Live-Cell Coimaging of the Genomic RNAs and Gag Proteins of Two Lentiviruses

Cell culture, transfections, and plasmid constructions.Cells were cultured in Dulbecco's modified Eagles medium with 10% fetal calf serum. DNA transfection was carried out with calcium phosphate coprecipitation (293T, HeLa, and CrFK cells) or Fugene 6 (Roche Applied Science) for COS-7 cells.

Correct plasmid constructions were all confirmed by sequencing. To generate MS2-Cherry, we replaced GFP between BamHI and ClaI in pMS2-GFP with an amplicon generated using template pRSET-B-mCherry (52) with sense primer 5′ BamMS2Cherry (5′ATATGGATCCGCCAACCATGGTGAGCAAGGGC-3′) and antisense primer 3′ ClaNLSCherry (5′-ATATATCGATTTATACCTTTCTCTTCTTTTTTGGCTTGTACAGCTCGTCCATGCC-3′; restriction sites are underlined). For pCiRev, an FIV 34TF10 Rev cDNA was synthesized by reverse transcription-PCR (RT-PCR) with primers 5′-TACTCGAGAATGGCAGAAGGATTTGCAGC-3′ and 5′-AATCTAGACTAGTCCATAAGCATTCTTTC-3′ and inserted into pCi (Invitrogen) between XhoI and XbaI. pMS2-YFP and pMS2-GFP were kindly provided by R. Singer (18).

pFIVψ-24 was generated from a previously described FIV vector system (32, 43, 44) by inserting 24- of the 19-nt MS2 stem-loops (18) by blunt end ligation of a BamHI-BglII fragment into the AvrII site of FIV transfer vector pCT26 (Fig. 1). The presence and correct orientation of all of the stem-loops were verified by sequencing. FIVΔψ-24 was created by replacing a BglII-SphI fragment containing the human cytomegalovirus (CMV) immediate-early gene promoter, R repeat, U5, leader sequences, and 308 bp of Gag with the corresponding BglII-SphI fragment from pGINWF-G0 (28), resulting in the net deletion of all FIV Gag sequences.

• Open in new tab
• Download powerpoint

FIG. 1.

System components. Three fluorescent protein chimeras (YFP, GFP, and mCherry) with the MS2 coat protein were used. pFIVψ-24 is derived from pCT26, an FIV transfer vector RNA that is packaged with wild-type efficiency (28, 29). pFIVψ-24 contains 24 MS2 phage coat binding sites to which MS2-XFP fusion proteins can bind. Each stem-loop is 19 nt long. In pFIVΔψ-24, the proximal 230 nt of gag are deleted, which abolishes encapsidation (28, 29). pFP93 is the FIV packaging plasmid used to derive the pFIVgag-XFP plasmids, where XFP is CFP, mCherry, or DsRed. pHIVψ-24 contains the same 24 MS2 stem-loops as pFIVΔψ-24. In pHIVΔψ-24, the packaging signal (SL1 to SL4) was deleted. pHIVgag-CFP was derived by replacing pol with CFP, in frame with gag, in pLP1 (Invitrogen). IRES, internal ribosome entry site.

In FIVψ-24-CTE and FIVΔψ-24-CTE, the RRE was replaced by four copies of the constitutive transport element from MMPV (5). First, one copy of a CTE was made by annealing and extending the following two oligonucleotides: 5′ Xho/Nhe-CTE (5′-ATATCTCGAGCTAGCACCTCCCCTGTGAGCTAGACTGG ACAGCCAATGACGGGTAAGAGAGTGACATTTCTCACTAACCTAAGA CAGGAGGGCCGTCAAAGCTA-3′) and 3′ Sal-CTE (5′-ATATGTCGACAC ACATCCCTCGGAGGCTGCGCCTGTCTTAGGTTGGAGTGATACATTT CTTGTCACTATTACCCGTCATTGGATTAGGCAGTAGCTTTGACGGCCCTCCTG-3′). The fragment was digested with XhoI and SalI and cloned into pBSSK digested with XhoI, resulting in pBS-CTE1x. The CTE was further multimerized by ligating the CTE-XhoI/SalI fragment into pBS-CTE1x linearized with XhoI until pBS-CTE4x was obtained. A 750-bp fragment from pBS-CTE4x with a blunted KpnI site and an SmaI site was cloned into the SphI and XmnI sites of FIVψ-24 or FIVΔψ-24, replacing the RRE with four copies of the CTE, resulting in FIVψ-24-CTE and FIVΔψ-24-CTE, respectively.

The previously described FIV packaging plasmid pFP93 (33, 47) was used to generate pFIVgag-CFP (where CFP is cyan fluorescent protein), pFIVgag-Cherry, and pFIVgag-dsRed (where dsRed is Discosoma sp. red fluorescent protein [RFP]). In pFP93 the polymerase gene was replaced with either CFP or dsRed by fusing the fluorescent proteins to the C-terminal amino acids of Gag and introducing the following linker: SGGSGGTGGRDPPVAT. pFIVgag-Cherry was created by replacing CFP in pFIVgag-CFP with Cherry (BamHI-EcoRI fragment from pRSET-B-mCherry). pFIVgagG2A-CFP was generated by PCR replacing the myristoyl acceptor glycine at position two in Gag with alanine (Gag_G2A). pCT5efs is a derivative of pCT5, a full-length FIV proviral clone enabled for human cell expression by the replacement of the 5′ U3 with the human CMV immediate-early promoter (43, 44); for pCT5efs, a frameshift mutation was generated in env to abrogate Env cytopathicity (29). pCT5gag-CFP (diagramed in Fig. S1 in the supplemental material) was constructed by replacing an EcoRI-SapI fragment (nt 1872 to 5220) containing pol with the corresponding EcoRI-SapI fragment (nt 1872 to 3336) from pFIVgag-CFP containing the CFP gene.

pHIVψ-24 was constructed by inserting the 24 MS2 stem-loops as a BamHI-BglII fragment into pHR′-CMVLacZ (57) between the encapsidation signal and the RRE (Fig. 1). For pHIVΔψ-24 the packaging signal (SL1 to SL4) and 40 bp of the gag gene fragment were deleted. The HIV-1 packaging plasmid pLP1 (Invitrogen) was used to construct pHIVgag-CFP. First, PCR primers 5′-GAGACCATCAATGAGGAAGC-3′ and 3′-ATATCTGCAGCTCGAGCTAGCTTGGTGACGAGGGGTCGCTGCC-5′ were used to introduce unique NheI and XhoI sites at the 3′ end of Gag, followed by insertion of a CFP amplicon generated from eCFP-N1 with 5′ NheCFP (5′-ATATGCTAGCGGAGGTGGAGGTGTGAGCAAGGGCGAGGAG-′) and 3′ XhoCFP (5′-ATATCTCGAGTTACTTGTACAGCTCGTCCATG-3′; restriction sites are underlined). Then the CFP initiator methionine was deleted, and an SGGGG linker was incorporated between Gag and CFP. pHIVgagG2A-CFP was created by PCR replacing the glycine at position two in Gag with an alanine.

pCiYU2-Tat101 expresses the 101-amino-acid Tat. Random-hexamer-primed cDNA was made from RNA isolated from macrophages infected with the YU2 strain. A Tat cDNA amplicon was generated with the primers 5′ XbaYU2 (5′-ATATTCTAGAATGGAGCCAGTAGATCCTAACCTAG-3′) and 3′ NotYU2-101 (5′-TAAATGCGGCCGCTCCGTAGGTCGACCCAGATAATTGC-3′) and cloned into the XbaI and NotI sites of pCI. The correct construct was confirmed by sequencing. CMVΔR8.9 expresses HIV-1 Gag-Pol, Tat, and Rev (57). pLP2 (Invitrogen) expresses HIV-1 Rev.

Virion production and Western blotting.Viral particles were purified as described in Kemler et al. (28). Forty-eight hours after 293T cells were transfected, medium was centrifuged at 828 × g for 10 min in a Sorvall tabletop centrifuge, and supernatant was filtered through a 0.45-μm-pore-size filter. The virus was then pelleted through a 20% sucrose cushion for 2 h at 112,700 × g in an SW28 rotor. The viral pellet was resuspended in TNE buffer (10 mM Tris, pH 7.5, 100 mM NaCl, 1 mM EDTA). Concentrated virus was lysed in 5× radioimmunoprecipitation assay (RIPA) buffer (150 mM NaCl, 1% NP-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) containing protease inhibitors (Complete Mini; Roche). Cells were lysed in 1× RIPA buffer. Eight micrograms of cell lysate and equal amounts of viral lysates were separated on a reducing 10% SDS-polyacrylamide gel and transferred to a polyvinylidene difluoride membrane (Immobilon-P; Millipore). The membranes were blocked in 0.2% I-block (Tropix)-0.1% Tween and then probed with a monoclonal anti-GFP antibody (1:5,000; BD Biosciences), a polyclonal anti-dsRed antibody (1:16,000; BD Biosciences), or with a monoclonal antibody (MAb) to HIV-1 p24 (AG3.0; NIH AIDS Research and Reference Reagent Program). For secondary antibodies, horseradish peroxidase (HRP)-conjugated goat anti-mouse or HRP-conjugated goat anti-rabbit antibody (1:1,000; Calbiochem) was used. Bands were visualized with Lumi-Light detection reagent (Roche).

Indirect immunofluorescence and live-cell imaging.HeLa cells were transfected with 1 μg of pFIVgag-dsRed, fixed with 4% paraformaldehyde for 10 min, washed with phosphate-buffered saline (PBS), quenched with 50 mM NH₄Cl for 10 min, washed with PBS, permeabilized with 0.2% Triton X-100 for 10 min, washed, and blocked with 1% bovine serum albumin (BSA)-PBS for 15 min. Cells were then incubated with monoclonal antibodies against CD63, Lamp2, EEA1 (1:100; BD Biosciences), or DDX6 (1:150; Bethyl Laboratories) for 1 h at 37°C. After cells were washed with PBS, they were blocked again and then incubated with an anti-mouse Alexa Fluor 488-conjugated antibody (1:1,000; Molecular Probes) for 30 min. Slides were analyzed on a Zeiss LSM 510 laser scanning confocal microscope using a ×100 (oil) objective. MAb PAK3-2C1 (Custom Monoclonal Antibodies International, Sacramento, CA) was used for indirect immunofluorescence for wild-type Gag. For live-cell imaging, HeLa cells or COS-7 cells were seeded in Lab-Tek glass-bottom chamber slides (Nunc) and observed 20 or 48 h posttransfection on a Zeiss LSM 510 confocal laser scanning microscope equipped with an Axiovert 200 M inverted platform using a Plan-Apochromat ×100 (1.4 numerical aperture) oil immersion objective. CFP was excited with an Argon laser at 458 nm, and emission was detected with a band-pass filter of 475 to 525 nm. mCherry was excited with a helium-neon laser at 543 nm, and emission was detected with a band-pass filter of 585 to 615 nm. YFP was excited with an Argon laser at 514 nm, and emission was detected with a band-pass filter of 530 to 600 nm.

In situ RNA hybridization.Cells were seeded on glass coverslips, and in situ RNA hybridizations were performed as described by Chartrand et al. (7). The 50-nt-long oligonucleotide used was complementary to a sequence between the MS2 stem-loops (5′-TtTCTAGGCAAtTAGGTACCTtAGGATCTAAtGAACCCGGGAATACtGCA-3′) and contained five amino-modified C6 deoxyribosylthymines (dTs; indicated by lowercase Ts) which were subsequently coupled to Cy3 fluorophore (Amersham), according to the manufacturer's protocol. An oligonucleotide complementary to a sequence in the FIV pol gene (5′-CAGCCtTGTGGTAGACtACACCACACAAAtCTCCTTCCtGGTCCCGCAtT-3′) was used. The oligonucleotides were purified by high-performance liquid chromatography (HPLC). Subsequent to the in situ RNA hybridization immunofluorescence was performed with MAb 414 (Covance) recognizing the nuclear pore complex and the anti-mouse Alexa Fluor 488-conjugated secondary antibody (Molecular Probes). Cells were imaged on a Zeiss LSM 510 confocal laser scanning microscope.

Labeling FIV genomic RNA with MS2-GFP does not inhibit packaging into particles.To investigate trafficking of the genomic RNA and Gag in living cells, we devised systems to fluorescently label both molecules. For the genomic RNA, we adapted the methods of Singer and colleagues (3, 18). GFP, YFP, or mCherry was fused to the RNA-binding coat protein of the RNA phage MS2 (indicated as MS2-XFP, where XFP is YFP, GFP, or mCherry) (Fig. 1A). A nuclear localization sequence (NLS) confines the MS2-XFP protein to the nucleus unless it associates with a cytoplasmically trafficked mRNA. To enable specific binding of MS2-XFPs to full-length HIV-1 or FIV lentiviral transfer vector RNAs, 24 tandem copies of the MS2 phage coat protein RNA binding element, a 19-nt stem-and-loop structure, were inserted downstream of the viral major splice donor (Fig. 1B and E). This positioning allows for exclusive visualization of unspliced genomic RNA.

Of the two lentiviruses, we began with FIV, using FIVψ-24 (Fig. 1B). To simultaneously track Gag, we fused either CFP, dsRed, or mCherry to the Gag C terminus. The resulting expression constructs, pFIVgag-CFP, pFIVgag-RFP, and pFIVgag-mCherry, function purely to supply Gag and Rev in trans and lack all genomic RNA encapsidation signals as well as all cis-acting elements needed for reverse transcription and integration (Fig. 1D). To determine whether Gag-XFP fusion proteins were incorporated into virions, we expressed them alone or together with unlabeled Gag-Pol (pFP93) (Fig. 1C). As shown in Fig. 2A for FIV Gag-RFP, a single 90-kDa band corresponding to the predicted size of the intact fusion protein was detected in cell lysates (data not shown) and in purified virions (lane 3). If wild-type Gag-Pol was coexpressed with Gag-RFP, protease cleavage occurred, resulting in detection of the 90-kDa precursor and smaller RFP-containing bands (Fig. 2A, lane 4).

• Open in new tab
• Download powerpoint

FIG. 2.

Encapsidation of MS2-GFP-labeled genomes. Encapsidation of MS2-GFP requires the viral RNA encapsidation signal and is proportional to genomic RNA input. (A) 293T cells were transfected with 10 μg of pFIVgag-dsRed, 10 μg of pCT5efs (proviral clone, expressing wild-type Gag-Pol), or 5 μg of each. Purified virions were lysed and separated on a 10% SDS-PAGE gel and probed with a mouse anti-GFP antibody. (B) 293T cells were transfected with the indicated amounts of DNA (μg). Purified virions were lysed and separated on a 10% SDS-PAGE gel and probed with a mouse anti-GFP antibody (lanes 1 to 5) or a rabbit anti-dsRed antibody (lanes 6 to 10).

To determine whether genomic RNAs with bound MS2-GFP molecules are incorporated into viral particles, we coexpressed Gag-RFP, the FIVψ-24 genomic RNA, and MS2-GFP in 293T cells. Immunoblotting of purified virions with an anti-GFP antibody showed that the 45-kDa MS2-GFP protein was incorporated into particles (Fig. 2B, lanes 2 and 3). In addition, MS2-GFP incorporation was increased when the amount of FIVψ-24 transfer vector was doubled (lane 2 versus lane 3), indicating a quantitative effect of RNA input on encapsidation of MS2-GFP. Importantly, MS2-GFP incorporation into virions was dependent on the viral RNA's having the encapsidation elements. No MS2-GFP incorporation was detected in virions produced with a packaging signal mutant (28) transfer vector, FIVΔψ-24 (Fig. 2B lanes 4 and 5; the region deleted from FIVψ-24 is shown in Fig. 1B). This was consistently the case despite equivalent budding of virions (lanes 7 to 10) and equivalent intracellular expression of MS2-GFP (data not shown). Thus, a Gag-XFP fusion protein is capable of specifically packaging viral RNA with bound MS2-XFP. Furthermore, these data show that the C-terminal positioning of a fluorescent protein (thus, in proximity to NC), does not compromise RNA packaging properties of Gag.

Subcellular distribution of FIV Gag in live cells.We then investigated the distribution of FIV Gag-XFPs in living cells. In HeLa (Fig. 3A) and COS-7 cells (Fig. 3B, C, and D), Gag-CFP had a dispersed and nonhomogenous cytoplasmic distribution, and it was readily seen to accumulate at the plasma membrane in HeLa and COS-7 cells (arrows). This subcellular distribution is similar to the reported distribution for an HIV-1 Gag-CFP fusion protein (14, 26). The same results were obtained with Gag-mCherry (Fig. 3D), demonstrating that the XFP fused to Gag does not influence its trafficking. Importantly, the subcellular distribution of fluorescently protein-labeled Gag was equivalent to the distribution of unmodified wild-type FIV Gag (Fig. 3E). High-resolution images were obtained and clearly identified Gag at the plasma membrane (Fig. 3D, enlarged lower image). Discrete plasma membrane accumulations were more prominent in COS-7 (Fig. 3B, C, and D) or 293T cells (see Fig. S4 in the supplemental material) than in HeLa cells. However, if the amount of transfected pFIVgag-CFP was doubled in HeLa cells, abundant Gag-CFP was also readily seen at the plasma membrane in these cells (Fig. 3A, inset).

• Open in new tab
• Download powerpoint

FIG. 3.

FIV Gag localization. (A to D) Subcellular localization of FIV Gag-CFP or FIV Gag-mCherry. Solid arrows indicate plasma membrane localization of Gag-CFP or Gag-Cherry; hollow arrowheads indicate nuclear envelope localizations of Gag. Live-cell confocal imaging was performed 20 h or 48 h after transfection with 2 μg (or 4 μg; inset photo) of pFIVgag-CFP (A and B), CT5gag-CFP, in which the Gag-expressing RNA is also encapsidated (see Fig. S1 in the supplemental material for a diagram) (C), or pFP93gag-Cherry (D). (E) Distribution of unmodified FIV Gag in fixed cells. HeLa cells were transfected with pFP93 and fixed, and indirect immunofluorescence was performed with mouse anti-FIV Gag (PAK3-2C1). (F to H) HeLa cells were transfected with pFIVgag-RFP, fixed, and stained with antibodies against CD63 (F), Lamp2 (G), or EEA1 (H). Scale bars, 10 μm and 2 μm (enlargements).

Strikingly, FIV Gag was also seen consistently at the nuclear envelope in all cell types (Fig. 3B, open arrowheads and enlargement). This finding is analyzed further below. Note also that, as shown in Fig. 3C, producing Gag from the RNA that is encapsidated rather than producing it in trans did not influence the Gag-CFP pattern. Here, Gag-CFP was expressed within a full-length FIV proviral DNA in which only pol is replaced by CFP (see Fig. S1 in the supplemental material); data in subsequent figures show that genomes with an intact ψ [ψ(+)] supplied in trans also do not affect Gag locations while the presence of Gag markedly affects ψ(+) but not ψ-deficient [ψ(−)] RNA locations.

Numerous studies have found that HIV-1 Gag localizes to intracellular compartments containing the late endosomal markers CD63 and the lysosome-associated membrane protein Lamp2 (41, 42, 53). Various interpretations have been put forth, including an endosomal assembly model, but recent studies have counter-proposed that such endosomal localization is mainly due to internalization by endocytosis or phagocytosis of Gag from the plasma membrane (13, 17, 26, 55). We asked whether FIV Gag traffics similarly to HIV-1. In HeLa cells a substantial fraction of FIV Gag-RFP colocalized with CD63-containing vesicles (Fig. 3F) as well as with Lamp2-containing membranes (Fig. 3G). In clear contrast, Gag-RFP did not localize to EEA1-containing early endosomes (Fig. 3H). These distributions are consistent with those recorded by other investigators for HIV-1 Gag (41, 42, 53).

MS2-XFP labeling in live cells reveals FIV genomic RNA accumulation at the nuclear envelope.We next investigated cellular trafficking of FIV genomic RNA using the MS2 system (Fig. 4). When expressed alone, MS2-YFP was confined to the nucleus in a diffusely homogeneous distribution, as expected because of its NLS (Fig. 4A). An FIV Rev expression plasmid was constructed and validated biochemically (see Fig. S2 in the supplemental material). MS2-YFP remained unchanged in distribution when coexpressed with either FIV Rev (Fig. 4B) or a transfer vector, FIVψ (Fig. 4C). In contrast, coexpression of MS2-YFP and an FIV transfer vector RNA having 24 tandem copies of the phage coat protein-binding element (FIVψ-24) resulted in a clearly different, nonhomogeneous (although still exclusively nuclear) distribution, generally with a speckled appearance consistent with focal intranuclear cotrafficking of genomic RNA and MS2-YFP (Fig. 4D).

• Open in new tab
• Download powerpoint

FIG. 4.

Subcellular localization of MS2-YFP-labeled genomic FIV RNA in live cells. HeLa (A to G and K) or COS-7 (H to J and L) cells were transfected with various combinations of plasmids, live-cell confocal imaging was performed, and representative images are shown: pMS2-YFP alone (A); pMS2-YFP and pCiRev, which expresses functional FIV Rev (see Fig. S2 in the supplemental material) (B); pMS2-YFP and pFIVψ (C); pMS2-YFP and pFIVψ-24 (D); pMS2-YFP, pFIVψ-24, and pCiRev (E); pMS2-YFP, pFIVψ-24, and pFP93 (F); and pMS2-YFP, pFIVΔψ-24, and pFP93 (G). COS-7 cells were transfected with the following: pMS2-mCherry, pFIVψ-24, and pFP93 (H); pMS2-mCherry and FIVψ-24-CTE, where the RRE was replaced with the CTE of MPMV (I); MS2-mCherry and FIVΔψ-24-CTE (J). Arrows indicate accumulation of FIV genomic RNA at the nuclear envelope. (K and L) pGFP-Nup153, pMS2-mCherry, pFIVψ-24, and pFP93 were transfected into HeLa cells (K) or COS-7 cells (L). In both K and L, the following are shown: GFP-Nup153 (i), MS2-Cherry-labeled FIV genomic RNA (ii), and merge (iii). In panel L, frame iv is a Nomarski image. Magnifications are equivalent in panels A to K. Scale bar, 10 μm.

The distribution of the viral genomic RNA changed drastically when FIV Rev was coexpressed (Fig. 4E). Labeled RNA was exported from the nucleus and migrated to the cytoplasm, where it distributed throughout the cytoplasmic space in a speckled, punctate pattern. Notably, the genomic RNA was seen to accumulate around the nucleus in a rim-like pattern, suggesting a nuclear envelope location (e.g., Fig. 4E). Colocalization with the nucleoporin Nup153 confirmed this (Fig. 4K and L). Cells consistent with a range of temporal progressions in the Rev-dependent RNA export pathway were observed. For example, in many cells the majority of the MS2-YFP protein had been exported from the nucleus with trafficking RNA, and the remaining intranuclear MS2-YFP was highly altered as well, acquiring a speckled appearance. A similar RNA distribution was observed when the source of Rev was the FIV packaging plasmid pFP93, which expresses both Rev and unlabeled Gag-Pol (Fig. 4F). In COS-7 cells MS2-mCherry-labeled viral genomic RNA was also exported from the nucleus and accumulated in the cytoplasm in a speckled, punctate pattern (Fig. 4H), showing that the trafficking of the RNA was independent of the cell type or the XFP fused to the MS2 protein. In addition, note that the distribution of MS2-YFP-labeled RNA lacking the packaging signal (FIVΔψ-24) was the same as the distribution of wild-type RNA (Fig. 4G). This RNA migrated to the cytoplasm and also accumulated around the nuclear envelope.

Nuclear export of FIV genomic RNA was also induced when the RRE was replaced by four copies of the constitutive transport element (CTE) from Mason Pfizer monkey virus (MPMV) (5). The MS2 label was distributed throughout the cytoplasm and also accumulated around the nuclear envelope (Fig. 4I), indicating that the latter property was not Rev specific but common to both Rev- and CTE-mediated viral RNA export pathways. As with Rev/RRE export, the distribution of a CTE-dependent RNA lacking the packaging signal was similar to one containing it (Fig. 4J).

To verify that the distribution of the MS2 labeling reveals true RNA localization and not artifact induced by binding of the MS2-YFP protein, we performed RNA in situ hybridization experiments. HeLa cells were transfected with pFIVψ-24 in the presence or absence of pCiRev and then fixed and hybridized with a Cy3-labeled oligonucleotide probe that recognizes the RNA sequence linking MS2 binding loops (Fig. 5). Consistent with the results in living cells, FIVψ-24 RNA was confined to the nucleus in the absence of Rev (Fig. 5A). If Rev was coexpressed, FIVψ-24 RNA was exported from the nucleus and accumulated in the cytoplasm in a speckled pattern (Fig. 5B and C). Note that if cells were treated with RNase H before the hybridization with the Cy3 oligonucleotide probe, no signal was detected, demonstrating that the hybridization observed was due to binding of the probe to RNA and not DNA (Fig. 5D). Untransfected cells (Fig. 5F) or cells transfected with FP93 (Fig. 5E), which lacks the stem-loops, resulted in background staining. In contrast to results of the live-cell experiments, we did not observe cells where the nucleus was completely devoid of RNA, which is likely due to limiting amounts of MS2-XFP protein molecules versus genomic RNAs. The MS2-XFP-RNA complexes have a predicted protein-to-RNA stoichiometry of 48:1 (MS2 binds as a dimer). In addition, Fusco et al. showed that on average 33 MS2 proteins are bound to one mRNA (18).

• Open in new tab
• Download powerpoint

FIG. 5.

Detection of FIV genomic RNA by in situ RNA hybridization. HeLa cells were transfected with pFIVψ-24 alone (A and D) or together with pCiRev (B and C), transfected with pFP93 (E) or left untransfected (F). Cells were fixed and hybridized with a Cy3-labeled oligonucleotide probe recognizing the sequence between the MS2 stem-loops. In panel D cells were treated with RNase H before the hybridization. In panel G HeLa cells were transfected with pCT5efs (a full-length FIV proviral clone with one mutation, a frameshift in env that abrogates Env cytopathicity). Cells were fixed and hybridized with a Cy3-labeled oligonucleotide probe recognizing a sequence in the pol gene, followed by indirect immunofluorescence with MAb 414, an antibody detecting the nuclear pore complex. Scale bar, 10 μm.

In situ hybridization was also done directly for full-length wild-type FIV genomic RNA. HeLa cells were transfected with a full-length FIV proviral clone, fixed, and hybridized with a Cy3 oligonucleotide probe recognizing a sequence within the pol gene. Corroborating the MS2 experiments, this showed genomic RNA accumulation around the nuclear envelope (Fig. 5G). The RNA colocalized with an antibody to nuclear pore complexes, confirming that nuclear envelope accumulation of MS2-XFP-labeled RNA in living cells (Fig. 4) is not a result of artifact induced by bound MS2-XFP.

Colocalization of FIV Gag-CFP and MS2-XFP-labeled genomic RNAs at the plasma membrane requires Gag and an intact encapsidation signal.Having validated the RNA and Gag systems, we combined them to image FIV RNA and Gag simultaneously in live HeLa cells (Fig. 6). In the absence of a viral genome, MS2-YFP was strictly nuclear, as in Fig. 4A to C, whether or not Gag and/or Rev was also present (data not shown). When Gag-CFP, Rev, and MS2-YFP were coexpressed with FIVψ-24 RNA, colocalization was clear and abundant, as were marked shifts in MS2-YFP location (Fig. 6). The labeled RNA trafficked to the cytoplasm and associated with Gag in discrete cytoplasmic locations (Fig. 6A and B).

• Open in new tab
• Download powerpoint

FIG. 6.

Colocalization of FIV Gag-CFP and MS2-YFP-labeled FIV genomic RNA in live cells. HeLa cells were transfected with various combinations of plasmids. Bar, 10 μm. In the absence of a viral RNA, MS2-YFP had the appearance shown in Fig. 4A to C. (A) pFIVgag-CFP, pMS2-YFP, and pFIVψ-24. The nuclear envelope region (box, right panel) is shown enlarged below (scale bar, 5 μm). (B) pFIVgag-CFP, pFP93, pMS2-YFP, and pFIVψ-24. (C and D) pFIVgag-CFP, pFP93, pMS2-YFP, and pFIVΔψ-24. Live-cell confocal imaging was performed 48 h after transfection, and representative images are shown.

Note that substantial colocalization of FIV Gag and MS2-labeled FIV genome RNA occurred at the nuclear envelope (Fig. 6). Focal coaccumulations of the ψ(+) genomic RNA with Gag were also seen in apposition to the cytoplasmic face of the nuclear envelope, perhaps at nuclear pores, where they extended into the perinuclear cytoplasm (Fig. 6A). Fig. S3 and Movie S1 in the supplemental material show time-lapse imaging of an MS2-Cherry-labeled RNA granule forming with Gag-CFP in the proximity of the nuclear envelope, followed by movement away from the nuclear envelope into perinuclear cytoplasm. In the cell shown in Fig. 6A, the majority of the RNA has been exported from the nucleus, but a spectrum of phenotypes consistent with different stages was observed. For example, Fig. 6B shows a cell with substantial amounts of nuclear MS2-YFP fluorescence and less colocalization of Gag-CFP and RNA in the cytoplasm, representing an earlier stage of RNA export. In contrast, when the genomic RNA lacked the encapsidation signal, cytoplasmic colocalization of the two signals was much less (Fig. 6, compare panels A and B with panels C and D).

In HeLa cells, we did not detect abundant steady-state colocalization of FIV Gag-CFP and MS2-labeled RNA at the plasma membrane (Fig. 6). This is consistent with the relatively less frequently observable plasma membrane accumulation of FIV Gag-CFP in HeLa cells demonstrated in Fig. 3A. However, Gag-XFP and genomic RNA colocalization was very evident at the plasma membrane in COS-7 cells (Fig. 7) and 293T cells (see Fig. S4 in the supplemental material). We compared viral genomic RNAs with and without the full encapsidation sequence in these cells. In COS-7 cells transfected with FIV Gag-CFP, MS2-mCherry, and FIVψ-24, prominent accumulation of the FIV RNA at the plasma membrane was observed, and this colocalized with Gag-CFP (Fig. 7D and E; see also Fig. S4A). In important contrast, FIVΔψ-24 RNA did not traffic to the plasma membrane although Gag-CFP accumulated there (Fig. 7F; see also Fig. S4B). This result was corroborated by quantitation of the different distribution patterns observed of FIV Gag-CFP and genomic RNA with (ψ) or without (Δψ) the packaging signal (Fig. 8). Gag and ψ-intact RNA colocalized at the plasma membrane in 17.1% of cells (Fig. 8C). In contrast, when the viral RNA was ψ(−), Gag still reached the plasma membrane in a similar percentage of cells (14.5%), but colocalization with the RNA was never seen (Fig. 8D). None of the other distribution patterns exhibited a difference between FIVψ-24 and FIVΔψ-24 RNA (Fig. 8A, B, E, F, and G). Thus, plasma membrane trafficking of the RNA genome requires that it contain the packaging signal. This encapsidation signal-dependent trafficking also required the presence of Gag, since it was not observed when Rev protein alone was coexpressed with the RNA (Fig. 7C). Coexpression of MS2-mCherry with the FIV genomic RNA but no Gag or Rev resulted in nuclear redistribution of the RNA (Fig. 7B), as seen in HeLa cells (Fig. 4D).

• Open in new tab
• Download powerpoint

FIG. 7.

Plasma membrane accumulation of FIV genomic RNA depends on an intact packaging signal in COS-7 cells. Cells were transfected with the following: pMS2-mCherry (A); pMS2-mCherry and pFIVψ-24 (B); pMS2-mCherry, pFIVψ-24, and pciRev (C); pFIVgag-CFP, pFP93, pMS2-mCherry, and pFIVψ-24 (D); pFIVgag-CFP, pMS2-mCherry, and pFIVψ-24 (E); or pFIVgag-CFP, pMS2-mCherry and pFIVΔψ-24 (F). Live-cell confocal imaging was then performed. Arrows indicate plasma membrane accumulation of FIV Gag-CFP (E and F) and labeled genomic FIV RNA (E). Micrographs shown are representative of independently performed experiments. Scale bar, 10 μm.

• Open in new tab
• Download powerpoint

FIG. 8.

Quantitation of the colocalization of FIV Gag-CFP and MS2-mCherry-labeled FIV genomic RNA. COS-7 cells were transfected with pFIVgag-CFP, pMS2-mCherry, and pFIVψ-24 or pFIVΔψ-24. Twenty hours after transfection, cells were observed under the microscope. Cells showing the following phenotypes were counted: only RNA was expressed (A); only Gag was expressed (B); Gag and RNA are in the cytoplasm and also at the plasma membrane (C); Gag is in the cytoplasm and at the plasma membrane, and RNA is in the cytoplasm but not at the plasma membrane (D); both Gag and RNA are in the cytoplasm (E); both Gag and RNA are in the cytoplasm and around the nucleus (F); Gag is in the cytoplasm, and RNA is around the nucleus (G). Results are from two independent experiments. n = 356 cells for FIVψ-24; n = 308 cells for FIVΔψ-24.

Tracking the HIV-1 genomic RNA: targeting of the plasma membrane and endosomal vesicles requires Gag and an intact encapsidation sequence.To gain comparative insight, we also applied the MS2 system to HIV-1. The 24 stem-loops were similarly inserted in an HIV-1 vector, and a packaging-deficient version was created by deleting the four adjacent stem-loops (SL1 to SL4) upstream of the 5′ portion of the gag gene and 40 bp of the gag gene (Fig. 1E). To visualize HIV-1 Gag, we fused CFP to its C terminus (Fig. 1F).

Similar to FIV Gag-CFP, HIV-1 Gag-CFP is incorporated into viral particles (data not shown). Also similar to FIV, MS2-GFP incorporation was dependent on an intact HIV-1 encapsidation element (see Fig. S5, lanes 3 and 4, in the supplemental material). An HIV-1 RNA lacking the 24 MS2 stem-loops (HIV-lacZ) did not enable MS2-GFP particle incorporation (see Fig. S5, lane 2). To investigate the subcellular distribution of Gag-CFP in live cells, we transfected HeLa or COS-7 cells with pHIVgag-CFP together with plasmids encoding HIV-1 Tat and Rev (Fig. 9A and C) or unlabeled HIV-1 Gag-Pol, Tat, and Rev (Fig. 9B and D). HIV-1 Gag-CFP showed a punctate cytoplasmic distribution and strong accumulation at the plasma membrane in both HeLa and COS-7 cells (Fig. 9A to D). This is in contrast to FIV Gag-CFP, which showed less intense signal at the plasma membrane in HeLa cells (compare to Fig. 3A). In COS-7 cells HIV-1 Gag-CFP accumulated along the plasma membrane and in an intracellular compartment (Fig. 9C and D). These intracellular HIV-1 Gag-CFP puncta may reflect Gag-CFP endocytosed from the plasma membrane (26).

• Open in new tab
• Download powerpoint

FIG. 9.

Intracellular localization of HIV Gag-CFP or MS2-mCherry-labeled HIV-1 genomic RNA in HeLa and COS-7 cells. HeLa cells (A, B, and E to H) or COS-7 cells (C, D, and I to L) were transfected with various combinations of plasmids: pHIVgag-CFP, pCiTat, and Rev (pLP2) (A and C); pHIVgag-CFP and pCMVΔR8.9 (B and D); pMS2-mCherry, pHIVψ-24, pCiTat, and Rev (pLP2) (E and I); pMS2-mCherry, pHIVψ-24, and pCMVΔR8.9 (F and J); pMS2-mCherry, pHIVΔψ-24, and pCMVΔR8.9 (G and K); pMS2-Cherry (H and L). Live-cell confocal imaging was performed 20 h after transfection. Arrows in panels C, D, and J point to intracellular puncta, and arrows in panels F and G indicate accumulation of HIV-1 genomic RNA at the nuclear envelope. Representative images are shown. Bar, 10 μm.

We next investigated the distribution of HIV-1 genomic RNA. In both HeLa and COS-7 cells, MS2-mCherry protein labeled the HIV-1 RNA and demonstrated efficient export from the nucleus and migration to the cytoplasm (Fig. 9E to G and I to K), while MS2-mCherry expressed alone was confined to the nucleus (Fig. 9H and L). In HeLa cells (Fig. 9E to G) the HIV-1 RNA showed a dispersed cytoplasmic distribution. In some cells, the RNA could be seen accumulating at the nuclear envelope (Fig. 9E to G, arrowheads), similar to FIV (Fig. 4), but this was considerably less frequently detected than with FIV. Similar overall patterns were observed when only Tat and Rev were present (Fig. 9E) and when unlabeled Gag-Pol was added (Fig. 9F).

In COS-7 cells, the HIV-1 genomic RNA was distributed throughout the cytoplasmic space in the presence of Tat and Rev (Fig. 9I). However, when unlabeled Gag-Pol was additionally provided, many cells accumulated RNA in the cytoplasm within relatively large focal punctate structures (Fig. 9J; compare with I). Importantly, trafficking of HIV-1 genome RNA to these cytoplasmic structures required both Gag and an intact encapsidation signal since RNAs lacking the encapsidation signal were distributed diffusely in the cytoplasm and never accumulated in such puncta (Fig. 9K; compare with J). Note that the COS-7 cell intracellular puncta with HIV-1 RNA (Fig. 9J) resemble closely the intracellular compartment in which HIV-1 Gag-CFP accumulated in these cells (Fig. 9C and D). We therefore investigated if the genomic RNA and CFP-labeled Gag colocalized. Indeed, when COS-7 cells were cotransfected with pHIVgag-CFP, pCMVΔR8.9, pMS2-mCherry, and packaging signal-intact HIVψ-24 RNA, many cells showed accumulation of Gag-CFP and the RNA in the same punctate cytoplasmic structures (Fig. 10A). In contrast, this colocalization was not observed with HIV-1 RNA with a deletion of ψ (Fig. 10C; here the Gag puncta still form, but without the RNA). Quantification of these differences is shown in Fig. 10D. Like FIV, the intracellular HIV-1 Gag puncta colocalize with the late endosomal markers CD63 and Lamp2; however, they did not localize with the P body marker DDX6 (RCK/p54) (data not shown). Taken together, these specific ψ(+) genomic RNA internalizations show that Gag and genomic RNA are internalized together from the plasma membrane into endosomes, and they show that the MS2 system can track the genomic RNA faithfully in the cell, even after the plasma membrane anchoring event. They show that it is not just Gag but also nascent virions with encapsidated genomes that become internalized. Both HIV-1 Gag-CFP and the labeled RNA could be seen coaccumulated at the plasma membrane (Fig. 10B, arrowhead). However, in contrast to findings in FIV, we could not detect coaccumulation of HIV-1 Gag and RNA at the nuclear envelope, suggesting either that this does not occur or that its kinetics are too rapid to allow visualization.

• Open in new tab
• Download powerpoint

FIG. 10.

HIV Gag-CFP and MS2-mCherry-labeled HIV-1 genomic RNA colocalize in intracellular puncta in COS-7 cells. COS-7 cells were transfected with various combinations of plasmids. (A and B) pHIVgag-CFP, pCMVΔR8.9, pMS2-mCherry, and pHIVψ-24. Arrows in panel A point to intracellular puncta, and arrowheads in panel B point to plasma membrane accumulation. (C) pHIVgag-CFP, pCMVΔR8.9, pMS2-mCherry, and pHIVΔψ-24. Live-cell confocal imaging was performed 20 h after transfection. Scale bar, 10 μm. (D) Quantitation of colocalization of HIV-1 Gag-CFP and MS2-mCherry-labeled HIV-1 genomic RNA in intracellular puncta in COS-7 cells. A total of 381 puncta in 34 cells were evaluated for ψ(+) RNA, and 324 puncta in 29 cells were evaluated for ψ(−) RNA. Puncta were scored for containing both Gag and RNA, exclusively Gag (RNA absent), or exclusively RNA (Gag absent). The average of two independent experiments is shown.

To further understand the molecular requirements, we tested Gag mutants that cannot be myristoylated (mutant G2A, which we introduced into both FIV and HIV-1 Gag proteins) and therefore are unable to accumulate at the plasma membrane (20). As shown in Fig. 11B, myristoylation-deficient FIV Gag_G2A-CFP and FIVψ-24 genomic RNA did colocalize around the nucleus, as with the normally myristoylated Gag. However, FIV Gag_G2A-CFP did not accumulate at the plasma membrane, and it was not able to bring the FIVψ-24 RNA to the plasma membrane. HIV-1 Gag_G2A-CFP did not accumulate at the plasma membrane or in cytoplasmic puncta, and the HIV-1 genomic RNA was distributed diffusely in the cytoplasm (compare Fig. 11C and D). These results are further evidence that lentiviral genomic RNAs are trafficked to (or secured at) the plasma membrane by Gag and that Gag and the genomic RNA recycle together.

• Open in new tab
• Download powerpoint

FIG. 11.

FIV and HIV MS2-mCherry-labeled genomic RNAs do not accumulate at the plasma membrane in the presence of a myristoylation-defective Gag. COS-7 cells were transfected with the following: FIV Gag-CFP, MS2-mCherry, and FIVψ-24 (A); FIV Gag_G2A-CFP (myristoyl acceptor glycine at position two changed to alanine), MS2-Cherry, and FIVψ-24 (B); HIV Gag-CFP, MS2-Cherry, and HIVψ-24 (C); HIV Gag_G2A-CFP (myristoyl acceptor glycine at position two changed to alanine), MS2-mCherry, and HIVψ-24 (D). Live-cell confocal imaging was performed 20 h after transfection. Results shown are representative images. Scale bars, 5 μm (A and B) and 10 μm (C and D).