct in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced viral RNA copies per 1000 cells was determined for cell lysate. (E) Lysate from D was also analyzed by velocity sedimentation, and the number of unspliced viral RNA copies per 1000 cells was determined for each gradient fraction and normalized to inputs in A. (F) Gradient fractions from E were subjected to IP with αGFP and the number of unspliced viral RNA copies per 1000 cells was determined for IP eluates from each gradient fraction and normalized to inputs in A, with units indicated on left side Y-axis. Gradient fractions were also directly analyzed for 18S rRNA, to mark the position of the 40S small ribosomal subunit. Quantity of 18S rRNA in gradient fractions is expressed as a distribution (% of RNA in all fractions), with units indicated on the right side Y-axis. Positions of kD markers are shown to the right of blots. Brackets at top show S value markers, and dotted lines demarcate assembly intermediates based on their migrations in the corresponding Gag WB. Error bars show SEM from duplicate samples. Data in each column are from a single experiment that is representative of three independent replicate experiments.

Fig 5. The ~80S assembly intermediate formed by WT Gag is the first assembly intermediate to contain unspliced viral RNA in human 293T cells and in the absence of PuroHS treatment.

(A) 293T cells transfected with the indicated construct (Set II construct in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced viral RNA copies per 1000 cells was determined for cell lysate. (B) Lysate from A was analyzed by velocity sedimentation, and the number of unspliced viral RNA copies per 1000 cells was determined for each gradient fraction. (C) Gradient fractions from B were subjected to IP with αGFP, and the number of unspliced viral RNA copies per 1000 cells was determined for IP eluates from each gradient fraction. (D) COS-1 cells transfected to express WT Gag GFP and the V1B genome (Set II constructs in Fig 1A) were harvested under standard conditions without PuroHS treatment, and the number of unspliced viral RNA copies per 1000 cells was determined for cell lysates. (E) Lysate from D was analyzed by velocity sedimentation, and the number of unspliced HIV-1 RNA copies in the equivalent of 1000 cells was determined in each gradient fraction. (F) Gradient fractions from E were subjected to IP with αGFP, and the number of unspliced HIV-1 RNA copies in the equivalent of 1000 cells was determined for IP eluates from each fraction. Positions of kD markers are shown to the right of blots. Brackets at top show S value markers, and dotted lines demarcate assembly intermediates based on their migrations in the corresponding Gag WB. Error bars show SEM from duplicate samples. Data in each column are from a single experiment that is representative of two independent replicate experiments.

Fig 6. Analysis of unspliced viral RNA in the ~80S and ~500S assembly intermediates from chronically infected H9 T cells.

(A) Human H9 T cells chronically infected with the indicated provirus (Set III construct in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for cell lysates. (B) Lysate from A was also analyzed by velocity sedimentation, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for each gradient fraction. (C) Gradient fractions from B were subjected to IP with αABCE1 and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for IP eluates from each gradient fraction. Positions of kD markers are shown to the right of blots. Brackets at the top show S value markers, and dotted lines demarcate assembly intermediates based on their migrations in the corresponding Gag WB. Error bars show SEM from duplicate samples. Data in each column are from a single experiment that is representative of three independent replicate experiments.

Fig 7. Gag Zip fails to associate with viral-RNA-containing granules, but is rescued by a heterologous viral-RNA-binding domain.

(A) COS-1 cells transfected with indicated constructs (Set II constructs in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced viral RNA copies per 1000 cells was determined for cell lysates. (B) Lysates from A were analyzed by velocity sedimentation, and the number of unspliced HIV-1 RNA copies per 1000 cells was determined for each gradient fraction and normalized to inputs in A. (C) Gradient fractions from B were subjected to IP with αGFP, and the number of unspliced viral RNA copies per 1000 cells was determined was determined for IP eluates from each gradient fraction and normalized to input in A. (D) COS-1 cells transfected with indicated constructs (Set II constructs in Fig 1A) were harvested following PuroHS treatment, and the number of unspliced viral RNA copies per 1000 cells was determined for cell lysates. (E) Lysates from D were analyzed by velocity sedimentation, and the number of unspliced viral RNA copies per 1000 cells was determined for each gradient fraction and normalized to inputs in A. (F) Gradient fractions from E were subjected to IP with αGFP, and the number of unspliced viral RNA copies per 1000 cells was determined for IP eluates from each gradient fraction and normalized to inputs in A. Positions of kD markers are shown to the right of blots. Brackets at top show S value markers, and dotted lines demarcate assembly intermediates based on their migrations in the corresponding Gag WB. Error bars show SEM from duplicate samples. Data in each column are from a single experiment that is representative of two independent replicate experiments.

Fig 8. Gag-DDX6 colocalization in situ upon provirus expression.

(A) PLA was used to detect regions in which Gag is within 40 nm from DDX6 in situ, with concurrent αGag IF for quantification of intracellular Gag levels. An experimental schematic for PLA methods is shown. For PLA experiments, 293T cells were transfected with the proviruses shown in diagram (Set 1 constructs in Fig 1A). (B) The average number of PLA spots per cell was determined for all Gag-positive cells in five randomly chosen fields, and normalized to Gag levels. C) Representative images. From left to right for each construct: Gag IF (green) with DAPI-stained nuclei (blue), Gag-DDX6 PLA signal (red) with DAPI-stained nuclei (blue), and a merge of all three. Merge demonstrates that PLA spots are mainly in Gag-expressing cells. Inset in PLA panel shows a high magnification view of the cell to the right of the inset. Scale bars, 5 μm. Data are representative of three independent replicate experiments. Error bars show SEM (n = 5 fields). +++ indicates a significant difference relative to WT (p≤0.001).

Fig 9. Gag-ABCE1 colocalization in situ upon provirus expression.

(A) PLA was used to detect regions in which Gag is within 40 nm from ABCE1 in situ, with concurrent αGag IF for quantification of intracellular Gag levels. An experimental schematic for PLA methods is shown. For PLA experiments, 293T cells were transfected with the proviruses shown in diagram (Set 1 constructs in Fig 1A). (B) The average number of PLA spots per cell was determined for all Gag-positive cells in five randomly chosen fields, and normalized to Gag levels. C) Representative images. From left to right for each construct: Gag IF (green) with DAPI-stained nuclei (blue), Gag-ABCE1 PLA signal (red) with DAPI-stained nuclei (blue), and a merge of all three. Merge demonstrates that PLA spots are mainly in Gag-expressing cells. Inset in PLA panel shows a high magnification view of the cell to the left of the inset. Scale bars, 5 μm. Data are representative of three independent replicate experiments. Error bars show SEM (n = 5 fields). + indicates a significant difference relative to WT (p≤0.01).

Fig 10. Complexes containing DDX6 colocalized with Gag G2A are far more numerous than DDX6-containing P bodies.

(A) Gag-DDX6 PLA was used to identify regions in which Gag is within 40 nm from DDX6 in situ, with concurrent αDDX6 IF for quantification of P body foci. An experimental schematic for PLA methods is shown (see text for details). 293T cells were transfected with the indicated G2A provirus (Set 1 constructs in Fig 1A) or mock transfected. (B) The average number of P body foci per cell (green bar) or PLA spots per cell (red bar) was determined for all Gag-positive cells in five randomly chosen fields. (C) Representative images. From left to right for Gag G2A and Mock: DDX6 IF to detect P body foci (green) overlaid with nuclei (blue), PLA signal (red) overlaid with nuclei (blue), and a merge of all three. Insets in DDX6 IF panels in the first column show higher gain/higher magnification versions of cells above the insets, with specific P bodies indicated by arrows at different angles to allow P bodies in insets to be matched to P bodies in low power fields. The dotted oval shows an example of three small granular foci that are too small to be P bodies (i.e. not visible in low gain/low power image), but are visible as discrete small foci when gain and magnification are increased. Insets in merged panels show higher gain/higher magnification versions of cells to the right, with specific P bodies indicated by arrows at different angles to allow P bodies in inset to be matched to P bodies in low power fields. Scale bar, 5 μm. Data are representative of three independent replicate experiments. Error bars show SEM (n = 5 fields).

Fig 11. The association of DDX6 with unspliced viral RNA is observed in situ at PM assembly sites for WT Gag but not for Gag Zip.

(A) HeLa cells expressing MCP-GFP were transfected with V1B genomes that contain MS2 binding sites (Set IV constructs in Fig 1A) and express WT Gag (shown), Gag Zip, or MACA. Cells were analyzed by double-label IEM, using αDDX6 and αGFP, which allowed detection of DDX6 (large gold) and MCP-GFP-labeled unspliced viral RNA (small gold), respectively. All early and late assembly sites at the PM were identified in ten cells (~250 μm of PM total per group), and scored for labeling of unspliced viral RNA, DDX6, and double labeling. (B) Graph shows the percentage of all early and late PM assembly events that display DDX6 labeling (total DDX6+), labeling of unspliced viral RNA (total viral RNA +), or double labeling (viral RNA+/DDX6+). (C) Images show representative assembly sites at the PM for each group, with symbols indicating early or late PM assembly sites that are single-labeled for either unspliced viral RNA or DDX6 (dark arrows), double-labeled (asterisks), or unlabeled (open arrows). Dotted lines outline the PM. Scale bars, 200 nm. Error bars show SEM (n = 10 cells). ++ indicates a significant difference between WT Gag and Gag Zip (p<0.005). For additional data, see S1 Table.

Fig 12. Hypothetical model for how packaging of unpsliced HIV-1 RNA could be intiated within a subclass of host RNA granules.

(A) MACA Gag, which lacks NC, fails to associate with RNA granules. In contrast, the targeting-defective Gag G2A mutant associates with ~80S RNA granules that contain unspliced viral RNA, leading to formation of the ~80S assembly intermediate in the cytoplasm, without further progression in the assembly pathway. In our studies, this ~80S assembly intermediate was the first assembly intermediate in which Gag associates with unspliced HIV-1 RNA, raising the possibility that the ~80S assembly intermediate is the packaging intiation complex. WT Gag also associates with RNA granules and forms the ~80S putative packaging initiation complex in the cytoplasm, which is then targeted to the PM where Gag multimerizes to complete assembly. RNA granule proteins dissociate from the fully assembled capsid prior to virus budding and release. Like WT Gag, Gag Zip targets to RNA granules, but because Gag Zip contains a protein-protein dimerization domain in place of the viral-RNA-binding NC domain, it is unable to target to the subset of ~80S granules that contains unspliced viral RNA. Ultimately, Gag Zip also undergoes PM targeting, assembly, dissociation of RNA granule proteins, and VLP release; however, Gag Zip VLPs lack viral RNA because the viral-RNA-binding-deficient Gag Zip may have targeted to the “wrong” subset of ~80S granules. Note that we have not defined the components of the RNA granule that contains unspliced viral RNA in each setting; for simplicity, we show the components we have defined for unspliced viral RNA in the context of WT Gag. (B) The two-step model for initial targeting of Gag to viral-RNA-containing RNA granules is shown in more detail. Only a subset of ~80S RNA granules contain viral RNA. While MACA Gag does not associate with any RNA granule subsets, G2A and WT Gag target to the subset of RNA granules that contains unspliced viral RNA. Gag Zip targets to RNA granules regardless of whether they contain unspliced viral RNA; however, a heterologous viral-RNA-binding domain (MCP) is able to rescue Gag Zip localization, allowing it to target to ~80S granules containing unspliced viral RNA.