Disparate Contributions of Human Retrovirus Capsid Subdomains to Gag-Gag Oligomerization, Virus Morphology, and Particle Biogenesis

ABSTRACT

The capsid domain (CA) of the retroviral Gag protein is a primary determinant of Gag oligomerization, which is a critical step for immature Gag lattice formation and virus particle budding. Although the human immunodeficiency virus type 1 (HIV-1) CA carboxy-terminal domain (CTD) is essential for CA-CA interactions, the CA CTD has been suggested to be largely dispensable for human T-cell leukemia virus type 1 (HTLV-1) particle biogenesis. To more clearly define the roles of the HTLV-1 CA amino-terminal domain (NTD) and CA CTD in particle biogenesis, we generated and analyzed a panel of Gag proteins with chimeric HIV-1/HTLV-1 CA domains. Subcellular distribution and protein expression levels indicated that Gag proteins with a chimeric HIV-1 CA NTD/HTLV-1 CA CTD did not result in Gag oligomerization regardless of the parent Gag background. Furthermore, chimeric Gag proteins with the HTLV-1 CA NTD produced particles phenotypically similar to HTLV-1 immature particles, highlighting the importance of the HTLV-1 CA NTD in HTLV-1 immature particle morphology. Taken together, these observations support the conclusion that the HTLV-1 CA NTD can functionally replace the HIV-1 CA CTD, but the HIV-1 CA NTD cannot replace the HTLV-1 CA CTD, indicating that the HTLV-1 CA subdomains provide distinct contributions to Gag-Gag oligomerization, particle morphology, and biogenesis. Furthermore, we have shown for the first time that HIV-1 and HTLV-1 Gag domains outside the CA (e.g., matrix and nucleocapsid) impact Gag oligomerization as well as immature particle size and morphology.

IMPORTANCE A key aspect in virus replication is virus particle assembly, which is a poorly understood process for most viruses. For retroviruses, the Gag structural protein is the primary driver of virus particle biogenesis, and the CA CTD is the primary determinant of Gag-Gag interactions for HIV-1. In this study, the HTLV-1 capsid amino-terminal domain was found to provide distinct contributions to Gag-Gag oligomerization, particle morphology, and biogenesis. This study provides information that will aid efforts for discovery of therapeutic targets for intervention.

INTRODUCTION

Retrovirus assembly is an intricate and poorly understood process that requires coordination of viral proteins, host cell proteins, and viral RNA. The Gag polyprotein plays a critical role in the retrovirus assembly process as the major retroviral structural protein (1, 2). The Gag protein orchestrates viral RNA (vRNA) binding, formation of an immature Gag lattice, and plasma membrane (PM) interactions, among many other roles (3, 4, 5). Expression of Gag in the cell is necessary for the formation of virus-like particles (VLPs) (6). Although VLPs cannot mature and are therefore noninfectious, they are morphologically similar to immature virions and are frequently used as a model system to study the functional role of Gag in particle-producing cells (7, 8, 9, 10).

Across the different orthoretroviral genera, Gag polyproteins vary in size and sequence but share three common domains: the amino-terminal matrix (MA) protein, the capsid (CA) protein, and the carboxy-terminal nucleocapsid (NC) protein (11). These domains are structurally distinct and connected by peptide linkers, which also vary in size and sequence among the different Gag proteins. MA is typically responsible for PM and vRNA binding (12, 13, 14). The CA domain mediates Gag-Gag dimerization as well as higher-order oligomerization and can be divided into amino-terminal and carboxy-terminal subdomains (CA NTD and CTD), which are independently folded (15). Finally, NC is responsible for vRNA binding and also plays a role in stabilizing CA-CA interactions (16). In addition to the MA, CA, and NC, some Gag proteins contain other domains. For example, human immunodeficiency virus type 1 (HIV-1) codes for spacer peptide 1 (SP1), spacer peptide 2 (SP2), and p6 (17, 18).

Although HIV-1 and human T-cell leukemia virus type 1 (HTLV-1) are both pathogenic human retroviruses, they have many distinct molecular and pathogenic features. The HIV-1 and HTLV-1 Gag proteins both have structurally conserved CA domains, sharing a β-hairpin turn and a centralized coiled-coil-like structure of six α helices in the CA NTD as well as four α helices in the CA CTD (19, 20, 21, 22). Despite the structural conservation between HIV-1 and HTLV-1 CA regions, the CA subdomains appear to play substantially different roles in assembly. For HIV-1, the prime determinants of CA-CA dimerization are W316 and M317, found in the CA CTD (23, 24). Unlike the case for HIV-1, the residues responsible for HTLV-1 CA-CA interactions have not yet been determined, but studies using large insertion and deletion mutations have implicated the HTLV-1 CA NTD as being indispensable for particle assembly (19). In contrast to HIV-1, mutations in the HTLV-1 CA CTD had little or no effect on VLP budding (19).

The roles of the HTLV-1 CA NTD and CTD in virus assembly and particle morphology are poorly defined. Based on comparative cryogenic transmission electron microscopy (cryo-TEM), HIV-1- and HTLV-1-like particles have distinct sizes and Gag lattice morphologies (2, 25, 26), but it remains unclear how the CA subdomains contribute to these differences. To more clearly define the roles of the HTLV-1 CA NTD and CTD in particle biogenesis, we generated and analyzed a panel of Gag proteins with chimeric HIV-1/HTLV-1 CA domains. The subcellular distribution and released protein expression levels indicated that Gag chimeras with an HIV-1 CA NTD/HTLV-1 CA CTD did not result in Gag oligomerization or particle budding regardless of the parental Gag background. Without CA-driven dimerization, the chimeric-CA Gag proteins in the HTLV-1 background trafficked to the PM, while chimeras in the HIV-1 background remained largely in the cytoplasm. This observation suggests differences in trafficking patterns between HIV-1 and HTLV-1 Gag. Additionally, chimeric Gag proteins with the HTLV-1 CA NTD produced particles morphologically similar to that of HTLV-1 particles as determined by cryo-TEM, revealing an important role for the HTLV-1 CA NTD in HTLV-1 immature particle morphology. Particles produced from chimeric-CA Gag proteins were morphologically distinct from either parental wild-type (WT) particle, implying that determinants outside CA also contribute to particle size and morphology. Taken together, these observations provide evidence that the HTLV-1 CA NTD can functionally replace the HIV-1 CA CTD, but that the HIV-1 CA NTD cannot replace the HTLV-1 CA CTD. This indicates that the HTLV-1 CA subdomains provide distinct contributions to Gag oligomerization, virus morphology, and particle biogenesis.

RESULTS

The objectives of this study were to dissect the functional roles of HIV-1 and HTLV-1 CA domains and subdomains. Given that chimeric constructs have been informative for evaluating the roles of Gag regions in past studies (27, 28, 29, 30), we created chimeric Gag proteins in which the CA region was fully or partially substituted for the equivalent sequences from HIV-1 or HTLV-1 (Fig. 1). Vectors were constructed with C-terminal enhanced yellow fluorescent protein (eYFP; i.e., Gag-eYFP), mCherry (i.e., Gag-mCherry), or hemagglutinin (HA) tags (i.e., Gag-HA). Gag expression constructs producing a nonlabeled Gag protein are referred to as untagged Gag.

FIG 1.

HIV-1 and HTLV-1 chimeric Gag expression vector constructs. Chimeric HIV-1 and HTLV-1 Gag expression vector plasmid constructs were engineered using codon-optimized gag genes. The panels in the figure show the gene sequence for the engineered Gag proteins analyzed in this study. HIV-1 gag sequences are represented in white boxes, and HTLV-1 gag sequences are represented in gray boxes. Vectors were created either without a C-terminal tag (i.e., untagged Gag) or with a C-terminal tag fused in-frame with Gag to produce proteins for fluorescence microscopy (eYFP and mCherry) or immunoblot analyses (HA). (A) WT Gag proteins, with the first and last 4 residues in the capsid (CA) domain indicated. (B) Gag CA chimeras with the entire HIV-1 or HTLV-1 CA domain removed and replaced. (C) Gag CA subdomain chimeras with the N-terminal domain or the C-terminal domain of CA replaced. (D) The parental HIV-1 vector MIG and the engineered chimeric Gag derivative with the HTLV-1 CA.

Chimeric Gag subcellular localization.

To investigate the subcellular localization of the chimeric Gag proteins compared with WT parental proteins, chimeric Gag-eYFP constructs were transiently transfected into HeLa cells. Both HIV-1 and HTLV-1 WT Gag proteins were localized around the cell perimeter as well as in the cytoplasm in a characteristic punctate distribution pattern, likely indicative of Gag oligomerization (Fig. 2A). To quantitatively characterize the subcellular distribution of the WT and chimeric Gag-eYFP proteins, the relative area of punctate fluorescence was calculated and compared to that of HTLV-1 WT (Fig. 2B). Although both chimeric Gag proteins with fully replaced CA domains (HIV/HTLV CA and HTLV/HIV CA) had a phenotypic punctate pattern, they each had significantly less punctate fluorescence than HTLV-1 WT, indicating a greater prevalence of diffuse fluorescence. Nonetheless, both HIV/HTLV CA and HTLV/HIV CA have fluorescence associated with puncta at approximately 50% the rate of HTLV-1 WT, suggesting Gag puncta formation at the PM.

FIG 2.

Subcellular localization of HIV-1 and HTLV-1 Gag CA chimeric proteins. HeLa cells were transiently cotransfected with Gag-eYFP and untagged Gag constructs at a 1:4 weight ratio. Cells were then fixed and stained with DAPI and Actin Red 555 approximately 24 h posttransfection. (A) Confocal microscopy analysis. Representative images of HeLa cells transfected with HTLV-1 WT, HIV-1 WT, HIV/HTLV CA, or HTLV/HIV CA. Scale bar, 10 μm. (B) Fluorescence area associated with distinct Gag puncta. The relative percentage of fluorescent area associated with puncta was determined by using particle analyses in ImageJ. Error bars indicate SD. The Student's t test was used to compare fluorescence percentages to that of HTLV-1 WT Gag. *, P < 0.001.

The subcellular localizations of Gag proteins containing either a chimeric CA NTD or CA CTD were analyzed in a manner similar to that for HIV/HTLV CA and HTLV/HIV CA (Fig. 3). We predicted that Gag proteins containing either the HTLV-1 CA NTD or the HIV-1 CA CTD would have the ability to oligomerize and therefore form Gag puncta, based upon literature identifying these domains as indispensable for particle formation (19, 23). HIV/HIV-C and HTLV/HIV-C both contain the HTLV-1 CA NTD and HIV-1 CA CTD in the HIV-1 and HTLV-1 Gag background, respectively (i.e., HIV-1 MA, SP1, NC, SP2, p6, or HTLV-1 MA and NC). Given that these chimeric proteins contain both HTLV-1 CA NTD and HIV-1 CA CTD, it is not surprising that they had a Gag punctate distribution similar to that of HIV/HTLV CA and HTLV/HIV CA. In contrast, HIV/HIV-N and HTLV/HIV-N, which contain the HIV-1 CA NTD and HTLV-1 CA CTD, resulted in little or no punctate Gag fluorescence and instead appear to have a diffuse fluorescence pattern, suggesting a lack of oligomerization. This observation provides one line of evidence for the presence of amino acid sequences located in distinct domains required for oligomerization, since these chimeric Gag proteins do not contain either the HIV-1 CA CTD or HTLV-1 CA NTD.

FIG 3.

Subcellular localization of HIV-1 and HTLV-1 CA subdomain chimeric proteins. HeLa cells were transiently cotransfected with Gag-eYFP and untagged Gag constructs at a 1:4 weight ratio. (A) Confocal microscopy analysis. Representative images showing subcellular localization of Gag-eYFP in cells transfected with HTLV-1 WT, HIV/HIV-C, HIV/HIV-N, HTLV/HIV-N, or the HTLV/HIV-C Gag expression construct. Scale bar, 10 μm. (B) Fluorescence area associated with distinct Gag puncta. Percentage of fluorescent area associated with puncta relative to that of HTLV-1 WT Gag was determined using particle analyses in ImageJ. A minimum of at least 20 cells were analyzed for each construct. Error bars indicate SD. The Student's t test was used to compare fluorescence percentages to that of HTLV-1 WT values. *, P < 0.05; ****, P < 0.0001.

It is also important to note that both the HTLV-1 and HIV-1 CA subdomains are separated by flexible 3- and 4-residue linkers, respectively (31). In the case of HIV-1, these linker residues have been shown to be important for mature CA core formation and infectivity (26, 32). To our knowledge, little is known about the role of the HTLV-1 CA linker domain in virus assembly and particle maturation. Based upon the design of the chimeric CA subdomain Gag constructs, inferences about the role of the CA linker domains in Gag oligomerization and particle biogenesis cannot be fairly made based on our analyses. Both HIV/HIV-C and HTLV/HIV-C contain the HTLV-1 CA linker as well as part of the HIV-1 CA linker, while HTLV/HIV-N and HIV/HIV-N contain part of the HIV-1 CA linker.

PM localization of chimeric Gag proteins.

HTLV-1 and HIV-1 Gag proteins have markedly different Gag-Gag interactions and PM trafficking phenotypes based on previous studies (33, 34). While HIV-1 Gag-Gag interactions occur in the cytoplasm, HTLV-1 Gag-Gag interactions initiate at the PM. Given these differences, we predicted that the loss of CA-driven oligomerization would perturb PM trafficking in different ways for HIV/HIV-N and HTLV/HIV-N, which have identical CA domains but distinct Gag backgrounds (Fig. 2; see also Fig. 5). To examine this, HeLa cells were cotransfected with Gag-eYFP and mCherry fluorescent protein, which served as a marker for the cytoplasm. Using dual-color z-scan microscopy, the fluorescence signal of Gag-eYFP at the PM was analytically separated from that of Gag-eYFP localized in the cytoplasm, as shown in Fig. 4. The dual-color z-scan technique uses vertical motion (a so-called z-scan) of the laser focus to measure fluorescent intensity versus cell height at individual locations within a cell. A deconvolution model was applied to the fluorescent intensity traces resulting from each z-scan (35, 36). The model was used to compare, within constraints imposed by the deconvolution, the vertical distribution of the mCherry fluorescence to that of Gag-eYFP. This comparison determined whether Gag-eYFP was distributed in a manner consistent with a PM-bound or exclusively cytoplasmic protein at the location of each z-scan. Z-scans then were performed at up to 4 locations in the cytoplasm of each cell. If 4 z-scan locations were found to have no detectable PM-bound Gag, the cell was considered PM negative. If one or more z-scan location had detectable Gag-eYFP at the PM, the cell was considered PM positive. Furthermore, any cells with Gag-eYFP puncta were considered PM positive.

FIG 5.

Differential membrane association of the HIV/HIV-N and HTLV/HIV-N Gag proteins. Dual-color z-scan fluorescence fluctuation spectroscopy was performed in HeLa cells by using YFP-tagged versions of either HIV-1 WT (A), HIV/HIV-N (B), HTLV-1 WT (C), or HTLV/HIV-N (D). Cells were then visualized approximately 12 h posttransfection.

FIG 4.

Dual-color z-scan PM binding assay. Fluorescence z-scans were performed on HeLa cells coexpressing Gag-eYFP and mCherry by moving the two-photon excitation volume vertically through the cytoplasm of individual cells. The resulting dual-color fluorescence z-scan traces were fit to a model that decomposes the Gag-eYFP fluorescence trace into corresponding PM-bound and cytoplasmic Gag-eYFP fluorescence traces. This model used the mCherry fluorescence trace (not shown) to locate the top and bottom edges of the cytoplasm (dashed lines). (A) The decomposition of the z-scan trace of a cell expressing HIV-1 Gag-eYFP recovered a cytoplasmic signal and a zero-valued PM signal, indicating the lack of detectable PM-bound Gag-eYFP. (B) The decomposition of the z-scan trace of a cell expressing HTLV-1 Gag-eYFP recovered fluorescence signals corresponding to both cytoplasmic and PM-bound Gag-eYFP, indicating that the cell is PM positive.

Similar to previous results, HIV-1 WT Gag exhibited a concentration-dependent shift from the cytoplasm to the PM (Fig. 5A), indicative of a concentration-dependent change from Gag monomer to oligomer (34). Unlike HIV-1 WT Gag, HTLV-1 WT Gag was found at the PM (Fig. 5C) at cytoplasmic concentrations where HIV-1 Gag was largely monomeric in the cytoplasm (37). The chimeric HIV-1 CA NTD/HTLV-1 CA CTD Gag in the HTLV-1 background (HTLV/HIV-N) behaved similarly to that of HTLV-1 WT Gag (Fig. 5D). Since HTLV/HIV-N was not predicted to oligomerize, this supports past data indicating that HTLV-1 Gag does not require dimerization for PM transport. In contrast, HIV/HIV-N was exclusively found in the cytoplasm at up to 1.3 μM, the maximum concentration evaluated for HIV-1 WT (Fig. 5B), indicating that the majority of HIV/HIV-N Gag was unable to translocate to the PM. This agrees well with previous findings that indicate HIV-1 WT Gag-Gag interactions must occur before efficient Gag translocation to the PM (34).

HTLV-1 CA NTD and HIV-1 CA CTD drive particle production.

To verify the VLP production levels, Gag-HA constructs were transiently transfected into 293T/17 cells and immunoblots were performed with cell lysates and supernatants from transiently transfected 293T/17 cells. Approximately 48 h posttransfection, supernatant and soluble cell lysates were collected and analyzed by immunoblotting (Fig. 6). As anticipated, both chimeric Gag proteins that did not form Gag puncta (HIV/HIV-N and HTLV/HIV-N) resulted in near-complete abrogation of Gag expression in the supernatant, indicating little or no particle production (Fig. 6C). These proteins were readily detected in cell lysates, indicating that the proteins were being expressed. The four other chimeric Gag proteins had comparable levels of Gag-HA in the supernatant compared to that of HTLV-1 WT Gag. Based upon these observations, the chimeric Gag proteins containing either the HTLV-1 CA NTD or the HIV-1 CA CTD were able to assemble and produce VLPs. It should be noted that the WT Gag proteins were not readily detectable in cell lysates, possibly due to the isolation of the soluble fraction, which likely contains fewer Gag proteins than the PM fraction.

FIG 6.

Immunoblot analysis of chimeric Gag proteins. 293T/17 cells were transfected with Gag-HA-tagged expression plasmid constructs, and the soluble fractions of cell lysates and culture supernatants were analyzed. Protein levels of Gag CA chimeras (A and B) and Gag subdomain CA chimeras (C and D) were detected by using a mouse monoclonal anti-HA antibody and an HRP-conjugated secondary antibody. Gag protein levels from cell culture supernatant samples were quantified by densitometry analysis and plotted relative to that of HTLV-1 WT Gag expression. Error bars indicate SD. Student's t test was used to compare expression levels to that of HTLV-1 WT Gag. ***, P < 0.001; ****, P < 0.0001.

HTLV-1 CA NTD drives puncta formation in the absence of HIV-1 CA CTD dimerization.

To determine whether HTLV-1 CA NTD could drive oligomerization and Gag puncta formation in the absence of HIV-1 CA CTD-CA CTD interactions, the WM316,317AA mutations were introduced into both HIV/HIV-C and HTLV/HIV-C. Since the WM mutation abrogates HIV-1 CA CTD dimerization, the HIV/HIV-C WM and HTLV/HIV-C WM constructs contain only one putative oligomerization domain, the HTLV-1 CA NTD.

Introduction of the WM mutations into HIV-1 WT Gag predictably resulted in diffuse fluorescence due to the lack of oligomerization (Fig. 7). However, HIV/HIV-C WM formed Gag puncta at a level that was not significantly different from WT levels, indicating that the HTLV-1 CA NTD could rescue the HIV-1 WM mutant phenotype. Intriguingly, HTLV/HIV-C WM, which has a CA domain identical to that of HIV/HIV-C WM, had a diffuse Gag fluorescence phenotype and did not exhibit a WT Gag phenotype. These data indicate that the parental Gag backgrounds of the two chimeric proteins impact the ability of HTLV-1 CA NTD to oligomerize and form Gag puncta. When the HTLV-1 CA NTD is in the presence of HIV-1 MA, SP1, NC, SP2, and p6, it is sufficient for Gag oligomerization. Previous studies have established that HIV-1 Gag domains exhibit functional redundancy, wherein single defects do not abrogate spherical particle assembly and production (38). It is formally possible that HTLV-1 Gag does not have similar functional redundancy, indicated by the inability of HTLV-1 CA NTD to drive puncta formation in the absence of HTLV-1 CA CTD.

FIG 7.

Subcellular localization of WM mutant Gag protein. HeLa cells were cotransfected with Gag-eYFP and untagged Gag expression constructs at a 1:4 ratio. (A) Confocal microscopy analysis. Representative images of the subcellular localization of the HTLV-1 Gag WT, HIV-1 Gag WT, HIV-1 WM Gag, HIV/HIV-C WM Gag, and HTLV/HIV-C WM Gag proteins are shown. Scale bar, 10 μm. (B) Fluorescence area associated with distinct Gag puncta. The relative fluorescent area associated with puncta was determined using ImageJ particle analyses. Error bars indicate SD. Student's t test was used to compare the fluorescence percentage to that of HTLV-1 WT values. *, P < 0.0001.

Full-length CA domains contribute to Gag colocalization and copackaging.

We next investigated whether the CA domain or subdomains impacted the subcellular colocalization and VLP copackaging of chimeric Gag proteins with WT Gag proteins. To analyze the colocalization of WT and chimeric Gag proteins, mCherry-tagged WT Gag and eYFP-tagged chimeric Gag constructs were cotransfected into HeLa cells. Pearson's coefficients were used to evaluate the degree of colocalization between mCherry- and eYFP-tagged Gag proteins (Fig. 8B). While HIV-1 WT and HTLV-1 WT did not appear to colocalize, the chimeric Gag proteins with fully replaced CA domains colocalized well and almost exclusively with the WT Gag that shared the same CA domain (i.e., HIV/HTLV CA with HTLV-1 WT). This observation supports a previous report that indicated the CA domain is the prime determinant of Gag colocalization (28). However, while HIV/HIV-C and HTLV/HIV-C have identical CA domains, only HIV/HIV-C was observed to colocalize with HIV-1 WT. Furthermore, we did not observe any evidence that cotransfection with either WT Gag rescued the diffuse Gag fluorescence phenotype of HIV/HIV-N or HTLV/HIV-N (Fig. 8B and 9A).

FIG 8.

Colocalization and copackaging of YFP-tagged chimeric Gag and mCherry-tagged WT Gag proteins. (A) Confocal microscopy analysis of colocalization of chimeric and WT Gag proteins. HeLa cells were cotransfected with Gag-eYFP, Gag-mCherry, and respective untagged Gag expression constructs (1:1:1:1 weight ratio). Scale bar, 10 μm. (B) Statistical analysis of fluorescence protein colocalization. The subcellular colocalization of WT and chimeric Gag proteins was quantitatively analyzed in at least 15 cells for each construct using ImageJ, and the colocalization correlation was analyzed using Pearson's coefficient. Error bars indicate SD. The Student's t test compared colocalization of HIV-1-mCherry cotransfections for each YFP-tagged WT or chimeric construct. *, P < 0.01. (C) Confocal microscopy analysis of Gag protein copackaging in virus-like particles (VLPs). 293T/17 cells were cotransfected with Gag-eYFP plasmids, WT Gag-mCherry plasmids, and respective untagged Gag expression constructs (1:1:1:1 weight ratio). Harvested VLPs were visualized by confocal microscopy. Scale bar, 200 nm. (D) Statistical analysis of fluorescence protein copackaging. The percentage of Gag copackaging was determined by the number of particles containing both the mCherry and YFP signals divided by the total number of particles analyzed.

FIG 9.

Representative images of colocalization and copackaging of YFP-tagged chimeric Gag and mCherry-tagged WT Gag proteins. (A) Confocal microscopy analysis of colocalization of chimeric and WT Gag proteins. HeLa cells were cotransfected with Gag-eYFP, Gag-mCherry, and respective untagged Gag expression constructs (1:1:1:1 weight ratio). Scale bar, 10 μm. (B) Confocal microscopy analysis of Gag protein copackaging in virus-like particles (VLPs). 293T/17 cells were cotransfected with Gag-eYFP plasmids, WT gag-mCherry plasmids, and respective untagged Gag expression constructs (1:1:1:1 weight ratio). Harvested VLPs were visualized by confocal microscopy. Scale bar, 200 nm.

We next sought to confirm that colocalized Gag proteins budded together into the same VLP by analyzing the supernatant of cotransfected 293T/17 cells. The resultant VLPs were imaged with confocal microscopy. In most cases, Gag proteins that colocalized in HeLa cells also copackaged into particles together at a rate similar to that of homospecies cotransfections (Fig. 8C and D). Unsurprisingly, neither HIV/HIV-N-eYFP nor HTLV/HIV-N-eYFP copackaged due to few visible particles containing eYFP (Fig. 8D and 9B).

Morphological differences between WT and chimeric Gag-based VLPs.

HIV-1 and HTLV-1 WT Gag-based VLPs have established and distinct electron-dense regions characteristic of a Gag lattice (25). HTLV-1-like particles contain flat regions of Gag lattice electron density that do not follow the curvature of the lipid bilayer (Fig. 10A). HIV-1 VLPs, however, have a Gag lattice electron density that exclusively follows the curvature of the lipid bilayer (26, 39, 40, 41). We next characterized the morphologies of the chimeric Gag-based VLPs that were shown to produce particles: HIV/HTLV CA, HTLV/HIV CA, HIV/HIV-C, and HTLV/HIV-C.

FIG 10.

Analysis of virus-like particle morphology by electron microscopy. (A) Thin-section transmission electron microscopy and cryo-TEM analysis. Thin-section and cryo-TEM were performed on either VLP-producing 293T/17 cells or purified VLPs, respectively. The arrow indicates electron densities characteristic of an unstructured Gag assembly. Colored circles indicate qualitative Gag electron density morphology scoring. Orange represents curved lattice and curved bilayer; pink represents flat lattice and flat bilayer; green represents flat lattice and curved bilayer; blue represents no lattice. The scale bar for thin-section EM is 200 nm. The scale bar for cryo-TEM is 100 nm. (B) Particle diameter analysis. VLP diameters were measured using ImageJ, and particle diameters were averaged across at least 50 different VLPs. The mean diameters and standard errors of the means (SEM) are shown. Student's t test was used to compare the VLP diameters to those of HIV-1 WT and HTLV-1 WT. *, P < 0.01 compared with HIV-1. ^, P < 0.01 compared with HTLV-1. (C) The morphologies of the electron-dense Gag lattices of the VLPs. The percentage of particles is shown as the averaged result of the analysis from two independent reviewers. The color codes are the same as those for the dots in panel A.

Intriguingly, the four chimeric Gag proteins produced morphologically distinct sets of VLPs containing Gag lattice electron densities (Fig. 10A and C). Furthermore, all three chimeric Gag proteins that contained the HTLV-1 CA NTD (HIV/HTLV CA, HIV/HIV-C, and HTLV/HIV-C) produced VLPs with flat regions of Gag lattice electron density. Unlike the HTLV-1 WT Gag lattice, both HIV/HTLV CA and HIV/HIV-C VLP populations contained nonspherical particles, with flat regions of Gag lattice electron density correlating with flat regions of the lipid bilayer. HTLV/HIV-C, however, had flat regions of Gag lattice that did not follow the curvature of the lipid bilayer, maintaining a spherical particle shape similar to that of HTLV-1 WT. This indicates that the HIV-1 MA domain impacts the ability of the lipid bilayer to curve in the presence of flat Gag lattice regions.

The only chimeric Gag-based VLP that did not appear to have a distinct Gag lattice was HTLV/HIV CA, which instead had a poorly visible Gag lattice electron density, distinguishable only as electron-dense regions beneath the lipid bilayer (Fig. 10A, arrow). These regions did not appear to contain an organized Gag lattice due to the unresolved nature of the electron-dense region. Furthermore, many particles observed by both thin-section and cryo-TEM lacked any apparent electron-dense regions beneath the lipid bilayer, which could indicate a lack of Gag lattice electron density, but the two-dimensional nature of our cryo-TEM limits our ability to be certain. The lack of apparent, organized Gag lattice in HTLV/HIV CA VLPs indicates either the requirement of stabilizing interactions present in HTLV-1 CA or destabilizing forces in the MA and NC.

It has previously been suggested that the CA domain dictates VLP size in addition to morphology (28). This observation is supported by both HIV/HTLV CA and HTLV/HIV CA VLPs, which were not statistically different in size from the WT Gag sharing the same CA (Fig. 10B). HIV/HIV-C and HTLV/HIV-C particles, however, share the same CA and yet were markedly different in size, not only from each other but also from either WT Gag. These data support the conclusion that the MA, NC, and other domains also have an impact on VLP size and morphology.

Infectivity and morphology of chimeric HIV-1 with HTLV-1 CA.

Since the chimeric Gag proteins were able to form an electron-dense Gag lattice, we sought to investigate whether these proteins could be cleaved and create infectious virus particles. Based on the many factors that influence infectivity, we hypothesized that the chimeric Gag proteins were unlikely to form infectious particles. In order to determine if Gag chimeras would impact virus infectivity, we engineered the HTLV-1 CA domain into an HIV-1 vector, leaving the HIV-1 CA cleavage sites intact (Fig. 1D). It should be noted that a corresponding HTLV-1 vector with HIV-1 CA was not tested due to the low infectivities of the parental constructs (42). In most expression systems, transfection of a vector containing authentic HTLV-1 sequence does not result in substantial gene expression or the production of detectable particles. Pseudotyped vector virus was produced via cotransfection of chimeric HIV-1 vector and vesicular stomatitis virus glycoprotein (VSV-G) into 293T/17 cells. Viral supernatants were collected and subjected to both quantitative PCR (qPCR) and immunoblotting to test for virus production and Gag cleavage. Based on quantification of RNA copy number, chimeric vector particle supernatants contained approximately 50-fold less virus than the parental construct (Fig. 11A). When the titer of the viral supernatant was determined by RNA copy number, the chimeric vector virus did not have readily measured infectivity, indicating a severe infectivity defect (Fig. 11B).

FIG 11.

Analysis of infectivity and particle morphology of HIV-1 vector expressing chimeric Gag protein. The HTLV-1 CA was swapped with the HIV-1 CA in a single-cycle infectivity HIV-1 vector with a Gag WT to create Gag HIV/HTLV CA. (A) Analysis of particle release. Particle release was analyzed by measuring viral RNA in released vector virus particles by using qPCR. NTC, no template control. (B) Particle infectivity analysis. MAGI cells were challenged with HIV-1 Gag WT or Gag HIV/HTLV CA or were subjected to mock infection with cell culture supernatants containing either ∼1.5 × 10⁶, ∼3.0 × 10⁶, or ∼6.0 × 10⁶ RNA copies of each vector. Infectivity was determined by flow cytometry analysis for expression of GFP and mCherry, which are both encoded by the HIV-1 vector. (C) Immunoblot analysis. Proteolytic cleavage of Gag was determined by harvesting the soluble cell lysates and cell culture supernatants from 293T/17 cells transfected with the viral constructs approximately 48 h posttransfection. The p24, p17, and tubulin proteins were evaluated by detection with an HRP-conjugated secondary antibody. (D) Virus particle analysis by cryo-TEM. Single-cycle infectivity HIV-1 vectors expressing either Gag WT or Gag HIV/HTLV CA were transfected into 293T/17 cells, and virus particles purified 48 h posttransfection. Representative cryo-TEM images are shown. Scale bar, 100 nm.

To determine the reason for the relatively poor particle production and lack of infectivity, we analyzed the parental and chimeric vector virus particles by immunoblotting. Since p17 and other Gag cleavage products were detected in the chimeric vector virus supernatants, the chimeric Gag appears to undergo cleavage, albeit in a different pattern than the parental vector (Fig. 11C). Although we attempted to probe HTLV-1 p24 cleavage, the anti-HTLV-1 p24 antibody did not detect HTLV-1 CA, perhaps due to a lack of or changes in epitope binding sites due to the replacement of 10 HTLV-1 amino acids with HIV-1 sequence to preserve the cleavage sites. Additionally, less p17 was detected in the supernatant from the chimeric Gag, corroborating the results of the qPCR. It is presently unclear why the chimeric Gag vector produced fewer virus particles than the WT Gag vector, but it suggests a clear defect(s) in the particle assembly pathway. This is supported by cryo-TEM analysis of the chimeric and parental vector viruses. Clearly visible and characteristic electron-dense conically shaped cores were observed in virus produced from the parental vector (43). Chimeric vector virus particles were slightly smaller than those reported for authentic HTLV-1 WT particles (101.8 nm with a standard deviation [SD] of ±32.1 nm) (7). Furthermore, evaluation of these particles revealed the lack of an apparent core, which may explain the lack of particle infectivity and also implies a defect(s) in the particle assembly pathway (Fig. 11D). Since HIV/HTLV CA VLPs contained a Gag lattice (Fig. 11A), we hypothesize that the assembly deficiency does not necessarily stem from the oligomerization of the full-length Gag but from the inability of CA to oligomerize to form a mature core, which requires different CA-CA interactions from those for Gag lattice formation. This strongly suggests that there are some other nucleating factors necessary for HTLV-1 CA core formation.

DISCUSSION

Chimeric human retroviral Gag proteins have been utilized in this study to provide the first thorough experimental evidence that establishes the HTLV-1 CA NTD as a key determinant for HTLV-1 CA-CA interactions (and Gag-Gag oligomerization), which are required for virus particle assembly and release. This is in contrast to HIV-1 Gag-Gag oligomerization, where the HIV-1 CA CTD is key for CA-CA interactions. In particular, our studies show, for the first time, that replacement of the HIV-1 CA NTD with the HTLV-1 CA NTD can rescue the ability of HIV-1 Gag to oligomerize and form a lattice structure in the context of the dimerization-defective HIV-1 Gag mutant WM316,317AA. We also provide evidence indicating that the HTLV-1 CA CTD performs a functional role in HTLV-1 WT stability and/or oligomerization, which is in contrast to the single dimerization domain of HIV-1 WT. In the presence of a dimerization-defective HIV-1 CA CTD, HTLV-1 CA NTD is not sufficient to drive oligomerization. It is possible that HIV-1 Gag contains more functional redundancy throughout the protein, wherein defects may be present in one functional domain and still allow for particle formation (38). Based on these observations, the general findings presented in this study indicate that the HTLV-1 CA NTD can functionally replace the HIV-1 CA CTD, and conversely the HIV-1 CA CTD can functionally replace the HTLV-1 CA NTD.

Amino acid alignments indicate that the HIV-1 and HTLV-1 CA NTDs are relatively similar, with well-conserved amino acids comprising approximately 50% of HIV-1 and HTLV-1 CA NTD sequences (19). Additionally, a comparison of the HTLV-1 CA NTD nuclear magnetic resonance (NMR) structure with the HIV-1 CA NTD structure revealed a high degree of structural similarity with some differences in the orientation of the β-hairpin (44). Given the structural similarities of HIV-1 and HTLV-1 CA, it is intriguing that the CA domains of these two human retroviruses play such different roles in assembly. This observation emphasizes the importance of the divergent sequences in the two Gag protein domains, such as the aspartate residues orienting the β-hairpin (44, 45, 46).

Previous studies have indicated that the retroviral CA domain dictates VLP size, morphology, and copackaging of Gag proteins into VLPs (28). While the findings of our study indicated that the HIV-1 and HTLV-1 full-length CA domains define VLP size, Gag colocalization, and copackaging into particles, we also demonstrated that other Gag domains can impact HIV-1 and HTLV-1 particle morphology. In particular, none of the chimeric Gag proteins analyzed in our study formed puncta at the same rate as HTLV-1 WT, indicating that Gag domains other than CA contribute to Gag puncta formation. Surprisingly, VLPs produced from all four of the Gag CA chimeras contained Gag lattice electron densities, and all four of these sets of VLPs had different particle morphologies. Neither the HIV/HTLV CA nor the HTLV/HIV CA were identical in morphology compared to corresponding parental Gag-based VLPs despite the similarities in size. Furthermore, the HIV/HIV-C and HTLV/HIV-C Gag proteins contained identical CA regions (HTLV-1 CA NTD followed by the HIV-1 CA CTD) yet resulted in the production of VLPs of substantially different sizes and morphologies, highlighting the alternative and as-of-yet unestablished roles of MA, NC, and other Gag domains in VLP size and morphology. It is possible that morphological differences in the Gag-based HIV/HIV-C and HTLV/HIV-C particles are due primarily to the MA domain. HTLV/HIV-C particles have flat regions of Gag lattice but are largely spherical. In contrast, the lipid bilayer of HIV/HIV-C particles remains close to the flat regions of the Gag lattice, leading to a nonspherical VLP. Taking these observations in total, we propose that the flat and nonspherical regions of the chimeric Gag-based VLPs are caused by HIV-1 MA-derived factors. One possible interpretation is that the tight viral PM interaction of the HIV-1 MA in the HIV/HIV-C and HIV/HTLV CA chimeric Gag proteins results in the deformation of the lipid bilayer. A second possible interpretation is that the C terminus of the HIV-1 MA is less flexible than the C terminus of deltaretroviral MA domains, which would allow the flat regions of HTLV-1 lattice to have less of an impact on the VLP shape (47).

The observation that Gag domains outside CA impact VLP size and morphology is in contrast to previous observations with other Gag proteins with chimeric CA domains, but we have presented the first report of chimeric Gag proteins from the deltaretrovirus and lentivirus genera. In particular, past observations of chimeric Gag proteins have been made for lentivirus- and alpharetrovirus-derived Gag proteins (28). The established differences in VLP assembly and morphology among different retroviral genera may offer an explanation for the differences observed in our present and previous studies (25, 28). Furthermore, it is important to note that the CA subdomain chimeras contain marked alterations to the WT CA sequence due to the introduction of CA domains from another genus, which limits our ability to extrapolate our interpretations to WT Gag. Future studies that evaluate the role of MA, NC, and other Gag domains in particle size and morphology using physiologically relevant constructs should be a priority.

We have also demonstrated that the HIV-1 and HTLV-1 Gag proteins translocate to the PM in distinctly different manners. It has been observed that HTLV-1 Gag translocates to the PM as a monomer at low cytoplasmic concentrations, while HIV-1 Gag can only translocate to the PM after reaching a higher cytoplasmic concentration, which triggers dimerization and PM targeting (3, 33, 34, 37, 48, 49). In addition, we have demonstrated that the Gag HTLV/HIV-N expression construct, which lacks a CA subdomain capable of oligomerization, binds strongly to the PM. In contrast, HIV/HIV-N, which also lacks a CA subdomain capable of oligomerization, is found in the cytoplasm and was not strongly associated with the PM, even at very high cytoplasmic concentrations. Taken together, our data support the general observation that HIV-1 Gag does not readily target the PM in the absence of Gag-Gag interactions (33). Furthermore, our data, along with previous observations, support the idea that HIV-1 Gag-Gag interactions should be considered a critically important determinant of HIV-1 Gag translocation to the PM, similar to the highly basic region of MA and the presence of phosphatidylinositol-4,5-bisphosphate [PI(4,5)P₂] (50, 51, 52). PI(4,5)P₂ has been shown to be dispensable for HTLV-1 particle assembly and budding, and our data provide evidence that Gag-Gag oligomerization is also unnecessary for PM targeting (53), indicating that CA-CA interactions have markedly different effects and roles in HIV-1 and HTLV-1 Gag translocation to the PM and in particle assembly. Further studies are needed to determine whether HTLV-1 has an alternative, compensatory mechanism(s) for Gag translocation to the PM.

MATERIALS AND METHODS

Plasmids, cell lines, and reagents.

Codon-optimized HIV-1 and HTLV-1 gag genes in the pN3 and peYFP-N3 vectors have been previously described (25, 54). WT and chimeric Gag expression constructs were engineered into vectors with C-terminal eYFP, mCherry, and HA tags using 5′ HindIII and 3′ BamHI restriction sites or were untagged. The HTLV-1 envelope (Env) expression plasmid, CMV-ENV, was kindly provided by Kathryn Jones (55). Viruses were pseudotyped with the VSV-G expression construct, pHCMV-G, a kind gift from J. Burns (University of California, San Diego) (56). HeLa and HEK293T/17 cells lines were obtained from the ATCC (Manassas, VA) and maintained in Dulbecco's modified Eagle medium supplemented with 10% fetal clone III (GE Healthcare Lifesciences, Logan, UT). U373-MAGI-CXCR4_CEM cells were obtained from Michael Emerman through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH (57).

Chimeric Gag expression constructs.

Chimeric Gag plasmids were engineered using mammalian codon-optimized HIV-1 and HTLV-1 sequences (Fig. 1A). A chimeric codon-optimized Gag expression plasmid containing HIV-1 CA (HIV-1 Gag_133–361) and the HTLV-1 MA and NC (HTLV-1 Gag_{1–130,346–429}) (HTLV/HIV CA) was engineered using a modified version of overlap extension PCR cloning (58). Briefly, the CA region of a codon-optimized HIV-1 Gag expression construct was PCR amplified with primers containing the HTLV-1 MA and NC sequence on the 5′ and 3′ ends, respectively. An HTLV-1 codon-optimized Gag expression plasmid was then mutagenized using the chimeric PCR product as primers and a QuikChange II XL site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). The chimeric codon-optimized Gag expression plasmid containing HTLV-1 CA (HTLV-1 Gag_131–345) and HIV-1 MA, SP1, NC, SP2, and p6 (HIV-1 Gag_{1–132,362–500}) (i.e., HIV/HTLV CA) was engineered using the same method. Codon-optimized Gag plasmids with chimeric CA subdomains (Fig. 1C) were engineered using gBlocks gene fragments (Integrated DNA Technologies, Coralville IA). An HTLV-1 Gag plasmid containing the HIV-1 CA NTD (HTLV-1 Gag_{1–130,265–429} and HIV-1 Gag_133–278) (i.e., HTLV/HIV-N) was constructed using a 1,116-bp gene fragment with engineered 5′ HindIII and 3′ XcmI sites. The HTLV-1 Gag plasmid containing the HIV-1 CA CTD (HTLV-1 Gag_{1–264,345–429} and HIV-1 Gag_279–362) (HTLV/HIV-C) was engineered using a 791-bp gene fragment with engineered 5′ PstI and 3′ BamHI sites. The HIV-1 Gag plasmid containing the HTLV-1 CA NTD (HIV-1 Gag_{1–132,279–500} and HTLV-1 Gag_131–265) (i.e., HIV/HIV-C) was engineered with an 801-bp gene fragment containing 5′ HindIII and 3′ BssHII sites. Finally, the HIV-1 Gag plasmid containing the HTLV-1 CA CTD (HIV-1 Gag_{1–278,362–500} and HTLV-1 Gag_131–264) (i.e., HIV/HIV-N) was engineered with a 464-bp gene fragment containing 5′ SbfI and 3′ BstEII sites.

The HIV-1 WM, HIV/HIV-C WM, and HTLV/HIV-C WM constructs were engineered through PCR mutagenesis of the HIV-1 CA CTD residues W316 and M317 to A316 and A317. The HIV-1 WM, HIV/HIV-C WM, and HTLV/HIV-C WM gag sequences were then inserted into peYFP vectors using 5′ HindIII and 3′ BamHI sites.

Chimeric gag sequence was engineered into the HIV-1-based vector HIV-1 MIG (59), a single-cycle env-negative HIV-1 NL4-3 reporter vector expressing both the mCherry and GFP fluorescence proteins by swapping the HIV-1 CA with the HTLV-1 CA (HIV/HTLV CA MIG) using overlap extension PCR cloning. The HTLV-1 CA was amplified with primers containing the HIV-1 CA cleavage sites at the 5′ and 3′ ends (Fig. 1D). The pNL4-3 MIG was then mutagenized using the QuikChange II XL site-directed mutagenesis kit and the PCR product as primers.

Subcellular localization and confocal microscopy.

HeLa cells were cultured in six-well plates on no. 1.5 standard glass coverslips coated with poly-l-lysine, and experiments were performed as previously described (25). Cells were transiently transfected with eYFP-tagged Gag and untagged Gag expression plasmids at a 1:4 weight ratio using GenJet, version II (SignaGen, Gaithersburg, MD). Subcellular localization of Gag-eYFP was detected using a Zeiss LSM 700 confocal laser scanning microscope using a Plan-Apochromat 63×/1.40-numeric-aperture (NA) oil objective at 1.2× zoom (Carl Zeiss, Oberkochen, Germany).

Confocal cell image analysis.

The degree of Gag assembly in a punctate pattern was quantified by determining the area associated with punctate fluorescence and expressing this as a percentage of the total fluorescence area. The percentage of fluorescent area associated with puncta was analyzed using ImageJ software (NIH, Bethesda, MD). Cell perimeter was determined using Actin Red 555 staining, and total fluorescence area in pixels was calculated within the cell perimeter. Fluorescence area associated with puncta in pixels was determined using ImageJ “Analyze Particles” with a circularity setting of 0.95 to 1.00. Percent fluorescence associated with puncta was determined by dividing the area associated with puncta by total fluorescence area. Percent fluorescence area associated with puncta was calculated for at least 15 individual cells from 3 individual preparations. Percent punctate fluorescence for each cell value was set relative to the mean HTLV-1 WT value of all 3 replicates.

Colocalization was evaluated in 15 single cells using the ImageJ plugin Coloc 2, and the Pearson's coefficient was recorded. Cell perimeters were determined using Acti-stain 670. Cell outlines were created in ImageJ with the wand tool. Colocalization analyses were performed in a minimum of 15 cells from 3 individual replicates.

Gag copackaging into VLPs.

Fluorescent VLPs were produced through cotransfection of eYFP-tagged and mCherry-tagged Gag plasmids with untagged Gag plasmids at a 1:1:1:1 weight ratio in 293T/17 cells using GenJet, version II. Supernatant was filtered through 0.2-μm filters after 48 h and diluted in phosphate-buffered saline (PBS) at 1:20. Diluted supernatant (100 μl) was added to Lab-Tek II 8-well chambered coverglasses (Thermo Scientific, Waltham, MA) coated in poly-l-lysine. Polybrene (EMD Millipore, Darmstadt, Germany) was added at a 1:100 ratio. Images of fluorescent VLPs were acquired using a Zeiss LSM 700 with a Plan-Apochromat 100×/1.40-NA oil objective at 2.5× zoom (Carl Zeiss) and stored as 8-bit images for later analysis with routines written in IDL 8.3 (Excelis Visual Information Solutions, Inc., Boulder, CO).

Gag copackaging analysis.

The fluorescent color of VLPs was assessed to determine whether each VLP contained detectable amounts of both eYFP- and mCherry-labeled Gags. VLPs were termed “mixed” if their dual-color fluorescence intensities were statistically incompatible with a null hypothesis where the VLP contains only mCherry- or eYFP-labeled Gags. Briefly, the population of all VLPs in the dual-color confocal images was defined by searching the images for intensity peaks that were statistically distinct from the image background. For each VLP, the fluorescence intensity fraction, f = F_mCh/(F_mCh + F_eYFP), was calculated, where F_mCh and F_eYFP are the fluorescent intensities for the mCherry and the eYFP signal of the VLP integrated over an area of size ∼2ω₀. The radial beam waist, ω₀, of the confocal point spread function was determined from the spatial autocorrelation of the VLP images. The uncertainty in the intensity fraction, Δf, was calculated based on Poisson counting statistics of the original dual-color intensity images. This experimentally determined intensity fraction was then compared to the value of f for VLPs containing only mCherry- or eYFP-labeled Gag proteins. VLPs were scored as mixed if the comparison resulted in a statistical difference with a 5-sigma threshold. The percent copackaged VLPs was determined from a population sample containing several hundred VLPs.

PM binding fluorescence z-scan measurements.

Dual-color z-scans were performed in HeLa cells cultured in 8-well plates (Nunc, Inc.) and transiently cotransfected with Gag-eYFP and fluorescent protein mCherry plasmids at a 1:2 weight ratio using GenJet, version I (SignaGen), 12 h before measurement. The cell medium was changed to PBS immediately before measurement. Cellular z-scans were performed on a two-photon microscope as described previously using a vertical scan distance of 24 μm over a period of 5 s (60).

A z-scan measurement consists of an axial scan of the two-photon beam focus through the cytoplasm of the cell (Fig. 2). The fluorescently labeled Gag excited during the scan gives rise to a z-scan intensity profile (Fig. 2B and D), which provides information about the axial distribution of the Gag-eYFP at the scan location (61). Fitting of the z-scan intensity profile separates the fluorescence contributions from the cytoplasmic and the PM-bound protein populations (35, 36). For a cell without Gag-eYFP at the PM (Fig. 4A), the z-scan fit recovers a purely cytoplasmic component (Fig. 4B); for a cell with Gag-eYFP at the PM (Fig. 4C), the z-scan fit recovers a PM component in addition to the cytoplasmic Gag-eYFP component.

Analysis of the dual-color z-scan intensity profiles was performed in two steps. First, the intensity profile of the red channel (mCherry) was corrected for spectral cross talk from the eYFP signal and fit to determine the location and thickness of the cytoplasmic volume occupied by mCherry (61). Second, the intensity trace of eYFP was fit to a model that allows Gag-eYFP to be located at the ventral and dorsal PM and in the cytoplasm with the cytoplasmic location and thickness determined by the fit of the mCherry z-scan intensity profile.

Cells were selected for z-scan measurements by exposure to epifluorescence light to check the protein expression level. Repeated z-scan measurements at each location were used to estimate the standard deviation of the Gag-eYFP PM intensities. Cells were deemed PM positive for fluorescence if PM-associated Gag-eYFP was detectable with a 1% statistical confidence level at any of up to four z-scan locations. Cells were also deemed PM positive for fluorescence if Gag-eYFP puncta were visible in epifluorescence imaging. Conversely, a PM-negative cell had no visible fluorescent puncta, and the z-scans of at least four locations failed to identify PM-bound Gag. The cytoplasmic Gag-eYFP concentration was determined for each measured cell as previously described (35). Because determination of concentration from thin sections is challenging (60), we removed cells from the final data set if all z-scans passed through locations with cytoplasmic thickness of <2 μm.

Immunoblot analysis.

293T/17 cells were transfected with HA-tagged Gag plasmids or MIG plasmids using GenJet, version II. After 48 h, supernatant was filtered through 0.2-μm filters, and VLPs or viruses were concentrated by ultracentrifugation at 150,000 × g for 90 min through an 8% OptiPrep cushion (Sigma-Aldrich) in a 50.2 Ti rotor (Beckman). VLP or virus pellets were lysed in PBS with 1.0% Triton X-100. Cells were washed with PBS before they were lysed with 1.0% Triton X-100 in PBS. Lysates were clarified via centrifugation at 16,000 × g for 1 h to obtain the soluble fraction.

Lysates and supernatants were subjected to electrophoresis on ExpressPlus PAGE 4 to 20% gels (GenScript, Piscataway, NJ) and transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA). HA-tagged Gag was detected with anti-HA antibody diluted 1:1,000 in PBS (16B12; BioLegend, San Diego, CA). Gag proteins were detected with 1:1,000 rabbit polyclonal anti-p17 (lot 130088; AIDS Reagent Program, Division of AIDS, NIAID, NIH, from Paul Spearman and Lingmei Ding) or 1:1,000 mouse monoclonal anti-HIV-1 p24 (24-4; Santa Cruz Biotechnology, Dallas, TX). Gamma tubulin was detected with 1:1,000 anti-tubulin mouse monoclonal antibody (GTU-88; Sigma-Aldrich). Membranes were washed before incubation with either 1:10,000 horseradish peroxidase-conjugated goat anti-mouse IgG (Thermo Fisher) or 1:5,000 horseradish peroxidase-conjugated donkey anti-rabbit (Jackson ImmunoResearch, West Grove, PA). Protein expression was detected with a ChemiDoc Touch system (Bio-Rad, Hercules, CA) and analyzed with Image Lab, version 5.2.1 (Bio-Rad). Relative HA-tagged Gag expression levels were calculated by normalizing to the tubulin density and setting relative to WT HIV-1 or HTLV-1 Gag.

Production of virus and VLPs for thin-section TEM and cryo-TEM.

VLPs were produced by cotransfection of 293T/17 cells with untagged Gag plasmids and CMV-ENV at a 10:1 weight ratio using GenJet, version II. Viruses were produced by cotransfection of 293T/17 cells with the HIV-1 vector pHIV-1-MIG and pHCMV-G at a 10:1 ratio. After 24 to 48 h, the supernatant was collected and filtered through 0.2-μm filters. Virus particles or VLPs were prepared as previously described by ultracentrifugation on an Opti-Prep gradient (25).

Thin-section TEM of 293T/17 cells producing VLPs.

Thin-section TEM was performed as previously described (25). Sections were stained and examined with a JEOL 1200 EX II transmission electron microscope (JEOL, Tokyo, Japan).

Cryo-TEM of VLPs and virus particles.

EM grids used were Lacey/Formvar 300 mesh, Lacey carbon 300 mesh (EMS, Hatfield, PA), R2/2 Holey carbon 200 mesh, and Multi A Holey carbon 200 mesh (Quantifoil, Germany). Grids were glow-discharged and loaded on an FEI MarkIII Vitrobot system (FEI Company, Hillsboro, OR). Purified and concentrated VLP suspension (3 to 4 μl) was loaded on the carbon side of a grid and manually blotted before being plunge-frozen in ultracooled liquid ethane. Grids were imaged on a Tecnai F30 FEG transmission microscope (FEI Company, Hillsboro, OR) operating at 300 kV. Images were taken at ×49,300 magnification at an electron dose of ∼25 electrons/Ų and −4 to −8 μm defocus values using a Gatan 4,000 by 4,000 charge-coupled device (CCD) camera (Gatan Inc., Pleasanton, CA). VLP and virus particle diameters were measured using ImageJ software. Two perpendicular diameters were measured and averaged for each particle. The morphologies of all particles were qualitatively characterized by two independent reviewers as curved lattice, curved bilayer; flat lattice, flat bilayer; flat lattice, curved bilayer; or no lattice.

Production of viral stocks for qPCR and infectivity assays.

Vector virus stocks of the HIV-1 vector parent (i.e., HIV-1 MIG WT) and mutant expressing the Gag chimera (i.e., HIV/HTLV CA MIG) were produced by cotransfecting 293T/17 cells with HIV-1 vector plasmid and pHCMV-G at a 10:1 ratio using GenJet, version II. Vector virus supernatants were collected 48 h posttransfection, filtered through a 0.2-μm filter, treated with DNase I for 2 h at 37°C, aliquoted, and stored at −80°C.

Determination of copy number with RT-qPCR and infectivity.

Viral RNA was isolated from virus stocks in duplicate using a High Pure viral RNA kit (Roche, Indianapolis, IN). cDNA was synthesized from viral RNA using an iScript cDNA synthesis kit (Bio-Rad) and stored at −20°C. SYBR green qPCR reagents (Bio-Rad) were used to detect an amplicon in HIV-1 p19, and the threshold cycle (C_T) was used to determine copy number by interpolation to an HIV-1 vector plasmid standard curve. To determine infectivity, MAGI cells were plated in 24-well plates and infected with viral supernatant volumes containing approximately 1.5 × 10⁶, 3.0 × 10⁶, or 5.0 × 10⁶ copies of RNA. Medium was replaced 24 h posttransfection. Cells were washed and collected for flow cytometry 72 h postinfection. Viral infectivity was determined by plotting the percentage of infected cells (mCherry⁺ or GFP⁺) over total cells.