Cytoplasmic Parvovirus Capsids Recruit Importin Beta for Nuclear Delivery

Parvoviruses are small DNA viruses that deliver their DNA into the postmitotic nuclei, which is an important step for parvoviral gene and cancer therapies. Limitations in virus-receptor interactions or endocytic entry do not fully explain the low transduction/infection efficiency, indicating a bottleneck after virus entry into the cytoplasm. We thus investigated the transfer of parvovirus capsids from the cytoplasm to the nucleus, showing that the nuclear import of the parvovirus capsid follows a unique strategy, which differs from classical nuclear import and those of other viruses.

ABSTRACT

Parvoviruses are an important platform for gene and cancer therapy. Their cell entry and the following steps, including nuclear import, are inefficient, limiting their use in therapeutic applications. Two models exist on parvoviral nuclear entry: the classical import of the viral capsid using nuclear transport receptors of the importin (karyopherin) family or the direct attachment of the capsid to the nuclear pore complex leading to the local disintegration of the nuclear envelope. Here, by laser scanning confocal microscopy and in situ proximity ligation analyses combined with coimmunoprecipitation, we show that infection requires importin β-mediated access to the nuclear pore complex and nucleoporin 153-mediated interactions on the nuclear side. The importin β-capsid interaction continued within the nucleoplasm, which suggests a mixed model of nuclear entry in which the classical nuclear import across the nuclear pore complex is accompanied by transient ruptures of the nuclear envelope, also allowing the passive entry of importin β-capsid complexes into the nucleus.

IMPORTANCE Parvoviruses are small DNA viruses that deliver their DNA into the postmitotic nuclei, which is an important step for parvoviral gene and cancer therapies. Limitations in virus-receptor interactions or endocytic entry do not fully explain the low transduction/infection efficiency, indicating a bottleneck after virus entry into the cytoplasm. We thus investigated the transfer of parvovirus capsids from the cytoplasm to the nucleus, showing that the nuclear import of the parvovirus capsid follows a unique strategy, which differs from classical nuclear import and those of other viruses.

INTRODUCTION

The vast majority of DNA viruses replicate in the nucleus. This requires that the viral genome traverse the cytoplasm toward the nucleus and subsequently pass the nuclear envelope (NE). The viral DNA is surrounded by viral proteins that form a protective capsid around it. After cytoplasmic entry, the DNA-containing capsids exploit active transport toward the NE using cytoplasmic dynein for their movement along microtubules (1–3). A few viruses, such as human T-cell leukemia virus type 1 (HTLV-1) and papillomaviruses (4, 5), need NE degradation to gain access to the nucleus, but the vast majority infect nondividing cells. This requires passage through the nuclear pore complexes (NPCs), which are the only aqueous connections between the cytoplasm and nucleus. Due to the 39-nm size limit of the pore (6), the capsids of most viruses, such as the nonenveloped adenoviruses or herpes simplex virus 1 (HSV-1), disintegrate at the cytoplasmic side, requiring either a direct or an indirect interaction with the NPC (7). The released viral genomes mostly remain attached to karyophilic viral proteins and are then actively imported (e.g., adenoviral genomes), or the genomes pass the NPC by repulsion forces upon capsid opening (HSV-1) (8–11).

Active nuclear import is needed for macromolecules larger than 2.6 to 9 nm (12–15). It is mediated by nuclear transport receptors of the importin beta (Imp β) superfamily (16–18). In the classical pathway, Imp β attaches via Imp α to a nuclear localization signal (NLS) exposed on the cargo surface (19, 20). Imp α provides the NLS-binding domain and an Imp β-binding domain, whereas Imp β mediates the subsequent interactions with the NPC that drive translocation through the pore into the nuclear basket (21, 22). The basket is a filamentous structure on the nuclear face of the NPC, and in metazoan cells, the nuclear basket is composed of three different nucleoporins (Nups): Tpr, Nup50, and Nup153 (23). This import process is terminated at the C terminus of Nup153, where the import complex is dissociated by the GTP-bound form of ras-related nuclear protein (Ran) (24–26), which binds to Imp β. While the Imp β-RanGTP complex becomes directly recycled back to the cytoplasm, the released cargo and Imp α diffuse deeper into the nucleus (21, 27), from where Imp α becomes exported into the cytoplasm in a complex with the cellular apoptosis susceptibility gene protein (CAS) and RanGTP (28, 29). This pathway also exists in an Imp α-independent form in which the cargo exposes an Imp β-binding domain, allowing direct Imp β binding. Furthermore, nuclear import can be mediated by other functional Imp β homologues such as transportin (30, 31), and the nuclear import of macromolecules and viruses independent of transport receptors has been described (32–35).

Parvovirus capsids (18 to 26 nm) and hepatitis B virus (HBV) capsids (36 nm) are smaller than the physical diameter of the NPC, which allows the passage of intact capsids across the NPC. HBV capsids use Imp α/β-mediated nuclear import (36), but instead of diffusing deeper into the nucleoplasm, the capsids remain attached to Nup153 (37, 38). Parvoviruses comprise protoparvoviruses such as canine parvovirus (CPV) and parvovirus H1 (PV-H1), the latter of which was used in the first phase I/IIa clinical trial for glioblastoma (39, 40). The second group of parvoviruses, constituting the dependoparvovirus genus, includes adeno-associated viruses (AAVs), which are widely used as gene therapy vectors (41).

Most parvoviruses are composed of two viral proteins (VPs), VP1 and VP2, and of nonstructural protein 1 (NS1), which is bound to the viral DNA. They enter cells via clathrin-mediated endocytosis, requiring acidification in late endosomes for productive infection (42–45). The acidification changes the parvovirus structure, leading to the exposure of the N terminus of VP1, which is hidden in the virion. This sequence comprises a cluster of four basic amino acids (MAPPAKRARRGLV in CPV), fulfilling the minimal sequence requirement for a classical NLS (K-K/R-X-K/R) (21, 46–49). It further exhibits a phospholipase A2 domain, which is essential for endosomal release and the progression of infection (2, 50–54) but dispensable for the disintegration of the nuclear envelope in AAV (50).

Parvoviruses reach the perinuclear region of the nucleus by dynein-mediated transport at ∼1 h postinfection (p.i.) (55, 56). The subsequent steps of their NPC interaction have remained controversial. The exposure of a putative NLS supports classical nuclear transport using Imp α/β, which suggests nuclear entry of the intact capsid. This conclusion is supported by analyses of fluorescence fluctuations showing a correlation of CPV capsid and Imp β movements across the NE (55). However, the need for importins is not unequivocally clear, as an unconventional Imp α/Imp β-independent nuclear import motif was found on minute virus of mice (MVM), mediating the nuclear import of at least preassembled capsid subunits (57, 58). In contrast, Porwal et al. reported that a direct attachment of PV-H1 and AAV2 to Nups caused local NE degradation, activating key enzymes in mitosis (59). NE fenestration allows the entry of macromolecules such as IgG (59) but also of papillomaviruses, which have a diameter of 55 nm (5).

To differentiate between these scenarios, which are important for understanding the restricting factors of parvovirus infection, we investigated the CPV capsid interplay with Imp β and the NPC during early infection. Our studies demonstrated the need for Imp β for infection but also showed NE disintegration, allowing us to propose a mixed model of nuclear import, which is consensual with previously contradictious observations.

RESULTS

Cytoplasmic microinjection of Imp β antibodies inhibits CPV infection.

To determine whether Imp β transport is important for capsid nuclear import and the progression of infection, we studied the effect of cytoplasmically comicroinjected anti-Imp β monoclonal antibody (MAb) on infection. Infected cells were detected by the nuclear emergence of CPV capsid proteins. Microinjected cells were identified by comicroinjection with fluorescent dextran (150 kDa; radius of gyration [r_g] = 13 nm) that remains cytoplasmic (60). Imaging at 16 and 24 h p.i. demonstrated that the proportion of nuclear capsid protein-positive cells in the presence of anti-Imp β MAb was strongly reduced (Fig. 1A and B). In nonmicroinjected control cells, capsid proteins accumulated in the nucleus, excluding the nucleoli (61). Finally, in microinjected cells infected for 4 h, viral proteins were detected in the cytoplasm. At this early stage of infection, before the expression of viral proteins, cytoplasmic viral proteins are most likely associated with the entering viral capsids (Fig. 1C). Microinjection of a control antibody (Ab), anti-mouse IgG, does not affect the nuclear import of capsids (2). Fluorescent dextran was not observed in the nucleoplasm, and we conclude that neither the microinjection of the antibodies nor infection induced significant damage to the NE.

FIG 1.

Effect of importin β inhibition on the progression of infection. Importin β (Imp β)-mediated nuclear transport was inhibited by the cytoplasmic microinjection of importin β antibody (Imp β MAb). (A to C) Confocal microscopy images of microinjected cells at 16 h (A), 24 h (B), and 4 h (C) p.i. The infected cells were detected with an anti-VP2 Ab followed by anti-rabbit Alexa Fluor 633 (white), and microinjected cells were identified by coinjection of 500-kDa FITC-dextran (green). The distributions of endogenous Imp β immunostained with Imp β1 MAb and anti-mouse Alexa Fluor 555 are shown as inverted grayscale images in the third column from the left. Bars, 10 μm. (D) Fraction of Imp β antibody-microinjected cells containing nuclear VP2 at 16 h (n = 18) and 24 h (n = 33) p.i. normalized by the fraction of nuclear VP2-positive nonmicroinjected cells (n = 23 at 16 h p.i. and n = 62 at 24 h p.i.). Error bars represent the standard deviations.

In nonmicroinjected cells, Imp β was located in the cytoplasm and at the NE, while the microinjection of anti-Imp β MAb led to its nuclear localization (Fig. 1A and B, gray-and-white panels). This is in agreement with a previous study showing that Imp β antibodies preclude the formation of carrier‐cargo complexes by preventing Imp β and Imp α interactions and subsequent reexport from the nuclear basket (62).

The number of microinjected and nonmicroinjected cells with nuclear capsid proteins was analyzed. The occurrence of nuclear capsid proteins in anti-Imp β-microinjected cells was 77% lower at 16 h p.i. and 74% lower at 24 h p.i. than in control cells (Fig. 1D). This suggests that the anti-Imp β MAb-induced inhibition of the nuclear import of capsids and/or capsid proteins inhibited the progression of infection.

Nuclear entry of CPV is NPC dependent.

Next, we investigated whether CPV infection requires Imp β-dependent nuclear import for passage across the NPC. To address this question, we microinjected HBV capsids and wheat germ agglutinin (WGA) into the cytoplasm prior to infection. HBV capsids block in particular Imp α/β-dependent nuclear import via their interaction with Nup153 (38), and WGA completely blocks active nuclear transport for an hour, with decreasing blocking efficiency after that (about 50% 4 h after microinjection) (63). We analyzed the effect of NPC blockage on the progression of CPV infection and the initiation of replication by detecting the nuclear presence of the viral replication protein NS1. Immunostaining revealed a significant suppression or a complete lack of nuclear NS1 proteins in the presence of HBV capsids at 16 h p.i. (Fig. 2A and B). Quantitative analysis showed an 86% decrease in the number of nuclear NS1-positive HBV-microinjected cells and a 67% decrease in the number of nuclear NS1-positive WGA-microinjected cells, which suggests that CPV has to pass the NPC in order to initiate infection. Since NPC blockage by WGA occurs only at the beginning of the infection before significant viral protein production takes place, the observed decrease of nuclear NS1 later during infection is probably due to the blockage of capsid entry into the nucleus and not due to the blockage of de novo-synthesized capsid proteins or nonstructural proteins.

FIG 2.

Effect of HBV capsid- or WGA-induced blockage of NPCs on the progression of infection. Shown is the progression of infection after NPC blockage caused by the cytoplasmic microinjection of HBV capsids and WGA. (A and B) Confocal images showing the emergence of the replication protein NS1 recognized with an anti-NS1 MAb (white) in cells microinjected with HBV capsids (A) and WGA-Alexa Fluor 488 (B) at 16 h p.i. Cells microinjected with HBV capsids were identified by coinjection of 500-kDa FITC-dextran (green). Bars, 10 μm. (C) Fractions of HBV antibody-microinjected cells (n = 16 microinjected and n = 19 nonmicroinjected cells) and WGA-microinjected cells (n = 96 microinjected and n = 142 nonmicroinjected cells) containing nuclear NS1 at 16 h p.i. normalized by the fraction of nuclear NS1-positive nonmicroinjected cells. Error bars represent the standard deviations.

Based on approximately 4,000 NPCs per cell (64–67), the number of microinjected HBV capsids corresponded to an ∼1:1 ratio of HBV/NPC. This concentration was chosen as previous studies showed that it leads to a reduction of the nuclear import of Imp α/β-dependent karyophilic cargos by 70% (38). This occurs via capsid binding to the nuclear basket-localized Nup153, which is crucial for importin β-cargo dissociation. In contrast, transportin-mediated nuclear import, which is essential for cell viability, remains unchanged (38). Higher capsid concentrations would have caused the blockage of the central channel, as we showed previously by electron microscopy (38). In contrast, 4,000 HBV capsids are far below the number of karyophilic cargos per cell, which practically excludes indirect effects by competing with physiological nuclear import. As HBV capsids at a concentration of 1 capsid per NPC mainly localize in the nuclear basket and not within the channel (38), our finding further indicates that the inhibition of infection was caused not by a steric block of the NPC but rather by blocking the access of CPV capsids to Nup153.

Intracellular capsid interactions with Imp β.

According to the classical nuclear import of karyophilic cargos, CPV capsids should bind Imp α/β in the cytoplasm and pass through the pore into the nuclear basket, where Imp β dissociates, followed by direct recycling into the cytosol. We first verified the capsid-Imp β interaction by coimmunoprecipitation, which showed that CPV from infected cells is in fact bound to Imp β (Fig. 3A). As the precipitation was done at 1 h postinfection, which is before viral proteins are made (68), we conclude that Imp β interacted with incoming capsids and not with progeny ones.

FIG 3.

Capsid-Imp β interactions. The interaction of capsid and Imp β was analyzed by coimmunoprecipitation at 1 h p.i. (A) Immunoblot detection of viral capsid proteins precipitated with Imp β MAb (left) and Imp β precipitated with capsid protein Ab (right). Lysates were from infected and noninfected cells (mock). The arrows indicate positions of the VP2 capsid protein (67 kDa) (left) and Imp β (97 kDa) (right). The proximity of capsid and Imp β was determined with a PLA. (B) Views of cellular middle sections of confocal images showing infected cells labeled with antibodies against capsid proteins and Imp β at 1, 2, 4, and 6 h p.i. (C) Quantitative analysis of the time-dependent change in the number of PLA signals detected at 1, 2, 4, and 6 h p.i. (n = 17, 12, 5, and 2). The error bars represent the standard errors of the means. (D) Positive and negative controls showing PLA signals of the VP2 capsid proteins and intact capsids in noninfected cells (left) and infected cells at 1 h (middle) and 4 h (right) p.i. (E) Technical controls using the two PLA probes without (left) and with Imp β MAb (middle) and capsid Ab (right). RAb, rabbit polyclonal antibody. Bars, 10 μm.

To further analyze the temporal and spatial location of capsid-Imp β interactions, we used an in situ proximity ligation assay (PLA) (69). This method allows the detection of single intermediate or direct interactions between two proteins in their native form. The interactions, even weak or transient, are detectable when the distance between the proteins is less than 40 nm. We analyzed the capsid-Imp β PLA signal distribution at 1, 2, 4, and 6 h p.i., considering that newly synthesized viral capsid proteins can be disregarded for at least 4 h p.i. (68). Figure 3B shows that the majority of signals were located in the perinuclear cytoplasm, but a few signals were also found in the nuclear area. Quantitative analysis at various times after infection showed a temporal increase in the number of PLA signals (Fig. 3C). At 1 h p.i., the PLA signal density in the two-dimensional (2D) maximum-intensity projections of the cells was 0.025 ± 0.008 PLA signals/μm² (mean PLA signal/area ± standard deviation), with 19.0 ± 1.4 PLA signals per cell. At 2 and 4 h p.i., an increase in the interaction was detected (0.06 ± 0.020 PLA signals/μm² and 60 ± 30 PLA signals/cell, and 0.12 ± 0.04 PLA signals/μm² and 100 ± 50 PLA signals/cell, respectively). At 6 h p.i., the mean PLA signal density slightly decreased (0.08 ± 0.04 PLA signals/μm² and 90 ± 8 PLA signals/cell), which was, however, within the range of the variability between individual cells. The positive control of the PLA signal between antibodies against VP2 capsid protein and intact capsids at 1 h p.i. verified the specificity of the PLA signal (Fig. 3D). When the capsids were labeled with antibodies against capsid proteins and intact capsids at 1 h p.i., the total numbers of signals were 0.09 ± 0.04 PLA signals/μm² and 88 ± 30 PLA signals/cell (n = 9 cells). The negative viral capsid antibody control in noninfected cells (Fig. 3D) and technical controls with PLA probes only in infected cells at 1 h p.i. (Fig. 3E) indicated that the background level was low (approximately 7.18 PLA signals/cell [n = 170] and 0.09 PLA signals/cell [n = 40], respectively).

To study the nuclear and cytoplasmic distributions of the interaction, we analyzed three-dimensional (3D) confocal microscopy images of PLA signals at 1 h p.i. The analysis showed that the majority of the PLA signals were located in the cytoplasm, but some of the signals were located inside the nucleus (Fig. 4A). To quantitate the signal localization, the PLA signals were segmented, and their lateral distances to the NE, defined by the border of DAPI (4′,6-diamidino-2-phenylindole) staining, were calculated (Fig. 4B and C). The analysis showed that at 1 h p.i., on average, 13 ± 7 signals were located in the nucleus, and 90 ± 50 signals were located in the cytoplasm. The number and density of cytoplasmic signals increased from the cell periphery toward the NE. This is in agreement with previous studies showing that parvovirus capsids are transported into the nucleus in a similar time frame after infection (2, 56, 57, 70). The accumulation of PLA signals in the nuclear periphery further implies that a structural change in the acidic environment, which allows importin binding, is preserved after capsid neutralization. Quantitative analysis showed that within the nucleus, the highest numbers and densities were found close to the NE, but interestingly, a small portion of signals were also located deeper, within 0.75 to 6.0 μm from the NE.

FIG 4.

Cellular distribution of capsid-Imp β interactions. Capsid and Imp β interactions were visualized in infected cells. (A) Confocal sections of DAPI and PLA signals at 1 h p.i. The intranuclear localization of PLA signals is shown by xy and yz slices of the segmented nucleus and PLA signals, obtained by visualizing the PLA signals, in black, in the white nucleus. The yz slice is taken along the line shown in red. Bars, 5 μm. (B) Number of segmented PLA signals at 1 h p.i. as a function of the lateral (xy) distance from the nuclear envelope (n = 16). (C) Density of PLA signals at 1 h p.i. as a function of the lateral distance from the nuclear envelope (n = 16). The negative distance values denote the distance to the cytoplasmic side and positive values denote the distance to the nuclear side of the nuclear envelope. The error bars show the standard errors of the means.

In summary, these findings indicate that the nuclear entry of capsids is preceded by the cytoplasmic interaction of capsid with Imp β and is followed by the nuclear access of capsids, which, in contrast to classical karyophilic cargos, were still interacting with Imp β.

DISCUSSION

The nuclear entry of parvoviral capsids is controversial, as previous observations indicate classical nuclear import via transport receptors or by transient permeabilization of the NE following attachment to the NPC.

The first model requires a structural change upon virus entry leading to the exposure of a classical NLS on the N terminus of VP1. As all parvoviruses exhibit diameters below the transport limit of the nuclear pore, it was hypothesized that the capsid-Imp β complex follows the classical nuclear import pathway, which comprises the dissociation of the complex in the nuclear basket after binding to Nup153 (71–73). This would result in capsid entry into the nucleus while Imp β becomes recycled back to the cytoplasm. In agreement with this model, our results show that anti-Imp β antibody microinjection strongly inhibited infection. Inhibition was not complete, indicating the presence of functional Imp β molecules, which in turn are required to maintain cell viability. Noteworthy, the need for Imp β for CPV infection does not exclude the requirement for Imp α as an adaptor molecule. Further support for this model comes from our observation that no significant entry of 150-kDa dextran was observed, indicating that the NE remained intact. The conclusion that CPV capsids need direct or indirect access to Nup153 for infection is also congruent, as Imp β dissociates from cargos after interaction with Nup153 within the basket (72).

The second model, based on NE permeabilization, neither requires nor excludes VP1 N-terminal domain exposure during early infection. Coimmunoprecipitation studies showed that parvoviruses interact directly with some NPC proteins, enhancing the exposure of the N-terminal VP1 domain. The identified nucleoporins localize in all parts of the NPC, including cytoplasmic filaments (Nup358), the central channel (Nup62), and the nuclear basket (Nup153) (59). These findings, however, do not exclude that Imp β could mediate attachment, possibly prior to a direct capsid interaction with the NPC. This would be similar to what occurs during the transport of HBV capsids (36, 37), where Imp β allows the passage of the capsid through the central channel of the NPC, which is a requirement for its interaction with Nup153. In the experiments leading to this model, nuclear permeabilization was observed not only after cytoplasmic microinjection of different parvoviruses (AAV2 and PV-H1) but also after infection (PV-H1) allowing the nuclear entry of polyomaviruses (5). Evidently, our observation that at least no significant amounts of 150-kDa dextran entered the nucleus seemingly contradicts pore formation in the NE in particular, since antibodies (a typical IgG is ∼14.2 nm in diameter) entered the nucleus after parvoviral microinjection (59, 74, 75). It must be considered, however, that the numbers of microinjected capsids in these experiments were much higher than what can be achieved with infections and that nuclear permeabilization is transient, allowing maintenance of the nuclear-cytoplasmic gradient of, e.g., RanGTP, which is crucial for cell viability (76, 77). In support of nuclear permeabilization, we observed a fraction of nuclear CPV capsids associated with Imp β. This observation indicates that at least these capsids had not interacted with Nup153, which is a prerequisite for Imp β dissociation.

Collectively, our data allow the proposal of a composite model combining the key observations which led to the previously proposed nuclear entry models (Fig. 5). After cell entry (Fig. 5A), acidification in the endosomes (Fig. 5B) triggers structural changes leading to the exposure of the VP1 N-terminal domain, which comprises phospholipase A2 and the NLS on the capsid surface (2, 78). The phospholipase activity disintegrates the endosomal membrane, allowing the interaction of the NLS with Imp α and β after endosomal opening (Fig. 5C) or after capsid release from the endosome during capsid transport using microtubules (79–82) (Fig. 5D). Imp β then allows binding to the NPC and passage through the pore (Fig. 5E), which is in agreement with the concomitant movement of Imp β and CPV that we observed previously (54). The parts of the capsid that are not masked by Imp β interact with Nup358, Nup62, and Nup153 (Fig. 5F). Once translocation is terminated in the nuclear basket by the interaction of Imp β with Nup153, RanGTP dissociates the import complex, leading to the recycling of Imp β into the cytoplasm (Fig. 5G). Concomitantly, the NPC-bound capsids disintegrate the NPC/NE (Fig. 5H), probably by permeabilizing the nuclear membrane, which triggers Ca²⁺ release, a known initiator of NE degradation in mitosis (59). The holes in the NE then allow the passive entry of cytosolic capsid-Imp complexes into the nucleus (Fig. 5I).

FIG 5.

Schematic representation of the entry and nuclear import model of CPV capsids. The magnification shows events in the nuclear pore. Capsids are shown as blue hexagons containing the viral single-stranded DNA (ssDNA) genome (green circles). The dashed lines indicate degradation. Further details are given in the text. Gray, cytoplasm; white, nucleus; orange, Imp α; dark blue, Imp β; green ellipses, nucleoporins.

While our data and the resulting model allow consistent interpretations of previous findings, we must admit that we cannot conclude whether the NPC-bound or the free nuclear capsids initiate infection. The data, however, support a new and unique model of virus transport across the NE combining features of phylogenetically distant viruses such as HTLV-1 and papillomaviruses, which need NE permeabilization during mitosis for nuclear entry, with those of, e.g., adeno- and herpesviruses, which bind to the NPC, triggering genome release and subsequent passage of the genome through the nuclear pore.

MATERIALS AND METHODS

Cell lines and viruses.

Norden laboratory feline kidney (NLFK) cells were cultured in Dulbecco’s modified Eagle medium (DMEM) with GlutaMAX supplemented with 10% fetal bovine serum (FBS), 1% nonessential amino acids, and 1% penicillin-streptomycin (Gibco, Thermo Fisher Scientific, Waltham, MA) at 37°C in the presence of 5% CO₂. Canine parvovirus (CPV) type 2 was a generous gift from Colin Parrish (Cornell University, Ithaca, NY) and derived from an infectious plasmid clone, p265, by transfection of NLFK cells as previously described (83, 84). The viruses were grown, isolated, and concentrated as described previously (82). For infection, the cells were inoculated with CPV (multiplicity of infection [MOI] of 1 to 2) and incubated at 37°C in a humidified incubator (5% CO₂) for the duration of the infection.

Antibodies.

CPV VP2 proteins were detected with a rabbit antibody (Ab), and intact capsids were detected with a mouse monoclonal antibody (MAb) from Colin Parrish (Cornell University, Ithaca, NY). The viral nonstructural protein NS1 was localized by a MAb obtained from Caroline Astell (85). Primary antibodies were detected with Alexa Fluor 488-, 555-, or 633-conjugated anti-mouse/rabbit IgGs (Thermo Fisher Scientific). In microinjection studies, an Imp β1 MAb (anti-KPNB1 [clone 3E9], catalogue number ab2811; Abcam, Cambridge, UK) was used to interfere with the docking of the Imp-substrate complex to the NPC.

Confocal microscopy of fixed cells.

For laser scanning confocal microscopy, cells were grown on glass coverslips, fixed at set time intervals with 4% paraformaldehyde (PFA), and permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) supplemented with 1% bovine serum albumin (BSA) and 0.01% sodium azide. The cells were embedded with ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI; Thermo Fisher Scientific). Images were acquired with an Olympus FV-1000 confocal microscope with a UPLSAPO 60× oil immersion objective (numerical aperture [NA] = 1.35). Images of 800 by 800 pixels were acquired with line averaging of 2 and with voxel sizes of 88 nm in the x and y dimensions and 150 nm in the z dimension (zoom factor of 2). Five-hundred-kilodalton fluorescein isothiocyanate (FITC)-dextran was excited with a 488-nm argon laser, and fluorescence was collected with a 510- to 540-nm-band-pass filter; Alexa Fluor 555 and PLA probes conjugated with Alexa Fluor 594 were excited with a 543-nm He-Ne laser, and fluorescence was collected with a 570- to 620-nm-band-pass filter; and Alexa Fluor 633 was excited with a 633-nm He-Ne laser, and fluorescence was collected with a 647-nm-long-pass filter. DAPI was excited by a 405-nm diode laser and monitored with a band-pass filter of 460 to 500 nm.

Immunoblotting and coimmunoprecipitation.

Cells were cultured on 56-cm² dishes. Infected and mock-infected NLFK cells were lysed with ice-cold hypotonic gentle lysis buffer (10 mM Tris-HCl [pH 7.5], 10 mM NaCl, 2 mM EDTA, 0.1% Triton X-100) supplemented with a protease inhibitor (250 μl/5 ml CEB, catalogue number P-2714; Merck KGaA, Darmstadt, Germany) on ice for 30 min. The suspension was centrifuged (10,000 × g for 15 min at +4°C), and the supernatant was decanted. Nuclei were resuspended in 0.5% Tween 20 in PBS with 500 mM NaCl, incubated on ice for 15 min, and centrifuged (10,000 × g for 15 min at +4°C). The collected supernatants were combined to produce a total cell lysate.

For coimmunoprecipitation assays, the total cell extracts (200 μl) were incubated with rotation at room temperature (RT) for 1 h with 50 μl (1.5 mg) protein G-Dynabeads (Thermo Fisher Scientific) coated with BS³ (Thermo Fisher Scientific)-cross-linked anti-VP or anti-Imp β (10 μg) antibodies freshly bound to beads in 200 μl PBS with Tween 20 for 10 min at RT with rotation. After washes, the Dynabeads-Ab-antigen complexes were resuspended in 100 μl washing buffer and transferred into a new tube. Target antigens were eluted by resuspending the complexes directly in preheated SDS sample buffer and heated for 10 min at 70°C. The eluates were boiled for 10 min and cooled to RT prior to gel (12%) loading. After electrophoresis, the proteins were transferred to nitrocellulose membranes. The membranes were blocked for 16 h at 4°C with 5% BSA in 1% Tween 20–1× TEN. For immunoblotting, primary antibodies against Imp β and capsid proteins were diluted in 1% Tween 20–1× TEN and incubated at RT for 1 h. For detection, horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse IgG (Bio-Rad) was diluted in 0.1% Tween 20–1× TEN (1:2,000) and incubated at RT for 1 h. A SuperSignal West Pro Pico chemiluminescence detection kit (Thermo Fisher Scientific) was applied for the detection of the proteins. The signal was collected quantitatively with a Chemidoc XRS hood (Bio-Rad, UK) equipped with a CoolSNAP HQ2 charge-coupled-device (CCD) camera. Saturation of the signal was avoided with exposure time adjustment.

In situ proximity ligation assay.

For in situ proximity ligation assays (PLAs) (69), cells were grown on 8-well chamber slides (Nunc Lab-Tek II chamber slide system; Thermo Fisher Scientific) to 80 to 90% confluence. The cells were fixed with 4% PFA in PBS and permeabilized with 0.1% Triton X-100 in PBS supplemented with 1% bovine serum albumin and 0.01% sodium azide. Analysis was performed with a Duolink II kit (Olink Bioscience, Uppsala, Sweden). Anti-mouse PLA probes were diluted in 3% bovine serum albumin in PBS and incubated on cells for 1 h at 37°C in a humidified chamber, followed by ligation and amplification according to the manufacturer’s protocol. The samples were embedded in ProLong Gold antifade with DAPI (Thermo Fisher Scientific). The signal was detected with an Olympus FV-1000 confocal microscope and a UPLSAPO 60× oil immersion objective (NA = 1.35). The images of single cells were 512 by 512 pixels with voxel sizes of 69 or 82 nm in the x and y dimensions and 150 nm in the z dimension (line averaging of 2).

The intracellular distribution of PLA signals was analyzed by segmenting the nucleus and PLA signals from the fluorescence microscopy images using minimum cross-entropy segmentation (86). The geometric centers of PLA signals were calculated, and their lateral (xy) distances to the border of the segmented nucleus were determined. The distance values of the centroids were sorted into 750-nm-wide bins, and the mean number and density of PLA signals for each bin were then calculated, resulting in a graph of PLA signal numbers and densities as a function of the distance from the NE. Because the axial resolution of confocal microscopy images is much lower than the lateral resolution, it was difficult to determine if PLA signals located near the top and bottom of the nucleus were inside the nucleus or not. For this reason, the lateral distance to the NE was selected as the distance metric, and only those PLA signals that were within 750 nm from the plane of the geometric center of the nucleus were accepted for analysis.

Microinjection.

For microinjection, the cells were cultured on 21.5-cm² glass-bottom culture dishes (Cultureware; MatTek, Ashland, MA) or on cover glasses (Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany) to 80 to 90% confluence. Microinjection was accomplished with a system comprised of a Transjector 5246 instrument and a Micromanipulator 5171 instrument (Eppendorf, Hamburg, Germany) on a Zeiss LSM510 confocal microscope (equipped with Nomarski differential interference contrast [DIC]). Microinjection capillaries (Femtotips) were purchased from Eppendorf AG (Hamburg, Germany). The injection pressure used was 250 hPa, leading to a cytoplasmic injection volume of 0.01 to 0.05 pl (87). Imp β MAb combined with injection buffer (PBS) was concentrated to 2.5 mg/ml. Based on previous estimations of the number of NPCs (∼4,000 NPCs/HeLa cell nucleus) (64, 68, 88), the cells were microinjected with HBV capsids at a ratio of 1 to 2 capsids/NPC. An Alexa Fluor 488 conjugate of WGA was diluted in the injection buffer and used at a concentration of 2.5 mg/ml. To identify the microinjected cells, 4 μl HBV capsids or anti-Imp β was combined with 1 μl of 500-kDa FITC-dextran (12.5 mg/ml; Molecular Probes, Thermo Fisher Scientific) and centrifuged for 15 min at 10,000 × g before filling them into the capillaries and injecting them into cells. HBV-microinjected cells were incubated at 37°C with 5% CO₂ for 30 min prior to infection. At the end of the duration of infection, cells were fixed with paraformaldehyde (4% in PBS), immunostained, and embedded in ProLong Gold antifade with DAPI (Thermo Fisher Scientific). The effects of microinjection-induced inhibition of nuclear transport were analyzed by calculating from confocal microscopy images the proportions of infected and noninfected cells, with or without microinjection, based on the presence of a nucleoplasmic NS1 or VP2 antibody signal.