Selective Removal of FG Repeat Domains from the Nuclear Pore Complex by Enterovirus 2A^pro

ABSTRACT

Enteroviruses proteolyze nuclear pore complex (NPC) proteins (Nups) during infection, leading to disruption of host nuclear transport pathways and alterations in nuclear permeability. To better understand how enteroviruses exert these effects on nuclear transport, the mechanisms and consequences of Nup98 proteolysis were examined. The results indicate that Nup98 is rapidly targeted for degradation following enterovirus infection and that this is mediated by the enterovirus 2A protease (2A^pro). Incubation of bacterially expressed or in vitro-translated Nup98 with 2A^pro results in proteolytic cleavage at multiple sites in vitro, indicating that 2A^pro cleaves Nup98 directly. Site-directed mutagenesis of putative cleavage sites identified Gly374 and Gly552 as the sites of 2A^pro proteolysis in Nup98 in vitro and in infected cells. Indirect immunofluorescence assays using an antibody that recognizes the N terminus of Nup98 revealed that proteolysis releases the N-terminal FG-rich region from the NPC. In contrast, similar analyses using an antibody to the C terminus indicated that this region is retained at the nuclear rim. Nup88, a core NPC component that serves as a docking site for Nup98, also remains at the NPC in infected cells. These findings support a model whereby the selective removal of Nup FG repeat domains leads to increased NPC permeability and inhibition of certain transport pathways, while retention of structural domains maintains the overall NPC structure and leaves other transport pathways unaffected.

IMPORTANCE Enteroviruses are dependent upon host nuclear RNA binding proteins for efficient replication. This study examines the mechanisms responsible for alterations in nuclear transport in enterovirus-infected cells that lead to the cytoplasmic accumulation of these proteins. The results demonstrate that the enterovirus 2A protease directly cleaves the nuclear pore complex (NPC) protein, Nup98, at amino acid positions G374 and G552 both in vitro and in infected cells. Cleavage at these positions results in the selective removal of the FG-containing N terminus of Nup98 from the NPC, while the C terminus remains associated. Nup88, a core component of the NPC that serves as a docking site for the C terminus of Nup98, remains associated with the NPC in infected cells. These findings help to explain the alterations in permeability and nuclear transport in enterovirus-infected cells and how NPCs remain functional for certain trafficking pathways despite significant alterations to their compositions.

INTRODUCTION

Members of the Picornaviridae are small nonenveloped viruses with positive-strand RNA genomes that are responsible for a variety of diseases in humans and animals (reviewed in reference 1). The family includes enteroviruses, such as poliovirus and rhinovirus, along with cardioviruses, such as encephalomyocarditis virus and Theiler's murine encephalomyelitis virus. Following entry of these viruses into the host cell, translation and replication of the RNA genome occur exclusively in the cytoplasm. Despite this spatial compartmentalization, evidence suggests that picornaviruses rely on activities provided by host nuclear RNA binding proteins for efficient translation and RNA replication. The strongest support for this comes from experiments using either in vitro or in vivo systems that have demonstrated interaction of host nuclear factors with viral proteins or RNA elements, along with modulation of translation and/or replication of the viral genome (2–8). Consistent with a role for host nuclear factors in viral replication, many have been shown to redistribute to the cytoplasm following infection (2, 4, 7–13). Redistribution of host nuclear RNA binding proteins is brought about by alterations in the nuclear transport machinery that occur during infection.

Nuclear transport describes the bidirectional movement of cargos across the nuclear envelope. This process is dependent upon an ∼100-MDa protein complex called the nuclear pore complex (NPC) that is found embedded in the nuclear envelope (reviewed in reference 14). NPCs exhibit 8-fold symmetry and consist of a central scaffold that spans the nuclear envelope and that is capped by cytoplasmic and nucleoplasmic ring structures. The nucleoplasmic ring serves as the anchor point for filaments that connect to a second, smaller ring to form a basket-like structure. The cytoplasmic ring has eight fibrils that extend outward into the cytoplasm. Despite their large size, NPCs are composed of only about 30 different proteins that are collectively called nucleoporins (Nups) (15, 16). About one-third of Nups are referred to as FG-Nups due to the presence of domains containing phenylalanine-glycine (FG) repeats. The regions rich in FG repeats are intrinsically disordered and are found lining the transport channel and in the peripheral NPC structures, such as the cytoplasmic fibrils and nuclear basket (17). The protein meshwork created by the interactions between these disordered FG repeat regions creates a barrier to diffusion of molecules larger than ∼40 kDa and also provides binding sites for transport receptors as they move cargo across the NPC (18, 19).

Enteroviruses and cardioviruses have both been shown to alter the NPC, although the mechanisms employed are quite different. Cardioviruses modify the phosphorylation status of Nups via the action of the Leader (L) protein, a small protein not coded for by enteroviruses (20–24). In addition, the L protein also binds to and interferes with the function of Ran, a small GTPase critical for the nuclear transport of most cargos (25). In contrast, enteroviruses, such as poliovirus and rhinovirus, induce the proteolysis of Nups (26, 27). The results of these disparate modifications are surprisingly similar: increased permeability of the NPC and inhibition of nuclear import of certain cargos (21, 23, 26–29). These changes bring about the net cytoplasmic accumulation of many, although not all, host nuclear factors and may also help to impede cellular responses to infection, thus ensuring efficient viral replication.

During poliovirus infection, several FG-Nups are proteolyzed, including Nup153, Nup62, and Nup98 (27, 30). In HeLa cell monolayers, cleavage of Nup153 and Nup62 occurs between 3 and 4 h postinfection (p.i.) and correlates temporally with the cytoplasmic accumulation of host nuclear factors and inhibition of nuclear import (27). In contrast, the majority of Nup98 is proteolyzed within 1 h following poliovirus infection (30). Furthermore, cleavage of Nup153 and Nup62 is prevented by the addition of guanidine, an inhibitor of viral RNA synthesis, while cleavage of Nup98 is unaffected (30). These results indicate differences in the mechanisms underlying proteolysis of Nup98 compared to other FG-Nups, although the significance of these differences is not understood. Despite the differences in the kinetics of cleavage and sensitivity to guanidine, the viral 2A protease (2A^pro) is the primary protease responsible for the proteolysis of Nup98, Nup153, and Nup62 (28, 30–34).

In this study, we have extended our analysis of the cleavage of Nup98 during enterovirus infection. The results indicate that cleavage of Nup98 following poliovirus or rhinovirus type 2 infection results in the release of the N-terminal FG domain of Nup98 from the NPC while the C-terminal domain is retained. Consistent with these observations, cleavage analysis of endogenous and in vitro-translated Nup98 using purified 2A^pro has identified two cleavage sites located between the N-terminal FG-rich region and the C-terminal region at amino acid positions Gly374 and Gly552. These results provide further insights into how enteroviruses proteolytically alter the structure of the NPC to increase permeability and modify host nucleocytoplasmic transport pathways.

MATERIALS AND METHODS

Cell culture and virus.

HeLa and HEK293 cells were maintained in monolayers in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM l-glutamine, and penicillin-streptomycin at 37°C in 5% CO₂. The HGP strain of human rhinovirus type 2 (HRV2) and the 1059 strain of HRV14 were purchased from the ATCC, and viral stocks were prepared by infection of HeLa monolayers. Mahoney type 1 poliovirus (PV) stocks were prepared as described previously (11). HeLa cells at 80% confluence were either mock infected or infected at a multiplicity of infection (MOI) of 50 for the indicated times. Virus was adsorbed for 30 min at 32°C (HRV2 and HRV14) or 37°C (PV) in phosphate-buffered saline (PBS) supplemented with 1 mM MgCl₂ and 1 mM CaCl₂. After adsorption, unbound virus was removed, and DMEM with 10% FBS, 2 mM l-glutamine. and penicillin-streptomycin was added.

Plasmids.

The full-length human Nup98 open reading frame (ORF) was isolated from pAlterMaxMycNup98 (35) by digestion with SalI and NotI and subcloned into the same sites in pET30c(+) (Novagen) to create pET30(+)-Nup98, which encodes full-length Nup98 with an N-terminal 6-His tag. This plasmid was used to overexpress Nup98 in BL21(DE3)RIPL cells. The Nup98 ORF was PCR amplified with NcoI and XhoI restriction sites and subcloned into the corresponding sites in pET28b(+) (Novagen) to create pET28b(+)-Nup98, which encodes full-length Nup98 with a C-terminal 6-His tag. This construct was used for in vitro translation.

For overexpression in HeLa cells, the Nup98 ORF was PCR amplified using primers that introduced a FLAG tag on the N terminus and subcloned into the BamHI and EcoRI sites of pCDNA3.1.

Protein purification.

The pET30(+)-Nup98 construct was transformed into BL21(DE3)RIPL, and Nup98 protein expression was induced by addition of 1 mM isopropyl-β-d-thiogalactoside (IPTG) when cultures reached an optical density at 600 nm (OD₆₀₀) of 0.6. After induction, cultures were additionally incubated for 4 h at 37°C, sonicated, and then clarified by centrifugation at 38,500 × g. Nup98 localized to the insoluble fraction was solubilized with buffer B (8 M urea, 0.1 M NaH₂PO₄, and 0.01 M Tris-Cl [pH 8.0]) and then purified on Ni-NTA (nitrilotriacetic acid) Spin columns as described by the manufacturer (Qiagen). Purified Nup98 was dialyzed twice for 4 h at 4°C and then overnight against 1 liter of refolding buffer (400 mM l-arginine, 0.4 M Tris-Cl [pH 8.0], 0.2 mM EDTA, 5 mM reduced glutathione, 0.5 mM oxidized glutathione). Nup98 was concentrated by placing the dialysis bag on a bed of polyethylene glycol 20,000 for 1 h, quantified using the Bio-Rad protein assay kit, and stored at −20°C.

Immunoblotting.

HeLa whole-cell lysates were prepared by washing cells twice with PBS, followed by a 20-min incubation on ice in Tx lysis buffer (36). The lysates were cleared by centrifugation at 20,000 × g for 5 min and quantified using the Bio-Rad protein assay kit. Equal quantities of protein were separated by SDS-PAGE, followed by transfer to a polyvinylidene difluoride (PVDF) membrane (Millipore). Nup98 was detected by monoclonal antibodies recognizing either the N-terminal region (amino acids [aa] 1 to 466; anti-Nup98N antibody [Ab]; Sigma-Aldrich catalog number N1038) or residues surrounding proline 654 (anti-Nup98C antibody; Cell Signaling Technology catalog number 2598). Purified monoclonal mouse anti-Nup62 antibody (BD Transduction Laboratories; catalog number 610498) and anti-Nup88 antibody (BD Transduction Laboratories; catalog number 611896) were used to detect Nup62 and Nup88, respectively. Mouse monoclonal antibody MS3 was used to detect nucleolin (37). eIF4G was detected using rabbit polyclonal antibody (Cell Signaling Technology; catalog number 2498). FLAG-Nup98 was detected using mouse monoclonal antibody (Sigma-Aldrich; catalog number F1804). Antibody-antigen complexes were detected using a horseradish peroxidase (HRP)-conjugated secondary antibody and chemiluminescence.

In vitro cleavage assay.

Uninfected HeLa whole-cell lysates were prepared as described above. HRV2 2A^pro was purified from bacteria as described previously (38). For cleavage assays, the indicated amount of HRV2 2A^pro was incubated with 25 μg of uninfected HeLa whole-cell lysates, in vitro-translated Nup98, or the indicated amount of purified Nup98 at 30°C for the indicated amount of time in 2A^pro reaction buffer (39). After the incubation, samples were electrophoresed and analyzed by immunoblotting or autoradiography as described above.

Mutagenesis and in vitro translation.

Point mutants were constructed at putative cleavage sites in Nup98 using the QuikChange II site-directed mutagenesis kit (Stratagene) and confirmed by sequencing. For in vitro translation, 0.5 μg of plasmid containing the wild-type (wt) or mutant Nup98 open reading frame was incubated with 1 μl of TnT reaction buffer (Promega), 12.5 μl of rabbit reticulocyte lysate, 0.5 μl of T7 RNA polymerase, 0.5 μl of 1 mM amino acid mixture without methionine, and 10 μCi of [³⁵S]methionine (>1,000 Ci/mmol; GE Healthcare) for 90 min at 30°C. After incubation, samples were stored at −80°C.

Nup98 expression in HeLa cells.

Lipofectamine 2000 reagent (Life Technologies) was used for transient transfection as recommended by the manufacturer. One day after transfection, cells were infected with HRV2 at an MOI of 10. Eighteen hours after infection, cells were harvested for immunoblotting as described above.

Indirect immunofluorescence.

Mock- or virus-infected cells growing on 12-mm glass coverslips were washed three times with PBS, fixed in 3% formaldehyde for 20 min at 25°C, washed three times with PBS, and permeabilized in methanol at −20°C for 5 min. Following fixation/permeabilization, the cells were washed three times with PBS and then blocked in base solution (PBS containing 2% bovine serum albumin and 0.05% Triton X-100) for 15 min at 25°C. Coverslips were inverted onto 40 μl of base solution containing diluted primary antibody (1:1,000 for anti-Nup98N antibody, 1:40 for anti-Nup98C antibody, and 1:80 for anti-Nup88 antibody) and incubated overnight at 4°C. The coverslips were then washed three times in base solution at 25°C and then inverted into 40 μl of base solution containing a 1:2,000 dilution of either Alexa Fluor 555-conjugated goat anti-rabbit, Alexa Fluor 488-conjugated donkey anti-rat, or Alexa Fluor 555-conjugated goat anti-mouse immunoglobulin (Invitrogen) and incubated for 1 h at 25°C. The coverslips were washed twice in PBS and once in PBS containing 0.2 μg/ml Hoechst 33258, drained, and mounted in Vectashield mounting medium (Vector Laboratories) onto glass slides. Images were acquired using either a Nikon Eclipse E1000 microscope with a 60× objective, a Hamamatsu Orca digital camera, and Metamorph software or with a Zeiss Axio Imager using a 63× objective and an Axiocam 506 camera and Zen Pro software.

RESULTS

Cleavage of Nup98 during human rhinovirus type 2 infection.

Previously, we showed that HRV2 2A^pro induced the proteolysis of Nup98 when added to uninfected lysates (30). However, the status of Nup98 following HRV2 infection had not been examined. Consequently, we performed immunoblotting assays on HRV2-infected HeLa cell lysates using antibodies to Nup98 and Nup62. Over repeated experiments, we observed that Nup98 was cleaved more rapidly than Nup62. Figure 1 shows a representative experiment where steady-state levels of Nup98 were reduced by more than 60% at 6 h p.i. and by more than 90% at 8 h p.i. (Fig. 1). Comparison to the levels of Nup62 revealed that they did not decline as rapidly, with levels dropping by only 40% and 73% at 6 and 8 h p.i., respectively (Fig. 1). Analysis of eIF4G in this setting confirmed the cleavage of this 2A^pro substrate and revealed that cleavage began by 4 h p.i. (slightly before cleavage of Nup98) and was essentially complete by 6 h p.i. These results indicate that, similar to what was seen in poliovirus- and HRV16-infected cells (30, 34), Nup98 is proteolyzed more rapidly than is Nup62 following HRV2 infection.

FIG 1.

Cleavage of Nup98 during HRV2 infection. Twenty micrograms of whole-cell lysates from mock- or HRV2-infected HeLa cells was prepared at the indicated times after infection. The whole-cell lysates were analyzed by immunoblotting with monoclonal anti-Nup98 antibody and anti-Nup62 antibody to detect Nup98 and Nup62, respectively. Monoclonal antibody 2498 was used to detect eIF4G. Nucleolin was detected by monoclonal antibody MS3. The relative amounts of Nup98 and Nup62 were determined using ImageJ and normalized to the amount of nucleolin. The amounts relative to mock-infected cells are indicated (% Remaining). The experiment was repeated at least 3 times, and representative results are shown.

2A^pro induces the cleavage of Nup98.

To better characterize Nup98 proteolysis, we performed in vitro cleavage assays with uninfected HeLa whole-cell lysates and purified HRV2 2A^pro, followed by immunoblotting with N- and C-terminal antibodies to Nup98. Our prior analysis had revealed a putative cleavage product (cp) of approximately 55 kDa when reactions were probed with an antibody to the N terminus of Nup98 (30). Figure 2A confirms the presence of the 55-kDa cleavage product (cp1) and reveals that a second cleavage product of approximately 34 kDa was also readily detected with this antibody (cp2). Closer inspection reveals that cp1 appeared very rapidly following the addition of 2A^pro (within 10 min), with levels remaining relatively constant until 2 h and then declining from 2 to 8 h. In contrast, cp2 was not readily detected until 1 to 2 h after the addition of 2A^pro, and levels increased slightly at later time points. These results suggest the cleavage site responsible for cp1 is recognized earlier than that responsible for cp2 and that cp2 may be derived by further proteolysis of cp1. Reprobing of these blots with an antibody to the C terminus (anti-Nup98C) revealed another cleavage product (cp3) that migrated around 35 kDa. This cleavage product did not comigrate with cp2 and thus represents a distinct proteolytic product. These results indicate that Nup98 is proteolyzed at multiple sites in a 2A^pro-dependent manner.

FIG 2.

HRV2 2A^pro induces proteolysis of Nup98 in whole-cell lysates. (A) 2A^pro cleaves Nup98 at multiple locations. A single lysate from uninfected HeLa cells was prepared, and 25 μg was incubated in cleavage buffer with or without 0.1 μg of purified HRV2 2A^pro for the indicated times at 30°C. After the reaction, the lysates were analyzed by immunoblotting with anti-Nup98N antibody and anti-Nup98C antibody to detect Nup98 cleavage, followed by anti-eIF4G antibody. Molecular mass markers are indicated in kilodaltons. (B) Differential kinetics of proteolysis in vitro for Nup98 and Nup62. The blot shown in panel A (anti-Nup98N) was stripped and reprobed with anti-Nup62 antibody. The band intensities of the remaining full-length Nup98 and Nup62 at each time point were measured using ImageJ. (C) Nup98 proteolysis is inhibited by an inhibitor of 2A^pro. Cleavage reactions were carried out as described above with or without 0.6 μg of purified HRV2 2A^pro and 70 μM MPCMK. MPCMK was incubated with 2A^pro for 20 min at 30°C before adding whole-cell lysates and incubating for 4 h. The reaction mixtures were analyzed by immunoblotting with anti-Nup98 and anti-nucleolin antibodies. (D) 3C^pro does not efficiently proteolyze Nup98. Cleavage reactions were carried out as described above with or without the indicated amount of HRV14 3C^pro for the stated time. The reaction mixtures were analyzed by immunoblotting with anti-Nup98N antibody and anti-NF-κB antibody. HeLa whole-cell lysates prepared from cells infected with HRV14 for 7 h were analyzed by immunoblotting as a positive control. Each experiment was repeated a minimum of 3 times, and representative results are shown.

Analysis of HRV2-infected cells indicated that Nup98 was proteolyzed earlier than Nup62 (Fig. 1). This could be due to Nup98 being more sensitive to proteolysis or simply being more accessible to the viral protease, for example, due to different locations at the NPC. To distinguish these possibilities, the proteolysis of Nup62 and Nup98 was examined in whole-cell lysates where spatial constraints imposed by the nuclear envelope and NPC had been removed. Within 1 h following the addition of 2A^pro, levels of Nup98 were reduced by nearly 50%, and they were essentially undetectable following a 2-h incubation (Fig. 2B). Nup62 cleavage was slower, requiring more than 2 h for levels to fall below 50% and more than 4 h for complete proteolysis. To confirm the role of 2A^pro proteolytic activity in Nup98 cleavage, methoxysuccinyl-Ala-Ala-Pro-Val-chloromethylketone (MPCMK), an inhibitor of 2A^pro (40), was added to cleavage reaction mixtures. Although complete cleavage of Nup98 occurred after a 4-h incubation with 2A^pro, the addition of 70 μM MPCMK inhibited this proteolysis substantially (Fig. 2C). Previously, we found that these conditions also prevented cleavage of Nup62 (33). These results suggest that Nup98 is more susceptible to the proteolytic activity of 2A^pro than Nup62 and may explain the faster cleavage of Nup98 seen during viral infection.

The enterovirus 3C protease (3C^pro) is responsible for polyprotein processing and cleavage of various host factors (12, 13, 41–46). In addition, expression of HRV16 3C in uninfected cells has been reported to lead to proteolysis of Nup153, Nup214, and Nup358 (47, 48). To test for a possible role of HRV 3C^pro in Nup98 cleavage, purified HRV14 3C^pro was incubated with uninfected HeLa whole-cell lysates (Fig. 2D). Analysis of HRV14-infected cell lysates confirmed that Nup98 is, as reported by others, proteolyzed following infection with the enterovirus (Fig. 2D, lane 7) (34). The addition of 0.3 or 1 μg of 3C^pro followed by incubation for as much as 8 h had very little effect on levels of Nup98 (Fig. 2D, compare lanes 1 and 4 with 5). Activity of 3C^pro in these reactions was confirmed by demonstrating efficient proteolysis of NF-κB, a known target of 3C^pro (Fig. 2D, lanes 1, 3, and 4) (44). Cumulatively, these results are in agreement with other reports and support the conclusion that 2A^pro is the viral protease responsible for proteolysis of Nup98 (32, 34).

Nup98 is directly proteolyzed by HRV2 2A^pro.

The above-mentioned results indicate that HRV2 2A^pro can induce the proteolysis of Nup98 in HeLa cell lysates (Fig. 2A). This could be due to direct cleavage of Nup98 by 2A^pro or indirectly due to activation of cellular proteases present in the lysate. To examine this, we incubated purified HRV2 2A^pro with bacterially expressed and purified Nup98. After a 1-h incubation with purified 2A^pro, complete cleavage of Nup98 was observed, with cleavage products of ∼55 kDa and ∼40 kDa clearly visible (Fig. 3A, lane 3). The sizes of these cleavage products correspond quite closely to those of cp1 and cp3, respectively, which were detected as shown in Fig. 2. In these assays, Nup98 was purified under denaturing conditions and refolded, making it possible that some cleavage sites could be masked or others revealed by improperly folded Nup98. To address this possibility, we examined the cleavage pattern of Nup98 synthesized in a different system. Radiolabeled Nup98 was produced by in vitro translation in rabbit reticulocyte lysates, incubated with 2A^pro, and analyzed by autoradiography (Fig. 3B). The results again revealed 2 cleavage products of ∼55 kDa (cp1) and ∼40 kDa (cp2), along with two other cleavage products of ∼35 kDa (cp3) and ∼60 kDa (cp4). If we compare the cleavage patterns from purified Nup98 (Fig. 3A) and in vitro-translated Nup98 (Fig. 3B), cleavage products corresponding to cp1 and cp3 were observed in both systems. In addition, the analysis of HeLa lysates incubated with 2A^pro (Fig. 2) identified 3 Nup98 cleavage products that were very similar in size to cp1, cp2, and cp3 seen with in vitro-translated Nup98. These results indicate that the cleavage patterns of Nup98 are similar in all three systems and that Nup98 is cleaved by direct interaction with 2A^pro at multiple sites.

FIG 3.

2A^pro cleaves Nup98 directly. (A) 2A^pro can cleave purified Nup98. Bacterially expressed and purified Nup98 was incubated with or without 0.3 μg of purified 2A^pro for 1 h at 30°C. After incubation, samples were electrophoresed, transferred to PVDF membranes, and stained with Coomassie blue. (B) 2A^pro can cleave in vitro-translated Nup98. Nup98 was generated by in vitro translation of pET28b(+)-Nup98 in the presence of [³⁵S]methionine. Two microliters of the translation reaction mixture was incubated in the presence or absence of 0.3 μg purified 2A^pro at 30°C. After incubation, samples were electrophoresed, transferred to PVDF membranes, and analyzed by autoradiography. Each experiment was repeated a minimum of 3 times, and representative results are shown. Molecular mass markers are indicated in kilodaltons. The locations of cleavage products are indicated. The asterisk indicates a nonspecific product synthesized in the in vitro translation reaction.

Identification of 2A^pro cleavage sites in Nup98.

The positions of amino acids around protease cleavage sites are designated using the format Pn…P2-P1-P1′-P2′…Pn′, where the cleavage site is located between P1 and P1′ (49). Most of the known 2A^pro cleavage sites show a strong preference for Gly at P1′; Thr or Ser at P2; and a hydrophobic amino acid, such as Ile or Leu, at P4 (Table 1) (50–52). Examination of the Nup98 amino acid sequence revealed two potential cleavage sites that conformed well to the 2A^pro consensus located at Gly374 and Gly552 (Table 1). Figure 4A shows a schematic representation of the predicted 2A^pro cleavage sites in Nup98 and the proposed cleavage products that would be generated following proteolysis at these sites. The predicted molecular masses of polypeptides spanning amino acids 1 to 374 (∼37 kDa) and 1 to 552 (∼55 kDa) are similar to the estimated masses of cp2 and cp1, respectively, which were identified in the vitro cleavage assays (Fig. 2 and 3). Furthermore, these products should be recognized by an antibody to the N terminus of Nup98, as shown in Fig. 2A (N-Terminal Nup98 Ab). Similarly, the molecular masses of polypeptides extending from amino acids 374 (60 kDa) or 552 (42 kDa) to the C terminus are similar to those of cp4 and cp3, respectively, which were detected with in vitro-translated Nup98 (Fig. 3B).

TABLE 1.

Predicted 2A^pro cleavage sites in Nup98

Sequence	Amino acid at positiona:
	P4	P3	P2	P1	↓	P1′	P2′	P3′	P4′
Consensusb	L/I	X	T/S	X		G	X	X	X
Nup98	L	T	T	F		G374	S	S	T
	L	Q	T	T		G552	T	A	K

^aNomenclature as described by Schechter and Berger (49), where the cleavage site (arrow) is located between P1 and P1′. Amino acids observed in other known 2A protease cleavage sites are indicated in boldface (50).

^bAs described by Sommergruber et al. (52, 79).

FIG 4.

Identification of 2A^pro cleavage sites in Nup98. (A) Schematic representation of Nup98 and 2A^pro cleavage products. Full-length Nup98 is shown, along with the approximate locations of G374 and G552. Fragments arising from cleavage at these sites are shown with predicted molecular masses and designations (cp numbers), consistent with the data presented. The numbers in parentheses are the numbers of methionine residues. (B) 2A^pro cleaves at positions 374 and 552. In vitro-translated radiolabeled wt or mutant Nup98 was incubated with 0.3 μg of purified HRV2 2A^pro for 1 h at 30°C. After incubation, the lysates were electrophoresed, transferred to PVDF membranes, and analyzed by autoradiography. Constructs are designated by the P1′ amino acid and its location in the Nup98 primary sequence, followed by the mutant amino acid inserted at that position. (C) In vitro-translated radiolabeled wild-type and double-mutant Nup98 were incubated with 0.3 μg of purified HRV2 2A^pro for 4 h at 30°C. The reaction mixture was analyzed by autoradiography as described above. Each experiment was repeated a minimum of 3 times, and representative results are shown. Nonspecific bands in all the lysates are indicated with asterisks. Nonspecific bands produced in lysates containing the double mutant are indicated with daggers. Molecular mass markers are indicated in kilodaltons.

The schematic in Fig. 4A predicts that mutation of G374 should result in the loss of cp2 and cp4 and a corresponding increase in the levels of cp1, which contains G374. Similarly, mutation of G552 should result in the loss of cp1 and cp3 and an increase in the amount of cp4. To test these predictions, we constructed mutant forms of Nup98 that destroyed the presumptive cleavage sites and compared the cleavage pattern to that of wild-type Nup98. Previous studies showed that replacement of Gly with Lys at P1′ prevents proteolytic cleavage by 2A^pro (52, 53). Consequently, we replaced Gly at 374 and 552 with Lys to construct both single-mutant (G374K and G552K) and double-mutant (G374K G552K) Nup98 constructs. Wild-type and mutant constructs were translated in vitro in the presence of [³⁵S]methionine, incubated with purified 2A^pro for 1 h, and analyzed by autoradiography. As expected, cleavage of wild-type Nup98 resulted in 4 cleavage products (cp1 to cp4) (Fig. 4B, lane 5). Consistent with the predictions described above and depicted in Fig. 4A, mutation of G374 resulted in the loss of cp2 and cp4, and the levels of cp1 were increased, demonstrating that cleavage at G374 gives rise to cp2 and cp4 (Fig. 4B, lane 6). Similarly, mutation of G552 resulted in the loss of cp3 and cp1, confirming that they arise due to proteolysis at G552 (Fig. 4B, lane 7). Furthermore, the double mutant (G374K G552K) failed to produce significant amounts of any cleavage products in these assays (Fig. 4B, lane 8). To confirm that G374 and G552 are the only cleavage sites recognized by HRV2 2A^pro in this system, reaction mixtures containing Nup98 and the Nup98(G374K G552K) double mutant were incubated for an additional 3 h. Even following this extended incubation, very little cleavage was observed in the double mutant (Fig. 4C, lane 4). In contrast, wild-type Nup98 was almost completely proteolyzed to cp2 and cp3 (Fig. 4C, lane 3). The reduced levels of cp1 and cp4 likely reflect their further proteolysis to produce cp2 and cp3, respectively. These results indicate that G374 and G552 are the primary cleavage sites recognized by HRV2 2A^pro in vitro.

Nup98 is cleaved at Gly³⁷⁴ and Gly⁵⁵² in infected cells.

To confirm that Nup98 is proteolyzed at G374 and G552 during virus infection, constructs expressing FLAG-tagged wild-type and mutant forms of Nup98 were transfected into cells and examined for cleavage. For these experiments, the Gly at positions 374 and 552 was mutated to Asp, as the Lys mutants were targets of ubiquitination when expressed in HeLa cells (N. Park and K. E. Gustin, unpublished data). Previous studies have shown that replacement of Asp at the P1′ position efficiently blocks proteolysis by 2A^pro (52). As expected, HRV2 infection following transfection resulted in efficient proteolysis of FLAG-Nup98 wt (Fig. 5A, compare lanes 3 and 4). In contrast, the double mutant (G374D G552D) was completely resistant to proteolysis in infected cells (Fig. 5A, compare lanes 5 and 6). Surprisingly, although wild-type FLAG-Nup98 was efficiently cleaved in HRV2-infected cells, cleavage of endogenous Nup98 appeared to be inhibited (Fig. 5A, lanes 2 and 4). Similarly, expression of the G373D G552D form of Nup98 further suppressed cleavage of endogenous Nup98 (Fig. 5A, lanes 2 and 6). Although the reason for these results is unknown, one possibility is that the amount of 2A^pro is limiting and that overexpression results in some Nup98 escaping degradation. To determine if 2A^pro proteins from other enteroviruses also cleave Nup98 at G374 G552, a similar experiment was carried out in poliovirus-infected cells. Figure 5B shows that FLAG-Nup98 harboring the G374D G552D mutations was resistant to cleavage by poliovirus 2A^pro (Fig. 5B, lane 6). The lack of proteolysis was not due to mislocalization, as immunostaining confirmed that the double mutant was properly localized to the NPC and colocalized with endogenous Nup98 in transfected cells (Fig. 5C). These results confirm that Nup98 is proteolyzed at G374 and G552 in infected cells and that these sites are recognized by the 2A^pro proteins from multiple enteroviruses.

FIG 5.

Gly³⁷⁴ and Gly⁵⁵² are sites of proteolysis in infected cells. (A) Plasmids encoding FLAG-tagged wild-type or double-mutant Nup98 were transfected into HeLa cells and 24 h later infected with HRV2. Whole-cell lysates prepared 18 h later were electrophoresed, transferred to PVDF membranes, and analyzed by immunoblotting with antibodies recognizing FLAG, Nup98, or nucleolin. Molecular mass markers are indicated in kilodaltons. (B) Same as panel A, except that the cells were infected with poliovirus for 4 h. unTx, untransfected; DM, double-mutant Nup98 (G374D G552D). (C) Cells were transfected with the FLAG-tagged double-mutant Nup98 (G374D G552D) and stained with antibodies to FLAG and Nup98. Untransfected cells are indicated with asterisks. Each experiment was repeated a minimum of 3 times, and representative results are shown.

The FG repeat domain of Nup98 is released from the NPC.

Previously, we found that proteolysis of Nup62 during poliovirus infection releases the N-terminal FG repeat domain from the NPC while the C-terminal region is retained (33). To determine if proteolysis of Nup98 also releases the FG repeat domain from the NPC, mock- or virus-infected cells were immunostained using antibodies that recognize either the N-terminal FG repeat domain (Nup98 N-Terminus) or the C terminus (Nup98 C terminus) (Fig. 6A). Figure 6B and C show that, as expected, both antibodies stain the nuclear rims of mock-infected cells. Following HRV2 infection, however, staining of the N terminus containing the FG repeat domain was reduced while the signal from the C terminus was unchanged (Fig. 6B). Figure 6B shows that even in cells showing significant levels of cytopathic effect (CPE), as evidenced by shrinkage of their nuclei, the C terminus of Nup98 was still detected at the nuclear rim, further illustrating the stable retention of this region of Nup98 at the NPC. Notably, incubation of Nup98 with 2A^pro resulted in the appearance of a 35-kDa cleavage product (Fig. 2A, cp3) derived from the C terminus that may account for these findings. To determine if other enteroviruses also removed the FG repeat domain of Nup98 from the NPC, we performed a similar analysis in poliovirus-infected cells. Figure 6C shows that proteolysis of Nup98 during poliovirus infection also results in release of the N-terminal FG repeat region and retention of the C-terminal domain. These data indicate that proteolysis following enterovirus infection induces the release of the N-terminal FG repeat domain of Nup98 from the NPC while the C-terminal region remains associated with the NPC.

FIG 6.

Nup98 association with the NPC in infected cells. (A) Schematic representation of Nup98 and regions recognized by two different anti-Nup98 antibodies. The GLFG-rich region (1 to 497) is located in the N terminus of Nup98. Regions recognized by anti-Nup98N antibody (1 to 466) and anti-Nup98C antibody (around proline 654) are indicated. (B) HRV2-infected cells. Indirect immunofluorescence assays were performed with anti-Nup98N antibody and anti-Nup98C antibody on mock- or HRV2-infected HeLa cells for the indicated time. (C) Poliovirus-infected cells. Poliovirus-infected HeLa cells were analyzed as described for panel A. Anti-Nup98N antibody and anti-Nup98C antibody were detected with fluorescein isothiocyanate (FITC) and tetramethyl rhodamine isocyanate (TRITC) filters, respectively. Hoechst-stained DNA (Nuclei) was visualized with a UV filter. All images were acquired with identical exposure times and adjustments. Each experiment was repeated a minimum of 3 times, and representative results are shown.

Nup88 remains at the NPC during viral infection.

The findings described above indicate that 2A^pro proteolyzes Nup98 at two sites, leading to the release of the N-terminal FG-rich region from the NPC while the C-terminal region is retained. Deletion of Nup98 from embryonic fibroblasts results in the release of several cytoplasmic-facing Nups from the NPC, including Nup88 (54). Since Nup88 interacts with a region of the C terminus of Nup98 (55) that is retained at the NPC in infected cells, we hypothesized that Nup88 should also be retained at the NPC under these conditions. To test this hypothesis, infected cells were coimmunostained with antibodies to Nup88 and the N or C terminus of Nup98. As expected, staining with an antibody to the N terminus of Nup98 was dramatically reduced, confirming that these were indeed infected cells (Fig. 7A). Costaining for Nup88 revealed that during poliovirus infection, staining was essentially unchanged compared to uninfected cells (Fig. 7A). Costaining for Nup88 and the C terminus of Nup98 revealed no change in the staining pattern for these proteins following infection and demonstrated that they both colocalize to the NPC (Fig. 7B). Nup88 at the NPC of infected cells could be intact Nup88 or a proteolytic fragment similar to the C-terminal domain of Nup98 (Fig. 6). To examine this possibility, the status of Nup88 in mock- or poliovirus-infected cells was analyzed by immunoblotting. After 4 or 6 h of infection with poliovirus, Nup98 was almost completely proteolyzed while the levels of Nup88 were almost the same as those seen in mock-infected cells (Fig. 6B). Although these results indicate that Nup88 is retained at the NPC in infected cells, it is currently not clear if this is due to its ability to interact with the C terminus of Nup98 or with other binding partners at the NPC.

FIG 7.

Nup88 remains associated with the NPC in infected cells. (A) Nup88 is located at the nuclear rim during poliovirus infection. Indirect immunofluorescence assays were performed with anti-Nup98N antibody and anti-Nup88 antibody on mock-infected HeLa cells and cells that had been infected with poliovirus for 4 h (PV4). Anti-Nup98N antibody and anti-Nup88 antibody were detected with FITC and TRITC filters, respectively. Hoechst-stained DNA (Nuclei) was visualized with a UV filter. All images were acquired with identical exposure times and adjustments. (B) Nup88 colocalizes with the C terminus of Nup98. As for panel A, except that cells were stained with anti-Nup98C antibody and anti-Nup88 antibody. Anti-Nup98N antibody and anti-Nup88 antibody were detected with TRITC and FITC filters, respectively. (C) Nup88 is not proteolyzed during poliovirus infection. Twenty-five micrograms of HeLa whole-cell lysates prepared after 6 h mock infection and 4 and 6 h poliovirus infection was analyzed by immunoblotting with monoclonal anti-Nup98 antibody and monoclonal anti-Nup88 antibody. Each experiment was repeated a minimum of 3 times, and representative results are shown.

DISCUSSION

The results presented above add to our understanding of how enteroviruses alter the NPC and disrupt nucleocytoplasmic trafficking during infection. First, the results show that, as was the case in poliovirus-infected cells, Nup98 is targeted for degradation earlier than other NPC proteins in HRV2-infected cells. Second, we used in vitro cleavage assays with purified proteins to demonstrate that 2A^pro cleaves Nup98 directly. Third we have identified the 2A^pro cleavage sites in Nup98 as G374 and G552 and confirmed that these sites are recognized by 2A^pro in infected cells. Fourth, we have shown that cleavage of Nup98 by 2A^pro results in the selective removal of the FG-containing N terminus of Nup98 from the NPC, while the C terminus remains associated. Lastly, we have shown that Nup88, a core NPC component that serves as a docking site for the C terminus of Nup98, remains associated with the NPC in infected cells. As discussed below, these results help to explain the alterations in permeability and nuclear transport and how NPCs remain functional for certain trafficking pathways despite significant alterations in their compositions.

Belov et al. found that expression of poliovirus 2A^pro in HeLa cells was sufficient to bring about alterations in nuclear envelope permeability (28). Subsequent work provided an explanation for this finding by showing that HRV 2A^pro directly cleaves Nup62 in vitro (33). Additional studies have extended these findings by showing that purified 2A^pro from all three rhinovirus species targets Nup98 and Nup153 (34) and that expression of 2A^pro alone is sufficient to bring about the proteolysis of these NPC proteins (32). The results presented here indicate that HRV2 and poliovirus 2A^pro proteins proteolyze Nup98 at two locations, G374 and G552. Cleavage products consistent with proteolysis at these sites were detected in infected cells and in two different in vitro systems (Fig. 2A and 3). Furthermore, Nup98 cleavage was blocked both in vitro and in infected cells by mutating the Gly at the P1′ position in these sites (Fig. 4 and 5), indicating that these are bona fide cleavage sites for 2A^pro. In support of this, the amino acids found on the N-terminal side of the cleavage site (P1 to P4) conform well to those previously reported in enterovirus 2A^pro cleavage sites (Table 1) (50). In addition, both sites contained the nearly universally conserved Gly residue at the P1′ position. In contrast, the remainder of the C-terminal residues diverge substantially from those typically found at sites cleaved by 2A^pro (50). However, prior work has shown that 2A^pro can efficiently cleave peptides extending only to the P2′ position (52).

HRV 3C^pro has also been implicated in proteolysis of Nups during enterovirus infection. Initial experiments showed that expression of HRV16 3C^pro or 3CD^pro in uninfected cells led to proteolysis of Nup153, along with partial proteolysis of Nup214 and Nup358 and alterations in nuclear transport (47, 48). Subsequent work confirmed that HRV16 3C^pro can proteolyze Nup153 but revealed no effect on levels of Nup98 or Nup62 (48). The results presented here using purified HRV14 3C^pro extend these findings to another rhinovirus species and confirm that although 3C^pro may contribute to proteolysis of some Nups, cleavage of Nup98 appears to be entirely dependent upon 2A^pro activity.

The alterations to the NPC in enterovirus-infected cells lead to the redistribution of host RNA binding proteins from the nucleus to the cytoplasm, where they can contribute to aspects of the viral life cycle, such as protein and RNA synthesis (reviewed in reference 56). To allow the accumulation of host nuclear factors in the cytoplasm, they must be provided a way to exit the nucleus and must be prevented from reentering. By removing (or significantly diminishing) the permeability barrier, proteins lacking export signals and not bound to structures in the nucleus can rapidly diffuse across the nuclear envelope and accumulate in the cytoplasm. However, if the nuclear import pathway used by a given protein is still functional, it will be rapidly imported back into the nucleus, resulting in little or no net accumulation in the cytoplasm. Consequently, proteolytic targeting of FG-Nups by enterovirus 2A^pro needs to accomplish two tasks: increased permeability to allow diffusion out of the nucleus and inhibition of nuclear import to prevent import back into the nucleus. Enteroviruses have evolved an elegant strategy to accomplish both of these tasks simultaneously by targeting a single functional motif found in the NPC.

Nup98 consists of a relatively unstructured N terminus containing multiple FG repeats separated by hydrophilic linkers. In contrast, the C terminus of Nup98 is highly structured and mediates recruitment and interaction of Nup98 with the NPC (55, 57). Cleavage of Nup98 by 2A^pro results in the release of the N-terminal FG repeat domain from the NPC while the C-terminal domain remains associated (Fig. 6B and C). FG repeat domains have been implicated in both forming the NPC permeability barrier and serving as binding sites for transport receptor-cargo complexes as they transit the central channel of the NPC (58). Patel et al. demonstrated that removal of the FG-rich region from the Saccharomyces cerevisiae homologues of Nup98 and Nup153 increased the permeability of the NPC (19). More recent work found that vertebrate Nup98 was essential for establishing the normal permeability barrier in NPCs assembled in vitro using extracts from Xenopus laevis oocytes (59). Thus, the loss of the FG-rich region of Nup98 likely explains the increased permeability of the nuclear envelope that is observed in poliovirus-infected cells (28). However, several other FG-containing Nups are also proteolyzed following enterovirus infection, including Nup153, Nup214, Nup358, Nup62, Nup58, and Nup54 (26, 27, 60). The combined removal of multiple Nup FG domains from the NPC may further erode the permeability barrier of the NPC.

Interaction between transport receptors and FG domains in Nups is important for efficient translocation of the receptor-cargo complex across the nuclear envelope (reviewed in reference 61). Importantly, analysis of numerous transport receptors revealed that they exhibit nonoverlapping affinities for the FG repeat regions of Nups (62–65). Consistent with these findings, the loss of FG repeat regions by targeted deletions in yeast or vertebrates has variable effects on nuclear transport and NPC function, with some pathways exhibiting greater sensitivity to the loss of one Nup over another, while other pathways are relatively insensitive to the loss of FG repeats (54, 66, 67). Thus, the loss of FG repeats from the NPC of enterovirus-infected cells provides an explanation for why certain nuclear transport pathways (importin-α/β and transportin-2) are inhibited while others (glucocorticoid receptor import and Crm1 export) appear to function normally in enterovirus-infected cells (27).

The retention of Nup structural domains may also contribute to maintaining NPC function in infected cells. The results presented above indicate that the C-terminal structural domain of Nup98 remains associated with the NPC. This region is required for targeting Nup98 to the NPC via its interaction with the scaffold Nups, Nup88 and Nup96 (55, 57, 68). Cleavage of Nup62 by 2A^pro follows a pattern remarkably similar to that described here for Nup98. 2A^pro cleavage sites in Nup62 are clustered between the N-terminal FG domain and the structured C terminus, and following poliovirus infection, the N terminus is released from the NPC while the C terminus is retained (33). The C terminus of Nup62 interacts with structured domains found in two other FG-Nups located in the central channel of the NPC and is thought to anchor this complex to the NPC (69, 70). Thus, the retention of Nup structural domains may stabilize the NPC scaffold and help to maintain transport function despite the loss of a significant number of FG repeat domains. Interestingly, the C-terminal domain of Nup62 is released from the NPC in HRV2-infected cells (33), perhaps indicating that the region is dispensable for maintaining NPC function. This is consistent with recent work using reconstituted nuclei in X. laevis egg extracts showing that the Nup62 complex is not absolutely required for NPC function (59). Currently, it is not clear if retaining NPC function during infection is important for viral infection. However, maintaining active export pathways would facilitate the cytoplasmic accumulation of the shuttling RNA binding proteins that contribute to viral replication (reviewed in reference 56).

Nup98 proteolysis in enterovirus-infected cells occurs well before cleavage of other Nups is detected (30, 34, 48). In addition, cleavage of Nup98 is insensitive to the inhibitor of viral replication guanidine hydrochloride (30). One possible explanation for the faster kinetics of Nup98 proteolysis arises from findings that Nup98 shuttles on and off the NPC and that the flexible FG domain can be detected on the cytoplasmic side of the NPC (71, 72), perhaps making Nup98 or the FG region more accessible to 2A^pro. However, this seems unlikely, given that Nup153 exhibits similar characteristics yet is proteolyzed much later during infection (73, 74). Furthermore, rapid Nup98 cleavage is apparent in vitro when spatial constraints imposed by cellular structures such as the nuclear envelope or NPC have been removed (Fig. 2B) (34). Thus, the mechanisms responsible for the rapid targeting of Nup98 remain to be determined.

While removal of Nup98 from the NPC would be predicted to alter NPC permeability and transport properties, recent work has suggested other possible reasons for the rapid proteolysis of Nup98 in infected cells. Work in Drosophila melanogaster has shown that, in addition to its role at the NPC, Nup98 also has a role in transcriptional induction during normal fly development and in the heat shock response (75, 76). More recently, Panda et al. showed that Nup98 is required for an effective antiviral transcription response in D. melanogaster (77). Combined with earlier work showing that Nup98 is itself induced by type I and II interferons, these results suggest the possibility that Nup98 may contribute to the antiviral response in mammals. In support of this possibility, Liang et al., found that Nup98 regulates the transcription of a subset of genes in human embryonic stem cells and neural progenitor cells (78). Interestingly, this study identified an N-terminal fragment (aa 1 to 503) that acted as a dominant-negative inhibitor of transcriptional activation of Nup98-dependent genes in neural progenitor cells. It is interesting to speculate that the N-terminal cleavage products generated by 2A^pro cleavage function in a similar manner to inhibit the antiviral response in humans.