Gag.

Punctate structures, characteristic of assembly complexes or budding viruses, were readily observed across the PM of COS7 cells expressing wild-type Gag (Fig. 2A). Cluster analysis revealed that ∼64% of the Gag molecules were associated with clusters with a mean radius of 33.8 ± 0.16 nm (Fig. 2 B and C), with ∼61% of the clusters having a radius between 20 and 30 nm and very few clusters with radius approaching 150 nm (see SI Appendix for more details). Thus, virion formation is likely not driven by delivery to the PM of large and/or late-stage assembly complexes preassembled within the cytoplasm. Rather, it appears to be a multistage process that takes place at the PM, in which Gag forms assembly intermediates of increasing size in a step-wise fashion. Presumably, this process is initiated from an assembly precursor with a radius less than or equal to ∼20 nm, which likely contains a small number of Gag molecules bound to vRNA which may form in the cytoplasm (28). Overall, these findings are consistent with previous biochemical assays showing that Gag assembly is a multistage process (29) and that the PM is the primary site for higher-order Gag multimerization and subsequent particle assembly (4).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

PALM imaging and cluster formation analysis of Gag and NC mutants. (A) Representative PALM images are shown. Individual spots represent single molecules. (Scale bar, 500 nm.) (B) Percentage of identified Gag molecules associated with clusters. Values represent mean ± SD determined from at least 11 cells. Cluster information was extracted from PALM images using a Hoshen–Kopelman-based algorithm (SI Appendix, SI Materials and Methods). (C) Cluster radius distribution of Gag (n = 15,450) and Gag_ZiL (n = 14,242). Inset shows mean ± SEM radius. (D) Cluster-density distribution of Gag (n = 15,450) and Gag_ZiL (n = 14,242). For each cell, the cluster densities were normalized with respect to the mean densities across the entire PM. Inset shows mean ± SEM density. (E) Gag cluster density (from D) plotted as a function of cluster radius. The red line represents the least square regression from a logistic curve-fitting (r² = 0.98). (F) Gag_ZiL cluster density (from D) plotted as a function of cluster radius. The red line represents the least square linear regression (r² = 0.96). (G) Gag and Gag_ZiL cluster density as a function of cluster radius, plotted together on the same scale. Note that no data were shown for ΔNC-Gag because of its lack of clustering. For B, C, and D, * represents significant difference from Gag (P < 0.05).

Based on in vitro assembly assays, it is thought that Gag continuously becomes more densely clustered during the progression of assembly (13, 30). To investigate this possibility in nanoscale at the PM, Gag cluster density was determined by PALM cluster analysis (SI Appendix, SI Materials and Methods) (Fig. 2D) and was plotted as a function of the cluster radius (Fig. 2E). Interestingly, Gag cluster density increased asymptotically with the cluster radius, reaching a plateau at a cluster radius of ∼80 nm. Thus, it appears that HIV-1 assembly is not a process in which Gag molecules interact with each other toward ever more densely packed clusters from nascence to completion, as inferred from in vitro assembly assays. Rather, Gag assembly may involve two distinct growth phases in sequence, separated by a transition in the cluster radius. In the initial phase, Gag molecules are recruited to assembly precursors to form sizable clusters in a manner that favors dense-packing. After the assembly grows to a certain size, the process shifts to a phase in which additional Gag molecules increase cluster size at a constant packing density (Fig. 2E). Supporting this idea, a critical cluster radius of ∼63 nm was determined for Gag assembly from a logistic curve-fitting of the cluster density versus cluster radius relationship (SI Appendix, Fig. S5).

Gag_ZiL.

PALM imaging revealed that Gag_ZiL formed clusters and exhibited typical viral-budding characteristics similar to wild-type Gag at the PM of COS7 cells (Fig. 2A), as expected given that the Gag_ZiL mutant is known to be assembly-competent. Approximately 73% of Gag_ZiL were found to be associated with clusters, comparable with wild-type Gag clustering (Fig. 2B). Despite this similarity, the nanoscale organization of Gag_ZiL was quite different from that of Gag. Specifically, Gag_ZiL clusters were on average smaller (Fig. 2C) and more densely packed than Gag clusters (Fig. 2D). Additionally, the density of Gag_ZiL clusters increased linearly with cluster radius and did not plateau, as seen in Gag assembly (Fig. 2F), and is significantly higher than that of Gag for all cluster radii measured (P < 0.05) (Fig. 2G). Thus, in contrast to Gag assembly, Gag_ZiL appears to assemble through a single, cluster-centric process, in which Gag_ZiL trimers interact extensively with one another to form growing clusters with cluster density increasing linearly with cluster radius.

ΔNC-Gag.

In contrast to Gag and Gag_ZiL, very few clusters were observed at the PM of COS7 cells expressing ΔNC-Gag (Fig. 2A). Cluster analysis revealed that ΔNC-Gag molecules did not form clusters (Fig. 2B). Thus, lack of NC significantly inhibited Gag’s ability to form clusters at the PM. These findings are consistent with the inability of ΔNC-Gag to form VLPs (Fig. 1 and SI Appendix, Fig. S3).

Gag.

Gag was found to exhibit diverse motions at the PM, including stationary, diffusion (Brownian or corralled), and directed transport (Fig. 3A). More than 90% of molecules were mobile, with the fraction of mobile molecules undergoing diffusion and directed transport determined to be 92 and 8%, respectively. For the molecules that underwent diffusion, Gag had an overall mean diffusion coefficient of 0.13 ± 0.00 µm²/s (Fig. 3B), consistent with previous measurements (19, 26). The diffusing molecules were observed to be equally divided between Brownian and corralled motion (Fig. 3C). Of the molecules in directed transport, the mean speed was measured to be 1.35 ± 0.05 µm/s (Fig. 3D), which is within the range of measurements previously made for membrane-bound proteins (31⇓–33). Furthermore, the molecules had a mean track acceleration of 0.03 ± 0.05 µm/s², which is not significantly different from zero acceleration (P = 0.60, two-tailed t test assuming equal variance) (Fig. 3E). Together, these findings suggest Gag assembly through NC-RNA interactions is regulated kinetically by both diffusion and active transport at a constant speed.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Dynamics of Gag and NC mutants at the PM, assessed by spt-PALM. (A) Representative full-track movements of Gag at the PM. Gag exhibits a wide range of motions including no movement, diffusion (Brownian or corralled), and directed transport. (B) Analysis of lateral diffusion. The histogram shows the distribution of diffusion coefficients of single complexes of Gag and NC mutants. Total tracks of Gag (n = 3,604), Gag_ZiL (n = 1,730), and ΔNC-Gag (n = 1,115) were acquired from at least 11 cells. Inset shows mean ± SEM diffusion coefficients. (C) Plot of the percentage of molecules showing either Brownian (blue bar) or corralled (yellow bar) diffusion of Gag and its NC mutants at the cellular PM. Data represent mean ± SD of at least 11 cells. (D) Analysis of directed-transport properties of Gag and NC mutants. The histogram shows the distribution of the speed of single trajectories acquired for Gag and NC mutants. Right shows mean ± SEM speed. (E) The histogram shows the distribution of the mean acceleration of single trajectories acquired for Gag and NC mutants. Right shows mean ± SEM acceleration. For D and E, total tracks of Gag (n = 598), Gag_ZiL (n = 430), and ΔNC-Gag (n = 519). * represents significant difference from Gag (P < 0.05).

Gag_ZiL.

As seen in cells expressing Gag, Gag_ZiL also exhibits a wide range of motions. Eighty-four percent of the Gag_ZiL molecules were found to be mobile, with 89 and 11% of the mobile molecules undergoing diffusion and directed motion, respectively. The mean diffusion coefficient of Gag_ZiL molecules was 0.09 ± 0.00 µm²/s, which was significantly slower than Gag (Fig. 3B). Nearly 40 and 60% of the diffusive molecules were Brownian and corralled, respectively (Fig. 3C). Thus, it appears that Gag_ZiL’s diffusion at the PM is more confined compared with Gag. Of the molecules in directed transport, the mean speed was 0.79 ± 0.04 µm/s, which was significantly lower than the mean speed of Gag (Fig. 3D). Notably, Gag_ZiL has a mean track acceleration of −0.46 ± 0.04 µm/s² (Fig. 3E). This suggests that Gag_ZiL molecules experienced a negative driving force to restrict their movement. Together, these results indicate that Gag_ZiL on average exhibits slower dynamic properties than Gag. This suggests the mobility of Gag_ZiL at the PM is highly restricted and is in good agreement with its propensity to form densely packed clusters, as revealed by our PALM cluster analysis (Fig. 2).

ΔNC-Gag.

In contrast to Gag and Gag_ZiL, nearly all of ΔNC-Gag were mobile (>99%). The diffusive and directed-transport fractions were 56 and 44%, respectively. Compared with Gag and Gag_ZiL, ΔNC-Gag showed more rapid diffusion, with a mean coefficient of diffusion of 0.31 ± 0.01 µm²/s (Fig. 3B). Among the diffusing molecules, 91% showed Brownian motion, and 9% showed corralled motion (Fig. 3C). Additionally, more than 82% of ΔNC-Gag molecules had diffusion coefficients greater than the mean diffusions of Gag (0.13 µm/s²) and Gag_ZiL (0.09 µm²/s). Of the molecules that traveled by directed transport, the mean speed and mean acceleration were 1.79 ± 0.06 µm/s (Fig. 3D) and 0.14 ± 0.08 µm/s² (Fig. 3E), respectively. Thus, compared with Gag and Gag_ZiL, ΔNC-Gag exhibits higher mobility at the PM, suggesting its movement is less constrained. This is consistent with its inability to undergo viral RNA-mediated cluster assembly and viral budding (Fig. 1 and SI Appendix, Fig. S3).