INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The complicated life cycle of human immunodeficiency virus type 1 (HIV-1) involves a dynamic interplay between viral and host factors. The ability of the virus to access the nucleus via active nuclear import is mediated by both viral and host proteins. This function distinguishes HIV-1 from oncoretroviruses, which rely on nuclear envelope disassembly during cell division for nuclear import (36, 53). HIV-1's primary targets in vivo are CD4⁺ T cells, most of which are resting, and terminally differentiated tissue macrophages, rendering the ability of HIV-1 to enter the nucleus of nondividing cells critical for viral pathogenesis and disease progression (41). Hence, there is great interest in understanding nuclear import of the HIV-1 preintegration complex (PIC), which consists of viral DNA and certain viral and host proteins (8, 42). Several viral proteins, including the matrix protein, are implicated in PIC nuclear localization (7, 28, 33, 60).

Matrix, a key component of the HIV-1 PIC, contributes to nuclear localization of the PIC and plays other crucial roles throughout the HIV-1 life cycle (6, 15, 24, 38, 50, 67). Matrix is a 17-kDa myristoylated protein derived from the extreme N terminus of the Gag precursor polyprotein (Pr55^Gag). Nuclear import of matrix is believed to be mediated by its two nuclear localization signals (NLSs) that resemble the canonical simian virus 40 T-antigen NLS (7, 60). However, the role of these NLSs and the mechanism of nuclear import are matters of debate (7, 21-23, 28, 52, 60). Recently, a nuclear export activity was suggested for matrix (16) which could override its NLS, facilitating nuclear export of unspliced viral RNA and cytoplasmic retention of Pr55^Gag during virion production. Late in the viral life cycle, prior to cleavage as part of Pr55^Gag, matrix orchestrates virion assembly and release by targeting the Gag proteins to the host cell membrane. The Gag proteins recruit viral genomic RNA, as well as viral and host cell proteins, into the newly budding virion (17, 57, 68). Soon after assembly, following its incorporation into the virion, Pr55^Gag is cleaved by the HIV-1 protease to generate mature p17 matrix (MA), p24 capsid (CA), p7 nucleocapsid, and p6.

As obligate intracellular parasites, viruses often recruit help from host cell factors. Interactions of matrix with cellular proteins have previously been described. These include HO3, a putative tRNA synthetase (40), HEED, the human homolog of mouse eed (51), translation elongation factor 1-alpha (13), and hIF2, a human homolog of bacterial translation initiation factor 2 (64). However, it remains unclear how these host proteins contribute to matrix's role in viral replication, and our understanding of how matrix operates is still incomplete.

To elucidate the mechanisms of matrix function, we looked for new cellular partners for matrix using a two-hybrid screen with matrix as bait and a human activated T-cell cDNA library as prey. We isolated a putative partial open reading frame (ORF) of unknown function that we designated virion-associated nuclear shuttling protein, or VAN. VAN is evolutionarily highly conserved in vertebrates, and its transcript is present in all human tissues tested. Here, we describe the characterization of the matrix-VAN interaction, VAN's nuclear-cytoplasmic shuttling capacity, and its potential role in the viral life cycle.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Two-hybrid screen. The two-hybrid screen was performed in a triple-reporter strain, Saccharomyces cerevisiae MaV103, bearing GAL promoter-dependent URA3, HIS3, and lacZ genes (MATa ade2-101 leu2-3,112 trp11 his3-200 gal4 gal80 pGAL1:HIS3 at lys2 pGAL1:lacZ at unknown locus SPAL10:URA3) (59). As bait we used the matrix protein derived from HIV-1_YU2, a macrophagetropic primary isolate derived from the brain of a patient suffering from AIDS dementia (63), fused at its N terminus to the GAL4 DNA-binding domain (GAL4DBD) of pPC97 (11). This vector contains the GAL4DBD and LEU2 selectable marker. Full-length YU2 MA (pKG104), JR-CSF MA, NL4-3 MA, AD MA, YU2 Vpr, YU2 Pr55^Gag, LAI Pr55^Gag, LAI Vpr, and LAI Nef were generated via standard PCR using the corresponding proviral DNAs as templates and appropriate primers. The PCR products were digested and cloned in frame into the SalI/BglII or SalI/NotI sites of pPC97. The inserts were sequenced, and protein expression was confirmed by Western analysis. The two-hybrid bait plasmid used for the initial screen was pKG104. The activation domain library, Alala2 (kindly provided by Joshua La Baer, Massachusetts General Hospital, Boston, Mass.) contains cDNA derived from phytohemagglutinin (PHA)-activated T cells fused to the coding region of the GAL4 transactivation domain (GAL4TAD) of pPC86, which has the TRP1 selectable marker.

Identification of full-length cDNA. Full-length VAN cDNA, pCMVSport-VAN (pKG148), was isolated by screening a human leukocyte cDNA library (SuperScript human leukocyte cDNA library; Life Technologies, Gaithersburg, Md.) using the GeneTrapper cDNA positive selection system (Life Technologies catalog no. 10421-014) as instructed by the manufacturer. The cDNAs identified were tested by Southern blotting for the ability to hybridize to a VAN-C-derived probe. Of the positive cDNAs, the longest was 3.3 kb (pKG148). It contains 262 bp of untranslated leader sequence followed by an ATG in a relatively good Kozak context with an in-frame stop codon nine nucleotides upstream of the ATG. The cDNA encoded a predicted ORF of 637 amino acids corresponding to a protein of ~72 kDa. The gene has been mapped to the long arm of chromosome 5 by the NCBI Unigene database (http://www.ncbi.nlm.nih.gov/UniGene/index.html).

Generation of plasmids, cell lines, and antibodies. An AvrII/NotI fragment of full-length VAN cDNA was cut out of pKG148 and cloned into the NheI/NotI sites of pCEP4 (Invitrogen) to generate pKG123 (pCEP4-VAN), the mammalian VAN expression plasmid. A second PCR product corresponding to the two-hybrid interacting partial cDNA was generated with KpnI and XhoI sites at its ends (using primers M27pCEP4-N [AATCGGGGTACCACCATGGGCGGCCGCAGAATTCGGCACGAGAGT] and M27pCEP4-C [TACCGCTCGAGTAGCACTAGTCTCGAGTTT]). The PCR product was then digested with KpnI/XhoI and ligated into the same sites in pCEP4 to generate pKG125 (pCEP4-VAN-C), the VAN-C mammalian expression vector.

GST-MA interaction with VAN-C'. A 1,300-bp fragment obtained from an XhoI/XmnI partial digest of pCMV Sport-VAN was ligated into the pET30c vector backbone (Novagen) digested with XhoI/EcoRV to generate pKG122, the His₆-tagged VAN-C' expression plasmid. The MA insert was excised from pKG104 by digesting it with SalI/NotI. The resulting insert was ligated into pGEX4-T2 (Pharmacia Biotech catalog no. 27-4581-01) also digested with SalI/NotI to generate pKG115, expression vector for glutathione S-transferase (GST)-MA fusion protein. Escherichia coli BL21 cells were transformed with pGEX4T-2, pKG115, and pKG122 and assayed for protein expression upon induction. One hundred milliliters of a 1:50 dilution of an overnight culture started from a single transformant was incubated at 37°C. At A₆₀₀ of ~0.6, 0.1 mM isopropyl--D-thiogalactopyranoside was added to induce protein expression. After induction for 3 h at 37°C, the cells were centrifuged at 8,000 rpm for 10 min and the pellets were frozen at 80°C. The pellets were thawed, and the GST fusion proteins were resuspended in 5 ml of lysis buffer (150 mM NaCl, 5 mM dithiothreitol, 2 mM EDTA, 50 mM Tris [pH 8.0], protease inhibitor cocktail [1,000× protease inhibitor cocktail contains 1 mg of leupeptin, 2 mg of antipain, 10 mg of benzamidine, 10,000 KIU of aprotinin, 1 mg of chymostatin, and 1 mg of pepstatin per ml plus 1 M phenylmethylsulfonyl fluoride], while the VAN-C'-expressing lysates were resuspended in 5 ml of binding buffer (5 mM imidazole, 500 mM NaCl, 20 mM Tris-HCl [pH 7.9], protease inhibitor cocktail). GST- or GST-MA-containing cell lysates were mixed with equal volumes of VAN-C' cell lysates and incubated for 4 h at 4°C; 500 µl of 50% glutathione-Sepharose 4B (Pharmacia Biotech) slurry was added to 500 µl of the lysate and incubated at 4°C for another 2 h. The beads were then washed three times: wash 1 with binding buffer; wash 2 with binding buffer with a final NaCl concentration of 350 mM, 400 mM, or 500 mM; and wash 3 with phosphate-buffered saline (PBS). The beads were resuspended in 1× Laemmli buffer and subjected to standard sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western analysis. Recombinant VAN-C', purified under denaturing conditions (8 M urea) over a nickel chelating column, was used as a size marker in lieu of the VAN-C-containing lysates used for the GST binding assay.

Northern analysis (human and mouse). VAN-C was used as a template to generate a ³²P-labeled probe as previously described (18) to probe a commercially available human multiple-tissue Northern blot (lots 53626 and 54756; Clontech, Palo Alto, Calif.) as instructed by the manufacturer. The lanes were normalized for equal amounts of -actin prior to loading.

Identification of mouse and chicken cDNAs. We used the BLAST algorithm to search the EST (expressed sequence tag) database and identified a number of overlapping mouse ESTs. We ordered four of these clones from the American Type Culture Collection and sequenced them. One of these (GenBank accession no. AA061107, from a mouse testis library) encoded a 2.4-kb insert that shared a high degree of identity with human VAN (hVAN) (P = 3.1e⁵⁵). The BLAST search also identified a chicken sequence in the 5' untranslated region of the reported chicken proto-Ets protein sequence (GenBank accession no. M23688 [61]) that shared a high degree of homology to hVAN. We assume this transcript represents a chimeric cDNA.

Immunofluorescence. HeLa cells (3 × 10⁶) were plated onto coverslips in a six-well plate 18 to 24 h prior to transfection. The cells were transfected with Lipofectamine Plus (Life Technologies) the following day in accordance with the manufacturer's recommendations. At 48 h posttransfection, the coverslips were rinsed in PBS and then fixed in PBS-4% paraformaldehyde (pH 7.2) for 20 min at room temperature. The coverslips were rinsed and permeabilized with PBS-0.2% Triton X-100 for 5 min. Fixed, permeabilized cells were blocked with PBS-5% normal goat serum for 1 h. They were incubated with primary antibody(ies) diluted in the blocking solution overnight at 4°C, washed in PBS three times for 15 min each, and then incubated with secondary antibody(ies) also diluted in the blocking solution for 45 min at room temperature. Coverslips were washed twice in PBS, then stained with 4',6-diamidino-2-phenylindole (DAPI; 1 µg/ml; Sigma) in PBS for 5', washed again in PBS, patted dry between two Kimwipes, mounted in Prolong mounting medium (Molecular Probes), and allowed to set at 4°C overnight prior to microscopic analysis. The antibodies were diluted as follows: affinity-purified anti-VAN, 1:50; anti-HIV IgIV (a human polyclonal antibody stock derived from HIV-1-seropositive individuals), 1:2,000; monoclonal antibody HA.11 (clone 16B12; Covance catalog no. MMS-101P), 1:2,000; anti--Gal (Life Technologies catalog no. 19929-017), 1:20; anti-p65 RelA (NFB; Santa Cruz Biotechnology catalog no. sc-109), 1:50; Cy3-conjugated affinity-purified goat anti-rabbit immunoglobulin G (IgG; Jackson ImmunoResearch Laboratories catalog no. 111-165-045), 1:5,000; anti-mouse IgG-fluorescein (Boehringer Mannheim catalog no. 1814222), 1:500; fluorescein-conjugated anti-rabbit IgG (Vector Laboratories catalog no. FI-1000), 1:100; and goat anti-human IgG-fluorescein conjugate (Pierce catalog no. 31530), 1:100.

Confocal microscopy. For the subcellular localization of VAN and leptomycin B (LMB) studies, samples were acquired with the Bio-Rad MRC 600 confocal laser scanning microscope system utilizing COMOS software (version 6.03) on a Compaq Deskpro with a Pentium processor (Intel). A krypton-argon laser (Bio-Rad) with excitation at wavelengths of 488 and 568 nm was used to obtain optical sections. Narrow-band emission filters (525 and 605 nm) were used to eliminate channel cross talk, and the pinhole aperture was set to obtain 0.5-µm z-plane sections (as determined by full-width half-maximum intensity values). Slides were imaged with a 60× oil immersion planar apochromatic objective lens (numerical aperture, 1.4) through a Nikon Optiphot microscope. All images shown were representative of the entire field.

Immunoblot analysis. Typically, 10⁷ cells were lysed in 100 µl of Laemmli buffer. Ten microliters of the cell lysate was run on a 12% SDS protein gel. Proteins were transferred to an Immobilon-P transfer membrane (Millipore) at 300 mA for 1.5 h at 4°C. The membrane was blocked with PBS-5% milk-0.05% Tween 20 and incubated with primary antibody at the appropriate concentration at 4°C overnight. The blot was washed three times for 15 min each in PBS-0.05% Tween 20 and then incubated with a 1:10,000 dilution of the appropriate horseradish peroxidase-conjugated secondary antibody (Amersham) for 45 min. The blot was again washed three times, and antibody binding was detected by ECL (enhanced chemiluminescence) immunoblotting detection reagents (Amersham). Antibodies were used at the following dilutions: murine anti-p17 HIV-1 monoclonal antibody (Immune Diagnostics, Inc., Bedford, Mass.), 1:1,000; affinity-purified rabbit anti-VAN, 1:100; anti-gp41 (T-32 hybridoma supernatant [a kind gift from Patricia Earl, NIH]; the epitope spans residues 597 to 613 of gp41 from HIV-1_IIIB) 1:500; mouse anti-human HIV p24 monoclonal antibody (Chemicon International, Inc., Temecula, Calif.), 1:100; and monoclonal antiactin (clone AC-40; Sigma), 1:100.

Preparation and characterization of viral stocks. All viral stocks were generated by transfecting 293T cells using a commercially available calcium phosphate DNA precipitation kit (Stratagene, La Jolla, Calif.) according to the manufacturer's recommendation. HIV-1_MN, HIV-1_MNNEF, and HIV-1_NL4-3 were grown in chronically infected H-9 cells as described elsewhere (4). HIV-1_MNNEF contains a deletion starting at nucleotide 8805 and resuming at nucleotide 9064, deleting the start of Nef at nucleotide 8811. For viral growth curves, the proviral DNAs used were pLAI and pNL4-3, a kind gift from Keith Peden. Transfections were performed in 100-mM-diameter dishes at 30 to 50% confluency, and virus-containing supernatants were harvested and concentrated by centrifugation through a 20% sucrose cushion 48 h posttransfection. Supernatants were then normalized for p24 content by enzyme-linked immunosorbent assay (ELISA) (Coulter [Miami, Fla.] ELISA kit), and stored at 80°C. Recombinant vaccinia viruses, vp1287 (Pr55^Gag), vp1290 (MA), and vp1289 (CA) were obtained from Virogenetics Corporation; vsc8 was a kind gift from Bernie Moss, NIH (10).

HIV-1 infection and viral growth curves. HeLa T4 transfectants (10⁶) were plated in each six-well plate. Each well was infected with 80 ng of p24 equivalents from HIV-1_NL4-3- or HIV-1_LAI-containing 293T cell supernatant for 3 h, after which the wells were washed six times with serum-free medium; 2 ml of fresh culture medium (Dulbecco modified Eagle medium 10% fetal calf serum 1% Pen-Strep, 2% L-glutamine) was added per well. To monitor viral replication, 200 µl was taken from each well at time zero and at 2, 4, 6, and 8 days postinfection and replaced with 200 µl of fresh culture medium. The supernatants were then analyzed by ELISA (Coulter ELISA kit) to determine the amount of p24 present as a measure of viral replication over an 8-day time course.

VAN in virions. Virion preparations normalized for total protein content (determined by Lowry assay) were digested in 10 mM Tris-Cl (pH 8.0)-1 mM EDTA for 18 h at 37°C either with or without subtilisin (1 mg/ml; Boehringer Mannheim, Indianapolis, Ind.) (48, 49). The reactions were stopped with phenylmethylsulfonyl fluoride (5 µg/ml, final concentration), and the treated virions were recovered by centrifugation through a 20% sucrose pad in PBS (Life Technologies) at 25,000 rpm for 1 h in an SW28Ti rotor (Beckman Instruments, Inc., Fullerton, Calif.). Treated virions were lysed in gel-loading buffer, applied to a 4 to 20% Tris-glycine SDS-polyacrylamide gel (Novex, San Diego, Calif.), and subjected to SDS-PAGE. Microvesicles were prepared and treated in a similar manner (3). For semiquantitative immunoblot analysis, the VAN signal from known amounts of HIV-1_MN (8 × 10⁹ particles by p24 concentration) was compared to that of a dilution series of known amounts of full-length recombinant VAN fusion protein immobilized onto a polyvinylidene membrane (quantitated by amino acid analysis carried out on a Beckman System 6300 after hydrolysis with 6 N HCl as previously described [33a]).

VAN-C in virions. 293T cells were plated (1.5 × 10⁶/10-cm-diameter petri dish) 18 to 24 h prior to transfection. The cells were cotransfected with VAN-C and proviral DNA plasmid (pHXB2/pAD) or empty vector, using a previously described method (2). At 48 h posttransfection, supernatants were harvested from the various cell cultures, and virions were purified as described elsewhere (9). Each fraction was tested for p24 antigen content to identify where the purified virus peaked; 20 ml of each fraction was then loaded onto a 15% denaturing polyacrylamide gel and analyzed by immunoblotting for the presence of matrix and VAN.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

VAN, a new matrix partner. To identify factors that interact with HIV-1 matrix and thereby mediate its functions, we performed a yeast two-hybrid screen. We fused the matrix protein from HIV-1_YU2, a macrophagetropic, primary patient isolate, to the C terminus of the GAL4DBD as bait. The library screened was derived from PHA-activated T-cell cDNA fused to the GAL4TAD. We identified five clones capable of activating transcription in a matrix-dependent manner. All five encoded a single partial cDNA corresponding to a putative partial ORF of unknown function (Fig. 1a) (GenBank accession no. D30755; recently isolated as Naf1 [Nef-associated factor 1] [27]) (see Discussion).

View larger version (40K): [in this window] [in a new window]   FIG. 1.   Matrix-VAN interaction. (a) Matrix interacts with VAN in the yeast two-hybrid system. In each case, four independent transformants were picked and tested for -Gal activity. The first cDNA listed is fused to the GAL4DBD (in plasmid pPC97), while the second is fused to the GAL4TAD (in plasmid pPC86). 1, controls a to d (left to right) containing the following cDNA inserts: none/none, human retinoblastoma/E2F1, rat c-Fos/mouse c-Jun, and full-length GAL4/none, respectively; 2, none/VAN-C; 3, YU2 MA/none; 4, YU2 MA/VAN-C; 5, NL4-3 MA/VAN-C; 6, LAI Nef/VAN-C; 7, YU2 Vpr/VAN-C. (b) In vitro binding of GST-MA fusion protein to His6-tagged VAN-C' is stable to 500 mM NaCl. Bacterially expressed GST-MA was immobilized on glutathione-agarose beads. Equal amounts of His6-tagged VAN-C'-containing bacterial lysates were incubated with GST and GST-MA immobilized on beads. The beads were washed, and the proteins that remained associated with the beads were analyzed by immunoblotting with an anti-His tag antibody. Recombinant VAN-C', purified over a nickel chelating column under denaturing conditions, was used as a size marker in lane 1; its position is indicated by the bar on the left. The concentration of NaCl in each wash is indicated at the bottom.

View larger version (33K): [in this window] [in a new window]   FIG. 2.   Sequence comparison. At the top are schematic representations of full-length VAN (amino acids 1 to 637), VAN-C (amino acids 312 to 637), and VAN-C' (amino acids 468 to 637). Yellow regions highlight predicted coiled-coil domains, each red bar shows the putative NLS, blue bars depict putative NESs, and each purple bar shows the His6 tag. The consensus leucine-rich NES is shown in blue. Critical residues are the hydrophobic residues indicated as , usually leucine, at positions 1 and 3 (numbered from right to left as indicated) and additional hydrophobic residues at positions 6 and 7 and positions 9 to 11. The hydrophobic residues are interspersed with charged residues, serines or threonines marked as "X". The arrowheads point to residues that were mutagenized to the sequence ASA. SV40, simian virus 40.

VAN binds GST-MA in vitro. To confirm the matrix-VAN protein-protein interaction, we used in vitro binding assays with a bacterially expressed GST-MA fusion protein and a smaller C terminal segment of VAN tagged with His₆ (VAN-C' [Fig. 2]). Bacterial lysates containing GST alone or GST-MA were tested in a GST pull-down assay. While VAN-C' binding to GST-MA was stable to 500 mM NaCl, GST alone bound very weakly to VAN-C' and only at lower salt concentrations (Fig. 1b), confirming that HIV-1 matrix binds VAN in vitro.

Sequence features and homologs of hVAN. Sequence analysis of the putative hVAN ORF revealed a protein enriched in acidic amino acid residues (12% glutamic acid), with four putative leucine zippers, and several predicted regions of extensive coiled-coil structure (~45% of the VAN ORF [Fig. 2]) indicative of a protein-protein interaction motif. A search of public genomic sequence databases showed that the VAN gene is located on chromosome 5q31-q33 (58). Further BLAST searches (1) with full-length human sequences identified partial mouse and chicken homologs (>70% amino acid sequence identity) to hVAN (Fig. 3), suggesting evolutionary conservation across the three species.

View larger version (60K): [in this window] [in a new window]   FIG. 3.   Amino acid sequence of hVAN aligned with the mouse ABIN1 (mABIN1), partial mouse (mVAN), and chicken (cVAN) sequences. Residues highlighted in black show identity; residues highlighted in gray show similarity. GenBank accession numbers: hVAN (previously described as Naf1), AJ011896; mouse cDNA, AAO61107; mouse ABIN1, AJ24778; chicken proto-Ets, M23688.

Expression of VAN in mammalian tissues. VAN mRNA was present as a single, fairly abundant 3.1-kb band in Northern blot analysis with poly(A)⁺ RNAs derived from a variety of human tissues (Fig. 4a). VAN protein expression was also examined in a number of cell lines and primary cells, as well as peripheral blood mononuclear cells (PBMC), by immunoblot analysis with anti-hVAN antibody. While resting T cells expressed low amounts of VAN, activated CD4⁺ and CD8⁺ T cells and T-cell lines (Jurkat and H-9) showed high levels of VAN expression. However, primary monocytes and in vitro-derived monocyte-differentiated macrophages showed only low levels of protein expression similar to those seen in HeLa (cervical carcinoma) and HEK293 cells (Fig. 4b). Immunoblot analysis of protein expression in a number of mouse tissues using anti-hVAN antibody also showed a ~70-kDa band, which was abundant in heart and skeletal muscle and expressed at lower levels in thymus, liver, kidney, brain, and intestinal tract (data not shown).

View larger version (40K): [in this window] [in a new window]   FIG. 4.   VAN expression. (a) VAN mRNA expression in several human tissues, determined by Northern blot analysis of polyadenylated RNA from human tissues using a probe specific for VAN mRNA. The size of the VAN transcript is 3.1 kb. (b) Expression of VAN protein in human cell lines and primary cells. Cell lysates from 1 million cells were analyzed by immunoblotting using an anti-VAN antibody. Lane 1, HeLa T4 cells transfected with pCEP4 (empty vector); lane 2, HeLa T4 cells transfected with VAN; lane 3, Jurkat (CD4+ T-cell line); lane 4, H-9 (CD4+ T-cell line); lane 5, Epstein-Barr virus-transformed B-cell line; lane 6, monocytes; lane 7, monocytes differentiated into macrophages in vitro; lane 8, resting CD4+ T cells; lane 9, PHA-activated CD4+ T cells; lane 10, resting PBMC; lane 11, activated PBMC; lane 12, activated CD8+ T cells.

Localization of VAN. To determine VAN's intracellular location, we examined HeLa cells overexpressing full-length VAN by indirect immunofluorescence using anti-hVAN antibody. The cells showed punctate, predominantly cytoplasmic staining (Fig. 5a, image 2). VAN did not colocalize with markers for endoplasmic reticulum, Golgi bodies, mitochondria, actin, or microtubules (data not shown). Untransfected cells showed a similar staining pattern, though the intensity of the signal was fainter. In contrast to the full-length cytoplasmic protein, overexpressed VAN-C was localized primarily in the nucleus (Fig. 5a, image 3).

View larger version (204K): [in this window] [in a new window]   FIG. 5.   Intracellular localization of VAN. (a) HeLa cells overexpressing VAN or VAN-C. HeLa cells were fixed, permeabilized, and stained with an anti-VAN antibody or with secondary antibody alone, as indicated. The lower row shows the phase image corresponding to each labeled image in the top row. 1, HeLa T4 cells; 2, HeLa T4 cells transfected with VAN; 3, HeLa T4 cells transfected with VAN-C. (b) HeLa cells treated with LMB. HeLa cells were incubated with 5 nM LMB for 6 h prior to fixation and staining. The top row shows cells stained with anti-VAN; the bottom row shows the corresponding phase images. Similar results were obtained when cells were treated with different concentrations of LMB (10 nM, 100 nM, and 1 µM) for 2 h (data not shown). (c) VAN is localized to the nucleus of vaccinia virus-Gag-infected HeLa cells. HeLa cells were infected with vaccinia virus encoding HIV-1 Pr55Gag, matrix, CA, or -Gal. Cells were then fixed, permeabilized, and stained with anti-VAN, anti-HIV-1 (anti-HIV IgIV), and anti--Gal antisera. VAN staining is shown in red; Gag staining is shown in green. Arrows indicate cells in which VAN and Gag are both nuclear localized. The white arrows track one cell across the panel, while the yellow arrows point to a second cell. No detectable background staining was observed with secondary antibody alone (data not shown).

VAN shuttles between the nucleus and cytoplasm in a Crm1-dependent manner. Given the distinct localization of full-length VAN to the cytoplasm and the VAN-C to the nucleus, it is possible that the protein shuttles between the two compartments. The large size of VAN (72 kDa) precludes passive diffusion across the nuclear membrane. The directionality of movement across the nuclear pore depends on the direct or adapter-mediated association between a specific signal on a substrate protein and its receptor on one side of the pore and dissociation on the other side (44). Sequence analysis of VAN revealed four putative leucine-rich NESs (19, 31, 62) distributed throughout the VAN sequence (Fig. 2) that might explain the full-length protein's cytoplasmic localization. NESs of this type are recognized by the protein Crm1/exportin 1 (20, 45). Nuclear export is mediated by complex formation between an NES-containing protein, Crm1, and Ran-GTP. LMB blocks formation of this trimeric complex, preventing nuclear egress of NES-containing proteins and trapping the proteins in the nucleus (65). To determine whether VAN export is Crm1 dependent, we treated HeLa cells with LMB prior to fixation (58) and then examined the localization of endogenous VAN by indirect immunofluorescence using anti-hVAN antibody. While VAN was predominantly cytoplasmic in untreated cells (Fig. 5b, image 2), it was dramatically relocalized to the nucleus in cells treated with LMB (Fig. 5b, image 3). These data suggest that VAN shuttles between the nucleus and the cytoplasm in a Crm1-dependent manner. However, when wild-type (untagged) VAN was overexpressed, it did not relocalize to the nucleus in response to LMB treatment but tended to form discrete cytoplasmic aggregates (data not shown), indicating that some factor involved in nuclear localization of VAN may be limiting.

Relocalization of VAN in cells overexpressing HIV-1 Gag. The shuttling properties of and interaction in vitro between VAN and matrix suggested that the two properties might colocalize intracellularly. To test this hypothesis, we infected HeLa T4 cells with vaccinia virus Gag and examined the infected cells overexpressing HIV-1 Gag by indirect immunofluorescence using antibodies to VAN (endogenous) and Gag (vaccinia virus expressed). While Pr55^Gag showed a diffuse cytoplasmic staining pattern with some membrane staining, matrix was localized primarily to the plasma membrane. However, in a fraction of cells, both Pr55^Gag (25 to 30% of cells) and matrix (5 to 10% of cells) were localized in the nucleus. Interestingly, the VAN signal paralleled the Gag signal; VAN was predominantly cytoplasmic, with some signal in the nucleus in cells in which Gag was cytoplasmic or membrane associated, but it localized to the nucleus of those cells in which Gag was nuclear (Fig. 5c). By contrast, VAN's cytoplasmic localization was unaffected in cells infected with vaccinia virus expressing HIV-1 CA or -Gal, attesting to the specificity of VAN relocalization for Pr55^Gag and matrix. These data independently validate VAN's ability to transit to the nucleus and provide further evidence for in vivo Gag-VAN association.

VAN is incorporated into HIV virions. We investigated whether the matrix-VAN interaction was strong enough to allow VAN incorporation into virions. This cannot be determined by directly examining the virion preparations since they can include contaminating proteins that are present in microvesicles, which copurify with virions, or may include proteins that nonspecifically adhere to the surface of the virus. To measure virion incorporation, we performed a protease protection assay (47) on virus particles produced from chronically HIV-1_MN-infected H-9 cell lines (4). Proteins outside the virion are digested by subtilisin, a nonspecific protease, while intravirion proteins are shielded from digestion by the viral membrane. Previously, such protease treatment was shown to eliminate >95% of contaminating nonviral cellular debris including microvesicles (45-47). The preparations were then either mock or subtilisin digested and centrifuged through a 20% sucrose cushion to separate virions from proteolyzed debris. Mock- and subtilisin-treated virus preparations were then analyzed by immunoblotting for the presence of the extracellular domain of the envelope protein (gp41), matrix, and VAN. The extracellular domain of gp41 was degraded by subtilisin, while matrix and VAN were subtilisin resistant, as expected. Proteins outside the virion were removed, while those inside the viral membrane were protected (Fig. 6a). VAN was readily detectable in the lanes containing mock- and protease-digested HIV-1_MN samples in this and repeat experiments. The VAN signal in the protease-digested lane was somewhat less intense than that in the mock-digested lane, which indicated that some of the VAN associated with virions was removed by subtilisin and therefore may have been present outside the virion. Since the H-9 microvesicle sample showed little signal in the VAN blot (Fig. 6a), it is most likely that the VAN removed by digestion had simply adhered to the exterior of the virions. Similar data were obtained using a Nef deletion variant of HIV-1_MN (Fig. 6a) as well as wild-type HIV-1_NL4-3 (data not shown). Together, these results show that the majority of VAN in virion preparations is inside the virion. Interestingly, the ratio of VAN to matrix appears somewhat higher in the wild-type virus preparation than in the Nef deletion mutant. Thus, while Nef is not required for incorporation, it may increase the efficiency of incorporation of VAN. Using semiquantitative immunoblotting analysis with a full-length VAN fusion protein standard, we estimate that there are ~80 molecules of VAN per wild-type HIV-1_MN virion after subtilisin treatment (data not shown).

View larger version (36K): [in this window] [in a new window]   FIG. 6.   (a) VAN is incorporated into HIV-1 virions. HIV-1 virions produced from two chronically HIV-1MN-infected H-9 cell lines were concentrated by centrifugation through two consecutive sucrose density gradients. The virus-containing pellets were resuspended, and approximately 8 × 109 virions, as estimated by p24 ELISA, were either mock treated [(M)] or treated with subtilisin, a nonspecific, lipid-impervious protease [(S)]. The purified viral pellets were then resuspended in PBS and analyzed by immunoblotting using antibodies to VAN, matrix, and the extracellular domain of gp41 (lanes 3 and 4). Positions of molecular weight markers are shown on the left; the antibody used is shown on the right. Microvesicles derived from mock-infected cells are shown as controls (lanes 5 and 6). Similar data were obtained from Nef-deleted HIV-1MN (lanes 1 and 2) as well as wild-type (wt) HIV-1NL4-3 (data not shown). (b) Purification profile of virions over a Sephacryl S-1000 column. Virions produced from 293T cells cotransfected with VAN-C and an infectious proviral HIV-1AD clone were concentrated by centrifugation through a 20% sucrose cushion. The concentrated virus was resuspended in PBS and run over a Sephacryl S-1000 size exclusion column. The fractions were harvested and analyzed for amount of p24 present in each fraction. (c) Immunoblot analysis of HIV-1-transfected cells. The cell lysates and supernatant fractions obtained from 293T cells transfected with VAN-C and proviral DNA (pHIV-1AD) were subjected to immunoblot analysis for the presence of matrix and VAN-C. 1 to 13 denote the virus-containing fractions as they came off the column. (d) Immunoblot analysis of mock-transfected cells. Lysates and supernatant fractions from 293T cells cotransfected with VAN-C and empty vector were treated as described for panels a and b. Similar results were observed using HIV-1NL4-3 (data not shown).

VAN overexpression inhibits viral replication. Since we found that VAN interacts with matrix in vitro and is packaged into HIV-1 virions in vivo, it seemed likely that perturbing cellular VAN expression levels could affect viral replication. We engineered HeLa T4 populations to overexpress either full-length VAN or VAN-C. Whereas virion association of VAN suggests a positive role for VAN in the viral life cycle, we observed a paradoxical inhibition of replication in cells overexpressing VAN (11-fold) and VAN-C (2-fold) compared to control mock-transfected cell populations (Fig. 7a). This experiment was repeated multiple times using HIV-1_NL4-3 as well as HIV-1_LAI, and with independently derived VAN- and VAN-C-overexpressing HeLa T4 populations, with comparable results. Given the fact that HIV-1 replicates very efficiently in activated T cells, which have high levels of VAN (Fig. 3b), it is unlikely that VAN is inhibitory to the virus at physiologically relevant concentrations. However, this inhibition provides further evidence for in vivo interaction between Gag and VAN.

View larger version (30K): [in this window] [in a new window]   FIG. 7.   Overexpression of VAN in HeLa T4 cells inhibits HIV-1 replication. (a) HIV-1 growth curve in HeLa T4 cells. HeLa T4 cells were either mock transfected or transfected with full-length VAN or VAN-C. The cell lines thus derived were then infected with HIV-1. Supernatants were sampled for p24 production over a course of 6 days. Viral replication is impaired in cells transfected with VAN (circles) and VAN-C (triangles) compared with mock-transfected cells (squares). Each time point represents the mean of three independent infections, and error bars show standard deviations. Similar results were obtained using HIV-1NL4-3. (b) HeLa T4 cell lysates from mock (pCEP4)-transfected (M) and VAN-transfected cells. Numbers above the lanes indicate fold dilutions of cell lysates from VAN overexpressors.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Matrix-VAN interactions. A two-hybrid screen identified VAN, a novel matrix-interacting protein. VAN interacts with both matrix and Pr55^Gag precursor polyprotein when they are expressed as heterologous fusion proteins in yeast (Fig. 1a; Table 1). The specificity of this interaction is corroborated by a GST pull-down assay (Fig. 1b). Moreover, VAN-matrix interaction likely occurs in vivo, based on incorporation of VAN into HIV-1 virions (Fig. 6a), indirect immunofluorescence studies showing Gag-dependent VAN relocalization (Fig. 5c), and inhibition of viral replication in cells overexpressing VAN (Fig. 7a).

Cellular localization. We performed indirect immunofluorescence studies to gain insight into possible functions of VAN. Both native and overexpressed full-length VAN (tagged and untagged) are predominantly cytoplasmic, while VAN-C is nuclear (Fig. 5b, image 2). However, a dramatic shift of endogenous VAN to the nucleus was noted in cells treated with LMB (Fig. 5b, image 3). These data are consistent with the hypothesis that VAN shuttles between the nucleus and the cytoplasm in a Crm1-dependent manner, confirming the presence of a functional NES associated with VAN.

Role in the HIV life cycle. While VAN's cellular functions await further characterization, its interaction with HIV-1 Gag suggests important functions as a host factor involved in the HIV-1 life cycle. HIV-1's primary target cells, such as resting CD4⁺ T cells and macrophages, have low levels of VAN. To compensate for the absence of VAN in target cells, the virus encapsidates VAN when it buds from activated T cells, expressing high levels of the protein. This suggests that VAN may play an important role for HIV-1, possibly in the very early stages of infection establishment in those host cells with very low endogenous levels of VAN. The estimate of ~80 molecules of VAN per virion closely approximates the number of matrix molecules calculated to be associated with the PIC (29). These results are consistent with a role for VAN early in the viral life cycle.

1-33

87-132

View larger version (66K): [in this window] [in a new window]   FIG. 8.   Relevance of the Gag-VAN interaction in the HIV-1 life cycle. The viral life cycle (black arrows) begins with attachment of the virus particle to cell surface receptors and ends with assembly and release of infectious retroviral particles from the cell. VAN (yellow) shuttles between the cytoplasm and the nucleus. However, at steady-state, VAN is localized predominantly in the cytoplasm. In a target cell the virus imports VAN, presumably by an interaction between VAN and Pr55Gag (carat) in the producer cell. VAN interacts with matrix (green circle) early in the viral life cycle, possibly facilitating nuclear targeting of matrix in association with the PIC and, by extension, nuclear import and retention of the HIV-1 PIC. Late in the viral life cycle at the time of assembly, as a Pr55Gag interactor, VAN may also play a role in exporting the newly synthesized unspliced genomic HIV-1 RNA (purple) and/or the nucleus-trapped Pr55Gag from the nucleus.