Characterization of Localization and Export Signals of Bovine Torovirus Nucleocapsid Protein Responsible for Extensive Nuclear and Nucleolar Accumulation and Their Importance for Virus Growth

ToVs are diarrhea-causing pathogens detected in many species, including humans. BToV has spread worldwide, leading to economic loss, and there is currently no treatment or vaccine available.

ABSTRACT

Torovirus (ToV) has recently been classified into the new family Tobaniviridae, although historically, it belonged to the Coronavirus (CoV) family. The nucleocapsid (N) proteins of CoVs are predominantly localized in the cytoplasm, where the viruses replicate, but in some cases the proteins are partially located in the nucleolus. Many studies have investigated the subcellular localization and nucleocytoplasmic trafficking signals of the CoV N proteins, but little is known about ToV N proteins. Here, we studied the subcellular localization of the bovine ToV (BToV) N protein (BToN) and characterized its nucleocytoplasmic trafficking signals. Unlike other CoVs, BToN in infected cells was transported mainly to the nucleolus during early infection but was distributed predominantly in the nucleoplasm rather than in the nucleolus during late infection. Interestingly, a small quantity of BToN was detected in the cytoplasm during infection. Examination of a comprehensive set of substitution or deletion mutants of BToN fused with enhanced green fluorescent protein (EGFP) revealed that clusters of arginine (R) residues comprise nuclear/nucleolar localization signals (NLS/NoLS), and the C-terminal region served as a chromosomal maintenance 1 (CRM1)-independent nuclear export signal (NES). Moreover, recombinant viruses with mutations in the NLS/NoLS, but retaining nuclear accumulation, were successfully rescued and showed slightly reduced growth ability, while the virus that lost the NLS/NoLS-mediated nuclear accumulation of BToN was not rescued. These results indicate that BToN uniquely accumulates mainly in nuclear compartments during infection, regulated by an R-rich NLS/NoLS and a CRM1-independent NES, and that the BToN accumulation in the nuclear compartment driven by NLS/NoLS is important for virus growth.

IMPORTANCE ToVs are diarrhea-causing pathogens detected in many species, including humans. BToV has spread worldwide, leading to economic loss, and there is currently no treatment or vaccine available. Positive-stranded RNA viruses, including ToVs, replicate in the cytoplasm, and their structural proteins generally accumulate in the cytoplasm. Interestingly, BToN accumulated predominantly in the nucleus/nucleolus during all infectious processes, with only a small fraction accumulating in the cytoplasm despite being a major structural protein. Furthermore, we identified unique nucleocytoplasmic trafficking signals and demonstrated the importance of NLS/NoLS for virus growth. This study is the first to undertake an in-depth investigation of the subcellular localization and intracellular trafficking signals of BToN. Our findings additionally suggest that the NLS/NoLS-mediated nuclear accumulation of BToN is important for virus replication. An understanding of the unique features of BToV may provide novel insights into the assembly mechanisms of not only ToVs but also other positive-stranded RNA viruses.

INTRODUCTION

The genus Torovirus (ToV) belongs to the order Nidovirales, family Tobaniviridae, subfamily Torovirinae, but historically, it has belonged to Coronaviridae (1). The order Nidovirales was recently subdivided into eleven families but previously consisted of only four families, Coronaviridae, Arteriviridae, Roniviridae, and Mesoniviridae, which are evolutionarily related enveloped and positive-stranded RNA viruses (1). ToVs exhibit bovine, equine, swine, and human infectivity and are involved in diarrheic and respiratory diseases (2–7). Epidemiological studies indicate that the bovine ToV (BToV) is distributed worldwide, causing significant economic losses within the agricultural industry (8–15), and the porcine ToV is similarly prevalent (16–19). There are no effective antiviral drugs or vaccines available for disease treatment and prevention.

The coronavirus (CoV) and torovirus (ToV) are structurally and morphologically similar but exhibit some differences (20–23). The CoV virion consists of four primary structural proteins: the spike (S), membrane (M), envelope (E), and nucleocapsid (N) proteins (24, 25). Some beta-CoVs also contain hemagglutinin-esterase (HE) protein (26). ToV samples from clinical feces are composed mainly of S, M, HE, and N proteins (23, 27–34) but lack E protein, which plays an important role in CoV assembly (35–37). Although a large number of studies have been performed on CoVs, ToVs are not well studied because they are difficult to propagate in cultured cells, with the one exception of equine torovirus (Berne virus [BEV]) (6). However, over the past decade, several bovine toroviruses (BToVs) in Japan have been successfully isolated and propagated using HRT-18 cells (27, 38–40), although cell-adapted BToV seems to lack a full-length HE protein (40, 41).

Because positive-stranded RNA viruses (including Nidovirales) replicate and assemble in the cytoplasm of infected cells, their structural proteins are generally transported to the cytoplasm. The N proteins of CoVs and related arteriviruses (AVs) accumulate predominantly in the cytoplasm, but in some cases are reported to be found in the nucleolus as well (42–51). In AVs, part of the N proteins of porcine reproductive and respiratory syndrome virus (PRRSV), equine arteritis virus (EAV), and lactate dehydrogenase-elevating virus (LDV) localizes to the nucleolus or nucleus during infection (42–46, 52). Similarly, in several CoV genera, including alpha-CoVs (transmissible gastroenteritis virus [TGEV] and porcine epidemic diarrhea coronavirus [PEDV]), beta-CoVs (mouse hepatitis virus [MHV]), and gamma-CoVs (infectious bronchitis virus [IBV]), N proteins are partially transported to the nucleolus (47–51). In other beta-CoVs (severe acute respiratory syndrome coronavirus [SARS-CoV]), N proteins are localized exclusively to the cytoplasm and are not found in the nucleus or nucleolus (53–55). These N proteins exhibit substantially similar distributions when they are expressed alone or fused with green fluorescent protein (GFP). It is important to note that these N proteins are distributed predominantly in the cytoplasm, even when they are also detected in the nucleolus. Moreover, only 10% and 20% of IBV-infected cells and PRRSV-infected MARC-145 cells, respectively, have N protein distributed in both the nucleolus and cytoplasm; the remaining cells have no detectable N protein in the nucleolus (44, 49). It is not clearly understood why some of these N proteins were actively localized to the nucleolus. However, those of PRRSV and IBV could be colocalized with the major nucleolar proteins fibrillarin (43) or nucleolin (51) and may be involved in viral pathogenesis or replication (56, 57) or disruption of host cell division (50), respectively. These findings suggest that N proteins in the nucleolus have biological significance.

Proteins that are distributed to both the nucleus/nucleolus and cytoplasm often contain two or three nucleocytoplasmic trafficking signals. The nuclear localization signals (NLSs) and nuclear export signals (NESs) are relatively well characterized, whereas nucleolar localization signals (NoLSs) remain poorly defined. NLSs can be divided into two classes: the classical NLSs (cNLSs) and the proline-tyrosine NLSs (PY-NLSs) (58–62). The cNLSs are further divided into monopartite and bipartite clusters of positively charged amino acids, such as lysine (K) and arginine (R). The consensus sequences of the monopartite and bipartite clusters are K-K/R-X-K/R (or “pat4” and “pat7” motifs) and (K/R)(K/R)X_10-12(K/R)_3/5, respectively, where X is any amino acid and (K/R)_3/5 means three K or R out of five consecutive amino acids (59, 60, 62). The PY-NLSs are defined by a combination of several criteria as follows: hydrophobic or basic N-terminal motif and R/K/H-X_3-5-P-Y motif at the C terminus, structural disorder, and overall basic charge (58, 61). Among CoVs and AVs, all N proteins that have been detected in the nuclear compartments contain more than one cNLS but not PY-NLSs. NESs are highly variable in sequence, but the consensus sequence can be categorized into six classes: 1a, 1b, 1c, 1d, 2, and 3 (63). Recently, five more classes have been added: 1a-R, 1b-R, 1c-R, 1d-R (64), and class 4 (65) (see Fig. 3B). These NESs are recognized by a chromosomal maintenance 1 (CRM1) protein, and the export of proteins dependent on these NESs is inhibited by leptomycin B (LMB) (66–69). Because the N proteins of CoVs and AVs in the nucleolus are not static but shuttle between the nucleus and cytoplasm, they often contain functional NESs. In fact, the N proteins of two AVs (PRRSV and EAV) have class 1a CRM1-dependent NES consensus sequences (Ф1XXXФ2XXФ3XФ4, where Ф represents L, V, I, F, M, W, C, T, or A, but W, C, T, and A are allowable only at a single position, and X is any amino acid; i.e., 108-THHTVRLIRV-118 and 26-LLRMFGQMRV-35, respectively; key amino acids [positions Ф1 to Ф4] are shown in bold and underlined), and their distribution patterns are affected by LMB treatment, indicating that their nuclear export is CRM1 dependent (42, 46). Likewise, in the case of CoVs, PEDV N protein contains overlapping class 1a or class 3 NES consensus sequences (Ф1XXXФ2XXФ3XФ4 or Ф1XXФ2XXXФ3XXФ4; i.e., 223-VAAVKDALKSLGI-235) and are sensitive to LMB treatment (47). Interestingly, even though IBV N protein contains a class 2 NES consensus sequence (Ф1XФ2XXФ3XФ4; i.e., 291-LQLDGLHL-297), their export from nuclear compartments is not inhibited by LMB, suggesting that the nuclear export mechanism is CRM1 independent (70). In contrast to NLSs and NESs, there is no obvious consensus sequence for NoLSs, but the signals of IBV, PEDV, and PRRSV N proteins have been identified (44, 47, 48).

FIG 3.

(A) Amino acid sequence of BToN (Aichi strain) and potential nuclear localization signals (NLSs) and nuclear export signals (NESs) based on bioinformatics analysis. The arginine (R)-rich region containing amino acid positions 31 to 62 is outlined in light blue, and R residues in this region are highlighted in red. The class 1d chromosomal maintenance 1 (CRM1)-dependent NES consensus sequence is outlined in gray, and Ф1 to 4 are highlighted in red. (B) Consensus sequences of NESs for each class. In each sequence, Ф can be F, I, L, M, or V (65, 66) but W, C, T, and A are allowable only at a single position in classes 1a to 1d, 2, and 3 (63). The table is modified based on reference 64. (C) Schematic diagram of the wild-type BToN (wtN) and BToN fused with EGFP at its C (N-EGFP) or N (EGFP-N) terminus and three repeated NLSs of simian virus 40 and single NLSs of nucleoplasmin fused with EGFP at the C terminus, respectively (SV40-EGFP and Nuc-EGFP). No, nucleolus; Nu, nucleus; C, cytoplasm.

Although many studies have been performed to investigate nuclear/nucleolar-cytoplasmic localization and intracellular trafficking signals for N proteins from CoVs and AVs, only two brief studies have been carried out for ToV (BEV) N proteins. Notably, these two studies had conflicting results—one group reported that BEV N protein of the infected cells was detected mainly in the nucleus (71), whereas the other group showed that the N protein was distributed only in the cytoplasm (72). Thus, even the intracellular localization of ToV N protein has not been fully elucidated. Here, we studied the subcellular localization of BToN in infected and transiently N-expressing cells and characterized its NLS/NoLS and NES. Unlike CoVs and AVs, in which these N proteins were localized primarily to the cytoplasm and partially to the nucleolus, the BToN was transported mainly to the nucleolus only during early infection. At later infection, it was distributed predominantly throughout the nucleoplasm rather than the nucleolus, with lesser quantities detected in the cytoplasm. In addition, we identified a unique R-rich NLS/NoLS and a CRM1-independent NES and showed the importance of NLS/NoLS-mediated nuclear accumulation for the viral life cycle, using recombinant BToVs.

RESULTS

Subcellular localization of BToN in infected HRT-18 cells.

To determine the intracellular localization of the BToN during infection, HRT-18 cells were infected with BToV (Aichi strain) at a multiplicity of infection (MOI) of 0.5 and then fixed and permeabilized. Indirect immunofluorescence (IFA) was performed using mouse anti-N antiserum (green) and fibrillarin rabbit monoclonal antibody MAb (red) as a nucleolar marker with secondary antibodies, and the nucleus was stained with Hoechst solution (blue). The stained cells were observed using confocal laser scanning microscopy (CLSM). As shown in Fig. 1A, at 6 h postinfection (hpi), BToN began to be detected and was predominantly colocalized with the nucleolar protein fibrillarin. BToN were distributed to the nucleoplasm at 8 hpi and then accumulated primarily and nonuniformly in the nucleoplasm rather than in the nucleolus. Limited quantities of BToN were generally detected in the cytoplasm during infection, although in subpopulations, BToN was somehow detectable in the cytoplasm (Fig. 1B). In contrast, the representative CoV MHV N proteins in infected DBT cells were detected only in the cytoplasm at 9 hpi, and these cells contained no detectable MHV N proteins in the nucleus or nucleolus (Fig. 1A), although partial N accumulation in the nucleolus has been reported elsewhere (50). In addition, other structural BToV M proteins began to be detected in the perinuclear area at 6 hpi and were distributed throughout the cytoplasm at 24 hpi (Fig. 1C). To clarify the subcellular localization of BToN, the relative fluorescence intensity of BToN in each subcellular compartment was analyzed. At early infection (4 to 6 hpi), the fluorescence peaks of BToN (green) corresponded well with those of fibrillarin (red), whereas during later infection (10 to 24 hpi), the green peak was separated from the red peak (Fig. 1D). This pattern of N protein accumulation in the nucleoplasm remained consistent, while infected cells and the quantity of BToN increased (Fig. 1E), until 24 hpi, when cytopathic effect began to occur. This result indicated that unlike N proteins from other CoVs and AVs, BToN was localized predominantly in the nucleolus only during early infection and was subsequently mainly excluded from the nucleolus. Instead, BToN accumulated in the nucleoplasm during later infection, and a small quantity was transported to the cytoplasm during infection.

FIG 1.

(A) Confocal microscopy images of the subcellular localization of bovine torovirus (BToV) nucleocapsid protein (BToN) in BToV-infected HRT-18 cells. The cells were fixed with 4% paraformaldehyde at the indicated hours postinfection (hpi) and permeabilized with 0.2% Triton X-100, and then BToN and fibrillarin were detected using mouse anti-N antiserum (green) and anti-fibrillarin rabbit MAb (red), respectively. The nucleus was stained with Hoechst solution (blue). (B) Cytoplasmic distribution of a small amount of BToN in the subpopulation of HRT-18 cells. BToV-infected HRT-18 cells were observed at 24 hpi. (C) Subcellular localization of M protein in BToV-infected HRT-18 cells. The cells were fixed at 4, 6, or 24 hpi, and BToV M protein was stained with mouse anti-M antiserum. (D) Relative fluorescence intensities of BToN (green peak) and fibrillarin (red peak) were analyzed along the red arrow using ZEN 2 software (blue lite edition; Zeiss Microscopy, Oberkochen, Germany). (E) Immunoblotting of BToN of BToV-infected HRT-18 cells. BToV-infected HRT-18 cells were lysed at the indicated hpi, and BToN was analyzed using sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to a membrane, and immunoblotted with rabbit anti-N antiserum. One-fifth of the sample volume prepared at 24 hpi was used to load the gel.

Subcellular localization of BToN in transiently N-expressing HRT-18 and COS-7 cells.

To investigate whether the subcellular localization of the BToN depended on the presence of other viral proteins, HRT-18 and nonpermissive COS-7 cells were transfected with pCAGGS-BToN, and the N-expressing cells were fixed at 42 h posttransfection (hpt) and then stained and observed as described above. The experiment was repeated at least three times, and Fig. 2A shows two representative images. In HRT-18 cells, BToNs of many cells were colocalized with fibrillarin, which was similar to the pattern observed during the early stage of infection, whereas those of a few cells were sometimes distributed throughout the nucleus and nucleolus, which differed from its distribution pattern during the later stages of infection (Fig. 2A, left, and Fig. 2B). BToN in COS-7 cells appeared to be colocalized with fibrillarin and accumulated in the nuclear compartments but disrupted both the morphology of the nucleolus and the normal distribution of fibrillarin within the nucleus (Fig. 2A, right). Both cell types had minimally detectable BToN in the cytoplasm. In contrast, MHV N proteins in both types of cells were distributed throughout the cytoplasm but not in the nucleus or nucleolus, and no morphological disruption of the nucleolus was observed (Fig. 2A). The nucleoli were disrupted only in COS-7 cells expressing BToN. However, this effect may have been artificial, as similar results were observed in COS-7 cells for the coexpression of non-BToNs fused with enhanced GFP (EGFP) and the nucleolar marker DsRed-B23 (data not shown). Significant accumulation of proteins targeting the nucleolus seems to cause nonspecific nucleolar disruption, presumably because of overexpression. Thus, these results suggest that BToN has an intrinsic ability to be transported to and accumulate in the nucleolus but could be excluded from the nucleolus by the presence of other viral proteins or host metabolic changes due to infection.

FIG 2.

(A) Confocal microscopy images of the subcellular localization of transient N-expressing HRT-18 and COS-7 cells. Cells were transfected with pCAGGS-BToN and fixed at 42 h posttransfection (hpt). Two representative images are shown for each cell type. Detection of BToN and fibrillarin, staining of the nucleus, and relative fluorescence intensity analysis (B) were carried out as described in the legend to Fig. 1.

Bioinformatics analysis of potential NLSs or NoLSs and NESs of BToN.

Because the BToN was localized mainly to the nucleus and nucleolus in infected HRT-18 cells or transiently N-expressing HRT-18 and COS-7 cells, potential NLSs or NoLSs and NESs were predicted using several motif prediction algorithms, such as PSORT II (73), cNLS mapper (74), NoD (75, 76), NES mapper (77), and LocNES (78). We found no cNLSs using cNLS mapper, but PSORT II found two potential overlapping monopartite cNLS motifs—a pat4 motif (34-RRRR-37) and a pat7 motif (31-PRFRRRR-37) (Fig. 3A). The 47-RRNNSNQNRSRQSRPR-62 sequence, which was similar to the bipartite cNLS motif, was regarded as a putative NLS, although neither algorithm identified it as cNLSs (Fig. 3A). Overall, the region at amino acid positions 31 to 62 containing 12 R (and no K) residues was defined as an R-rich region (Fig. 3A). NoD, an NoLS predictor that uses an artificial neural network, predicted 24-QYPMGFQPRFRRRRNPGFRPMFQRRNN-50 as an NoLS that partially overlaps an R-rich region. NES mapper and LocNES predicted several motifs, but both algorithms shared two potential NESs, 18-VVAMPI-23 and 153-IATFTIKVAL-162, which were designated NES1 and NES2, respectively. NES2 was completely concordant with the class-1d CRM1-dependent NES consensus sequence (Ф1XXФ2XXXФ3XФ4) (Fig. 3B) (64).

Determining intracellular distribution of BToN fused with EGFP in N-expressing HRT-18 and COS-7 cells using QFIA.

To identify the nucleocytoplasmic trafficking signals of the BToN, we first established an image-based quantitative fluorescence intensity analysis (QFIA) that could quantitatively evaluate the subcellular localization of BToN in living cells. To this end, we constructed BToN fused with EGFP at its N or C terminus (EGFP-N and N-EGFP) (Fig. 3C) and three repeats of well-characterized monopartite NLS of simian virus 40 (SV40) and a single bipartite NLS of nucleoplasmin fused with EGFP at their C termini (SV40-EGFP and Nuc-EGFP, respectively) as a control, as well as nucleolar marker protein B23 fused with DsRed at its N terminus to facilitate observations in living cells. HRT-18 cells were cotransfected with cDNA encoding EGFP fusion protein (green) and DsRed-B23 (red). The nuclei of these cells were stained with Hoechst solution (blue) at 38 to 44 hpt and observed using CLSM. Macromolecules smaller than 40 kDa can passively diffuse to both the nucleus and cytoplasm through nuclear pore complexes (NPCs). Therefore, it is possible that wild-type (wt) BToN (18.6 kDa) fused with EGFP (27 kDa) and deletion mutants thereof might pass through NPCs. Thus, in the absence of trafficking signals, the distribution pattern would be similar to that of EGFP. As shown in Fig. 4, both N-EGFP and EGFP-N were highly colocalized with the nucleolar marker DsRed-B23, in a manner similar to that of BToN expressed alone. SV40-EGFP was distributed throughout the nucleus with accumulation in the nucleolus. Nuc-EGFP was uniformly distributed throughout the nuclear compartment and partially detected in the cytoplasm, whereas EGFP was distributed throughout the nucleoplasm and cytoplasm but tended to be excluded from the nucleolus. We next evaluated relative fluorescence intensity in the subcellular compartments as described in Materials and Methods. As relatively uniform distribution patterns were observed among cells within the same samples and field, QFIA of HRT-18 was carried out by calculating the average fluorescence intensity ratio (nucleolus/nucleus/cytoplasm) for the EGFP fusion proteins. These average ratios were 10:1.6:0.2 (N-EGFP; n = 47), 10:2.2:0.9 (EGFP-N; n = 31), 10:5.3:0.8 (SV40-EGFP; n = 49), 9.8:10:5.3 (Nuc-EGFP; n = 53), and 6.1:10:9.6 (EGFP; n = 49), which corresponded well with the CLSM observations described above.

FIG 4.

Live-cell imaging of the subcellular localization of BToN fused with EGFP at its C or N terminus (N-EGFP and EGFP-N, respectively) in HRT-18 cells. The cells were cotransfected with each of pCAGGS-N-EGFP, pCAGGS-EGFP-N, pCAGGS-SV40-EGFP, pCAGGS-Nuc-EGFP, and pEGFP-N1 (green) and with DsRed-B23 (red). The nucleus was stained with Hoechst solution (blue). The transfected cells were observed under confocal microscopy at 38 to 44 hpi. (Middle panel) Relative fluorescence intensity of EGFP fusion protein (green peak), DsRed-B23 (red peak), and Hoechst stain (blue peak). (Lower panel) Relative fluorescence intensity of EGFP fusion protein (green peak). Blue or red horizontal lines correspond to blue and red peaks in the middle panel and indicate localization to the nucleus and nucleolus, respectively. To quantitatively evaluate the subcellular localization of EGFP fusion protein, the average ratio of EGFP fluorescence intensity in the nucleolus (No; red), nucleus (N; blue), and cytoplasm (C; black) was calculated. The highest value of EGFP fluorescence intensity among the cellular compartments was set to 10. N, number of analyzed cells.

On the other hand, three differences were observed for COS-7 cells. First, even single expression of EGFP-N or N-EGFP sometimes disrupted nucleolar morphology. To reduce nucleolar disruption, COS-7 cells were not cotransfected with DsRed-B23. Second, the distribution patterns and relative fluorescence intensities of EGFP-N or N-EGFP differed significantly among cells in the same fields of view. Therefore, images were captured using at least two or three different exposure times to include a range of fluorescence intensities. Third, unlike in HRT-18 cells, the patterns of distribution differed between N-EGFP and EGFP-N (Fig. 5). Among cells expressing N-EGFP, 87% showed N protein accumulation in the nuclear compartments. In contrast, 26% of cells expressing EGFP-N exhibited accumulation in the nuclear compartments, while 67% of cells showed N protein exclusion from the nuclear compartments into the cytoplasm with or without accumulation in the nucleolus (Fig. 5). As described below, BToN has functional NLS/NoLS and NES at the N and C terminus, respectively, but the former is stronger under normal conditions. However, the addition of EGFP to the N terminus of BToN may seal NLS/NoLS of the N terminus, resulting in stronger NES activity of EGFP-N proteins in a subpopulation of COS-7 cells.

FIG 5.

Live-cell imaging of the subcellular localization of BToN fused with EGFP at its C or N terminus in COS-7 cells. The protocol was similar to that described in the Fig. 4 legend, except for cotransfection of pCAGGS-DsRed-B23. Because the distribution patterns differed significantly among cells of the same mutant, several criteria were set to categorize patterns into five types as described in Materials and Methods. Each type was further divided into subtypes depending on the purpose. Representative images of N-EGFP, EGFP, and EGFP-N with their actual ratios are shown. The images were obtained from at least three independent experiments. N, number of analyzed cells.

Because there were various distribution patterns for EGFP-N and N-EGFP, QFIA of COS-7 cells was carried out by categorizing these distribution patterns into five types as described in Materials and Methods (see Fig. 5). These results indicated that the distribution patterns for N-EGFP fell mainly into the nucleolar/nuclear accumulation types (types 1 and 2), whereas those of EGFP-N fell considerably into the nuclear export types (types 4 and 5) (Fig. 5). The distribution pattern for EGFP was categorized as type 3 (EGFP type), whereas those of SV40-EGFP and Nuc-EGFP were type 1 and type 1/2, respectively. These results were also consistent with the CLSM observations of COS-7 cells. All results indicated that QFIA of HRT-18 and COS-7 cells was a reasonable means of evaluating the subcellular localization of BToNs. In addition, it is noteworthy that EGFP-N proteins in COS-7 cells were excluded from nuclear compartments, likely owing to their NES activity.

Involvement of arginine clusters of R-rich regions with the nuclear and nucleolar localization of BToN.

We identified NLS/NoLS using N-EGFP in HRT-18 and COS-7 cells, given the possibility that the addition of EGFP to the N terminus of BToN (EGFP-N) may affect NLS/NoLS activity. To determine whether the C-terminal truncation of BToN alters its distribution patterns, several deletion mutants were constructed (Fig. 6). The living cells expressing BToN mutants were observed via CLSM, and QFIA was performed as described above. The HRT-18 cells expressing mutants 123, 82, 72, and 62 containing full R-rich regions and lacking NES2 had distribution patterns similar to those of parent N-EGFP, although BToN mutants tended to be distributed in the nucleoplasm/cytoplasm as the truncation extended (Fig. 6). In contrast, the cells expressing mutants 51 and 41, which still contain the pat4 and pat7 cNLSs of the R-rich region and NES1, had distribution patterns identical to that of EGFP (Fig. 6). Similar results were obtained from QFIA of COS-7 cells, although more COS-7 cells expressing mutants 123, 82, and 72 shifted from type 2 to type 1 than cells expressing parent N-EGFP, and the cells expressing mutant 62 exhibited slightly different patterns (Fig. 7). These results indicated that amino acids at positions 1 to 62 of BToN containing a full R-rich region play an important role in the accumulation in nuclear/nucleolar compartments and that NES1 and pat4/7 cNLS alone did not seem to be functional. We created an additional deletion mutant, the Δ40 mutant, in which 40 amino acids from its N terminus were truncated such that it contained the putative NLS but not the pat4/7 NLS. The distribution pattern of the Δ40 mutant was similar to that of N-EGFP in HRT-18 cells (Fig. 6) but partially shifted from type 1 to type 4 or 5 in COS-7 cells (Fig. 7), indicating that proteins with only putative NLS were able to accumulate in the nuclear and nucleolar compartments. We aimed to identify the minimum sequence requirement for the nuclear and nucleolar accumulation. However, amino acids at positions 1 to 82 in BToN, which contained the R-rich region, seemed to become unstable when both ends were truncated. In fact, immunoblotting analysis of these BToN mutants using anti-EGFP antibody detected several bands, likely due to protein degradation (data not shown). Thus, we could confirm only that the BToN requires amino acids at positions 20 to 82 as a minimum sequence requirement. With this sequence, the protein remained relatively stable and had a distribution pattern similar to that of N-EGFP (Fig. 6 and 7). Thus, it suggests the importance of the 20 to 62 residues that overlapped between amino acids at positions 1 to 62 and 20 to 82.

FIG 6.

(Left) Schematic diagram of deletion mutants based on N-EGFP. The box in light blue at the bottom shows the amino acids of the R-rich region, and the main localization of each mutation is shown. (Right) Live-cell imaging of the subcellular localization of deletion mutant proteins in HRT-18 cells and the results of quantitative fluorescence intensity analysis (QFIA). Transfection, relative fluorescence intensity analysis, and calculations were carried out as described in the Fig. 4 legend.

FIG 7.

Results of QFIA based on live-cell imaging of the subcellular localization of deletion and alanine-substituted mutations in COS-7 cells. Mutant designations are indicated above the graphs. Transfection, relative fluorescence intensity analysis, and categorization of each distribution pattern type were carried out as described in the Fig. 5 legend.

Next, to investigate whether arginine clusters in the R-rich region were involved in nuclear/nucleolar localization, these arginine clusters were replaced with alanine residues (Fig. 8). To avoid the appearance of NES activity by suppressing NLS/NoLS activity with alanine substitutions, mutant 123 (lacking NES2) was used as a template. Each arginine cluster was designated R1, R2, or R3, where R1 is identical to pat4 cNLS and R2 and R3 consist of parts of the putative NLS (Fig. 3A). The R1, R2, and R3 mutants in HRT-18 cells were associated with decreased BToN accumulation in the nucleolus compared with that of mutant 123, but BToN was still localized primarily in the nucleus, whereas the cells expressing the R12 and R123 mutants exhibited protein distribution patterns similar to that of EGFP (Fig. 8). In COS-7 cells, the overall tendency was similar, with minor differences. The distribution patterns of the R1 and R3 mutants mainly or partially shifted from type 1 to type 2, but the distribution pattern of the R2 mutant had a small effect, suggesting that substitutions of R1 and R3 caused loss of accumulation in the nucleolus in COS-7 cells. These results suggested that each arginine cluster is involved in nucleolar accumulation and that the R1 cluster played an essential role in the COS-7 cells. Deletion mutants 41 and 51, which contained only the pat4 cNLS (R1 cluster), failed to accumulate BToN in the nuclear or nucleolar compartments. However, the substitution mutant (R3) that contained pat4 cNLS exhibited localized accumulation in the nuclear compartments, implying that both the R1 and R3 clusters are involved in nuclear accumulation. Collectively, these data suggested that the BToN contained a functional NLS/NoLS, likely at amino acid positions 31 to 62 (R-rich region) and that arginine clusters were involved in this activity.

FIG 8.

(Left) Schematic diagram of alanine-substituted mutations based on mutant 123. Twelve arginine residues in the R-rich region (light blue bar) are represented by white letters, and the three R clusters are designated R1, R2, and R3. The main localization of each mutation is shown. (Right) Live-cell imaging of the subcellular localization of substituted mutations in HRT-18 cells and results of QFIA. Transfection, relative fluorescence intensity analysis, and calculations were carried out as described in the Fig. 4 legend.

Identification of the region containing a functional NES in the BToN and its CRM1-independent NES activity.

Because EGFP-N proteins revealed the presence of NES activity in COS-7 cells, BToN must contain a functional NES. However, the NLS/NoLS assay described above suggested that NES1 was not functional. Thus, we focused on NES2 (153-IATFTIKVAL-162), located at the C terminus of the BToN, and constructed several deletion mutants in which the full R-rich region was removed, and the protein was fused with EGFP at its C or N terminus (Fig. 9A). The resulting BToN mutants were expressed in COS-7 cells, and QFIA was performed. To clarify the differences among the mutants, the type 4 category, which represents greater protein accumulation in the cytoplasm than in the nucleus (i.e., N protein export from the nucleus to the cytoplasm), was further divided into three subtypes: 4a (C/Nu = 1.2 to <1.8), 4b (C/Nu = 1.8 to 4.5), and 4c (C/Nu > 4.5) (Fig. 9B). The EGFP-Δ129 mutant exhibited the strongest NES activity, and the other mutants containing NES2 regions exhibited moderate activity, whereas the mutant without the NES2 region did not exhibit any NES activity (Fig. 9C). Because the NES activity of EGFP-Δ129 was stronger than that of Δ129-EGFP or EGFP-Δ82, the addition of EGFP to the C terminus of the BToN or the addition of amino acids 82 to 129 to EGFP-Δ129 may suppress NES activity. It should be noted that among cells expressing EGFP-Δ129, cells that exhibited strong NES activity (type 4c, C/Nu > 4.5) exhibited higher expression levels (i.e., stronger fluorescence) than the other cells (Fig. 10A). Highly expressed BToN was distributed to the cytoplasm nonuniformly (Fig. 9B), which sometimes disrupted nuclear morphology or rarely destroyed the nucleus (Fig. 10B). HRT-18 cells expressing EGFP-Δ129 showed a similar tendency as COS-7 cells, but their NES activity was relatively weak and highly expressed BToN was localized to part of the cytoplasm (Fig. 10C).

FIG 9.

(A) Schematic diagram of deletion mutants fused with EGFP at their C or N termini. (B) Live-cell imaging of the subcellular localization of EGFP-Δ129 in COS-7 cells. To clarify the differences between the mutant profiles, type 4 was further divided into three subtypes, 4a (C/Nu = 1.2 to 1.8), 4b (C/Nu = 1.8 to 4.5), and 4c (C/Nu > 4.5), for the NES analysis. Representative images of type 4b and type 4c patterns with actual ratios are shown. (C) Results of QFIA using live-cell imaging of the subcellular localization of these mutant proteins in COS-7 cells.

FIG 10.

(A) Different expression levels among COS-7 cells expressing EGFP-Δ129 within a single field. Cells exhibiting strong expression were likely to be categorized as type 4c. Images of the same field were taken at low or high exposure because of strong EGFP intensity. (B) Nuclei of cells expressing EGFP-Δ129 with strong intensity sometimes exhibited morphological changes (arrowhead) or, more rarely, were destroyed (arrow). (C) Distribution of EGFP-Δ129 in HRT-18 cells. Strongly expressed EGFP-Δ129 tended to be localized to a part of the cytoplasm (arrow).

We also created an EGFP-Δ152 mutant to identify the minimum sequence requirement for NES activity. This mutant maintained NES activity, suggesting that amino acids 152 to 163 had functional NES. However, since the NES activity of this mutant decreased (data not shown), we used EGFP-Δ129 for further studies. The NES2 sequence of EGFP-Δ129 was completely identical to the class 1d CRM1-dependent NES consensus sequence (Fig.3B), strongly suggesting that LMB treatment affected its distribution. To determine whether NES2 was CRM1 dependent (LMB sensitive), we first constructed the well-known HIV1 Rev protein, which contains a single NLS and NES and uses a CRM1-dependent pathway (79), fused with EGFP at its N terminus (EGFP-hRev) as a control. The COS-7 cells were transfected with EGFP-hRev and treated with or without 20 nM LMB at 36 hpt, incubated for 6 h, and then observed using CLSM. Without LMB treatment, the EGFP-hRev proteins were detected mainly in the nucleolus and cytoplasm. With LMB treatment, they accumulated predominantly in the nucleolus. QFIA clearly showed that protein distribution patterns shifted from type 4 or 5 to type 1 after LMB treatment, suggesting that LMB treatment blocked the CRM1-dependent export pathway (Fig. 11). Next, we compared the distribution patterns of EGFP-Δ129, EGFP-N, N-EGFP, and EGFP in COS-7 cells with or without 20 nM LMB treatment for 6 h. Unexpectedly, the distribution patterns of EGFP-Δ129 were not affected by LMB treatment (Fig. 11). Those of EGFP-N and N-EGFP did not change significantly but shifted slightly from type 1 or 2 to type 4 or 5. These results suggested that the NES2 region of the BToN serves as a functional NES in a CRM1-independent manner.

FIG 11.

Effects of leptomycin B (LMB) treatment on the subcellular localization of EGFP-fusion protein in COS-7 cells. Cells expressing EGFP fusion protein or EGFP alone were treated with or without 20 nM LMB for 6 h. Two representative images are shown for the EGFP-Δ129 mutant for each treatment, one of which was taken at low exposure because of the strong fluorescence intensity (high expression). The relative fluorescence intensity analysis and categorization of each distribution pattern type were carried out as described in the Fig. 5 legend.

Contribution of F156 and V160 to nuclear export.

In spite of its CRM1-independent activity, the NES2 region corresponded exactly to the class 1d CRM1-dependent NES consensus sequence (Ф1XXФ2XXXФ3XФ4; 153-IATFTIKVAL-162). To determine whether key amino acids (Ф1 to 4) play an important role in NES activity, we created a comprehensive set of alanine-substituted mutants based on EGFP-Δ129 (Fig. 12, left). The resulting BToN mutants were expressed in COS-7 cells, and QFIA was performed. The COS-7 cells expressing three or four alanine-substituted mutants (3Aa to d or 4A mutants, respectively) exhibited distribution patterns similar to that of EGFP (Fig. 12, right). Notably, the distribution patterns of the 2Ad mutant, which contained F156A and V160A, were identical to that of EGFP. The other mutants (2Aa to c and 2Af) exhibited significantly decreased NES activity, and the 2Ac mutant, which contained I153A and L162A, retained more activity than other mutants in this group. Likewise, among single-alanine-substituted mutants, the 1Ab and 1Ac mutants with F156A or V160A exhibited weaker NES activity than the 1Aa and 1Ad mutants with I153A or L162A (Fig. 12). These results indicated that four amino acids were involved in NES activity, of which F156 and V160 were the most important.

FIG 12.

(Left) Schematic diagram of alanine-substituted mutations based on the EGFP-Δ129 mutant. The expanded region shows the amino acid sequence of the CRM1-dependent NES consensus sequence, and key amino acids are indicated in bold. (Right) Results of QFIA using live-cell imaging of the subcellular localization of these mutant proteins in COS-7 cells.

Importance of NLS/NoLS-mediated nuclear accumulation of BToN for virus growth.

To study the importance of BToN accumulation in the nuclear compartments for virus growth, we generated four pBAC plasmids, pBAC-BToV-R1, -R2, -R3, and -R13, in which arginine clusters of R-rich regions in NLS/NoLS were replaced with alanine residues, as described in Materials and Methods (Fig. 13A) (80). These bacterial artificial chromosome (BAC) plasmids were transfected into 293T cells, after which the supernatants were used to inoculate permissive HRT-18 cells. The viruses were then plaque purified three times.

FIG 13.

(A) Schematic diagram and generation of recombinant BToVs. rR1, rR2, rR3, and rR13 had N mutants for which the arginine residues of R1, R2, R3, and R13 were replaced with alanine residues. n.t., nucleotide; a.a., amino acid. (B) A representative plaque in the 6-well plate of rescued BToVs. (C) Confocal microscopy images of the subcellular localization of mutant N-expressing COS-7 cells. Cells were fixed at 42 hpt, and N detection was carried out as described in the Fig. 2 legend. Two representative images are shown for cells expressing R13 BToN. Strong fluorescence intensity (high expression) is indicated by an arrow. (D) Confocal microscopy images of the subcellular localization of BToV N and M proteins in rescued BToV-infected HRT-18 cells. Cells were fixed at 24 hpi. Detection of BToN was carried out as described in the Fig. 1 legend. (E) Immunoblotting of BToN in COS-7 cells expressing mutant BToN and HRT-18 cells infected with rescued BToV.

Three recombinant BToVs (rR1, rR2, and rR3) were successfully rescued. They displayed no significant differences in plaque morphology compared to the wild-type BToV (wtBToV) and recombinant wtBToV (rWT) (Fig. 13B), whereas one recombinant BToV (rR13) was not rescued. To investigate whether virus recovery was associated with NLS/NoLS-mediated nuclear accumulation of BToN, the subcellular localization of BToN in COS-7 cells that expressed N proteins independently was observed. Cells were fixed or lysed at 42 hpt and subjected to IFA and immunoblotting using mouse and rabbit anti-N antiserum, respectively (Fig. 13C and E). N proteins derived from rR1, rR2, and rR3 accumulated nonuniformly in the nuclear compartments of the transfected COS-7 cells in a manner similar to that of wtBToV (Fig. 13C) (data from rR2 and rR3 are not shown). In contrast, the distribution pattern of N proteins derived from rR13 that could not be rescued differed significantly (Fig. 13C). Typically, they were uniformly distributed in the nucleus and localized in the perinuclear region in the cytoplasm (Fig. 13C, R13 left) and sometimes showed strong intensity in the cytoplasm (Fig. 13C, R13 right). Although rR13-derived N proteins showed relatively weak fluorescence intensity overall, similar band densities were observed in immunoblotting (Fig. 13E). These results suggested that the rR13-derived N proteins lost the ability for NLS/NoLS-mediated nuclear accumulation, leading to failure of rR13 recovery.

Subcellular localization of BToN in HRT-18 cells infected with the rescued recombinant virus was also observed. HRT-18 cells were infected at an MOI of 0.5 and fixed or lysed at 24 hpi. The N proteins in the cells infected with rR1, rR2, and rR3 accumulated nonuniformly in the nuclear compartment, similar to those of the cells infected with wtBToV and rWT, along with the similar detection of the M proteins (Fig. 13D). Immunoblotting revealed that the band size of N proteins showed some differences in both infected and transient N-expressing cells. This suggested that arginine residues may undergo posttranslational modifications (Fig. 13E), but it seemed not to be well correlated with virus recovery. Finally, to compare growth kinetics, HRT-18 cells were infected with these recombinant BToVs at an MOI of 0.001. Three rescued BToVs had a slight reduction in growth (Fig. 14). wtBToV and rWT had peak titers of >1.0 × 10^7.2 50% tissue culture infective dose (TCID₅₀)/ml at 48 hpi, whereas that of rR2 was 1.0 × 10^6.5 TCID₅₀/ml at 48 hpi. The growth kinetics of rR1 and rR3 were relatively slow and peaked one day late, with a peak titer of 1.0 × 10^6.7 TCID₅₀/ml at 72 hpi. These results suggested that NLS/NoLS-mediated nuclear accumulation of BToN played an important role in virus growth.

FIG 14.

Growth kinetics of rescued BToVs. HRT-18 cells were infected at an MOI of 0.001. At the indicated time, supernatants of infected cells were harvested, and virus titers were determined based on the TCID₅₀. The mean and standard deviation of results of five independent experiments are shown (N = 5). Dashed line, detection limit; N.D., not detected.

DISCUSSION

Unique subcellular localization of BToN.

The previously reported N proteins from CoVs and AVs were partially detected in the nucleolus but predominantly localized in the cytoplasm. Furthermore, in some cases they were primarily detected only in the cytoplasm in infected cells, with just 10 to 20% of cells exhibiting localization in both the cytoplasm and nucleolus (44, 49). In contrast, BToN was mainly observed only in the nucleolus during early infection, but during later infection it was excluded from the nucleolus and accumulated nonuniformly in the nucleoplasm. Notably, a limited quantity of BToN was transported to the cytoplasm during infection. Interestingly, when BToN was expressed alone, it accumulated in the nucleolus, as it did during early infection (Fig. 2). This suggested that other viral proteins or host metabolic changes during viral infection may play a role in excluding BToN from the nucleolus. Two brief studies on the subcellular localization of equine torovirus BEV N proteins in infected cells reported conflicting results: one reported that N protein accumulated only in the cytoplasm, whereas the other reported that it accumulated in the nucleus (71, 72). It is premature to predict the precise localization of BEV N from this study, because of the sequence differences of the R-rich regions in BEV and BToV and the differences of their permissive cells. Further studies of BEV N localization will shed light on the unique characteristics of the ToV N protein.

The replication of CoVs and AVs is restricted to the cytoplasm, and budding and assembly occur in the endoplasmic reticulum (ER)-Golgi intermediate compartment (ERGIC) (81, 82) (and in the Golgi or ER in ToVs [20, 22]). N proteins are transported to the cytoplasm to form nucleocapsids to package the viral genome, which then migrate to the vicinity of the ERGIC or Golgi and ER to be incorporated into virions. To do this, it is believed that these N proteins are localized predominantly in the cytoplasm. Thus, the finding that even parts of CoV or AV N proteins are actively transported into the nucleolus is unique. Among other positive-stranded RNA viruses, it has been widely reported that a similar protein, core (C) protein of Flaviviridae (Flavivirus [FlaV]), is localized in the cytoplasm and nucleus or nucleolus (83–90). The nucleolus is a subnuclear membraneless compartment where several processes take place, including rRNA production and ribosome assembly, cell cycle regulation, apoptosis, DNA damage repair, pre-mRNA and noncoding RNA processing, and cellular stress response (75, 91–93). Because viruses often manipulate the cell cycle, apoptosis, and cell stress to facilitate their replication (94, 95), it is not surprising that cytoplasmic viruses exploit or hijack nucleolar functions or proteins, but it is surprising that many of the proteins targeting the nucleolus are nucleocapsid proteins (N and C proteins; i.e., structural proteins) of cytoplasmic RNA viruses (83–85).

Many studies have investigated the nuclear functions of N or C proteins in CoVs, AVs, and FlaVs. CoV N proteins, including IBV, MHV, and TGEV, are transported to the nucleolus (49, 50) in a cell cycle-dependent manner (96). CoV N proteins in the nucleolus may be involved in cell division. For instance, the expression of IBV N proteins, which colocalize with fibrillarin and interact with nucleolin, delays cell growth by disrupting cytokinesis (51). In addition, TGEV N protein arrests the cell cycle at the S and G₂/M phases, suppressing cell proliferation and apoptosis (97). Among related AVs, the N proteins of PRRSV, EAV, and LDV are transported to the nucleus or nucleolus (42, 44, 52), and PRRSV N proteins colocalize and interact with fibrillarin (43). The NLSs/NoLSs of PRRSV N proteins play an important role in viral replication and pathogenesis (56, 57). The C proteins of several FlaVs, including dengue virus (DENV), West Nile virus (WNV), Japanese encephalitis virus (JEV), and hepatitis C virus (HCV), are also partially localized in the nucleus or nucleolus (83–90). These C proteins interact with nucleolar proteins such as B23 (98, 99), HDM2 (100), and DDX56 (101) and are involved in viral replication (88, 102), assembly (101), pathogenesis (88), and apoptosis (99, 100). In the present study, we did not investigate the nuclear or nucleolar function of BToN but did show that it was colocalized with fibrillarin, like CoV and AV N proteins. However, it remains unclear whether N proteins in the nucleolus serve a common function among CoVs (or AVs) and ToVs.

The most surprising result of this study was that BToN accumulated predominantly in the nuclear compartments throughout the infection process, even though it is the main component of the nucleocapsid. This may be partially explained by the unique characteristics of ToVs. Actinomycin D (an inhibitor of cellular DNA transcription) and alpha-amanitin (an inhibitor of DNA-dependent RNA polymerase II) suppress BEV replication but do not suppress the replication of CoVs, AVs, and many other cytoplasmic RNA viruses, suggesting that ToV replication requires specific cellular genes or nuclear functions (21). Moreover, electron microscopy analysis of BEV and BToV (Breda virus) revealed characteristic tubular structures, likely formed of N proteins, that are morphologically similar to the nucleocapsid in the nucleus and cytoplasm of infected cells (20, 22, 103). Interestingly, similar observations were made for FlaVs but not for CoVs or AVs. Actinomycin D and alpha-amanitin inhibit JEV replication (104, 105), nucleocapsid-like HCV structures in the nucleus have been observed using electron microscopy (106), and C proteins from some FlaVs have been localized in nuclear compartments (87, 102). These findings suggest that BToN may have nucleolar/nuclear functions required for BToV replication, similar to FlaV C proteins. Such nucleolar/nuclear functions required for BToV replication might require large amounts of N protein in the nuclear compartment, whereas small amounts in the cytoplasm may be sufficient to form nucleocapsids. Alternatively, although the distribution of BToN in the cytoplasm was limited overall, we observed that the BToN in the subpopulation of BToV-infected HRT-18 cells was distributed to the cytoplasm to some extent (Fig. 1B). Thus, the tubular structure in the nucleus may be transported to the cytoplasm by an as-yet-unidentified mechanism only when necessary.

Nucleocytoplasmic trafficking signals of BToN.

We established image-based QFIA to identify the nucleocytoplasmic trafficking signals of the BToN. Using QFIA and a set of BToN mutants, we demonstrated that the R-rich region at amino acid positions 31 to 62 likely acts as an NoLS and that arginine clusters have an essential role in NoLS activity. Owing to the sequence diversity of NoLSs, there is only a small degree of similarity in the position, length, and content of basic residues of known NoLSs among CoV and AV N proteins (44, 46–48). NoLSs of BToN and PRRSV N proteins contains a relatively large number of basic residues and cNLS motifs (31-PRFRRRRNPG FRPMFQRRNN SNQNRSRQSR PR-62 and 41-PGKKNKKKNP EKPHFPLATE DDVRHHFTPS ER-72, respectively; boldfacer indicates basic residues, and underlining indicates cNLS motifs). Conversely, those of IBV and PEDV N proteins have a few basic residues and no cNLS motif (71-WRRQARFK-78 and 71-SNWHFYYLGT GPHGDLRYRT-90, respectively). This diversity may be responsible for the differences in the quantities in the nucleolus, timing of trafficking, and binding partners, ultimately resulting in different nucleolar functions.

The N or C proteins targeting the nucleus and/or nucleolus via NLSs/NoLSs often contains NESs, and their trafficking between the nucleus and cytoplasm may be regulated by both signals. CRM1 can recognize cargo proteins containing an NES and export cargo from the nucleus to the cytoplasm, and this pathway is inhibited by LMB (66–69). Although NES sequences are highly diverse, peptide library, bioinformatics, and structural analyses showed that CRM1-dependent NES consensus sequences can be divided into 11 classes (Fig. 3B) (63–65). Based on this information, the N proteins of both AV PRRSVs and EAVs have the class 1a CRM1-dependent NES consensus sequence, whereas the N protein of CoV PEDVs contains an overlapping class 1a or class 3 NES consensus sequence, and these N proteins are LMB sensitive (42, 46, 47). In FlaVs, the NES of LMB-sensitive HCV C proteins was first reported as a nonclassical NES but now fits the criteria of the class 3 NES consensus sequence (109-PTDPRRRSRNLGKVIDTLTCGFADL-133) (107).In the SARS-CoV, the N proteins contain overlapping NES consensus sequences, including those in classes 1b-R, 1d-R, and 3 (220-LALLLLDRLNQL-231), but it is unclear whether this region is functional or nonfunctional as an NES (54). Alternatively, another region (324-EVTPSGTWLT-334), which does not match any CRM1-dependent NES consensus sequence, was reported to serve as an NES, and actually SARS-CoV N proteins were not sensitive to LMB treatment (53, 108). Our study showed that the BToN contains the class-1d NES consensus sequence (Ф1XXФ2XXXФ3XФ4: 153-IATFTIKVAL-162) but is insensitive to LMB treatment, like the IBV N protein, which contains a class 2 NES consensus sequence (Ф1XФ2XXФ3XФ4: 291-LQLDGLHL-297) (70). Nevertheless, the substitutions of key amino acids (positions Ф1 to 4) had a strong effect on NES activity in both IBV and BToV N proteins. Recently, a structural comparison of 13 CRM1-bound NESs highlighted the importance of hydrophobic amino acids at positions Ф2 and Ф3 for CRM1 binding (64, 65). Similarly, we found that F156 and V160 of BToN at positions Ф2 and Ф3 were the main contributors to NES activity in spite of its CRM1-independent nature, whereas L296 and L298 at positions Ф3 and Ф4 were the main contributors in the IBV N protein (70). Although we demonstrated CRM1-independent NES activity of BToN, we could not rule out the possibility that a highly hydrophobic NES region of the EGFP-Δ129 mutant was structurally exposed by the truncation. This may have resulted in the formation of larger proteins that could not pass through the NPCs, leading to superficial insensitivity to LMB. In vitro binding assays of BToN or IBV N proteins with CRM1 proteins may clarify their CRM1 dependency.

Importance of NLS/NoLS-mediated nuclear accumulation of BToN for virus growth.

We recently established a reverse genetics system using a full-length infectious cDNA clone of BToV in a BAC (80). Using this system, three recombinant BToVs (rR1, rR2, and rR3) were successfully rescued and displayed slightly reduced growth ability compared to wtBToV and rWT. The N proteins from rescued BToVs had alanine substitutions in R-rich regions of NLS/NoLS but retained the ability to accumulate in the nuclear compartments, similar to that of wtBToV. In contrast, one of the recombinant BToVs (rR13) was not rescued and its N protein lost the capacity for NLS/NoLS-mediated nuclear accumulation. We could not rule out the possibility that the failure of rR13 recovery was due to the recovery efficiency of reverse genetics. However, there is no doubt that the growth or virus production of rR13 was drastically reduced or impaired because this system could rescue the recombinant BToV with about 1,000-times-reduced growth ability (80). The rR13-derived N protein was distributed in both the nucleus and cytoplasm and was strongly expressed in the cytoplasm in the subpopulation of COS-7 cells, suggesting that simply the presence of BToNs in the nucleus or cytoplasm is not sufficient for virus growth. The results indicate the importance of BToN nuclear accumulation mediated by NLS/NoLS. Analogous studies have been performed using recombinant PRRSV (56, 57), in which a fraction of the N proteins localized specifically to the nuclear compartments in virus-infected cells. The recombinant virus in which the N proteins did not exhibit nucleolar localization by introducing mutations in its NLS/NoLS was successfully rescued, although it grew to a titer 100-fold lower than that of the parent virus (57). Therefore, in cultured cells, the effect of NLS/NoLS-mediated nucleolar localization of PRRSV N proteins was less than NLS/NoLS-mediated nuclear accumulation of BToN. These results suggest that NLS/NoLS-mediated nuclear accumulation of BToN plays a very important role in virus growth.

Conclusion.

We observed that large amounts of BToN were transported to the nucleus and/or nucleolus throughout infection, which is unique among positive-stranded RNA viruses. This nuclear accumulation was regulated by unique NLS/NoLS and CRM1-independent NES and was important for virus replication. Further analysis of BToN, including its nuclear/nucleolar function, binding partners of its receptor/adapter proteins, and its CRM1 dependency, may provide new insight into the assembly mechanisms of not only neglected pathogen ToVs but also positive-stranded RNA viruses in general.

MATERIALS AND METHODS

Cells and viruses.

A human rectal adenocarcinoma subcell line (HRT-18 Aichi) (38–40), 293T, and COS-7 cells were maintained at 37°C and 5% carbon dioxide in Dulbecco’s modified minimum Eagle’s medium (DMEM) supplemented with 5 to 10% fetal bovine serum (FBS) [DMEM(+)] and penicillin–streptomycin. BToV (Aichi strain) was propagated in HRT-18 cells in DMEM without FBS [DMEM(−)], as FBS inhibits BToV infection.

Antibodies.

Mouse polyclonal anti-N and anti-M antisera and rabbit polyclonal anti-N antiserum were obtained by transfecting mice with pCAGGS-BToN using in vivo electroporation as previously described (109). The mice were immunized using the synthetic peptide of M-protein (2-FETNYWPFPDQAPN-15), and the rabbits were immunized using the synthetic peptide of N-protein (141-EVSSGDQETPHKIA-154). Fibrillarin (C13C3) rabbit MAb was purchased from Cell Signaling Technology (Tokyo, Japan). The secondary antibody horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG and Alexa Fluor 594-conjugated goat anti-rabbit IgG were purchased from Invitrogen (Tokyo, Japan). Fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG+IgM+IgA was from Rockland Immunochemicals (Pottstown, PA, USA). IgG+IgM was from Jackson Immuno Research (West Grove, PA, USA). Two FITC-conjugated goat anti-mouse antibodies were mixed 1:1 for subsequent use.

Plasmid construction.

The cDNA encoding BToN was cloned into a pTarget vector using the TA cloning technique. The GAATTC nucleotide sequence within the BToN cDNA was substituted to GAACTC (silent mutation) to remove the EcoRI site, and the BToN cDNA of pTarget was then subcloned into the pCAGGS vector using the EcoRI sites of each vector (pCAGGS-BToN). Site-directed mutagenesis to create BToN mutants was performed using standard PCR protocols with the overlap extension technique (110). The nomenclature and construct of each mutant are as follows. EGFP-N and N-EGFP represent enhanced green fluorescent protein (EGFP) fused to the N and C terminus of the BToN through linker sequences (PPVAT; cca ccg gtc gcc aca), respectively (111). In the NLS analysis, because wild-type (wt) BToN consists of 163 amino acids (Fig. 3A) and the deletion mutants were constructed using pCAGGS-N-EGFP as a template, “mutant 123” represents the truncation of 40 amino acids from the C terminus of the BToN and EGFP was fused to its C terminus. In the NES analysis, Δ82-EGFP or EGFP-Δ82 represents the truncation of 82 amino acids from its N terminus and EGFP was fused to its C or N terminus, respectively. The alanine-substituted mutants were constructed using pCAGGS-123 or -EGFP-Δ129 as a template. All mutants contained linker sequences between N mutant proteins and the EGFP. The PCR-derived region of the final construct was verified via DNA sequencing. To prepare several control proteins, the cDNA of three repeated NLSs of simian virus 40 (monopartite: DPKKKRKV, three repeats) and single NLSs of nucleoplasmin (bipartite: KRPAATKKAGQAKKKK), HIV1-Rev (GenBank accession no. D86068.1), and B23 (accession no. M28699.1), all of which were flanked with two different restriction sites, were chemically synthesized and subcloned into pCAGGS, pCAGGS-EGFP, or pEGFP-N1 vectors using restriction sites. The resulting proteins were three repeated NLSs of SV40 and a single NLS of nucleoplasmin fused with EGFP at their C termini (SV40-EGFP, Nuc-EGFP) expressed from the pCAGGS vector, HIV1-Rev fused with EGFP at its N terminus (EGFP-hRev) from the pEGFP1-N1 vector, and B23 fused with DsRed at its N terminus (DsRed-B23) from the pCAGGS vector.

Immunofluorescence staining of BToV-infected HRT-18 cells and N-expressing cells.

Thirty-five-millimeter glass-base dishes (12-mm glass area) (Iwaki, Shizuoka, Japan) were seeded with 1.6 × 10⁵ HRT-18 cells. The following day, cells were washed with phosphate-buffered saline (PBS) twice and infected with BToV at a multiplicity of infection (MOI) of 0.5 and incubated for 1 h at 37°C. The cells were then washed with PBS and incubated in DMEM(−). Transient N-expressing HRT-18 or COS-7 cells were prepared by transfection with pCAGGS-BToN as described below. Cells were fixed with 4% paraformaldehyde (PFA) for 20 min at room temperature (RT) and permeabilized with 0.2% Triton X-100 for 20 min at RT. The fixed cells were incubated with mouse anti-N antiserum and fibrillarin rabbit MAb at 4°C overnight and then incubated with two FITC-conjugated anti-mouse antibodies and Alexa Fluor 594-conjugated goat anti-rabbit IgG for 2 h at RT. The nuclei of cells were stained with Hoechst 33258 solution (Dojindo Lab, Tokyo, Japan) for 20 min at RT. PBS washing was performed twice between each step. Stained cells were observed using confocal laser scanning microscopy (CLSM). As a control for CoV, mouse brain tumor DBT cells were infected with MHV (A59 strain) at an MOI of 1.0 and fixed, permeabilized, and stained as described above, except with MHV N MAb and FITC-conjugated anti-mouse IgG (Invitrogen).

Live-cell imaging of HRT-18 and COS-7 cells expressing BToN mutants fused with EGFP.

HRT-18 or COS-7 cells in 35-mm glass-base dishes were prepared at a density of 0.4 × 10⁵ or 0.5 × 10⁵ cells per dish, respectively. The following day, these cells were transfected with 0.5 μg pCAGGS-BToN mutants fused with EGFP using Lipofectamine 3000 reagents (0.75 μl Lipofectamine 3000 and 0.5 μl P3000 reagent) (Thermo Fisher Scientific, Tokyo, Japan), with 0.2 μg pCAGGS-DsRed-B23 for HRT-18 cells or 0.75 μl TransIT-X2 dynamic delivery system reagent (Clontech Takara, Tokyo, Japan) for COS-7 cells. Cells were observed at 38 to 44 h posttransfection (hpt) using CLSM and were incubated with Hoechst 33258 solution for 20 min before observation. The broad range of observation times was necessary, as it takes at least 30 min to observe one sample.

CLSM observations.

Confocal images of stained BToV-infected HRT-18 cells and COS-7 and HRT-18 cells expressing BToN or BToN mutants fused with EGFP were taken using an LSM 710 laser scanning microscope (Carl Zeiss, Oberkochen, Germany) with ZEN 2 software (Carl Zeiss) under a 40× or 63× oil immersion lens objective. The EGFP, Alexa Fluor 594 (or DsRed), and Hoechst stain were excited at 488, 561, and 405 nm, respectively. Because both stained BToN and BToN fused with EGFP showed relatively uniform fluorescence intensity in HRT-18 cells, only one exposure was used per field. In contrast, for COS-7 cells expressing BToN fused with EGFP, the fluorescence intensity differed markedly among cells, and images needed to be captured using at least two or three different exposure times to include a range of fluorescence intensities.

Image analysis and QFIA.

To quantitatively assess the relative fluorescence intensity of N proteins localized in the cytoplasm, nucleus, and nucleolus, images were analyzed using the lite blue edition of ZEN 2. As shown in Fig. 1, 2, and 4, the red peaks (fibrillarin or DsRed-B23), blue peaks, and area surrounding the blue peaks were regarded as the nucleolus, nucleus, and cytoplasm, respectively. The red arrow representing intensity was drawn to include the cytoplasm, nucleus, and at least one nucleolus for each cell, and relative fluorescence intensities were evaluated. Quantitative fluorescence intensity analysis (QFIA) was performed by different strategies, depending on cell type. For HRT-18 cells, cells were coexpressed with each EGFP fusion protein and DsRed-B23, and QFIA was carried out by calculating the average ratio of the relative EGFP fluorescence intensity in the nucleolus, nucleus, and cytoplasm, as the distribution pattern of EGFP fusion proteins did not differ significantly among cells. The highest value of EGFP fluorescence intensity among the cellular compartments was set to 10. For COS-7 cells, because the coexpression of DsRed-B23 and EGFP fusion proteins disrupted nucleolar morphology and caused an unusual distribution of DsRed-B23, only the EGFP fusion proteins were expressed. Fortunately, because the nucleoli of COS-7 were easily observed as one to three clear circles in the nucleus when the nucleolar compartments were maintained, these circles were regarded as nucleoli. As the distribution patterns of BToN fused with EGFP differed significantly among COS-7 cells, QFIA was performed by categorizing cells into five types, based on the setting criteria (Fig. 5). Type 1 represented accumulation in the nucleolus, which can be described as No > Nu > C or No > Nu = C (where No, Nu, and C represent the EGFP fluorescence intensity of the nucleolus, nucleus and cytoplasm, respectively). The criteria for type 1 were set to C/No < 0.4 and C/Nu < 1.0. Type 2 represented accumulation in the nucleus without clear nucleolar accumulation (Nu > C; the criteria were C/Nu < 0.8 and No was unclear or disrupted). Type 3 showed distribution of fluorescence throughout the cytoplasm and nucleus, similar to the pattern of EGFP (Nu = C; the criterion was C/Nu = 0.8 to 1.2). Types 4 and 5 represented accumulation in the cytoplasm without and with clear nucleolar accumulation, respectively (type 4, C > Nu, C/Nu > 1.2; type 5, C > Nu, No > Nu, C/Nu > 1.2, and No/Nu > 1.5). To clarify the differences in distribution patterns among BToN mutants, types were further divided into subtypes depending on the purpose. For example, type 4 was divided into the subtypes 4a (C/Nu = 1.2 to 1.8), 4b (C/Nu = 1.8 to 4.5), and 4c (C/Nu > 4.5) in the NES analysis (Fig. 9). These criteria were determined from the analysis of the total data set. Nearly all the distribution patterns belonged to one of these types, and the results reflected the differences in subcellular localization of N proteins among mutants. Approximately 40 HRT-18 cells or more than 80 COS-7 cells expressing BToN fused with EGFP were analyzed from each of at least three independent experiments, and we selected cells that had relative fluorescence intensities of a green peak within the appropriate dynamic range. In the results, n refers to the number of analyzed cells.

Immunoblotting.

BToV infection was performed as described above, except that the cells were seeded in the wells of 48-well plastic plates. At the indicated time, infected cells were lysed with sample buffer (50 mM Tris, 2% sodium dodecyl sulfate, 0.1% bromophenol blue, 10% glycerol, and 1% 2-mercaptoethanol) and boiled for 5 min. N proteins from samples were analyzed via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to polyvinylidene difluoride (PVDF) membranes (Merck Millipore, Tokyo, Japan), and then incubated with rabbit anti-N antiserum and HRP-conjugated goat anti-rabbit antibodies. Protein bands were visualized with using ECL Prime Western blotting detection reagent (GE Healthcare, Tokyo, Japan) on a Light Capture II instrument (ATTO, Tokyo, Japan).

LMB treatment.

COS-7 cells were transfected with mutant cDNA. At 36 hpt, growth medium was replaced with DMEM(+) containing 20 nM LMB (Cell Signaling Technology, Tokyo, Japan), and cells were incubated for 6 h. At 42 hpt, cells were observed using CLSM as described above. During all procedures, including nuclear staining and observation, 20 nM LMB was present in the medium.

Construction of pBAC-BToV with mutations in NLS/NoLS.

A BAC clone carrying full-length cDNA of the BToV genome, pBAC-BToV, was generated using a Red/ET recombination system counterselection BAC modification kit (Gene Bridges, Heidelberg, Germany), as described previously (80). To generate recombinant mutant BToVs, using the same kit and pBAC-BToV as a backbone, four pBAC plasmids were created, pBAC-BToV-R1, -R2, -R3, and -R13, in which arginine clusters of the R-rich region in NLS/NoLS were replaced with alanine residues (Fig. 13A).

Recovery of recombinant BToVs.

One day before experimentation, 293T and HRT-18 cells were seeded in 24-well plates at a density of 1.5 × 10⁵ and 2.0 × 10⁵ cells/well, respectively. 293T cells were transfected with 2.0 μg of pBAC-BToV mutants using 1.5 μl of Lipofectamine 3000 and 1.0 μl of P3000 reagent (Thermo Fisher Scientific, Waltham, MA, USA) and incubated at 37°C. The next day, DMEM(+) was replaced with DMEM(−) to remove FBS. At 3 days postinfection (dpi), the supernatant from the transfected 293T cells was used to inoculate permissive fresh HRT-18 cells, after which the virus was plaque purified three times. Purified virus was added to HRT-18 cells with 500 μl of DMEM(−) in 24-well plates, and after 2 dpi, the supernatant was harvested and stored at −70°C as the master virus. Next, 100 μl of master virus was used to inoculate HRT-18 cells in 10 ml of DMEM(−) in a 10-cm dish; these viruses were harvested after the appropriate cytopathic effect was observed and were stored as working virus. The recombinant BToVs rescued from these BAC plasmids were designated rR1, rR2, rR3, and rR13 (Fig. 13A).

Plaque assay.

HRT-18 cells were prepared in 6-well plates at a density of 1.0 × 10⁶ cells/well. The next day, the cells were washed twice with PBS, inoculated with recombinant BToVs, and incubated at 37°C for 1.5 h. Unabsorbed virus was removed, and cells were overlaid with MEM (Sigma-Aldrich, St. Louis, MO, USA) containing 0.65% agarose, 2 nM l-glutamine, 10 mM HEPES, and penicillin–streptomycin. The cells were then incubated at 37°C until the appropriate plaque size was observed. Cells were fixed with 10% formaldehyde and stained with crystal violet.

Growth kinetics of recombinant BToVs.

Confluent HRT-18 cells in 24-well plates were washed twice with PBS and infected with recombinant BToVs at an MOI of 0.001. After 1.5 h of adsorption at 37°C, unadsorbed virus was removed, cells were washed twice with PBS, and 0.5 ml of DMEM(−) was added. Virus in the supernatant was collected at 2, 24, 48, and 72 hpi. Virus titers in the culture medium were determined using HRT-18 cells in a 96-well plate based on the TCID₅₀, as previously described (112).