Featured Post

Avira Antivirus Pro - Review 2020 - PCMag India

Image
Avira Antivirus Pro - Review 2020 - PCMag IndiaAvira Antivirus Pro - Review 2020 - PCMag IndiaPosted: 11 Jun 2020 12:00 AM PDTEvery computer needs antivirus protection, and one way companies can support that aim is to provide free antivirus to the masses. But these companies can't survive unless some users shell out their hard-earned cash for paid antivirus utilities. Piling on pro-only tools and components is one way companies encourage upgrading to a paid antivirus. Avira Antivirus Pro adds several components not available to users of Avira Free Security, but they don't really add much value. The biggest reason to pay for it is if you want to use Avira in a commercial setting, which isn't allowed with the free version.Avira's pricing is undeniably on the high side, with a list price of $59.88 per year for one license, $71.88 for three, and $95.88 for five. Admittedly, it seems to be perpetually on sale; just now, the one-license price is discounted to $44.99. That…

Minute Virus of Canines NP1 Protein Governs the Expression of a Subset of Essential Nonstructural Proteins via Its Role in RNA Processing - Journal of Virology

Minute Virus of Canines NP1 Protein Governs the Expression of a Subset of Essential Nonstructural Proteins via Its Role in RNA Processing - Journal of Virology


Minute Virus of Canines NP1 Protein Governs the Expression of a Subset of Essential Nonstructural Proteins via Its Role in RNA Processing - Journal of Virology

Posted: 29 Mar 2017 12:00 AM PDT

ABSTRACT

Parvoviruses use a variety of means to control the expression of their compact genomes. The bocaparvovirus minute virus of canines (MVC) encodes a small, genus-specific protein, NP1, which governs access to the viral capsid gene via its role in alternative polyadenylation and alternative splicing of the single MVC pre-mRNA. In addition to NP1, MVC encodes five additional nonstructural proteins (NS) that share an initiation codon at the left end of the genome and which are individually encoded by alternative multiply spliced mRNAs. We found that three of these proteins were encoded by mRNAs that excise the NP1-regulated MVC intron immediately upstream of the internal polyadenylation site, (pA)p, and that generation of these proteins was thus regulated by NP1. Splicing of their progenitor mRNAs joined the amino termini of these proteins to the NP1 open reading frame, and splice site mutations that prevented their expression inhibited virus replication in a host cell-dependent manner. Thus, in addition to controlling capsid gene access, NP1 also controls the expression of three of the five identified NS proteins via its role in governing MVC pre-mRNA splicing.

IMPORTANCE The Parvovirinae are small nonenveloped icosahedral viruses that are important pathogens in many animal species, including humans. Minute virus of canine (MVC) is an autonomous parvovirus in the genus Bocaparvovirus. It has a single promoter that generates a single pre-mRNA. NP1, a small genus-specific MVC protein, participates in the processing of this pre-mRNA and so controls capsid gene access via its role in alternative internal polyadenylation and splicing. We show that NP1 also controls the expression of three of the five identified NS proteins via its role in governing MVC pre-mRNA splicing. These NS proteins together are required for virus replication in a host cell-dependent manner.

INTRODUCTION

Parvoviruses use multiple mechanisms to maximize the coding potential of their compact genomes, including alternative transcription initiation, alternative splicing, alternative polyadenylation, and alternative translation initiation mechanisms (13).

Infection with minute virus of canines (MVC), a member of the genus Bocaparvovirus (4), can cause abortion and stillbirth in pregnant dogs, as well as mild gastroenteritis and respiratory disease in puppies (58). MVC generates a single pre-mRNA from a promoter at the left-hand end of the genome (P6) that is processed via alternative splicing and alternative polyadenylation into multiple nonstructural and capsid-encoding transcripts (9, 10). As with other parvoviruses, an open reading frame (ORF) in the left half of the genome encodes nonstructural (NS) proteins, while an ORF in the right half encodes the capsid proteins VP1 and VP2 (2). The bocaparvoviruses also encode a genus-specific protein, NP1, from a small ORF spanning the center of the genome (9, 11). As seen for other parvoviruses, the NS proteins of MVC were predicted to play multifunctional roles during infection; inspection of the MVC NS ORF identifies a DNA binding/endonuclease domain within their shared amino terminus, a helicase domain in the relative center of the ORF, and a zinc finger motif at its carboxyl terminus. MVC NS proteins have been shown to help initiate and sustain the replication of the viral genome, are required for virus packaging, and mediate a number of important virus-host cell interactions (1218). NP1 is a distinct Bocaparvovirus nonstructural protein that modulates RNA processing via the suppression of the internal polyadenylation signal (pA)p located in the middle of the genome. It is also required for the splicing of the 3D/3A intron that lies immediately upstream of (pA)p (19). Both of these processes are necessary to gain proper access to the capsid gene ORF (19, 20).

We have found that MVC generates a greater diversity of NS proteins than has been previously identified; alternative splicing of RNAs from the MVC NS region led to the generation of five NS proteins. Three of these proteins were generated from spliced mRNAs that fused their carboxyl termini in frame with the carboxyl terminus of NP1. The mRNAs encoding these proteins utilize the 3D/3A intron and, consistent with our previous characterization of the role of NP1 in splicing this intron, we found that generation of these mRNAs was dependent upon NP1. All NS isoforms share the same initiating AUG and the origin binding/endonuclease domain that is conserved among Parvoviridae NS1 and Rep proteins. Splice site mutations that prevented expression of all three of the proteins derived from RNAs using the 3D/3A intron inhibited virus replication in a host cell-dependent manner. Thus, in addition to its role in accessing the viral capsid genes, NP1 is also necessary for efficient expression of a subset of essential NS proteins by virtue of its role in RNA processing.

RESULTS

MVC encodes a larger spectrum of nonstructural proteins than previously determined.In previous studies, we identified two proteins encoded by the left-end nonstructural gene of MVC, in addition to the centrally encoded NP1 protein (20). These NS proteins migrated at ∼84 and ∼66 kDa, respectively. We demonstrated that the ∼66 kDa protein was encoded by a doubly spliced RNA, R3, as diagrammed in Fig. 1C. While the derivation of the ∼84-kDa protein was not determined, we speculated (in retrospect, incorrectly), based on the predicted molecular mass, that it was encoded by the uninterrupted ORF that spanned the entire NS region present within the unspliced RNA that had been described (Fig. 1C, R1).

FIG 1

MVC encodes multiple NS isoforms during infection of WRD cells and transfection of 293T cells. (A) (Left) Immunoblots, using antibodies directed against the NS ORF (epitope depicted as a star in panel C) or NP1 (epitope depicted as a circle in panel C), of cell lysates of either mock-infected cells (M; lane 1) or WRD cells infected with MVC at a multiplicity of infection (MOI) of 7 harvested 48 h postinfection (lane 2). (Right) Immunoblots, using antibodies directed against the NS ORF (star), NP1 (circle), or tubulin, of 293T cell lysates harvested 48 h posttransfection with either pIMVC WT (lane 4) or pIMVC 427 TAA (terminating the NS ORF as described in the text) (lane 5) or mock transfection (lane 3). The bands correlating with NS isoforms (NS-100, NS-84, NS-66, and NS-50) and NP1 are indicated on the right, and molecular mass markers are on the left. (B) Reverse transcriptase PCR of total 293T cell RNA extracted 48 h posttransfection with pIMVC WT, primed with oligo(dT), and reverse transcribed with (lanes 1 to 3; each lane represents an individual complete experiment) and without (lane 4) RT. Amplicons (1,683 bp, 1,224 bp, 678 bp, and 464 bp) generated with MVC NS ORF-specific forward (nt 868) and NP1 ORF-specific reverse (nt 3097) primers, depicted by horizontal arrows in the transcription profile schematic in panel C, are shown on the right. The size markers (lane M) are depicted on the left. (C) Transcription profile of MVC showing the alternative RNA-processing events that generate MVC nonstructural protein (NS isoforms and NP1)-encoding transcripts (R1 to R6) and structural protein (VP1 and VP2)-encoding transcripts (R7 and R8). The NS- and NP1-encoding transcripts are illustrated as alternatively polyadenylated transcripts using the proximal (pA)p (R1 to R6s) or distal (pA)d (R1 to R6l) polyadenylation site. The P6 promoter, splice donors (D) and acceptors (A), and the proximal (pA)p and distal (pA)d polyadenylation sites are shown, along with relevant nucleotide landmarks within the MVC genome (GenBank accession number FJ214110.1). The position of the NS1 termination mutant (pIMVC 427 TAA) is shown with an ×, while the positions of the 5′ (nt 868) and 3′ (nt 3097) primers used in the RT-PCR analysis shown in panel B are depicted with horizontal arrows. The NS ORF (nt 403 to 2727) and NP1 ORF (nt 2537 to 3097) are indicated. The epitopes detected by the antibodies used to detect the NS proteins (NS ORF amino acids 687 to 700: PKKQRKTEHKVLID) and NP1 (NP1 amino acids 1 to 13: MSTRHMSKRSKARSR) are indicated by the star and circle, respectively. The triangle indicates the HA epitope at the NP1 ORF carboxyl termini in pIMVC 3097HA WT and pIMVC 3097HA NP1m constructs used for the experiments shown in Fig. 3. The predicted (pred. MW) and observed (obs. MW) molecular weights of the NS isoforms (NS-100, NS-84, NS-50, and NS-40), NP1 proteins, and capsid proteins (VP1 and VP2) are indicated with the corresponding number of amino acid residues (774 aa, 715 aa, 439 aa, 382 aa, 309 aa,186 aa, 703 aa, and 571 aa) encoded by their mRNAs.

A more detailed examination using an antibody specific to an epitope within the carboxyl-terminal region of the MVC NS ORF (nucleotides [nt] 403 to 2727), indicated by the star in Fig. 1C, allowed us to identify 4 proteins that utilize the NS gene region during virus infection (Fig. 1A, lane 2). (A fifth NS protein is discussed below.) In addition to the previously identified abundant ∼84- kDa and ∼66-kDa proteins, this antibody revealed NS proteins of ∼100 kDa and ∼50 kDa that appeared at lower abundance. Expression of these four proteins was also detected following the transfection of the infectious clone of MVC in 293T cells (Fig. 1A, lane 4). Insertion of a terminating TAA codon at nt 427 in the infectious clone of MVC confirmed that, following transfection of 293T cells, all four NS proteins (but not NP1) utilize the previously identified NS gene AUG at nt 403 (Fig. 1A, lane 5).

Three of the newly identified NS proteins are generated from RNAs that excise the 3D/3A intron and join the NS ORF to the NP1 ORF.To identify the coding regions for these newly identified NS proteins we first performed reverse transcriptase (RT) PCR assays to identify spliced RNAs from the NS region that could potentially encode these proteins. Using probes spanning nt 868 to 3097 (diagrammed in Fig. 1C), RNA generated by the infectious clone pIMVC generated distinct RT-PCR species of 1,683 nt, 1,224 nt, 678 nt, and 464 nt (Fig. 1B) following transfection of 293T cells. These products were individually cloned into bacterial plasmids, and sequencing identified them as corresponding to the RNA species R2, R3, R4, and R5, respectively, as diagrammed in Fig. 1C. Of note, RNAs R2, R4, and R5 utilize the viral 3A acceptor and thus putatively fused the carboxyl-terminal region of NP1 in-frame with an upstream NS ORF.

The individual cDNAs representing the various splicing patterns were then cloned in place into the full-length clone of MVC. These constructs were transfected into 293T cells, and extracts were analyzed by immunoblotting using the same antibody to the C-terminal region of NS described above. The 1,224-nt cDNA, representing the R3 transcripts (Fig. 1B), generated a single protein with a mobility of ∼66 kDa, as expected (Fig. 2A, lane 5). Similarly, the 678-nt cDNA, representing the R4 transcripts (Fig. 1B), generated a single protein with a mobility of approximately 50 kDa (Fig. 2A, lane 6). As expected, expression of the 46-nt cDNA, which represents the R5 transcripts, did not generate a protein detected by this antibody (data not shown). This is addressed further below.

FIG 2

MVC NS-84, NS-50, and NS-40 mRNAs are alternatively spliced, using the 3A acceptor, into the NP1 ORF. (A) Lysates of 293T cells taken 48 h following transfection of pIMVC WT (lane 2), pIMVC 1Am (lane 3), R2 cDNA (lane 4), R3 cDNA (lane 5), R4 cDNA (lane 6), and R5 cDNA (lane 7) were analyzed by immunoblotting with antibodies directed against NS (star in Fig. 1C) and tubulin. Molecular mass markers are indicated on the left. (B) Lysates of 293T cells taken 48 h following transfection with pIMVC WT (lane 2) or CMV-3XF constructs expressing R3 cDNA (lanes 3 and 6), R4 cDNA (lanes 4 and 7), and R5 cDNA (lanes 5 and 8) isoforms were analyzed by immunoblotting using either anti-NS antibody (epitope depicted by a star in Fig. 1C) (lanes 1 to 5) or anti-FLAG (lanes 6 to 8). (C) Schematic of wild-type MVC nt 2524 to 2574 and the pIMVC NSNP1fus mutant. The T nucleotide (nt 2536) deleted in the NS1 ORF to generate the NSNP1fus mutant fusing the NS and NP1 ORFs is italicized and boldface. Reading frames are indicated by lines above and below the sequences. (D) Lysates of 293T cells taken 48 h posttransfection with pIMVC WT (lane 2) and the pIMVC NS1NP1fus mutant (lane 3) were immunoblotted using antibodies to the NS ORF (star in Fig. 1C). The bands correlating with each NS1 isoform are indicated on the right.

When we expressed the largest (1,683-nt) cDNA, representative of the R2 transcripts, we were surprised to see the generation of a protein of ∼50 kDa (data not shown). Upon further examination, we found that as expressed, this RNA had been further spliced at both upstream introns, yielding an R4-like transcript. In a separate set of experiments, we fortuitously found that silent mutation of the 1A donor site, constructed so as not to change the NS amino acid sequence, prevented all splicing of the MVC pre-mRNAs. This mutant generated only a single protein with a mobility of approximately 100 kDa (Fig. 2A, lane 3). The reason that the mutant did not undergo additional splicing has not yet been determined. However, the result confirmed that the unspliced R1 transcripts encode a protein of 100 kDa rather than the largest of the proteins we had identified in our previous study (i.e., the 84-kDa species), which had been our previous assumption based solely on the predicted molecular mass. Cloning the 3D/3A cDNA into the 1A mutant background to prevent further splicing (generating, in essence, the R2 transcript diagrammed in Fig. 1C) produced a single protein of 84 kDA (Fig. 2A, lane 4). These results confirmed that the 100-kDa, 84-kDa, 66-kdA, and 50-kDa proteins detected (Fig. 1) were generated by the R1, R2, R3, and R4 classes of transcripts, respectively.

As mentioned, sequencing of the 464-nt cDNA indicated that it lacked the epitope used for detection in the experiments shown in Fig. 1A and 2A. Thus, we cloned the 464-nt cDNA—and, in addition, the 1,224-nt and 678-nt cDNAs (encoding the 66- and 50-kDa proteins, respectively) as controls—into a cytomegalovirus (CMV)-driven expression vector containing a FLAG tag at the amino terminus of the NS open reading frame. As seen previously, immunoblotting of extracts generated following transfection of these constructs using the anti-NS antibody detected only the 66- and 50-kDa proteins (Fig. 2B, lanes 3 to 5), while the anti-FLAG antibody identified the 40-kDa protein as the product of the 464 cDNA (Fig. 2B, lanes 6 to 8), which represented the R5 transcripts diagrammed in Fig. 1C.

To confirm that the 84-kDa, 50-kDa, and 40-kDa NS proteins utilize a different carboxyl-terminal exon than the 66-kDa and 100-kDa proteins, we explored the prediction that deletion of a single nucleotide that fuses the NS ORF to the NP1 ORF (within the 3D/3A intron, nt 2536), would increase the sizes of only the 66-kDa and 100-kDa NS proteins (the mutation is diagrammed in Fig. 2C [cf. Fig. 1C]). The results following transfection of this mutant (NSNP1fus [Fig. 2D, lane 3]) demonstrate this to be the case. The levels of the 84- and 50-kDa proteins generated by the NS/NP1 fusion shown in Fig. 2D, lane 3, are relatively low, likely because as a fusion protein, a significant proportion of the NP1 present is mutated and poorly functional. NP1 is required for splicing of the 3A/3D intron (19), and the potential role for NP1 in this observation can be explained as discussed below.

NP1 governs expression of three NS proteins via its role in splicing the 3D/3A intron.Since NP1 is required for splicing of the 3A/3D intron (19), it seemed likely that NP1 may govern the expression of the 84-kDa, 50-kDa, and 44-kDa NS proteins. To investigate this possibility, we analyzed expression following transfection of 293T cells of the wild-type (WT) and NP1 mutant MVC infectious clones bearing a hemagglutinin (HA) epitope inserted at nt 3097 at the carboxyl terminus of the NP1 ORF (Fig. 1C, triangle). For these experiments, we assayed both RNA production, using RNase protection assays, and protein production, using antibodies directed either to HA or to the previously described epitope at the carboxyl terminus of the NS ORF (Fig. 1C, star). The HA insertion itself had only a slight effect on splicing of the 3D/3A intron in the presence of otherwise wild-type NP1 (Fig. 3A, compare lane 5 to lane 2). Mutation of NP1 [the previously described mutant bearing an insertion of 5 proline residues, which abrogated NP1's roles in 3D/3A splicing and suppression of (pA)p (19)] led to a decrease in splicing of the 3D/3A intron, both in the wild type, as expected, and to a slightly lesser extent in the HA-tagged construct (Fig. 3A, compare lane 6 to lane 3). As can be seen in Fig. 3B, corresponding to the decrease in splicing of the 3A/3D intron seen in the NP1 mutant constructs, the levels of expression of the 84-kDa and 50-kDa proteins were significantly reduced (the 40-kDa protein was poorly expressed in this experiment), as detected by the anti-HA antibody at the carboxyl terminus of the NP1 exon (Fig. 3B, compare lane 3 to lane 2). Reduced expression of the 84-kDa and 50-kDa proteins was also seen in the NP1 mutant transfection when detected by the antibody to the carboxyl terminus of NS (which does not detect the 40-kDa protein) (Fig. 3B, compare lane 6 to lane 5). These results demonstrate that NP1, which is required for the excision of the 3D/3A intron, governs the expression of the 84-kDa and 50-kDa (and, we expect, also the 40-kDa) proteins that are derived from mRNAs from which the 3D/3A intron is excised.

FIG 3

MVC NP1 is required for the expression of NS-84, NS-50, and NS-40. (A) RNase protection of 20 μg of RNA extracted from 293T cells 48 h following transfection with pIMVC WT (lane 2), pIMVC NP1m (lane 3), pIMVC 3097HA WT (lane 5), and pIMVC NP1m 3097HA (lane 6), using the 2A/3D probe that spans nt 2344 to 2550 (shown in Fig. 1C). The sizes of the probe (243 nt) and protected RNAs (206 nt, 164 nt, and 105 nt) are shown on the left. Bands reflecting RNA species unspliced through this region (RT) or spliced at the second intron acceptor but not at the third intron donor (2Aspl/3DUnspl) and RNAs spliced at the second intron acceptor and also the third intron donor (2Aspl/3Dspl) are indicated on the right. (B) Samples taken from 293T cells transfected with pIMVC 3097HA WT (lanes 2 and 4) and pIMVC NP1m 3097HA (lanes 3 and 6) were subjected to immunoblot analysis using antibody directed against HA (lanes 1 to 3) or antibody directed against NS (epitope depicted as a star in Fig. 1C) (lanes 4 to 6). Tubulin was monitored as a loading control. The bands that correspond to each NS isoform are indicated on the right.

The individual NS proteins exhibit a complex localization pattern.Generation of vectors that individually expressed the various NS isoforms allowed a general determination of their localization and perhaps provides insights into their potential functions during the viral life cycle. At 48 h following transient transfection of replication-competent pIMVC in Walter Reed 3873D (WRD) cells, immunostaining, using the anti-NS antibody, whose epitope is represented by the star in Fig. 1C, demonstrated that when expressed together, the 100-kDa, 84-kDa, 66-kDa, and 50-kDa proteins exhibited a diffuse distribution in the nucleoplasm (Fig. 4A row 1). Replication-defective pIMVC 1Am and pIMVC 1AmΔ3DA constructs, which individually encode R1 and R2 transcripts that are translated into NS-100 and NS-84, respectively, exhibited both nuclear and cytoplasmic NS localization. The isoforms have a partial-to-complete diffuse nucleoplasmic distribution with multiple cytoplasmic aggregates (Fig. 4A, rows 2 and 3). NS-100 seemed to form more round cytoplasmic aggregates, while the NS-84 isoforms generated more interlacing filamentous aggregates in the nucleus and cytoplasm. WRD cells transiently expressing NS-66 alone displayed a strong, diffuse nucleoplasm distribution of NS (Fig. 4A, row 4), while cells expressing NS-50 had a diffuse, faint nuclear distribution, apparently excluding nucleoli and multiple concentrated, discrete, punctate spherical bodies within the nucleoplasm (Fig. 4A, row 5). These results indicated that while NS-100, NS-84, NS-66, and NS-50 isoforms were primarily nuclear when generated together from the complete pIMVC WT clone, when generated individually, their distribution and localization were quite different. Whether these differences are due to differential posttranslational modifications of the proteins or interactions between the proteins themselves is currently being investigated.

FIG 4

MVC NS isoforms exhibit differential localization and distinct morphology in transiently transfected WRD cells. (A) WRD cells were transfected with pIMVC WT or pIMVC constructs expressing cDNAs individually encoding each NS1 isoform (pIMVC 1Am-NS-100, pIMVC R2 1AmΔ3DA-NS-84, pIMVC R3-NS-66, and pIMVC R4-NS-50), as indicated on the left, and stained 48 h later with anti-NS antibody (epitope depicted as a star in Fig. 1C). Nuclei were visualized with DAPI staining. Representative images are shown. (B) Immunofluorescence of WRD cells 48 h posttransfection with CMV-3XF pIMVC R5 encoding the NS-40 isoform and CMV-3XF pIMVC R3 encoding the NS-66 isoform, using anti-FLAG antibodies. Nuclei were counterstained with DAPI. Representative images are shown.

As mentioned previously, the NS antibody utilized in the experiments described above does not recognize NS-40. Thus, to determine the localization of individually expressed NS-40, we compared the localization of N-terminally FLAG-tagged NS-66 and NS-40 following immunostaining (Fig. 4B) using the anti-FLAG antibody (also used for the experiment shown in Fig. 2B). FLAG-tagged NS-66 was again seen to be diffusely nuclear (Fig. 4B, row 2), consistent with our observation shown in Fig. 4A, row 4; however, NS-40 was predominantly cytoplasmic (Fig. 4B, row 1). There is a potential bipartite nuclear localization signal (PSRKRTSSDETINSPPKKQRK; nt 2414 to 2478) that is located in the carboxyl ends of the NS-100, NS-84, NS-66, and NS-50 ORFs but absent in NS-40. Thus, the lack of the bipartite nuclear localization signal in NS-40 may account for its predominantly cytoplasmic localization.

The NS proteins whose expression was governed by NP1 were required for MVC replication in canine, but not human, 293T cells.Previous results have shown that mutations that prevent the production of all the MVC NS proteins prevent virus replication following transfection in both canine and 293T cells and that neither the MVC 100-kDa nor the 66-kDa NS protein can sustain MVC replication alone (10, 20). Additional experiments have shown that, as seen for certain other parvoviruses, mutations that prevent MVC VP1 and VP2 production still allow wild-type levels of mRF production (20; O. O. Fasina and D. J. Pintel, unpublished data). To determine the importance in replication of the NP1 ORF containing NS-84, NS-50, and NS-40 proteins, we inserted into the MVC infectious clone a mutation that debilitated both the donor and acceptor of the 3D/3A intron without changing the NS-100 and NS-66 amino acid sequence (3DAm, diagrammed in Fig. 5A). This mutation would also be expected to prevent capsid protein production, which requires mRNAs spliced using the A3 acceptor. As expected, this mutant generated the NS-100 and NS-66 proteins, but not NS-84, NS-50 (Fig. 5B), or NS-40 (data not shown), following transfection of 293T cells. A similar expression profile was seen in WRD cells, which transfect poorly (data not shown). Following transfection, the pIMVC 3D/3A mutant (3DAm) failed to replicate in both canine cell lines tested: WRD (Fig. 5C, compare lanes 3 and 6 to lanes 2 and 5) and Madin-Darby canine kidney (MDCK) (Fig. 5D, compare lanes 3 and 6 to 2 and 5) cells. In contrast, however, the 3DAm mutant replicated to levels similar to those of wild-type pIMVC following transfection of 293T cells (Fig. 5E, compare lanes 4 and 6 to lanes 3 and 5; 293T cells are not susceptible to direct MVC [re]infection). These results indicated that while the NS-84, NS-50, and NS-40 proteins were essential for replication of MVC in canine cells, they were dispensable for replication in 293T cells.

FIG 5

MVC NS-84, NS-50, and NS-40 are required for viral genome replication in a host cell-dependent manner. (A) Schematic representation of the MVC transcription profile as described for Fig. 1C. The locations of the third-intron donor (3D) and acceptor (3A) mutations in pIMVC 3DAm are designated by × and are also underlined for clarity. (B) Lysates taken from the 293T cells 48 h following transfection with pIMVC WT (lane 2) and pIMVC 3DAm (lane 3) were subjected to immunoblotting using antibodies directed against NS (epitope designated by the star in Fig. 1C) and tubulin. The MVC NS isoforms are shown on the right. (C and D) Southern blots of total DNA extracts from canine WRD (C) or MDCK (D) cells taken 72 h posttransfection with pIMVC WT (lanes 2 and 5) or pIMVC 3DAm (lanes 3 and 6), or infected with MVC (VI) at an MOI of 7 (panel C, lanes 1 and 4), or input plasmid (panel D, lanes 1 and 4), transferred from 1% agarose gels. The DNAs shown in lanes 4 to 6 were treated with DpnI to differentiate transfected input plasmid from monomer (mRF) and dimer (dRF) replicative intermediates, indicated on the left. (E) Southern blots of total DNA extracts from human 293T cells taken 48 h posttransfection of pIMVC WT (lanes 3 and 5) or pIMVC 3DAm (lanes 4 and 6). Lysates from MVC-infected WRD cells (lane 1) and input plasmid (lane 2 and 7) were included as controls. Parallel DNA samples were treated with DpnI (lanes 5 to 7) to differentiate transfected input plasmid from monomer (mRF) and dimer (dRF) replicative intermediates, shown on the left.

DISCUSSION

Here, we expand the identification of MCV NS proteins to five and show that three newly identified MVC NS proteins (84 kDa, 50 kDa, and 40 kDa) are generated from mRNAs spliced at the 3D/3A intron. In each case, splicing of 3D/3A fuses the 3′ ends of the NS proteins to the NP1 ORF. NP1, which is required for efficient splicing of the 3D/3A intron, governs the expression of these NS proteins by virtue of its enhancement of 3D/3A excision. Thus, in addition to controlling capsid gene access via its role in alternative polyadenylation of the internal polyadenylation site, (pA)p (20), and splicing of the 3D/3A intron (19), NP1 also controls the expression of three of the five identified NS proteins via its role in splicing. Mutations that prevent expression of these three proteins inhibit virus replication in a host-dependent manner. It is not yet fully clear if there is a difference in NS protein expression from RNAs that utilize (pA)p or read-through to the distal polyadenylation site, (pA)d.

Similarities between the 84-kDa, 50-kDa, and 40-kDa MVC NS proteins and the NS2 proteins of minute virus of mice (MVM) bear mentioning. Similar to the case for MVC, splicing of the RNA generating MVM NS2 joins the amino-terminal MVM NS1 ORF to a separate ORF at the carboxyl end of the NS gene (2124). Additionally, as is the case for the minor MVC NS proteins, MVM NS2 has been shown to be required for replication of MVM in a host-dependent manner. While restricted in murine host cells, MVM NS2 mutants replicate in a wide variety of transformed cells from various species (2528), and perhaps analogously, the MVC mutants that do not produce the 3D/3A spliced NS proteins are restricted for growth in canine cells yet replicate efficiently in human 293T cells. The extent of host range replication for these mutants is currently being investigated. Additionally, the Tattersall laboratory has pointed out that MVM NS2 and human bocavirus (HBoV) NP1 both bear Crm1 binding and 14-3-3 binding sites, and HBoV NP1 can complement certain early functions of MVM NS2 (29). Together, these results suggest an intriguing similarity in the small NS proteins of parvoviruses.

Surprisingly, the MVC NS proteins migrate differently on SDS-PAGE gels than expected based on their molecular masses, as predicted from their amino acid compositions (Fig. 1C). This led us previously to assume that the protein migrating at approximately 84 kDa was generated from the uninterrupted NS ORF (Fig. 1C). It also initially confounded our mapping of the various MVC NS proteins until cDNAs of each were generated. Why the migration of these proteins is so different than their predicted molecular masses is not yet clear. The unexpected migrations observed may certainly be due to posttranslational modifications; the NS proteins generated by other parvovirus genera have been shown to be highly modified by phosphorylation and acetylation (3034).

The conserved N-terminal region of the MVC NS proteins encodes a histidine-hydrophobic-histidine (HUH) endonuclease motif (35, 36), also conserved in other parvoviral nonstructural proteins (13, 36), which is required for site-specific DNA cleavage and ligation during rolling-hairpin replication (12, 37). The active site of the MVC HUH motif is predicted to be H122, L123, and H124, which are present in all the MVC NS isoforms (Fig. 4B), suggesting that all MVC NS isoforms may interact with the viral genome. As in other viral HUH motif-containing proteins (12, 14, 35, 38), there is an SF3 helicase domain with characteristic Walker A (FYGPASTGKTN; amino acids [aa] 423 to 433), B (LICWWEEC; aa 464 to 471), C (KCIMGGTQFRIDRK; aa 482 to 495), and C′ (PQTPLIISTNHNI; aa 503 to 515) motifs (39) downstream of the HUH motif in NS-100 and NS-84 proteins (amino acids 1 to 696 are identical in NS-100 and NS-84); however, it is removed by alternative splicing and absent in the NS-66, NS-50, and NS-40 proteins, suggesting distinct differences in function. Although the NS-100 and NS-84 proteins contain HUH endonuclease and SF3 helicase domains, they have distinct carboxyl-terminal ends that likely contribute to important differences. Furthermore, NS-84, NS-50, and NS-40 share a conserved 19-amino-acid carboxyl-terminal domain with NP1.

In silico analysis also suggests a strong bipartite nuclear localization signal (PSRKRTSSDETINSPPKKQRK; nt 2414 to 2478) that is located in the carboxyl end of all the NS isoforms, except the NS-40 protein. We found that FLAG-tagged NS-66 displays a diffusely nuclear localization, while the NS-40 protein is predominantly cytoplasmic, perhaps suggesting that the lack of the bipartite nuclear localization signal may alter its cellular localization.

It has recently been shown that interactions between intrinsically disordered domains of viral and host proteins can modulate certain host cellular responses to virus infection, such as the DNA damage response (40). In silico analysis showed that MVC NS-84, NS-50, and NS-40 share a highly disordered domain in their carboxyl termini with NP1 (data not shown). It is conceivable that these regions may play important roles in their function. Although the 3D/3A mutant replicates well in 293T cells, virus is not produced because capsid protein production is prevented. Mutants that prevent the production of NS-84, NS-50, and NS-40 yet allow capsid protein production are currently being constructed. These mutants should aid further analysis of the roles of these NS proteins during virus infection.

Parvoviruses exhibit surprisingly wide variability in how their genomes are expressed (4148). The bocaparvoviruses, which have a single promoter, make extensive use of RNA processing to properly express their genetic information and encode a genus-specific protein, NP1, that participates in this processing (19, 20). In addition to enhancing access to the capsid-coding gene, MVC NP1, by virtue of its role in splicing, also controls the expression of three NS proteins that play important roles in virus replication. Thus, MVC NP1 influences viral RNA processing for expression of both its nonstructural and structural genes, adding to the variety of strategies utilized by parvoviruses to maximize their genome coding potentials.

MATERIALS AND METHODS

Cells and viruses, infections, and transfections.All experiments were carried out as specified in WRD canine cells, MDCK cells, or human 293T cells, which were propagated as previously described (17, 49) in Dulbecco's modified Eagle medium (DMEM) with 10% or 5% fetal calf serum. The MVC used in this study was the original strain (GA3), obtained from Colin Parrish at Cornell University. The infectious molecular clone pIMVC, GenBank accession number FJ214110.1, used for transfection and generation of mutants was constructed by J. Qiu and described previously (10). Transfections of WRD, MDCK, and 293T cells were performed with either Lipofectamine Plus (Life Technologies) or LipoD293 transfection reagent (SignaGen Laboratories).

Immunoblot analyses.Cells grown and transfected in 60-mm dishes with various constructs as described in the text were harvested 48 h posttransfection and lysed in either Laemmli buffer with 2% SDS, RIPA buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.5% sodium deoxycholate, and 0.1% SDS), or WL lysis buffer (50 mM Tris, pH 8.0, 400 mM NaCl, 1 mM EDTA, 0.5% NP-40, and 10% glycerol), which contained protease and phosphatase inhibitors. Bradford assay of cell lysates was carried out to determine the total protein concentration in each lysate. SDS-PAGE and Western blotting were performed as previously described (20).

RNase protection assays.Total RNA was isolated using the TRIzol reagent (Invitrogen), and RNase protection assays were performed as previously described (50). The 2A/3D probe (nt 2344 to 2550), which spanned the MVC second acceptor at nt 2386 and the third donor at nt 2490, was used to analyze the splicing across the MVC second and third introns.

Antibodies.Polyclonal antibodies directed against the MVC NS epitope (NS ORF amino acids 687 to 700; PKKQRKTEHKVLID) and the MVC NP1 epitope (NP1 amino acids 1 to 13; MSTRHMSKRSKAR) were utilized for MVC NS protein immunoblotting. Although 4 of 13 amino acids of the NS protein antibody are missing in NS84 and NS50, the antibody detects NS100, NS84, NS66, and NS50 at ratios similar to those of the proteins tagged with FLAG or HA and detected with commercial anti-tag antibodies (Fig. 2B and 3B). Monoclonal antibodies against FLAG, HA, and tubulin epitopes (Sigma, St. Louis, MO) were used for immunoblotting, as previously described (20).

Immunofluorescence analyses.Cells grown in 6-well plates on coverslips and the transfected constructs described in the text were harvested 48 h posttransfection and fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 15 min. The cells were washed twice with PBS and then permeabilized with 0.5% Triton-X in PBS for 10 min. Primary and secondary antibodies were diluted in 3% bovine serum albumin (BSA) in PBS. Cells were incubated with primary antibodies for 1 h, followed by a wash with 3% BSA in PBS. This was followed by incubation with appropriate Alexa-488-tagged secondary antibodies. Nuclei were visualized with DAPI (4′,6-diamidino-2-phenylindole). The slips were mounted on slides in Fluoromount-G (Southern Biotech). Image analysis was carried out with a Leica TCP SP8 MP confocal microscope. All images were taken with a 63× objective.

Southern blotting.WRD and 293T cells were transfected with the pIMVC and pIMVC-derived mutant constructs described in the text. DNA was extracted 48 h posttransfection, and DNA replication was assayed by Southern blotting as described previously (17, 51), using a random-primed radiolabeled NotI-digested genomic fragment of pIMVC as the probe. Loading of samples was standardized using a Nanodrop spectrophotometer.

Plasmid constructs.The generation of the multiple plasmid constructs used in this study is described below.

(i) pIMVC (WT) and pIMVC 3Am.The wild-type infectious clone (WT) (10) and the third-intron acceptor mutant (3Am) (20) were generated as previously described.

(ii) pIMVC 1Am.Mutation of the splice acceptor site of the MVC first intron (nt 2199) was generated as an in-frame G-to-A substitution.

(iii) pIMVC 1AmΔ3DA.The pIMVC 1AmΔ3DA mutant was generated by site-directed mutagenesis by insertion of a gene block synthesized oligonucleotide (IDT) that lacked nt 2490 to 3037 between the EcoRI site (nt 2345) and the PvuII site (nt 3060). This generated a construct with deletion of the MVC third intron in a pIMVC1Am background.

(iv) pIMVC NS-66, pIMVC NS-50, and pIMVC NS-40.The pIMVC NS-66, pIMVC NS-50, and pIMVC NS-40 constructs were generated by cloning the respective cDNAs encoding each NS isoform into pIMVC WT with primers flanking the XbaI (nt 868) and PvuII (nt 3060) sites.

(v) 3XF NS-66, 3XF NS-50, and 3XF NS-40.the cDNAs encoding NS-66, NS-50, and NS-40 were cloned into the CMV 3×FLAG 7.1 expression vector (Sigma, St. Louis, MO).

(vi) pIMVC NSNP1fus.A single-nucleotide deletion at nt 2536 (underlined) fused the NS ORF to the NP1 ORF (TCGGCCGAGCAATATGTC to TCGGCCGAGCAAATG).

(vii) pIMVC NP1m.pIMVC NP1m, a functionally defective mutant of NP1, was generated by introducing five consecutive proline substitutions into the NP1 ORF at nt 2780 (the previously described 5×-pro mutant [20]).

(viii) pIMVC 3DAm.This clone was generated by debilitating the third-intron splice donor at nt 2491 via an in-frame G-to-A substitution in the NS ORF (underlined), in combination with the third-nucleotide substitutions between nt 3026 and 3040 (GTTCTCACACAGGAT to GTGCTGACACAAGAC), which surround the third-intron acceptor.

(ix) pIMVC 3097HA (WT, NP1m).The pIMVC 3097HA (WT, NP1m) mutant, created both in the wild-type and NP1 mutant infectious molecular clone backgrounds, was generated by site-directed mutagenesis, which introduced an HA epitope upstream of the NP1 termination codon.

Comments

Popular Posts

System detected an overrun of a stack-based buffer in this application [FIX] - Windows Report

Valorant anti-cheat lead answers many questions on Reddit - Millenium US