N protein is able to compete with PABP for binding to the poly(A) tail in vitro and in cells.

To address the question of whether N protein is able to compete with PABP for binding to the poly(A) tail in an environment in which they coexist in vitro, the ³²P-labeled poly(A) tail RNA probe was incubated with mixtures containing various molar ratios of N protein to PABP, followed by EMSA. The EMSA results of N protein or PABP binding to the poly(A) tail and the relative binding percentage are illustrated in the upper and lower portions of Fig. 2A, respectively. As shown in Fig. 2A, upper portion, at molar ratios of N/PABP from 65.6 to 5.7 in lanes 3 to 7 (with the increase of PABP), minor (complex 1, indicated by the white dot in lane 3) and major (indicated by the white asterisk in lane 3) RNA-protein complexes were observed. Since the major complexes in lanes 3 to 7 corresponded to N-RNA complex in lane 2, the preferential binding of the 65-nt poly(A) tail to N protein was determined at molar ratios between 5.7 and 65.6. With further increase of PABP (i.e., decreased molar ratio of N/PABP from 4.0 to 1.9 in lanes 8 to 10), the minor complex (complex 1, indicated by the white dot in lane 3) in lanes 3 to 7 became the major complex in lanes 8 to 10, suggesting that the major complex (complex 1) consists of PABP and the 65-nt poly(A) tail. Furthermore, with the increase of PABP in lanes 11 to 13, complex 1 almost disappeared; however, complex 2 appeared, which corresponded to the PABP-RNA complex in lane 14. Since the major complex in lanes 8 to 13 consists of PABP and the 65-nt poly(A) tail, the preferential binding of the 65-nt poly(A) tail to PABP was determined at molar ratios between 0.6 and 4.0 (lanes 8 to 13). Note that a small amount of N protein (~15% [Fig. 2A, lower portion]) still bound to the poly(A) tail when the molar ratio of N protein to PABP was from 3.0 to 4.0 (lanes 8 and 9). Based on these results, it was concluded that N protein can compete with PABP for binding to the poly(A) tail in vitro, even though at the same molar ratio (lane 12), PABP exhibits better binding affinity to poly(A) tails than N protein.

Open in a separate window

FIG 2

N protein competes with PABP for binding to the poly(A) tail. (A) (Top) In vitro analysis for preferential binding of the ³²P-labeled 65-nt poly(A) tail in an environment containing various molar ratios of N protein to PABP by EMSA (lanes 2 to 14). Lane 1, ³²P-labeled RNA only. Gels were spliced for labeling purposes. (Bottom) The relative binding percentages of N protein and PABP with the poly(A) tail were determined according to the results shown in the top portion. N/A, not applicable. (B) Identification of the binding of PABP and N protein with poly(A) tail in vivo. The ³²P-labeled 65-nt poly(A) tail was transfected into cells, followed by UV cross-linking and immunoprecipitation (IP) using an anti-PABP (left) or anti-N protein (right) antibody. The resulting products were analyzed by SDS-PAGE and autoradiographed. Values in panel A represent means ± SD from three independent experiments.

To determine whether N protein is able to bind to poly(A) tail in infected cells, the ³²P-labeled 65-nt poly(A) tail was transfected into BCoV- or mock-infected cells and UV cross-linked. Cell lysates were collected and an antibody against PABP or N protein was employed to immunoprecipitate PABP or N protein, followed by RNase treatment. As shown in Fig. 2B, left panel, antibody against PABP immunoprecipitated an ~70-kDa protein from mock-infected and BCoV-infected cells (lanes 4 and 5, respectively); however, in Fig. 2B, right portion, antibody against N protein immunoprecipitated an ~50-kDa protein from a BCoV-infected cell (lane 5) but not the mock-infected cell (lane 3). The results suggest that in addition to PABP, N protein is able to bind to the poly(A) tail in infected cells.

Determination of molar ratio of N protein to PABP in subcellular locations at different stages of infection.

As shown in Fig. 2A, the molar ratio of N protein to PABP plays a role in poly(A) tail binding preference. In addition, it has been suggested that coronavirus replication can occur in a modified membrane-associated compartment (32). It was therefore speculated that molar ratios in subcellular locations of coronavirus-infected cells at different stages of infection may also be decisive regarding PABP or N protein binding preference for the poly(A) tail. Thus, subcellular fractions of the cytosol and membrane were obtained at various time points of infection, and the amounts of N protein and PABP in each cellular fraction according to immunoblotting (Fig. 3A and ​andB,B, upper portions) were quantified based on a standard curve obtained from known amounts of the proteins. As shown in Fig. 3A and ​andB,B, middle and lower portions, the molar ratio of N protein to PABP in both the cytosol and membrane was low (~0.4) during the initial infection but increased (from ~0.4 to ~2.6 in the cytosol and from ~0.3 to ~10.5 in the membrane) at later infection stages. The results indicate that the amounts of N protein are increased in the both cytosol and membrane at the later time points of infection. Thus, based on the results shown in Fig. 2 and ​and3,3, we speculate that the poly(A) tail may preferentially bind with PABP during the initial infection but with N protein in the later infection, especially in membrane-associated structures.

Open in a separate window

FIG 3

Molar ratio of N protein to PABP in subcellular fractions during infection. (Top) N protein and PABP immunobloting analysis for the cytosol (A) or membrane (B). (Middle and bottom) Molar ratios of N protein to PABP and relative percentages between N protein and PABP, respectively. The amounts of N protein and PABP were measured as follows. Different known concentrations of N protein and PABP were identified by immunoblotting using antibodies against both N protein and PABP. The signals were scanned densitometrically and then plotted against the concentration to obtain a standard curve for the quantitation of N protein and PABP shown at the top. Values represent means ± SD from three independent experiments. hpi, hours postinfection.

BCoV N protein inhibits viral translation both in vitro and in vivo.

Because the poly(A) tail is able to bind to N protein with high affinity (Fig. 1) and in infected cells (Fig. 2), we hypothesized that such binding may prevent the poly(A) tail on coronavirus RNA from interacting with translation factors, leading to translation inhibition. To test the hypothesis, a BCoV defective interfering (DI) RNA, a surrogate for the coronavirus genome that has been extensively used for studies of coronavirus gene expression (33,–37) (Fig. 4A), was engineered to express enhanced green fluorescent protein (EGFP); the construct was designated DI-EGFP. For in vitro translation analysis, DI-EGFP with the 65-nt poly(A) tail was first incubated with various amounts of N protein (Fig. 4B) for 15 min to allow the binding of N protein to the 65-nt poly(A) tail on DI-EGFP and then added to a rabbit reticulocyte lysate (RRL) for another 90 min. A similar experiment was performed in which DI-EGFP was first incubated with PABP or GST. As shown in Fig. 4B, translation of DI-EGFP with the 65-nt poly(A) tail was inhibited with increasing amounts of N protein but not PABP or GST (data not shown). To test whether the inhibition was due to the effect of N protein on the RRL, various amounts of N protein were first incubated with RRL for 60 min, and then DI-EGFP with the 65-nt poly(A) tail was added. The translation efficiency of DI-EGFP, however, was not altered (data not shown), indicating that N protein at these concentrations had no effect on the translation efficiency of RRL. Accordingly, the reduced translation efficiency shown in Fig. 4B was due to the binding of N protein with DI-EGFP but not the effect of N protein on RRL. Furthermore, it has been shown that translation using RRL still occurs with an mRNA lacking a poly(A) tail, although the translation efficiency is affected (38). Consequently, we hypothesized that if the decreased translation efficiency was due to the binding of N protein to the poly(A) tail, translation efficiency of poly(A)-deficient DI-EGFP is not altered with increasing amounts of N protein. To test this, poly(A)-deficient DI-EGFP was generated, incubated with various amounts of N protein for 15 min, and then added to the RRL. As shown in Fig. 4C, the translation efficiency was not significantly affected with increasing amounts of N protein, suggesting that the decreased translation in Fig. 4B may have been mostly due to the interaction between N protein and poly(A) tail. Because N protein apparently had no effect on the translation efficiency of RRL (data not shown) and on translation efficiency of poly(A)-deficient DI-EGFP (Fig. 4C), the inhibitory effect of translation shown in Fig. 4B may be attributed to interaction between the poly(A) tail and N protein. Note that after in vitro translation in RRL the amounts of DI-EGFP at various concentrations of N protein were not significantly altered, indicating that the stability of DI-EGFP is not a factor affecting the translation efficiency. It was therefore concluded that N protein is able to inhibit viral translation by binding to the viral poly(A) tail in vitro.

Open in a separate window

FIG 4

Translation inhibition of coronaviral RNA by N protein. (A) Diagram of the BCoV genome, BCoV DI RNA, and its derivative DI-EGFP. DI-EGFP was employed for the following translation analyses. (B) (Left) In vitro-synthesized fusion protein (top) from 1 μg of input DI-EGFP RNA transcript with the 65-nt poly(A) tail (Ipt. DI.) (middle), which was preincubated first with 0, 2, and 4 μM N protein (bottom) and then subjected to in vitro translation in RRL. (Right) Relative levels of in vitro-synthesized DI-EGFP fusion protein. The values shown are relative to the amount of synthesis in the absence of N protein (i.e., 0 μM N protein). (C) (Left) In vitro-synthesized fusion protein (top) from 1 μg of input poly(A) tail-deficient DI-EGFP RNA transcript (middle), which was preincubated with 0, 2, and 4 μM N protein (bottom) and then subjected to in vitro translation in RRL. (Right) Relative levels of in vitro-synthesized DI-EGFP fusion protein. The values shown are relative to the amount of synthesis in the absence of N protein (i.e., 0 μM N protein). (D) (Left) Diagram showing the experimental procedures to determine the effect of N protein on the translation efficiency of DI-EGFP [with 65-nt poly(A) tail] in vivo. (Right) Immunoblot showing the synthesis of the fusion protein from DI-EGFP in the presence of N protein or His–β-actin at different times posttransfection. The levels of DI-EGFP RNA and 18S rRNA were similar between the groups at the same time point as quantified by RT-qPCR. (E) Relative levels of in vivo fusion protein synthesis based on the results of the right side in panel D. The values shown are relative to the amount of synthesis in the presence of His–β-actin at each time point. (F) (Left) Diagram showing the experimental procedures to determine the effect of N protein on the translation efficiency of BCoV nsp1 in vivo. (Right) Immunoblot showing the synthesis of BCoV nsp1 in the presence of N protein or His-β-actin at different times posttransfection. The levels of viral genome (BCoV gRNA) and 18S rRNA were similar between the groups at the same time point as quantified by RT-qPCR. (G) Relative levels of BCoV nsp1 in vivo based on the results on the right side in panel F. The values shown are relative to the amount of synthesis in the presence of His–β-actin at each time point. Values in panels B, C, E, and G represent means ± SD from three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (unpaired Student's t test). RRL, rabbit reticulocyte lysate; pt, posttransfection; N, N protein; β, His–β-actin.

To further assess whether translation inhibition by N protein also occurs in vivo, the N protein or His–β-actin transcript was transfected into HEK-293T cells, followed by transfection of DI-EGFP with the 65-nt poly(A) tail (Fig. 4D, left portion) or by infection of BCoV (Fig. 4F, left portion). Cell lysates were harvested and analyzed by immunoblotting to quantitate the translation efficiency of DI-EGFP and coronavirus nsp1 (representing genome expression). As shown in Fig. 4D, right portion, and Fig. 4E, inhibition of the DI-EGFP translation was observable in cells transfected with the N protein transcript at 3, 8, and 16 h in comparison with those transfected with the His–β-actin transcript, suggesting that N protein is able to inhibit translation of DI-EGFP in vivo. Similar inhibition results were also obtained for cells infected with BCoV (Fig. 4F, right portion, and Fig. 4G), suggesting that N protein can inhibit translation of the coronavirus genome. Note that the levels of DI-EGFP RNA (Fig. 4D) and viral genomic RNA (Fig. 4F) were similar between the groups at the same time point, as confirmed by reverse transcription-quantitative PCR (RT-qPCR) (data not shown). Therefore, based on the results for the in vivo binding of N protein to poly(A) tail (Fig. 2B) and the in vitro analyses shown in Fig. 4B and ​andC,C, the inhibitory effect of N protein on translation of DI-EGFP and BCoV in vivo may be at least partly attributable to the binding of N protein to the poly(A) tail on DI-EGFP and the BCoV genome.

BCoV N protein modulates translation of host mRNAs both in vitro and in vivo.

To examine whether the binding of N protein to the poly(A) tail of mRNA also inhibits host mRNA translation, a β-actin transcript with the 65-nt poly(A) tail was first incubated with N protein to form an N protein-poly(A) complex and then subjected to an in vitro translation assay with the RRL. As shown in Fig. 5A, expression of β-actin transcripts was inhibited with increasing amounts of N protein. As with the in vitro translation assay for DI-EGFP (Fig. 4), to further determine whether the inhibitory effect was due to the binding of N protein to the poly(A) tail, a poly(A) tail-deficient β-actin transcript was first incubated with various amounts of N protein, followed by the assay. However, inhibition was not observed (Fig. 5B), as no significant difference in expression of the poly(A) tail-deficient β-actin transcript was observed with increasing amounts of N protein. These results (Fig. 5A and ​andB)B) suggest that binding of N protein to the poly(A) tail of the β-actin transcript is a major factor leading to translation inhibition.

Open in a separate window

FIG 5

Translation inhibition of cellular mRNA by N protein. (A) (Left) In vitro-synthesized β-actin (top) from 1 μg of input β-actin RNA transcript with the 65-nt poly(A) tail (Ipt. β-act.) (middle), which was preincubated with 0, 2, and 4 μM N protein (bottom) and then subjected to in vitro translation in RRL. (Right) Relative levels of in vitro-synthesized β-actin. The values shown are relative to the amount of synthesis in the absence of N protein (i.e., 0 μM N protein). (B) (Left) In vitro-synthesized β-actin (top) from 1 μg of the input poly(A) tail-deficient β-actin RNA transcript (middle), which was preincubated with 0, 2, and 4 μM N protein (bottom) and then subjected to in vitro translation in RRL. (Right) Relative levels of in vitro-synthesized β-actin. The values shown are relative to the amount of synthesis in the absence of N protein (i.e., 0 μM N protein). (C) Effect of expressed N protein on translation of host mRNAs in vivo. After mock transfection or independent transfection of His–β-actin and N protein RNA transcripts into HEK-293T cells in the absence or presence of actinomycin D followed by [³⁵S]methionine, equal amounts of cell lysate were analyzed by SDS-PAGE and exposed to X-ray film (left) or stained with Coomassie blue (right). The levels of host mRNA (represented by GAPDH mRNA) and 18S rRNA shown on the left were quantified by RT-qPCR. (D) Relative levels of host protein synthesis based on the results on the left side of panel C. The values shown are relative to the amount of synthesis in the absence of transfection (i.e., mock transfection). Values in panels A, B, and D represent the means ± SD from three independent experiments. ***, P < 0.001 (unpaired Student's t test).

In addition to individual cellular mRNAs in vitro, inhibition of host mRNA translation by N protein was also examined globally in cells. For this, the N protein transcript or His-tagged β-actin transcript was independently transfected into HEK-293T cells for 1 h, after which the cells were incubated for 3 h in the presence or absence of actinomycin D. After addition of actinomycin D, the cells were labeled with [³⁵S]methionine for 8 h, and equal amounts of cell lysate were analyzed by SDS-PAGE. As shown in the left portions of Fig. 5C and ​andD,D, inhibition of host protein synthesis was not apparent in cells not treated with actinomycin D; however, with actinomycin D treatment, synthesis of host proteins in cells transfected with the N protein transcript was decreased by approximately one-third in comparison with that in cells transfected with the His-tagged β-actin transcript or mock transfected (Fig. 5C, left portion, and Fig. 5D, right portion). The fact that the efficiency of N protein inhibition was better in actinomycin D-treated cells than in untreated cells may indicate that in the absence of nascent mRNA synthesis, expressed N protein was involved in interaction with preexisting mRNAs, leading to inhibition of host protein synthesis. In addition, the levels of host mRNA (represented by glyceraldehyde-3-phosphate dehydrogenase [GAPDH] mRNA) were similar (Fig. 5C) between the groups treated with actinomycin D, as confirmed by RT-qPCR (data not shown), and thus were not affected by the expressed N protein. Thus, it was concluded that in addition to coronaviral RNA, N protein is also able to globally inhibit host mRNA translation, and based on in vitro results (Fig. 5A and ​andB),B), such inhibition in cells may at least partly result from the binding of N protein to the poly(A) tail.

Interactions among the poly(A) tail, N protein, and PABP.

To elucidate the possible mechanism by which interactions among the poly(A) tail, N protein, and PABP regulate gene expression, we first determined whether the poly(A) tail is able to interact with both the N protein and PABP using lysates of infected cell. For this, an 84-nt biotinylated RNA, consisting of 19 non-poly(A) nucleotides (containing biotin-conjugated uridine) followed by a 65-nt poly(A) tail, was synthesized and incubated with cell lysates, followed by a streptavidin pulldown assay and immunoblotting. As shown in Fig. 6A (lane 1), both the N protein and PABP were detected (indicated by asterisks), demonstrating poly(A) tail interaction. To ensure that the detection of the N protein and PABP was in fact due to interaction with the poly(A) tail and not with the 19 non-poly(A) residues, a biotinylated RNA containing the 19 non-poly(A) nucleotides was also used. However, neither was observed (data not shown) by immunoblotting, confirming that the 65-nt poly(A) tail, and not 19 non-poly(A) nucleotides, can interact with both the N protein and PABP from infected cell lysates. We next addressed whether the N protein is able to directly bind to the PABP by performing a pulldown assay in which purified His-tagged PABP (Fig. 6B, left portion) was bound to nickel-nitrilotriacetic acid (Ni-NTA) beads and mixed with purified untagged N protein (Fig. 6B, right portion). As shown in Fig. 6C, untagged N protein (left portion, lane 1) was copelleted by His-tagged PABP, suggesting that the N protein can physically bind to the PABP. Finally, we assessed whether the N protein is able to interact with the PABP from infected cell lysates by incubating Ni-NTA beads with His-tagged N protein and infected cell lysates and subjecting the eluate to immunoblotting with an antibody against PABP. As shown in Fig. 6D, a signal at ~70 kDa representing the cellular PABP was observed (left portion, lane 1), suggesting that the N protein is able to interact with cellular PABP from infected cell lysates. Consistently, the reciprocal pulldown assay with His-tagged PABP demonstrated that the PABP can interact with the N protein from infected cell lysates (Fig. 6D, right portion, lane 1). To determine whether RNA bridging is essential for such interaction, RNase treatment was included in the pulldown assay. As shown in Fig. 6D, both the PABP (left portion, lane 6) and N protein (right portion, lane 5) were detected, suggesting that RNA bridging is not a requirement for the interaction between the two proteins. It was concluded that the poly(A) tail can interact with both the N protein and PABP from infected cell lysates. In addition, the N protein is able to physically bind to the PABP in vitro and to interact with PABP from infected cell lysates.

Open in a separate window

FIG 6

Interactions between the poly(A) tail, N protein, and PABP. (A) Interactions of the poly(A) tail with N protein and/or PABP in mock-infected or infected cell lysates. Proteins from mock-infected or infected cell lysates interacting with a biotinylated poly(A) tail were pulled down by streptavidin, followed by immunoblotting using antibodies against PABP and N protein. The upper and lower asterisks in lanes 1, 9 (from streptavidin pulldown samples of infected cell lysates), and 3 (from infected cell lysates) indicate cellular PABP and coronaviral N protein, respectively; the asterisk in lane 4 (from mock-infected cell lysates) indicates cellular PABP; the asterisks in lanes 5 and 6 indicate E. coli-expressed N protein and PABP, respectively; the asterisk in lane 7 indicates untagged N protein; and the asterisk in lane 10 (from streptavidin pulldown samples of mock-infected cell lysates) indicates cellular PABP. (B) His-tagged PABP (left) and untagged N protein (right) were expressed in E. coli, analyzed by SDS-PAGE, and stained with Coomassie blue. (C) Pulldown assay to determine direct binding between PABP and N protein. His-tagged PABP was bound to Ni-NTA beads and mixed with untagged N protein. The pulldown materials were detected by immunoblotting using an antibody against N protein (left) or PABP (right). (D) Pulldown assay using His-tagged N protein (left) or His-tagged PABP (right) to assess its interaction with PABP or N protein, respectively, in infected cell lysates. Bound proteins from lysates were analyzed by immunoblotting with an antibody against PABP (left) or N protein (right). The arrow indicates the position of PABP (left) and N protein (right). PD, pulldown.

Interactions of the poly(A) tail and N protein with translation factor eIF4G and coronavirus replicase protein nsp9.

To further examine the role of the poly(A) tail and N protein in gene regulation, we first determined whether the poly(A) tail is able to interact with eIF4G and nsp9, a coronavirus replicase protein that is associated with polymerase nsp12 (20), is essential for replication (21), and is involved in the initiation of negative-strand RNA synthesis (22). To this end, the 84-nt biotinylated RNA described above [consisting of 19 non-poly(A) nucleotides followed by the 65-nt poly(A) tail] was used in the streptavidin pulldown assay. As shown in Fig. 7A, both eIF4G (left portion, lane 1) and coronavirus nsp9 (right portion, lane 1) were detected by immunoblotting, though they were not observed with the biotinylated RNA containing only the 19 non-poly(A) tail nucleotides (data not shown), suggesting that the 65-nt poly(A) tail is able to interact with eIF4G and coronavirus nsp9 from infected cell lysates. We next performed the pulldown assay with Ni-NTA beads to determine whether the N protein is able to interact with eIF4G in infected cells. In this case, eIF4G was not detected when using infected cell lysates treated with RNase or left untreated (Fig. 7B, lane 1 or 7, respectively) but was detected when using mock-infected cell lysates treated with RNase or left untreated (Fig. 7B, lane 2 or 8, respectively). To address whether the lack of eIF4G detection was due to the His-tagged N protein being outcompeted by endogenous N in infected cell lysates, a pulldown assay with protein G beads followed by incubation with an antibody against N protein was employed. Indeed, eIF4G was detected in the absence or presence of RNase (Fig. 7C, lane 1 or 6, respectively), suggesting that the N protein can interact with eIF4G from infected cell lysates without the assistance of RNA. Our results show that the N protein is able to bind to the poly(A) tail, yet it is possible that at least a portion of the detected coronavirus nsp9 in Fig. 7A is due to its interaction with the N protein. Another pulldown assay was performed to investigate this possibility. As shown in Fig. 7D, lane 2 and lane 7, nsp9 was detected in infected cell lysates in the absence or presence of RNase treatment, respectively, suggesting that the N protein is able to interact with nsp9 without an RNA bridge. These data suggest that both the poly(A) tail and N protein are able to interact with the translation factor eIF4G and replicase protein nsp9.

Open in a separate window

FIG 7

Interactions of poly(A) tail and N protein with cellular eIF4G and coronavirus nsp9. (A) The poly(A) tail interacts with cellular eIF4G and coronavirus nsp9. Infected cell lysates were incubated with the biotinylated poly(A) tail and pulled down by streptavidin, followed by immunoblotting with antibodies against eIF4G (left) and nsp9 (right). Coronavirus nsp9 (12 kDa) and cellular eIF4G (220 kDa), indicated by an arrow in lane 1 in the left and right blots, respectively, were identified. (B) Ni-NTA bead pulldown assay using the His-tagged N protein followed by immunoblotting with an antibody against eIF4G to determine interaction between N protein and eIF4G. The arrow indicates the position of 220-kDa eIF4G. (C) Protein G bead pulldown assay followed by immunoblotting with an antibody against eIF4G to determine interaction between N protein and eIF4G. The arrow indicates the position of 220-kDa eIF4G. (D) Ni-NTA bead pulldown assay using the His-tagged N protein followed by immunoblotting with an antibody against nsp9 to determine interaction between N protein and nsp9. The arrow indicates the position of 12-kDa nsp9.

Construction of plasmids and DNA templates for RNA probes.

The DNA templates 55 nt + 65A, 55 nt, 65A, 55 nt + 45A, 55 nt + 25A, 25A, and BCoV-65nts for synthesis of RNA probes were produced by PCR. The template for 65A and 25A (65 and 25 adenosine residues) were generated by PCR using a primer containing 65 and 25 thymidine nucleotides, respectively, and a primer containing T7 promoter sequence plus 3 guanosine residues. Therefore, except for 3 guanosine residues, there are no extra non-adenosine residues in both RNA probes after in vitro transcription. To synthesize a DNA template containing the 65-nt poly(A) tail and 19-nt non-poly(A) tail, a primer with a sequence of 5′-TGTAATACGACTCACTATAGGGCCAATTGAAGAAT-3′ and a primer with a sequence of 5′-T(65)GTGATTCTTCAATTGG-3′ were used for PCR. Constructs actin-65nts for the RNA probe and His-tagged β-actin for in vitro translation were amplified by RT-PCR using RNA extracted from HRT-18 cells. To construct DI-EGFP, the EGFP gene was inserted into BCoV DI RNA at the site between ORF 1a and the N protein gene. For this, a DNA fragment containing the EGFP sequence and HpaI and XbaI restriction enzyme sites was amplified by an overlap PCR mutagenesis procedure, digested with HpaI and XbaI, and ligated into HpaI- and XbaI-linearized pDrepI to create pDI-EGFP. pDI-EGFP contained full-length EGFP and N protein genes.

Expression of recombinant proteins.

For His-tagged N protein, pET32aN, which contains the BCoV N protein gene, was transformed into Escherichia coli BL21(DE3) cells, followed by inoculation into LB medium. The cells were then induced with isopropyl thio-β-d-galactoside, harvested by centrifugation, resuspended in phosphate-buffered saline (PBS), and then sonicated. The supernatant containing the recombinant protein was purified through the 6×His tag by immobilized-metal ion affinity chromatography with EDTA-resistant Ni Sepharose excel resin (GE Healthcare) and loaded on a nickel-chelating column (GE Healthcare). Fractions containing N protein were dialyzed and collected. Because the expressed BCoV N protein also contains His, Trx, and S tag-coding sequences, the resulting molecular weight is estimated to be ~65 kDa. To obtain N protein without the His tag, the tag along with Trx and S tags was removed using PreScission protease (GE Healthcare). To purify His-tagged PABP, pET28aPABP, which contains the PABP gene (GenBank accession no. NM_002568), was transformed into E. coli BL21(DE3) pLysS cells and the following procedures were similar to those for expression of N protein described above.

EMSA and K_d.

An in vitro transcription reaction for synthesizing ³²P-labeled RNA for electrophoretic mobility shift assay (EMSA) was carried out using T7 RNA polymerase and [α-³²P]ATP as specified by the manufacturer (Promega). To purify ³²P-labeled RNA, the synthesized ³²P-labeled RNA was separated on 6% sequencing gels, and passive elution was performed, followed by phenol-chloroform extraction. The ³²P-labeled RNA and N protein were added to a binding reaction mixture containing 20 mM HEPES (pH 7.5), 6 mM MgCl₂, 1.5 μM EGTA, 22.5 mM NaCl, 330 mM KCl, 36% glycerol, 3.6 mM dithiothreitol (DTT), 82.5 μg/ml of bovine serum albumin (BSA), and 36% glycerol and incubated for 15 min at 37°C with 1 U/ml of RNasin (Promega) (final concentrations for ³²P-labeled RNA and N protein were 1 nM and 5 nM, respectively). Reactions with unlabeled competitor at 1-, 10-, and 100-fold excesses and nonspecific yeast tRNA (0.1 mg/ml) were also performed in parallel. The RNA-protein complexes were resolved on a native polyacrylamide gel in TBE buffer (50 mM Tris, 45 mM boric acid, 0.5 mM EDTA) at a constant voltage at room temperature, dried, and analyzed by autoradiography. To determine the binding affinity, a fixed concentration of 0.2 nM ³²P-labeled RNA was titrated with protein (0, 14, 71, 143, 286, and 533 nM), and the bound RNA-protein complexes were separated from unbound RNA using an 8% polyacrylamide gel. Free and bound RNAs were quantitated and fit to the Hill equation: RNA bound = b × [P]n/(K_dⁿ +[P]ⁿ), where b is the upper binding limit, [P] is the protein concentration, n is the Hill coefficient, and K_d is the dissociation constant. GraphPad Prism was used. K_d was calculated based on at least three independent experiments.

UV cross-linking of RNA to N protein.

HEK-293T cells were mock infected or infected with BCoV. After 16 h of infection, the ³²P-labeled 65-nt poly(A) tail was transfected using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's instructions, and at 2 h posttransfection, cells were washed with PBS. Cells were subjected to irradiation on ice for 5 min at 254 nm with ~4,000 μW/cm² using a Spectrolinker (XL-1000; Spectrolinker). Cell lysates were collected and treated with RNase mix containig10 mM Tris (pH 7.5), 400 U/ml of micrococcal nuclease, 1 mM CaCl₂, 1% aprotinin, 2 mg/ml of leupeptin-pepstatin, 100 mM phenylmethylsulfonyl fluoride (PMSF), and 0.1 mg/ml of RNase A and RNase T1) at 37°C for 30 min. RNase-treated samples were centrifuged, and supernatants were collected and precleared for 1 h at 4°C by incubation with protein G beads (MagQu). The beads were then removed and immunoprecipitated with an antibody against N protein at 4°C overnight, followed by incubation with protein G beads for 4 h at 4°C using tilt rotation. After extensive washing, the RNA-protein complexes were resuspended in SDS-PAGE loading dye, resolved by SDS-PAGE, dried, and visualized by autoradiography.

Immunoprecipitation and pulldown assay.

His-tagged PABP (25 μg) was mixed with N protein in 100 μl of binding buffer containing 50 mM sodium phosphate (pH 7.4), 300 mM NaCl, and 0.02% Tween 20, and Qbeads-NTA-Ni (MagQu) were added. The mixture was incubated with tilt rotation for 30 min at room temperature. The beads were washed 3 times with 1 ml of binding buffer containing 50 mM sodium phosphate (pH 7.4), 300 mM NaCl, and 0.02% Tween 20. Proteins bound to the beads were eluted in SDS sample buffer, resolved by SDS-PAGE, and analyzed by Western blotting. The same method was employed to analyze proteins from cell lysates interacting with His-tagged N protein or His-tagged PABP in the presence or absence of RNase mix. Immunoprecipitation assay with N antibody bound to the protein G-coated magnetic beads followed by incubation with infected cell lysates was performed according to the manufacturer's instructions (MagQu). Proteins bound to the beads were analyzed by immunoblotting with antibody against eIF4G.

Biotinylated-RNA pulldown assays.

To synthesize RNA labeled with biotin, the DNA template containing the 65-nt poly(A) tail and 19 non-poly(A) tail nucleotides or only 19 non-poly(A) tail nucleotides was used for in vitro transcription with T7 polymerase (Promega) in the presence of a biotin-UTP labeling nucleoside triphosphate (NTP) mixture (Roche), as recommended by the manufacturer. After purification, biotinylated RNA was incubated with cell lysates in Tris-EDTA (TE) buffer. After incubation at room temperature for 30 min, a streptavidin suspension (MagQu) was added to the mixture and incubated for 30 min at room temperature, followed by three washes with binding buffer. The protein-associated beads were boiled with SDS-PAGE loading buffer for 5 min and analyzed by immunoblotting.

In vitro and in vivo translation assays.

Capped transcripts for in vitro translation were prepared using a T7 mMessage mMachine kit (Ambion) according to the manufacturer's protocol. For the in vitro translation assay, 1 μg of capped transcript was added to a mixture containing 17.5 μl of rabbit reticulocyte lysate (RRL; Promega), 20 U of RNasin RNase inhibitor (Promega), 1 μl of amino acid mixture minus methionine, and 20 μCi of [³⁵S]methionine. After incubation at 30°C for 1 h, the samples were resolved by 10% SDS-PAGE. The gel was then dried and exposed to Kodak XAR-5 film. The films were scanned and quantified with ImageJ software (NIH, Bethesda, MD). For a loading control, 1 μg of each capped transcript was resolved on a formaldehyde-agarose gel and stained with ethidium bromide, followed by band density quantitation using ImageJ software (NIH). For the effect of N protein on translation of DI-EGFP in vivo, HEK-293T cells were independently transfected with 3 μg of N protein or the His-tagged β-actin transcript using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's instruction. After 8 h of transfection, HEK-293T cells were transfected with 3 μg of DI-EGFP. Cell lysates were collected after 3, 8, and 16 h, and equivalent amounts of cell lysates were analyzed by immunoblotting. The amounts of translated products were normalized with loading control GAPDH and the amounts of DI-EGFP RNA quantified by RT-qPCR. For the effect of N protein on coronavirus translation in vivo, HEK-293T cells were independently transfected with 3 μg of N protein or the His-tagged β-actin transcript. After 8 h of transfection, HEK-293T cells were infected with BCoV. Cell lysates were collected at the time of postinfection as indicated in Fig. 4F, and equivalent amounts of cell lysates were analyzed by immunoblotting. The amounts of translated products were also normalized with loading control GAPDH and the amounts of BCoV genomic RNA quantified by RT-qPCR. For the effect of N protein on host protein synthesis, HEK-293T cells were mock transfected or independently transfected with N protein or the His–β-actin transcript. After 1 h, HEK-293T cells were incubated in medium in the presence or absence of actinomycin D (Thermo Fisher Scientific) for 8 h and incubated with methionine-free medium for 30 min followed by 20 μCi of [³⁵S]methionine for 1 h. The cells were then collected and equivalent amounts of cell lysates were analyzed by SDS-PAGE. The gel was exposed to X-ray film or Coomassie blue stained, followed by quantification with ImageJ software (NIH). The amounts of [³⁵S]methionine-labeled host proteins were then normalized with the amounts of Coomassie blue-stained proteins, and GAPDH mRNA was quantified by RT-qPCR.

We thank Wei-Li Hsu at National Chung Hsing University, Taiwan, for assistance with the RNA-protein interaction experiment and Chih-Jung Kuo at National Chung Hsing University, Taiwan, for help with protein purification. We also dedicate this work to David A. Brian (1941-2014), University of Tennessee, and thank him for many helpful discussions in the initial stage of the work.

This work was supported by grants MOST 105-2320-B-005-004 and MOST 106-2313-B-005-046-MY3 from the Ministry of Science and Technology, Taiwan, Republic of China.