|Home | About | Journals | Submit | Contact Us | Français|
Cytoplasmic dynein,together with its cofactor dynactin, transports incoming herpes simplex virus type 1 (HSV-1) capsids along microtubules (MT) to the MT-organizing center (MTOC). From the MTOC, capsids move further to the nuclear pore, where the viral genome is released into the nucleoplasm. The small capsid protein VP26 can interact with the dynein light chains Tctex1 (DYNLT1) and rp3 (DYNLT3) and may recruit dynein to the capsid. Therefore, we analyzed nuclear targeting of incoming HSV1-ΔVP26 capsids devoid of VP26 and of HSV1-GFPVP26 capsids expressing a GFPVP26 fusion instead of VP26. To compare the cell entry of different strains, we characterized the inocula with respect to infectivity, viral genome content, protein composition, and particle composition. Preparations with a low particle-to-PFU ratio showed efficient nuclear targeting and were considered to be of higher quality than those containing many defective particles, which were unable to induce plaque formation. When cells were infected with HSV-1 wild type, HSV1-ΔVP26, or HSV1-GFPVP26, viral capsids were transported along MT to the nucleus. Moreover, when dynein function was inhibited by overexpression of the dynactin subunit dynamitin, fewer capsids of HSV-1 wild type, HSV1-ΔVP26, and HSV1-GFPVP26 arrived at the nucleus. Thus, even in the absence of the potential viral dynein receptor VP26, HSV-1 used MT and dynein for efficient nuclear targeting. These data suggest that besides VP26, HSV-1 encodes other receptors for dynein or dynactin.
Virions, subviral particles, and viral proteins are actively transported during cell entry, assembly, and egress (17, 33, 64, 65, 73, 76). Early in infection many viruses use microtubules (MT) for efficient nuclear targeting, either for cytosolic transport of naked viral particles or for transport inside vesicles (16), e.g., herpes simplex virus type 1 (HSV-1) (77), human cytomegalovirus (58), human immunodeficiency virus (48), adenovirus (42, 80), parvoviruses (71, 79), simian virus 40 (61), influenza virus (41), or hepatitis B virus (29).
MT are polar cytoskeletal filaments assembled from α-/β-tubulin with a very dynamic plus-end and a less dynamic minus-end. N-type kinesins carry cargo towards the MT plus-ends and are involved in transport of viral particles to the plasma membrane during egress (37, 39, 66, 87). Cytoplasmic dynein in cooperation with its cofactor dynactin or C-type kinesins catalyzes transport towards MT minus-ends (60, 70, 85). In many cell types, the MT minus-ends are clustered at the MT-organizing center (MTOC), which is often in close proximity to the nucleus. Cytoplasmic dynein is required for nuclear targeting of human immunodeficiency virus reverse transcription complexes (48), and capsids of adenovirus (42, 80), canine parvovirus (79), and HSV-1 (18).
The HSV-1 virion consists of four structural components: a double-stranded DNA genome of 152 kbp, a capsid shell with a diameter of 125 nm, the tegument, and a membranous envelope (67). The major morphological units of the icosahedral capsid are (i) the UL6 master portal at 1 of the 12 vertices, (ii) 11 pentons at the remaining vertices made by five copies of the major capsid protein VP5, and (iii) 150 hexons at the capsid edges and surfaces which contain six copies of each of VP5 and the small capsid protein VP26 (54, 67, 82, 92). Moreover, all herpesviruses are characterized by a protein layer named the tegument that consists of about 20 different proteins and which is located between the envelope and the capsid. The bulk of the tegument is not icosahedrally ordered, but a small portion in the vicinity of the vertices shows icosahedral symmetry (34, 94).
HSV-1 infects many cell types by pH-independent fusion of the viral envelope with the plasma membrane, thereby inserting envelope proteins into the plasma membrane and releasing the tegument and the DNA-containing capsid into the cytosol (31, 55-57, 77). In addition, some cell types are productively infected after entry by endocytosis and fusion with the membrane of an early endosomal compartment (31, 51, 55-57). In epithelial and neuronal cells, the incoming capsids are transported along MT to the MTOC (40, 45, 77, 81). From the MTOC, capsids proceed to nuclear pores, where the viral genomes are injected into the nucleoplasm for viral transcription and replication (3, 59, 77).
Incoming HSV-1 capsids colocalize with dynein and its cofactor dynactin (18, 77). Moreover, blocking dynein function by overexpression of the dynactin subunit dynamitin inhibits capsid transport to the nucleus and immediate-early viral gene expression (18). Cytoplasmic dynein 1 consists of two dynein heavy chains, two dynein intermediate chains, several dynein light intermediate chains, and a series of dynein light chains of three different families (62, 85). The small HSV-1 capsid protein VP26 can interact in vitro with the 14-kDa dynein light chains of the Tctex family (DYNLT1 and DYNLT3) (19). MT-mediated nuclear targeting is considered to be essential for HSV-1 in cells such as neurons, where the presynaptic plasma membrane is located far away from the nucleus (40). However, in epithelial cells, depolymerization of the MT network inhibits HSV-1 infection significantly but not completely (45, 77). Thus, a structural viral protein that is not essential for virus replication, such as VP26 (14), may be responsible for the recruitment of dynein or dynactin to incoming capsids.
To study the viral requirements for HSV-1 capsid transport during cell entry, we analyzed the nuclear targeting of incoming HSV1-ΔVP26 deleted for VP26 (14) and also of HSV1-GFPVP26, which expresses a fusion protein of green fluorescent protein (GFP) and VP26 instead of VP26 (15). The GFPVP26 fusion protein does not interfere with virus replication in tissue culture and is incorporated into virions during capsid assembly (15). When Vero cells, which are infected by fusion of the viral envelope and the plasma membrane (56, 57, 77), were infected with HSV1-ΔVP26 or HSV1-GFPVP26 preparations of high-quality, cytosolic capsids accumulated at the nuclear envelope in an MT-dependent manner. Furthermore, less HSV1-ΔVP26 and HSV1-GFPVP26 capsids reached the nucleus when dynein function was inhibited by overexpression of the dynactin subunit dynamitin. Since even in the absence of the potential viral dynein receptor VP26 HSV-1 used MT and dynein/dynactin for efficient nuclear targeting, other HSV-1 proteins must be able to recruit dynein either directly or via its cofactor, dynactin, to the incoming capsids for MT transport.
BHK-21 cells (ATCC CCL-10) and PtK2 cells (ATCC CCL-56) were grown in medium containing 10% (vol/vol) fetal calf serum, and Vero cells (ATCC CCL-81) were grown in medium containing 7.5% fetal calf serum (Invitrogen, Paisley, United Kingdom). All media contained Eagle's minimal essential medium with 2 mM glutamine and nonessential amino acids (Cytogen, Sinn, Germany). Several structural herpesvirus proteins were detected by the anti-HSV-1 rabbit polyclonal antibody (PAb) Remus V (59); the major capsid protein VP5 was detected by the rabbit PAb NC-1 (10) or the mouse monoclonal antibodies (MAbs) LP12, 5C10, or 8F5 (63, 84); US11 was detected by mouse MAb 28 (68); VP16 was detected by rabbit PAb SW7 (89); and VP13/14 was detected by rabbit PAb R220 (90). The membrane proteins gB and gD were immunolabeled by rabbit PAb R68 and mouse MAb DL6, respectively (22, 23), the immediate-early proteins ICP4 and ICP0 were labeled by mouse MAbs 58S (72) and 11060 (24), respectively, and actin was labeled using the mouse MAb 1501 (Sigma-Aldrich, Taufkirchen, Germany). VP26 was visualized with a rabbit PAb generated against His6-tagged VP26 (provided by A. Helenius, ETH, Schweiz) or a rabbit PAb generated against amino acids 95 to 112 (14). To monitor incoming viral capsids, we also used preadsorbed (77) rabbit PAbs raised against DNA-containing (anti-HC) or empty capsids (anti-LC) (10).
Concentrated stocks containing secreted, extracellular virions of HSV-1 wild type (wt) strain F (ATCC VR-733), HSV-1 wt strain KOS (ATCC VR-1493), HSV1-ΔVP26 strain KOS expressing β-galactosidase instead of the small capsid protein VP26 (KΔ26Z) (14), or HSV1-GFPVP26 strain KOS, in which VP26 had been replaced by GFPVP26 (K26GFP) (15), were essentially prepared as described elsewhere (77) but using linear nycodenz (Axis-shield PoC; AS, Oslo, Norway) instead of sucrose gradients. Secreted extracellular viral particles were pelleted from the medium of cells infected with a multiplicity of infection (MOI) of 0.01 PFU/cell for 2 days for HSV-1 wt or 3 days for HSV1-ΔVP26 and HSV1-GFPVP26. The resuspended sediment was layered on top of a 10 to 40% (wt/vol) nycodenz gradient in MNT buffer (20 mM morpholineethanesulfonic acid, 100 mM NaCl, 30 mM Tris, pH 7.4) and spun at 50,000 × g for 2 h at 4°C (69). The virus band in the middle of the gradient was harvested, snap-frozen, and stored in single-use aliquots at −80°C.
We obtained the best virus preparations, as defined by low particle- and low viral genome-to-PFU ratios, when we harvested at an early time point when nearly all cells were rounded, but only detached from the substrate after knocking them off, and when we used nycodenz instead of sucrose gradients. Due to its high osmolarity, sucrose may induce shrinking of the enveloped viral particles upon entering the gradient and abrupt swelling and even rupture upon dilution of the virus-containing sucrose band in the inoculating medium. Compared to HSV-1 wt, HSV1-ΔVP26 assembles less efficiently (14), and both HSV1-ΔVP26 and HSV1-GFPVP26 developed cytopathic effects slower than HSV-1 wt (not shown).
In addition, we characterized the supernatants of infected cells stored at 4°C for 2 to 5 days without further concentration. Since we analyzed HSV-1 mutants attenuated in the efficiency of assembly and egress compared to HSV-1 wt (14, 15), virus was not harvested from cells infected at a very low MOI, as described by Everett et al. (25), but from cells synchronously infected with an MOI of 3 PFU/cell. However, these supernatants contained a higher proportion of particles deficient in efficient nuclear targeting than virus preparations purified on nycodenz gradients, and therefore the latter were used for all experiments described here except for those shown below in Fig. Fig.2a.2a. All virus preparations were plaque titrated on Vero cells (18, 77).
Gradient-purified virions were diluted 1,000-fold in DNase I reaction buffer (10 mM Tris-HCl, pH 7.5, 2.5 mM MgCl2, 0.5 mM CaCl2) to a final volume of 495 μl. To determine the amount of viral DNA protected in capsids or virions, the samples were incubated either without or with 5 units of protease-free DNase I (ABgene, Epsom, United Kingdom) for 30 min at 37°C and then further for 10 min at 75°C to inactivate the DNase. Twenty-five micrograms of salmon sperm DNA (Invitrogen) was added as carrier subsequent to DNase I inactivation. After the DNA had been purified using the QIAamp DNA blood mini kit (QIAGEN, Hilden, Germany), a real-time detection PCR was performed using the LightCycler FastStart DNA Master HybProbe kit (Roche Diagnostics, Mannheim, Germany), the forward sense primer 5′-CCACGAGACCGACATGGAGC-3′, the reverse antisense primer 5′-GTGCTYGGTGTGCGACCCCTC-3′, the fluorescein-coupled donor probe 5′-TGTTGGCGACTGGCGACTTTG-3′-fluorescein, and the R640-coupled acceptor probe R640-5′-TACATGTCCCCGTTTTACGGCTACCGG-3′-phosphate, which are all specific to the coding region of HSV-1 gB (gene UL27), and the Roche LightCycler 1.5. After one cycle of denaturation (95°C for 10 min), 55 cycles of amplification were performed (95°C for 10 s, 58°C for 15 s, and 72°C for 15 s). During the 58°C phases, the acceptor fluorescence was measured at 640 nm, and from these data the DNA concentrations of the probes were calculated using standards of known DNA concentrations.
To analyze the protein composition, extracellular virions were pelleted, sometimes further purified on nycodenz gradients, solubilized in sample buffer (50 mM Tris-HCl, pH 6.8, 1% [wt/vol] sodium dodecyl sulfate [SDS], 1% [vol/vol] β-mercaptoethanol, 5% [vol/vol] glycerol, 0.001% [wt/vol] bromphenol blue), and loaded onto SDS gels. Following SDS-polyacrylamide gel electrophoresis (SDS-PAGE), proteins were either stained with 0.1% (wt/vol) Coomassie brilliant blue R-250 and 0.1% (wt/vol) Coomassie brilliant blue G-250 in 10% acetic acid and 50% methanol or transferred to nitrocellulose membranes. The membranes were washed with phosphate-buffered saline (PBS) containing 0.1% (vol/vol) Tween 20 (PBS-T). After blocking in 5% lowfat milk in PBS-T, the membranes were incubated with primary antibodies and then with secondary alkaline phosphatase-conjugated goat anti-rabbit or anti-mouse immunoglobulin G (IgG) antibodies (Dianova, Hamburg, Germany). After washing, the membranes were stained with 0.2 mM nitroblue tetrazolium chloride and 0.8 mM 5-bromo-4-chloro-indolyl-3-phosphate in 100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 5 mM MgCl2.
To analyze HSV-1 immediate-early gene expression, confluent Vero cells grown in 10-cm culture dishes were preincubated for 1 h at 37°C without or with 50 μM nocodazole to depolymerize the MT network. The cells were then precooled in ice-cold CO2-independent medium (Invitrogen) with 0.1% bovine serum albumin (BSA) and inoculated for 2 h on ice with 5 PFU/cell in CO2-independent medium with 0.1% BSA without or with 50 μM nocodazole. Unbound virus was removed by washing with ice-cold RPMI with 25 mM HEPES and 0.1% BSA, and the cells were further incubated at 37°C and 5% CO2 in medium with or without nocodazole. After 3 or 4 h, the cells were washed with PBS, lysed in hot sample buffer, scraped, and resuspended. Samples were resuspended to shear DNA, stored at −20°C, and analyzed by immunoblotting as described above.
Viral particles were purified via nycodenz gradients and diluted to 1.6 × 106 PFU/μl with 25% nycodenz solution. After adsorption on carbon- and Formvar film-coated copper grids (400 mesh) and blocking with 10 mg/ml BSA in PBS, the samples were labeled with anti-gD (MAb DL6), rabbit anti-mouse IgG (Cappel, MP Biomedicals, Irvine, Calif.), and protein A-gold (10 nm; Dept. Cell Biology, Utrecht School of Medicine). Finally, all preparations were negatively contrasted using 2% uranyl acetate (Merck, Darmstadt, Germany) and analyzed with an EM 10 electron microscope (Carl Zeiss AG, Jena, Germany) at 80 kV.
For transient transfections, we used the plasmids pEGFP-C1 (BD Biosciences, Mountain View, CA) expressing enhanced GFP (EGFP) and pDynamitin-EGFP expressing dynamitin-GFP under the control of the cytomegalovirus immediate-early promoter (18). Vero cells were seeded in 24-well plates at a density of 3 × 104 cells/well for 19 h and then transfected with 0.3 μg/well DNA and 0.75 μl/well Gene Juice reagent (Merck Biosciences, Schwalbach, Germany). At 30 h posttransfection, the cells were infected with HSV-1.
PtK2 or Vero cells grown on coverslips in 24-well dishes were synchronously infected with HSV-1 in the absence or presence of 50 μM nocodazole (18, 77). To prevent synthesis of progeny virus, 0.5 mM cycloheximide (Sigma-Aldrich) was added. The cells were fixed with 3% paraformaldehyde, permeabilized with 0.1% Triton X-100, and labeled with antibodies as described elsewhere (18, 77). For the anti-VP26 labeling (anti-VP26, amino acids 95 to 112), we used 10% (vol/vol) human serum of an HSV-1-seronegative volunteer to block the HSV-1 Fc receptor, which has strong affinity for human IgG, decreasing affinity for rabbit, sheep, and goat IgG, and no reactivity for murine IgG (18, 20). In all other experiments (see Fig. Fig.4,4, ,6,6, ,8,8, and and9,9, below), 0.5% (wt/vol) BSA-PBS was used as blocking reagent. The specimens were analyzed with an Eclipse E800 microscope (Nikon Instruments, Kanagawa, Japan) equipped with the appropriate fluorescence filter sets. The cell margins and nuclei were visualized by phase contrast, and images were taken with a digital interline charge-coupled device camera (Micromax-1300Y; Princeton Instruments Inc.) controlled by the IPLab 3.2 software. Images were further processed using MetaMorph 5.0.5 (Universal Imaging Corporation, West Chester, Pa.) and Adobe Photoshop 6.0. The nuclei and cell margins were indicated by dashed or continuous white lines, respectively. To determine the efficiency of nuclear targeting (see Table Table1,1, below), we randomly selected more than 100 cells for each viral preparation to be analyzed and categorized them into three classes: (i) cells with most capsids around the nucleus, (ii) cells with many capsids at the nucleus and also many in the periphery, and (iii) cells with most capsids in the periphery.
To be able to compare HSV-1 wt with the mutants HSV1-ΔVP26 and HSV1-GFPVP26 in cell entry experiments, we first characterized the particle composition of the inocula. Preparations of viral mutants may contain higher numbers or different ratios of defective particles, which can be impaired at any stage of the viral life cycle, including cell entry. Different viral particles can be released from HSV-1-infected cells into the culture medium either by secretion or by cell lysis (Fig. (Fig.1)1) in the following forms: (i) vesicles containing viral membrane proteins but neither tegument nor capsid proteins, (ii) L-particles, which consist of tegument surrounded by an envelope with viral membrane proteins (78), (iii) empty or (iv) DNA-filled capsids, (v) enveloped and tegumented capsids which lack the viral DNA, and (vi) noninfectious or (vii) infectious virions. Antibodies raised against the viral membrane, tegument, or capsid proteins will detect particles as well as cellular structures containing the respective viral proteins. Using an HSV-1-specific PCR, the number of genomes can be quantified, whereas only infectious virions will form plaques. First of all, we determined the specificity of the generated anti-VP26 antibodies. Immunoblotting showed no anti-VP26-reactive bands for HSV1-ΔVP26, whereas in HSV-1 wt and HSV1-GFPVP26 samples, VP26-positive bands were detected at the expected molecular weights (Fig. (Fig.2a2a).
To identify inocula with the lowest possible amount of defective particles, several virus preparations were compared regarding their titer (PFU), genome content, protein composition, number of particles, and nuclear targeting. To reduce the amount of possible cellular contaminants, we isolated virions from the medium of infected cells and purified them on linear nycodenz gradients (except for the experiment shown in Fig. Fig.2a).2a). Our best HSV-1 wt and HSV1-GFPVP26 virus preparations had a genome/PFU ratio of 15 to 25, whereas the best HSV1-ΔVP26 preparation had a ratio of about 50 (Table (Table1).1). Preparations characterized by a low genome/PFU ratio (Fig. (Fig.2b,2b, lanes 1 to 4) also had lower amounts of viral proteins than preparations with a high ratio (lanes 5 to 7) when equal amounts of PFU were compared. The overall protein composition (Fig. (Fig.2b)2b) and the relative amounts of glycoproteins, tegument, and capsid proteins (Fig. (Fig.2c)2c) were similar, suggesting that neither deletion nor GFP tagging of VP26 significantly altered the ratios of the different particles in the gradient-purified virus preparations.
Electron microscopy demonstrated that preparations with a high genome/PFU ratio also had a higher particle concentration than those with a low ratio (Fig. 3a to e and l). The morphology of many particles was characteristic of herpesviruses with a capsid surrounded by a broken envelope, which therefore enabled negative contrasting of both capsid and envelope with uranium (Fig. 3f and g). In addition, there were many particles of a similar size that showed no internal contrast (Fig. 3h and i). Both populations were labeled with antibodies directed against the HSV-1 envelope proteins gD (Fig. 3f and h) and gB (not shown). Thus, these particles either represented vesicles containing viral membrane proteins or virions with an intact membrane which prevented an internal contrasting of the capsids. The latter seems to be more likely, since both populations were derived from the same fraction of the velocity gradient and had a very similar if not identical size. The ratio of particles with the typical herpesvirus morphology and those lacking internal contrast varied somewhat between preparations of the same sample, suggesting as well that some of the smooth particles indeed reflected virions. In addition, the preparations also contained capsids (Fig. 3j and k), of which some were also labeled for gD (Fig. (Fig.3j)3j) or gB (not shown), albeit with lower intensity than the viral membranes. This capsid labeling either represented unspecific background or may have indicated that small remnants of envelope proteins were bound to them. Attaching GFP to VP26 (HSV1-GFPVP26) led to a moderate production, and the deletion of VP26 (HSV1-ΔVP26) led to a higher production of noninfectious particles (Fig. (Fig.3l).3l). However, the preparations of the different HSV-1 strains contained the different particles in similar ratios, except that HSV1-GFPVP26 showed a particularly high proportion of particles with a vesicular morphology lacking internal contrast (Fig. (Fig.3m3m).
Next, we used immunofluorescence microscopy to analyze the nuclear targeting of different preparations. In all experiments analyzing the subcellular localization of incoming HSV-1 particles (see Fig. Fig.44 to to6,6, ,8,8, and and9,9, below), cycloheximide was added to prevent the synthesis of progeny virus. PtK2 cells were infected for 3 h and double-labeled with antibodies against capsids and gD (Fig. (Fig.4,4, a to h) . To quantify the efficiency of nuclear targeting, we classified the inoculated cells into three categories: (i) cells with most capsids at the nucleus, (ii) cells with many capsids at the nucleus and many capsids in the cell periphery, and (iii) cells with most capsids in the cell periphery (Table (Table1).1). Capsids derived from HSV-1 wt characterized by a low genome/PFU ratio (Table (Table1;1; Fig. Fig.4a)4a) reached the nucleus with high efficiency. The gD signal of virus preparations of lower quality (Fig. 4b and c) remained higher than the gD signal of high-quality virus preparations (Fig. 4a and g). Capsids of a preparation with a high genome/PFU ratio also accumulated at the nuclear envelope, but relatively more particles remained in the cell periphery (Table (Table1;1; Fig. 4b and c). Therefore, virus preparations with a low genome/PFU ratio were considered to be of higher quality than those with a high ratio. Similar results were obtained with preparations of the mutants (Table (Table1;1; Fig. 4e to h). However, of the best HSV1-ΔVP26 preparation (Fig. (Fig.4e),4e), relatively more capsids remained in the cell periphery compared to HSV-1 wt (Fig. (Fig.4a)4a) or HSV1-GFPVP26 (Fig. (Fig.4g4g).
Thus, for meaningful entry experiments, high-quality preparations of wt and mutant viruses had to be used. If we had compared an HSV1-ΔVP26 mutant preparation of low quality characterized by a high genome/PFU ratio (Fig. (Fig.4f)4f) with an HSV-1 wt preparation of high quality (Fig. (Fig.4a),4a), we would have erroneously concluded that VP26 was required for efficient nuclear targeting. However, the best HSV1-ΔVP26 preparations showed efficient nuclear targeting (Table (Table1;1; Fig. Fig.4e),4e), although less pronounced than HSV-1 wt (Fig. (Fig.4a)4a) and HSV1-GFPVP26 (Fig. (Fig.4g4g).
To further characterize the cell entry of high-quality preparations of HSV1-ΔVP26 and HSV1-GFPVP26, we analyzed the subcellular localizations of capsids and envelope proteins to distinguish virions from cytosolic capsids (Fig. 4i to k) and we used Vero cells, which are infected by fusion of the viral envelope with the plasma membrane (56, 57, 77). After 30 min of infection, the spotted capsid (red) and gD (green) signals were randomly distributed over the entire cell. Most capsids of HSV-1 wt (Fig. (Fig.4i)4i) and HSV1-GFPVP26 (Fig. (Fig.4k),4k), but only 50% of the HSV1-ΔVP26 capsids (Fig. (Fig.4j),4j), did not colocalize with gD. The fraction of particles containing capsid antigens and gD might either represent virions bound to the plasma membrane or virions inside endosomes. The early separation of capsid and glycoprotein suggested that the majority of HSV-1 wt and HSV1-GFPVP26 and about 50% of HSV1-ΔVP26 particles were cytosolic capsids.
At 30 min postinfection (p.i.) with HSV1-GFPVP26, the capsid (red) and GFPVP26 (green) signals only partially colocalized (Fig. (Fig.4l).4l). Several capsids did not contain any detectable GFP, and several GFP particles were not labeled by the anticapsid antibodies. To characterize the subcellular fate of VP26 and GFPVP26 after inoculation, infected cells were labeled with anti-VP26. At 30 min p.i. with HSV-1 wt (Fig. (Fig.5a)5a) or HSV1-GFPVP26 (Fig. (Fig.5c),5c), VP26-positive spots were randomly distributed over the entire cell, whereas no virus-specific signal was detected with HSV1-ΔVP26 (Fig. 5b and f). The VP26 localization resembled that of the capsids (not shown, but see Fig. 4i to l). The intensity of the VP26 labeling of individual particles varied with both viruses, but the signal of HSV-1 wt (Fig. (Fig.5a)5a) was generally slightly stronger than that of HSV1-GFPVP26 (Fig. (Fig.5c).5c). This either might reflect the lower copy number of GFPVP26 per capsid compared to VP26 (P. Desai, personal communication) or GFP might have masked VP26 epitopes. Most of the GFP spots of HSV1-GFPVP26 colocalized with the VP26 signal (Fig. 5c and d).
At 3 h p.i. most VP26-containing particles of HSV-1 wt had accumulated at the nuclear envelope (Fig. (Fig.5e),5e), like the spots detected by anti-VP5 or anticapsid antibodies (c.f. Fig. 4a to h with Fig. Fig.6,6, ,8,8, and and9,9, below). Compared to parallel labeling with anticapsid antibodies (not shown, but similar to Fig. Fig.6g,6g, below), fewer VP26- and GFP-positive than capsid antigen-containing spots were detected after HSV1-GFPVP26 infection. In other words, the anticapsid antibodies detected more viral particles than were visualized by anti-VP26 antibodies or the GFP signal. In addition, fewer VP26 (Fig. (Fig.5g)5g) and fewer GFP spots (Fig. (Fig.5h)5h) were detected at 3 h than at 30 min p.i. with HSV1-GFPVP26 (Fig. 5c and d). Most of the remaining GFP and VP26 signals colocalized at the nuclear envelope (Fig. 5g and h). The intensity of the GFP signal was much weaker at 3 h (Fig. (Fig.55 h) than at 30 min p.i. (Fig. (Fig.5d).5d). Thus, a fraction of the GFPVP26 dissociated from the incoming capsids during transport to the nucleus. Consistent with a potential function of VP26 as a viral dynein receptor (Douglas et al. ), VP26 remained bound to capsids until arrival at the nuclear pores.
Because the GFP signal was not detected on all HSV1-GFPVP26 capsids (Fig. (Fig.4l4l and and5),5), we used several antibodies to localize HSV1-GFPVP26 and HSV1-ΔVP26 incoming capsids. In contrast to HSV-1 wt (Fig. (Fig.6a),6a), the anti-VP5 MAbs 5C10 and 8F5 (not shown) did not label capsids of HSV1-ΔVP26 (Fig. (Fig.6b)6b) and only weakly labeled the ones of HSV1-GFPVP26 (Fig. (Fig.6c).6c). Most of these weak 5C10 spots colocalized with the GFP signal of HSV1-GFPVP26 (e.g., Fig. 6c and d). In contrast, the PAb NC-1 (not shown) and anti-VP5 MAb LP12 labeled capsids of HSV1-ΔVP26 (Fig. (Fig.6f)6f) and HSV1-GFPVP26 (Fig. (Fig.6g)6g) with higher efficiency than HSV-1 wt (Fig. (Fig.6e).6e). The GFP signal of HSV1-GFPVP26 (Fig. (Fig.6h)6h) colocalized with the LP12 labeling (Fig. 6g and h), but many nuclear capsids labeled by LP12 were not detected by GFP fluorescence (compare Fig. 6g and h). The PAbs anti-HC and anti-LC against DNA-containing and empty capsids, respectively, labeled capsids of the four tested HSV-1 strains with similar efficiencies (not shown in Fig. Fig.5;5; for anti-HC, see Fig. Fig.4,4, ,8,8, and and99).
Efficient expression of HSV-1 wt immediate-early genes requires MT (45, 46, 77). To test whether this was also true for HSV1-GFPVP26 and HSV1-ΔVP26, Vero cells were infected with 5 PFU/cell in the presence or absence of nocodazole, which reversibly depolymerizes MT (38). For all HSV-1 strains tested, viral gene expression at 3 h (Fig. (Fig.7)7) and 4 h p.i. (not shown) was strongly reduced by nocodazole.
This reduction in immediate-early gene expression could have been due to impaired nuclear targeting either of the HSV-1 transcription factor VP16, which is provided by the incoming tegument (7), or of the capsids, resulting in fewer viral genomes reaching the nucleoplasm (18, 77). Therefore, we analyzed the subcellular localization of incoming HSV-1 wt, HSV1-ΔVP26, and HSV1-GFPVP26 capsids in the presence and absence of MT. After 3 h in the presence of nocodazole, fewer capsids of both HSV1-ΔVP26 and HSV1-GFPVP26 reached the nuclear envelope (Fig. (Fig.8).8). Compared to HSV-1 wt (Fig. (Fig.8a)8a) and HSV1-GFPVP26 (Fig. (Fig.8c),8c), nuclear targeting of HSV1-ΔVP26 was less efficient (Fig. (Fig.8b).8b). This was most likely due to a higher proportion of HSV1-ΔVP26 defective particles (Table (Table1;1; Fig. Fig.22 to to4).4). In contrast to residual capsids in the cell periphery, the HSV-1 wt, HSV1-ΔVP26, and HSV1-GFPVP26 capsids at the nucleus or in the perinuclear area did not colocalize with gD (not shown). Since Vero cells are known to be productively infected by fusion of the viral envelope and the plasma membrane (56, 57, 77) and the capsids separated from the envelope early in infection (Fig. 4i to k), we concluded that the three strains used MT for efficient capsid transport to the nucleus.
Dynein and its cofactor, dynactin, catalyze MT-mediated nuclear targeting of HSV-1 wt (18, 77). To test whether they were also required for nuclear targeting of HSV1-ΔVP26 and HSV1-GFPVP26, we overexpressed the dynactin subunit dynamitin, which blocks dynein-mediated transport (5, 21). In cells overexpressing dynamitin-GFP (Fig. (Fig.9),9), fewer HSV-1 wt (Fig. (Fig.9a),9a), HSV1-ΔVP26 (Fig. (Fig.9b),9b), and HSV1-GFPVP26 (Fig. (Fig.9c)9c) capsids reached the nucleus compared to untransfected cells. Instead, many viral particles were still distributed over the entire cytoplasm or located in the cell periphery. In contrast, overexpression of GFP (Fig. 9d to f) had no effect on nuclear targeting of capsids. Thus, HSV1-ΔVP26, HSV1-GFPVP26, and HSV-1 wt used MT and dynein/dynactin for capsid transport to the nucleus.
Our characterization of the inocula by plaque titration, PCR, biochemical, and microscopic approaches revealed that different preparations of the same strain varied significantly in the particle/PFU ratio. Virus preparations with a higher genome/PFU ratio also contained more viral particles and were less efficiently targeted to the nucleus. Therefore, these virus preparations were considered to be of lower quality than preparations with a low genome/PFU ratio. Besides the source of the particles (infected cells or medium) and the gradient material (nycodenz or sucrose), the quality of virus preparations also depends on the cell type used for propagation, the virus strain, and the passage number of the virus (6, 11, 28).
Many groups have analyzed HSV-1 inocula by electron microscopy. Using calibrated latex beads as a concentration standard, particle/PFU ratios ranging from 5 to more than 1,000 have been reported (6, 11, 27, 28, 35, 36, 88), with a ratio of 5 being considered excellent and a ratio of up to 50 being acceptable (6, 35, 36). Moreover, Cai and Schaffer (6) reported a genome/PFU ratio of 33 and a particle/genome ratio of 1.7 for HSV-1 wt (KOS). Thus, our best HSV-1 wt preparations with a genome/PFU ratio of 15 to 25, and after removal of nonencapsidated DNA of 7 to 10, were among the best described so far.
According to these criteria, our best HSV1-ΔVP26 preparations were of lower quality than the best HSV-1 wt preparations, most likely due to a less efficient assembly of the mutant (14). Using an HSV1-ΔVP26 preparation of low quality, we would have erroneously concluded that VP26 was required for efficient nuclear targeting. Therefore, it was crucial to identify high-quality virus preparations, especially when the cell entry was analyzed. Incoming HSV1-ΔVP26 capsids of good preparations were efficiently transported to the nucleus, although less completely than capsids of HSV-1 wt or HSV1-GFPVP26.
The quality of virus preparations is crucial for analyzing cell entry but may also influence the outcome of animal experiments. Even as defective viral particles are not measured in regular plaque assays, the PFU concentration is often the only parameter used to standardize experiments using different viral strains. But the amount of noninfectious particles present in an inoculum may modulate the innate immune response, which is already triggered during virus entry and at this stage is independent of viral gene expression and replication (26, 52). Thus, the mode of inoculum preparation may significantly influence the outcome of animal experiments designed to determine the functional consequences of mutations.
After 30 min, the capsid signal of HSV-1 wt and HSV1-GFPVP26 had mostly separated from the gD signal, suggesting that it represented cytosolic capsids released into Vero cells by viral fusion at the plasma membrane (56, 57, 77). In contrast, the HSV1-ΔVP26 capsids colocalized more often with gD, reflecting a higher proportion of defective particles in this strain which either remained at the plasma membrane or were taken up by endocytosis. The 50% of HSV1-ΔVP26 capsids lacking gD most likely represented cytosolic capsids that were still distributed over the entire cytoplasm at this early time point.
HSV1-GFPVP26 capsids detected by antibodies only partially colocalized with the GFP signal. This lack of colocalization may be explained by (i) a dissociation of GFPVP26 from the incoming capsids during infection, (ii) a masking of capsid epitopes by GFPVP26, (iii) proteolysis of GFP, (iv) fluorescence resonance energy transfer between GFP as the donor and the lissamine-rhodamine-coupled secondary antibody detecting capsids, or (v) some capsids containing too little GFPVP26 to be detected with our microscopic setup. These scenarios are not mutually exclusive. VP26 remained bound to HSV-1 wt as well as to many HSV1-GFPVP26 capsids until arrival at the nucleus. However, the GFP signal decreased from 30 min to 3 h p.i., suggesting that indeed a fraction of GFPVP26 dissociated from incoming capsids during cytosolic passage.
Moreover, some anticapsid antibodies do not detect extracellular virions bound to the plasma membrane but cytosolic capsids after cell entry (77). This was explained by a subsequent removal of tegument during virus entry, which increases the accessibility of capsid antigens (32, 43, 77). This hypothesis agrees well with the observation that some GFP particles not labeled by anticapsid antibodies colocalized with antibodies directed against envelope proteins (data not shown). Since GFPVP26 is incorporated into capsids in varying amounts, resulting in heterogenous GFP labeling of the particles present in the inoculum (reference 74 and data not shown), the degree of masking may also diverge on different capsids.
Several anti-VP5 antibodies had different affinities for the capsids of HSV-1 wt, HSV1-ΔVP26, or HSV1-GFPVP26. In contrast to HSV-1 wt, HSV1-ΔVP26 did not and HSV1-GFPVP26 did only poorly display epitopes of the MAbs 8F5 and 5C10, while the MAb LP12 and PAb anti-NC1 labeled HSV1-ΔVP26 and HSV1-GFPVP26 with higher efficiency than HSV-1 wt. On the contrary, the PAbs anti-HC and anti-LC detected the different strains with similar affinities and were therefore mainly used to analyze the subcellular capsid localization.
VP26 is located on VP5 hexons but not on VP5 pentons, even if provided in an eightfold molar excess (4, 83, 95, 96), and its binding to mature, angularized capsids requires ATP (8). Similarly, MAb 8F5 and 5C10 epitopes are present on the hexons of mature capsids (9, 30, 47, 84), and their formation also requires ATP (12). Moreover, MAb 8F5 binds no longer to capsids from which VP26 has been removed by guanidinium hydrochloride (4). Together with our observation that MAbs 8F5 and 5C10 did not recognize HSV1-ΔVP26 capsids, these data suggest that VP26 induces conformational changes in VP5 which result in the formation of the 8F5 and 5C10 epitopes, rather than the formation of 8F5 and 5C10 epitopes being a prerequisite for VP26 binding.
The MAb LP12 may then have recognized an immature VP5 epitope that persisted on HSV1-ΔVP26 but was of low abundance on HSV-1 wt. The weak MAb 8F5 and 5C10 signals with HSV1-GFPVP26 may either be due to GFPVP26 masking these epitopes or to the weaker potency of GFPVP26 to induce conformational changes in VP5. The latter is supported by MAb LP12, which labeled HSV1-GFPVP26 capsids with higher efficiency than HSV-1 wt. However, these explanations may not be mutually exclusive, since GFPVP26 interfered with labeling of MAb 8F5 but not MAb 5C10 in capsid assembly assays (8).
Incoming cytosolic HSV-1 wt capsids are propelled by dynein and dynactin along MT from the cell periphery to the nucleus (18, 77). Although the small capsid protein VP26 can interact with dynein light chains (19), we have shown here that HSV1-ΔVP26 and HSV1-GFPVP26 capsids also required MT and dynactin for efficient nuclear targeting. Therefore, HSV1-GFPVP26 provides an excellent model to study MT-mediated transport in living cells. Our data agree with results from a mouse ocular model showing transport of HSV1-ΔVP26 to the trigeminal ganglia (14) and another study showing efficient nuclear targeting of the porcine alphaherpesvirus pseudorabies virus in the absence of VP26 in neurons (1). However, in these studies it was not determined which MT motors were responsible and whether cytosolic capsids or endocytosed virions were transported to the neuronal cell body.
As HSV-1 wt, HSV1-ΔVP26, and HSV1-GFPVP26 capsids separated as early as 30 min p.i. from the viral envelope, the particles transported in Vero cells which were infected by viral fusion at the plasma membrane (56, 57, 77) represented cytosolic capsids rather than endocytosed virions. Compared to HSV-1 wt and HSV1-GFPVP26, the nuclear targeting of HSV1-ΔVP26 was less efficient. This was most likely due to a higher proportion of defective particles in the HSV1-ΔVP26 inoculum which retained envelope antigens and remained in the cell periphery. Those particles of HSV1-ΔVP26 which reached the nucleus utilized, like HSV-1 wt and HSV1- GFPVP26, MT for nuclear targeting.
In addition to dynein, some C-type kinesins catalyze minus-end-directed MT transport during interphase (60) and could therefore catalyze HSV1-ΔVP26 transport to the nucleus. However, the overexpression of dynamitin reduced nuclear targeting of all three capsids, HSV-1 wt, HSV1-ΔVP26, and HSV1-GFPVP26. Overexpression of the dynactin subunit dynamitin disrupts the dynactin complex and reduces both dynein- and kinesin-2-mediated transport (13, 70). But, since kinesin-2 is a plus-end-directed MT motor (37), it is unlikely to be involved in minus-end-directed transport. To our knowledge, an effect of dynamitin overexpression on minus-end-directed kinesins has not been reported. Likewise, anti- dynamitin antibodies reduce binding of late endosomes to MT, which is mediated by dynein and kinesin-2, but not binding of early endosomes via the minus-end-directed kinesin KIFC2 and kinesin-1 (2). Thus, blocking dynactin function does not interfere with binding of minus-end-directed kinesins to their cargo. While we cannot formally exclude a role for minus-end-directed kinesins in nuclear targeting of HSV-1, it is more likely that dynein and dynactin catalyzed the transport of HSV1-ΔVP26 and HSV1-GFPVP26 capsids to the nucleus.
Moreover, in an in vitro binding assay, capsids of HSV1-ΔVP26 and HSV1-GFPVP26 bound dynein and dynactin as efficiently as HSV-1 wt (93). Furthermore, capsids whose outer tegument proteins have been removed bind dynein and dynactin more efficiently than nuclear capsids which expose VP26 but no tegument proteins on their surface or than capsids with a complete tegument (93). Such capsids which contain VP16, a potential linker between the inner and outer tegument (53, 86), as well as the inner tegument proteins VP1-3 and UL37, also showed the strongest in vitro motility along MT (93). Thus, VP26 is not sufficient for motor recruitment and MT motility, but inner tegument proteins and/or VP16 are required. Moreover, excessive tegument coating reduces transport, suggesting that the removal of outer tegument proteins either exposes a hidden motor receptor or removes a transport inhibitor (93).
Consistent with these biochemical results, most of the HSV-1 tegument and VP11/12, VP13/14, VP16, and VP22 of pseudorabies virus detach from the incoming capsid upon fusion at the plasma membrane, whereas the inner tegument proteins VP1-3, UL37, and US3 remain on the capsids (32, 43, 77, 91). The potential interaction between VP26 and the dynein light chains may play a role during HSV-1 assembly, when cytosolic capsids are transported to the site of secondary budding (49, 50). Nevertheless, during egress of pseudorabies virus, VP1-3 is also required for processive MT transport of GFPVP26-tagged particles, and UL37 enhances its efficiency (44). These proteins may, possibly in concert with VP26, either interact directly with MT motors or may be required for recruiting other tegument proteins involved in MT transport. In neurons, GFPVP26-tagged pseudorabies virus particles undergo bidirectional transport during entry and egress, with retrograde transport dominating during entry and anterograde transport dominating during egress (74, 75). The transport direction may be controlled by varying the tegument composition or by cellular or tegument-associated kinases (UL13 and US3) which may target tegument proteins or MT motors during different stages of the viral life cycle (43, 75).
In summary, the in vivo and in vitro experiments indicate that besides VP26, HSV-1 must encode at least one additional receptor for dynein or dynactin and that HSV-1 inner tegument proteins are likely candidates.
We thank S. Hübner and D. Petzhold, C.-H. Nagel, and J. Janus for assistance in PCR, plaque titration, and microscopy, E. Ungewickell for unlimited access to his electron microscopy laboratory, and R. Bauerfeind, M. Messerle, C.-H. Nagel, K. Theusner, and A. Wolfstein for helpful discussions (Hannover Medical School, MHH). We are grateful to P.G. Spear (Northwestern University, Chicago, Ill.) for HSV-1 wt (KOS) and to P. Desai (Johns Hopkins University, Baltimore, Md.) for HSV1-ΔVP26, HSV1-GFPVP26, and anti-VP26. J. Brown and W. W. Newcomb (University of Virginia Health System, Charlottesville), G. H. Cohen and R. J. Eisenberg (University of Pennsylvania, Philadelphia), R. D. Everett (MRC Virology Unit, Glasgow, United Kingdom), A. Helenius (Institute of Biochemistry, ETH, Zürich, Switzerland), and A. C. Minson (University of Cambridge, Cambridge, United Kingdom) generously donated antibodies.
We are supported by the German Research Council (DFG; So403/1 and So403/2). S.S. received an MHH fellowship of the State of Lower Saxony, Germany, and K.R. received a fellowship by the MHH Center of Infection Biology.