A Heterogeneous Nuclear Ribonucleoprotein A/B-Related Protein Binds to Single-Stranded DNA near the 5′ End or within the Genome of Feline Parvovirus and Can Modify Virus Replication

Abstract

Phage display of cDNA clones prepared from feline cells was used to identify host cell proteins that bound to DNA-containing feline panleukopenia virus (FPV) capsids but not to empty capsids. One gene found in several clones encoded a heterogeneous nuclear ribonucleoprotein (hnRNP)-related protein (DBP40) that was very similar in sequence to the A/B-type hnRNP proteins. DBP40 bound specifically to oligonucleotides representing a sequence near the 5′ end of the genome which is exposed on the outside of the full capsid but did not bind most other terminal sequences. Adding purified DBP40 to an in vitro fill-in reaction using viral DNA as a template inhibited the production of the second strand after nucleotide (nt) 289 but prior to nt 469. DBP40 bound to various regions of the viral genome, including a region between nt 295 and 330 of the viral genome which has been associated with transcriptional attenuation of the parvovirus minute virus of mice, which is mediated by a stem-loop structure of the DNA and cellular proteins. Overexpression of the protein in feline cells from a plasmid vector made them largely resistant to FPV infection. Mutagenesis of the protein binding site within the 5′ end viral genome did not affect replication of the virus.

Animal viruses have many specific interactions with their host cells which allow them to infect the cells, replicate their genomes, express their genes, and assemble new infectious particles. Autonomous parvoviruses have genomes of about ∼5,000 nucleotides (nt) of single-stranded (ssDNA) DNA which encode two genes and as few as four proteins, and they depend on the host cell for most of their replicative and metabolic additional functions (15, 16), while the adeno-associated viruses (AAV) of the genus Dependovirus also require functions supplied by helper viruses, including adenoviruses or herpesviruses (5). Successful parvovirus replication depends on the mitotic and transformed state of the cell, as well as on its specific differentiated phenotype and host of origin (13, 14).

Several different cellular proteins which interact with the genome of the autonomous parvoviruses or with nonstructural protein 1 (NS1) have been defined; they include the parvovirus initiation factors, which bind within the 3′-terminal palindrome of minute virus of mice (MVM) (12); HMG and SGT, which interact with NS1 (18, 19); and others that are not yet identified associated with various regions of the viral genome (49, 50). Members of the 14-3-3 family of proteins interact specifically with NS2 (8).

In the case of AAV, the associations of cellular proteins or proteins from the helper adenovirus or herpesviruses may occur through binding the larger nonstructural proteins Rep68 and Rep78 or directly with the viral DNA. Cellular proteins involved in DNA replication and gene expression that bind the Rep proteins include the TATA binding protein and the transcription coactivator PC4 (26, 55). Another protein involved in DNA replication and controlling the fill-in of the incoming viral DNA strand is a ∼53,000-Da protein termed the D-box binding protein which binds a sequence between nt 125 and 145 in the AAV genome, in a reaction that is regulated by tyrosine phosphorylation of the protein (41). A heterogeneous nuclear ribonucleoprotein (hnRNP)-related protein has been identified as binding to the NS1 protein of MVM (25). The hnRNP family includes many different members that bind DNA or RNA, and they may control gene expression, DNA metabolism, pre-mRNA processing, or nucleocytoplasmic transport (20, 24). More than 20 different hnRNPs have been identified in vertebrate cells; they are designated A through U, according to their positions in two-dimensional protein gel electrophoresis (20).

Few cellular proteins or other ligands that bind the parvovirus capsids in either their assembled or unassembled forms have been defined. Canine parvovirus (CPV) binds to a number of cellular proteins in virus-overlay protein blots (2, 3). Several parvoviruses, including the feline panleukopenia virus (FPV) and CPV, bind efficiently to carbohydrates including sialic acids on a number of glycoproteins (2, 51). AAV type 2 binds the heparan sulfate proteoglycan, and it also uses the α_Vβ₅ integrin and/or the human fibroblast growth factor receptor 1 as coreceptors for infection (40, 47, 48).

CPV and FPV are closely related parvoviruses that differ in host range, antigenic structure, and a number of other biological properties (37, 52). Both viruses efficiently infect cat cells, but only CPV infects dog cells. The viral determinants of host range include a small number of sequence differences in the coat protein gene, which result in differences in the structure of the surface of the capsid. Infection by FPV in the nonpermissive dog cells is blocked at an early stage of the replication cycle, between virus uptake from the cell surface and the amplification of the viral DNA within the nucleus (11, 27).

We used cDNA expression library screening to identify an hnRNP-related protein that binds viral DNA sequences near the 5′ end of the viral genome outside the capsid. Although that particular sequence near the 5′ end of the genome did not appear essential for virus replication, the protein also bound sequences in other regions of the genome, and overexpression of the protein in cells blocked successful FPV replication.

MATERIALS AND METHODS

cDNA library preparation and screening.

mRNA was isolated from the feline CRFK cell line, and poly(dT) priming was used to prepare cDNA. The cDNA was ligated to adapter sequences so that it could be cloned in a directional manner into the T7Select system (Novagen, Madison, Wis.). The products of cDNA sequences fused in frame with the 3′ end of the T7 phage coat protein gene are therefore displayed as fusion proteins on the surface of the phage capsid. The phage were grown and titrated in the capsid protein-complementing host BLT5615, and the phage library containing 10⁷ independent clones was amplified once. Full (DNA-containing) FPV capsids purified by repeated sucrose gradient banding at 5 μg/ml were bound to polystyrene 96-well plates, which were then incubated with 0.5% bovine serum albumin (BSA) in phosphate-buffered saline (PBS). A sample of 10⁹ phage were incubated with the virus in PBS with 2 mM CaCl₂, 1 mM MgSO₄, and 0.1 to 0.5% Triton X-100. After washing in the same buffer, the bound phage were grown in BLT5615 cells and then repanned sequentially four times on full FPV capsids. Inserts in individual phage were sequenced by using a T7Select capsid protein gene primer and automated DNA sequencing, and the sequences were used to search the GenBank database by using the BLAST 2.0 search algorithm (1).

Recombinant phage prepared from bacterial lysates were incubated with full or empty FPV capsids in 96-well trays; after washing, the bound phage were detected with an anti-T7 capsid antibody. In some cases the capsids were pretreated for 30 min with micrococcal nuclease (100 U/ml) or DNase I (100 U/ml) prior to use in enzyme-linked immunosorbent assay (ELISA).

Protein expression.

Several phage that bound the FPV capsids encoded the same DNA binding protein, and one clone (referred to here as DBP40) appeared to be missing only six amino acids from its amino terminus by comparison with closely related proteins in the database. That gene was recovered and expressed in Escherichia coli from the pET-28b (+) expression vector (Novagen) with both the polyhistidine tag and T7 epitope tag fused to its amino terminus. The expressed protein was purified by binding to a nickel column, eluted with 2 mM Tris-HCl (pH 7.9)–0.5 mM NaCl–1 M imidazole (53), and then dialyzed against 10 mM Tris-HCl (pH 7.9)–100 mM NaCl.

ssDNA binding and in vitro fill-in assays.

For electrophoretic mobility shift assays (EMSA), synthetic oligonucleotides representing sequences near the 3′ and 5′ ends of the viral genome (Table 1) were 5′-end labeled with polynucleotide kinase and [γ-³²P]ATP. The labeled oligonucleotides were incubated with 20 ng of the purified DBP40 in 20 mM HEPES-NaOH (pH 7.9)–100 mM NaCl–5 mM MgCl₂–1 mM dithiothreitol–10% glycerol–1 μg of poly(dI-dC)–1 μg of BSA and then electrophoresed in 6% nondenaturing polyacrylamide gels, which were dried and exposed to X-ray film (45). Competition EMSAs were performed as described above, but with 10- or 100-fold excess of unlabeled oligonucleotides added.

TABLE 1.

Oligonucleotide probes used in this study^a.

Oligo-nucleotide designation	Nucleotide sequence
Probe 1	TCTATAAGGTGAACTAACCTTACCATAAGT
Probe 2	ACTTATGGTAAGGTTAGTTCACCTTATAGA
Probe 3	CGTGAACTTGGTCAGTTGGTTCTAAAGAATGAT
Probe 4	CGCCCAAGTTTAAACACAACACAAACCGCCTAT
Probe 5	CTTTTAGAACCAACTGACCAAGTTC
CArG box	GATCCCATTTCCTAATTAGGTAAAAG
VLDL D box	GGGTTCGGATTTGGTAATGGAAGGTCCCC
(TAAGG)4	TAAGGGTAAGGGTAAGGGTAAGGG
(TAAGA)4	TAAGAGTAAGAGTAAGAGTAAGAG
Mutant 1	CGTTCTATTCAGTGAACTAACCTTACCATAAG
Probe 7	TATACTGAGGAAGTTATGGAGGGAGTAAATTGGTT
Probe 8	AACCAATTTACTCCCTCCATAACTTCCTCAGTATA

^aLocations of probes 1 to 5 in the FPV genome are shown in Fig. 3A. Mutant 1 was used to alter the predicted DBP40 binding sequence of FPV between nt 4892 and 4929 in the genome; probes 7 and 8 are shown in Fig. 6C. CArG box and VLDL D box represent sequences from the CBF-A and ssDBF binding sites, respectively. Likely binding motifs are underlined. 

To map other DBP40 binding sites in the genome, immune coprecipitation of FPV DNA fragments with the protein was performed. An FPV infectious clone in plasmid pGEM3Z (Promega, Madison, Wis.) was digested with DdeI and HinfI, and the 3′ ends were ³²P labeled with the Klenow fragment of E. coli DNA polymerase I (Gibco/BRL, Gaithersburg, Md.) and [α-³²P]dATP. After boiling for 10 min, the DNA was transferred to a dry ice-ethanol bath and incubated for 30 min with 0.1 μg of E. coli-expressed DBP40 at 30°C for 1 h; then the protein and any associated DNA were immunoprecipitated with antibody against the T7 epitope (Novagen). After incubation with 10 μg of proteinase K for 30 min, the DNA was electrophoresed in 2% alkaline agarose gels, which were then dried and exposed to X-ray films.

We examined the effect of DBP40 on the availability of FPV DNA as a template in a DNA fill-in assay. Viral ssDNA was prepared from full capsids by boiling for 10 min, then extracted with phenol and chloroform, and ethanol precipitated. The ssDNA was incubated with or without DBP40 for 30 min at 30°C in EMSA buffer with BSA as the carrier nonspecific protein. Two units of the Klenow fragment (Gibco/BRL), 50 μM each dATP, dGTP, dCTP, and dTTP, and 5 μCi of [α-³²P]dCTP (4 Ci/mmol) were added, and the reaction mixture was incubated for a further 20 min at 37°C; 20 mM EDTA and 1% sodium dodecyl sulfate were added to stop the reaction. After phenol-chloroform extraction, the DNA was ethanol precipitated and then digested with SnaI or BseRI, which cut nt 289 and 469 from the genomic 3′ end, respectively. The DNA was electrophoresed in neutral polyacrylamide gels, which were then dried and exposed to X-ray film or to phosphorimager plates (Fuji Medical Systems, Stamford, Conn.).

DNase I footprint assays were performed according to standard procedures (7). ssDNA fragments of the negative- and positive-strand orientations generated by asymmetric PCR using primers flanking the region from nt 268 to 507 were gel purified. After 5′-end labeling with [γ-³²P]dATP, 20,000 cpm of each probe was preincubated for 30 min with 20, 50, or 100 ng of DBP40; then 0.22 U of DNase I (Promega) was added. After incubation at room temperature for 1 min, 200 mM NaCl, 30 mM EDTA, 1% sodium dodecyl sulfate, and 100 μg of yeast tRNA per ml were added; then the mixture phenol-chloroform extracted and the DNA was ethanol precipitated. Samples were loaded on sequencing gels along with G+A ladders prepared according to the Maxam-Gilbert sequencing protocol (36).

Effect of protein expression on viral replication.

The DBP40 gene fused to the T7 epitope tag was cloned into the mammalian expression vector pcDNA3.1 (−) (Invitrogen, Carlsbad, Calif.), and that plasmid was used to transfect CRFK cells by electroporation. The cells were stained for the T7 epitope with a mouse monoclonal antibody and also with rabbit antibody against the capsid protein and then with specific anti-mouse or anti-rabbit immunoglobulin G (IgG) labeled with fluorescein isothiocyanate (FITC) or rhodamine isothiocyanate.

CRFK cells transfected with the DBP40 expression plasmid were incubated for 2 days and then inoculated with FPV. The cells were suspended by trypsinization 24 h later, fixed with 2.5% paraformaldehyde, and then incubated with 0.1% Triton X-100 and 0.5% BSA in PBS. The cells were incubated with rabbit serum against the FPV capsid protein and a mouse monoclonal antibody against T7 epitope for 1 h at 37°C and then washed twice with PBS. The cells were subsequently stained with FITC-conjugated goat anti-rabbit IgG and phycoerythrin-conjugated goat anti-mouse IgG and analyzed by flow cytometry.

For the cell cycle analysis, DBP40-transfected cells were fixed and permeabilized in methanol and acetone (1:1) and then stained with a monoclonal antibody against the T7 epitope for 1 h at 37°C followed by FITC-conjugated goat anti-mouse IgG. After secondary antibody staining, the cells were treated with 1 mg of RNase per ml at 37°C for 30 min, and then the DNA was stained with 0.01 mg of propidium iodide per ml–0.02% Triton X-100. The DNA content of the cells was examined with a FACScan flow cytometer (Becton Dickinson, San Jose, Calif.), and cell cycle parameters were obtained by using the ModFit LT program (Verity Software House Inc., Topsham, Maine).

Mutagenesis of the DBP40 5′-end binding site.

The infectious plasmid clone of FPV b strain was mutagenized to alter the predicted DBP40 binding site. The GeneEditor site-directed mutagenesis system (Promega) was used to replace the TCTATAAGGTGAACT sequence on the inboard side of the 5′-terminal palindrome with TCTATTCAGTGAACT, so that the sequence within both arms of the palindrome would be altered after virus replication (Table 1; Fig. 3D). The mutated sequence was recloned into the FPV infectious clone, and the region replaced was sequenced. The mutant plasmid was transfected into CRFK cells, and infectious virus yields were determined by 50% tissue culture infective dose in CRFK cells (42).

FIG. 3.

EMSA to test the binding of DBP40 to ³²P-labeled oligonucleotides (Table 1). (A) Positions of the oligonucleotide probes in the FPV genome. (B) EMSA of ³²P-labeled probes with 50 ng of DBP40. The labeled probes were incubated with the protein for 30 min and then electrophoresed in a 6% native acrylamide gel. (C) Competition EMSA between ³²P-labeled probe 1 and 10- or 100-fold excess of various oligonucleotides. (D) Position of the mutation in the inboard end of the FPV 5′-terminal palindrome. After replication of the DNA in cells, synthesis of the outboard arm of the palindrome from the inboard sequence in the plasmid results in the mutation in both arms.

Nucleotide sequence accession number.

The DBP40 sequence has been deposited in GenBank (accession no. AF153444).

RESULTS

hnRNP-related DNA binding protein interacts with FPV full viral particles.

After five rounds of panning of the recombinant library on full FPV capsids, 24 phage DNAs were prepared, sequenced, and compared with sequences in GenBank. Fifteen clones encoded proteins with high homology to A/B-type hnRNP proteins, and the genes were fused in frame with the T7 capsid gene, although with five different fusion sites. The phage bound to the viral DNA and not the capsid, as shown by the finding that they bound strongly to full capsid preparations but weakly to empty capsids, and the binding was reduced >90% by pretreatment of the capsids with micrococcal nuclease or DNase I (Fig. 1). The same phage bound in a similar manner to CPV full capsids (data not shown).

FIG. 1.

ELISA optical density at 450 nm (OD₄₅₀) of T7 phage 40 screened from the library by panning on full FPV capsids. The phage supernatant was incubated in wells coated with full FPV capsids, empty FPV capsids, or capsids pretreated with DNase I, nuclease A, or micrococcal nuclease. The bound phage were detected with a horseradish peroxidase-conjugated antibody against the T7 capsid protein.

The longest cDNA clone isolated was completely sequenced that clone showed high homology to members of the hnRNP A/B subfamily proteins; it was most similar to the CArG binding factor A (CBF-A) and the chicken apolipoprotein D-box binding factor (ssDBF) (Fig. 2) (30, 46).

FIG. 2.

Alignments of the translated product of the largest clone of the DBP40 gene with mouse CBF-A and chicken apoVLDL II promoter single-stranded D-box binding factor (ssDBF), the most similar sequences in the GenBank database. The shaded sequences indicate the residues which differ between DBP40 and either of the other two proteins.

Recombinant DBP40 binds specifically to the 5′ end of the virus genome.

The full capsid of FPV is very stable, and the viral DNA is packaged within the capsid with about 24 nt of the 5′ end of the genome outside of the capsid, most likely with NS1 bound to the 5′ end when it is first encapsidated (17). EMSA with purified recombinant DBP40 showed that it bound specifically to the single-stranded 5′-end sequence of the negative strand and to its complementary sequence, but no binding was seen to oligonucleotides representing 3′-end sequences or the genomic region corresponding to the D-box sequence of AAV (which is distinct from apoVLDL gene promoter D box) (Table 1) (39, 41). Several members of hnRNP A/B proteins have been identified as binding specifically to the telomeric DNA sequence TTAGGG and the complementary motif CCCTAA (21, 45). Comparing the sequence of the 5′-end upper probe with other identified specific binding sequences of hnRNP proteins showed a common motif of TAAGG, which is similar to the protected TTAGG sequence seen in the CArG box, TTTGG in the apoVLDL single-stranded D box, and TTAGG in the rat heptocyte telomeric sequence (21, 30, 46). Binding to the 5′-terminal probe was also competed by CArG box and apoVLDL gene promoter D-box oligonucleotides, by (TAAGGG)₄ tetramers, and to a much lesser degree by the mutated (TAAGAG)₄ tetramer oligonucleotide (Fig. 3). DBP40 appeared to have a similar affinity for the CArG box oligonucleotide compared with the 5′-end sequence, but competition was considerably lower with the TAAGAG tetramer, indicating that DBP40 has specificity for the TAAGG motif (Fig. 3).

A number of heat-denatured restriction fragments from the FPV infectious plasmid were immunoprecipitated when denatured DNA was incubated with purified DBP40. The regions of the genome precipitated most efficiently under these conditions included the 5′ end of the genome, nt 795 to 2015 and 2189 to 3249 (Fig. 4). The latter fragment contained three TAAGG sequence motifs, two of which were within the splice donor sequences of the viral gene transcripts.

FIG. 4.

Protein binding to various sequences from the FPV genome identified by immunoprecipitation of the protein-DNA complex. (A) The DdeI and HinfI restriction maps of the FPV genome in plasmid pGEM3Z. (B) The complete FPV genome in pGEM3Z was digested with DdeI or HinfI and the fragments were ³²P labeled. After denaturation, the DNA was incubated with 100 ng of purified DBP40 and then immunoprecipitated with an antibody against the T7 epitope fused to the N terminus of the protein. After incubation with proteinase K, the DNA was electrophoresed in 2% alkaline agarose gels, which were dried and exposed to X-ray film. Differing buffer conditions resulted in slightly slower migration of the digested DNA. Total DNA, original digestion; IP, the immunoprecipitated DNA.

DBP40 inhibited virus genome fill-in in vitro.

DBP40 added to an in vitro fill-in reaction of FPV DNA inhibited the production of second-strand DNA by about 40% (Fig. 5A). When the DNA was digested with SnaI (cleaving nt 289 from the 3′ end), incorporation into that fragment was seen to be reduced by ∼50%, while formation of double-stranded DNA (dsDNA) past the BseRI site (nt 469) fragment production was reduced by >90% (Fig. 5B), indicating that extension by polymerase was greatly attenuated between nt 289 and 469 in the presence of the added protein (Fig. 5).

FIG. 5.

Effect of DBP40 on the in vitro DNA filled-in of FPV ssDNA by the Klenow fragment of DNA polymerase I. Viral DNA recovered from purified virions was incubated with the polymerase in the presence of deoxynucleoside triphosphates, [³²P]dCTP, and either 0 or 50 ng of DBP40. (A) The product generated was electrophoresed in a 1% agarose gel, with the number of disintegrations per minute (in phosphorimager units) in the total DNA product shown. (B) dsDNA produced was digested with SnaI (nt 289) or BseRI (nt 469) and electrophoresed in a 5% nondenaturing acrylamide gel, which was exposed to X-ray film. Incorporation into the lower band is shown below each lane; size markers are indicated in base pairs. (C) Positions of SnaI and BseRI sites relative to the 3′-end palindrome of the FPV ssDNA genome.

To define the nucleotides that interact with DBP40 in this region, we next performed the DNase I protection studies. Since some of the hnRNP proteins, such as CBF-A (30) and qTBP42 (45), bind specifically to both 5′-TTAGG-3′ and its complementary sequence 5′-CCTAA-3′, and CBF-A protected both strands in the methylation interference studies, we generated both negative (viral ssDNA, complementary to mRNA) and positive ssDNA representing the sequences between nt 268 and 507 by asymmetric PCR and examined them for protection from DNase I cleavage by DBP40. The sequence on the negative strand between nt 296 and 330 was protected. Protection of positive strand in the region between nt 296 and 330 was hard to assess, as that region contained only a few sites susceptible to DNase I under the conditions of the assay (Fig. 6A). To further investigate the presence and specificity of binding, oligonucleotides representing each strand in the region of DNA protection were synthesized and labeled, and both bound DBP40 protein (Fig. 6C). Together, the footprinting and EMSA data suggested that DBP40 interacts with both positive and negative strands between nt 296 and 330. The positive-strand region contained a TATGG motif which was located within the sequence AGTTATGGAG and that differed in only one nucleotide from the AAV D-box sequence AGTGATGGAG (Fig. 6B).

FIG. 6.

Protein binding and protection of sequences within the FPV genome. (A) DNase I protection of the negative and positive strands of the region from nt 290 to 370 of the FPV genome by increasing amounts of DBP40, in the region where the block to DNA polymerization appeared to occur. Numbers represent nucleotides in the complete FPV genome. (B) Sequence of the FPV genome between nt 250 and 340, showing the approximate region protected by the bound DBP40. (C) EMSA showing binding of 20 ng of DBP40 to ³²P-labeled oligonucleotides containing the sequences from the negative (−ve) and positive (+ve) strands of the viral DNA in the region protected and competition with 10- and 100-fold excess of the same unlabeled oligonucleotide.

The sequence of the DBP40 binding region had a high likelihood of making a stem-loop structure in its single-stranded form, and the DNase I-protected sequences were within the predicted loop structures (Fig. 7B). That region has been associated with attenuation of transcription of the MVM genome (31, 32), and so the protected sequence of DBP40 was compared with the homologous sequences from MVM and LuIII virus and shown to differ in much of the sequence in that region (Fig. 7A).

FIG. 7.

(A) Alignment of FPV, MVM, and LuIII virus sequences in two regions of the genome identified as containing binding sequences for FPV in these studies. The sequence in FPV that appears to be the consensus binding sequence of the NS1 is underlined. The negative DNA strand sequences are shown in each case. Numbering is from the complete FPV genome sequence, starting at the left-hand end. (B) Predicted folding of the negative strand of the FPV DNA in the region of the genome between nt 260 and 360, as predicted by the mfold program (44). Arrows indicate the DNase I-sensitive sites. Numbering is from the complete FPV genome sequence, starting at the left-hand end.

DBP40 overexpression blocked virus replication in cells.

The T7-tagged DBP40 recombinant protein was primarily found in the nucleus when expressed in CRFK cells, although it was also found in the cytoplasm of some cells (Fig. 8A). When transfected cells were inoculated with FPV and examined for the coincidence of recombinant DBP40 expression and CPV infection, few cells expressing DBP40 were infected (Fig. 8A). Flow cytometry analysis showed that the nontransfected cells were infected with an efficiency of 12.5%, while only 1% of the DBP40-expressing cells became infected, indicating a >90% reduction in efficiency of infection by FPV after DBP40 expression (Fig. 8C). The experiments were repeated four times, with transfection efficiencies ranging from 5.75 to 8.46% and infection efficiencies ranging from 12.1% to 28.54%; all showed greater than 85% reduction of susceptibility in transfected cells. No difference in the cell cycle distribution was seen between the transfected and nontransfected cells (results not shown).

FIG. 8.

Infection of feline cells expressing DBP⋕40. The cells were stained for the T7 epitope fused to the N-terminus of the DBP⋕40 and for FPV capsid protein antigens. (A) Two different fields showing CRFK cells transfected with a plasmid expressing DBP⋕40, and then inoculated with FPV. The cells were then stained for T7-epitope on the DBP⋕40 and for the FPV capsid protein. Rhodamine (red) stained cells are FPV infected cells, and fluorescein (green) stained cells show T7 epitope fused to DBP⋕40. Yellow stain indicates overlapping cells, but none of the transfected cells were found to be infected by the fluorescent microscopy. (B) Flow cytometric analysis of the cells without DBP⋕40 transfection. The T7 epitope tagged with DBP⋕40 is shown in the vertical direction, and FPV antigen is shown in the horizontal direction, and DBP⋕40 expressing FPV-infected cells would be in the upper right field. The percentage of cells in each quadrant is shown. (C) Flow cytometric analysis of the cells with DBP⋕40 transfection, infected and analyzed as in (B).

Mutation of the DBP40 binding site near the 5′ end of FPV did not affect replication.

The sequence of the inboard 5′-terminal palindrome of the FPV infectious clone was replaced with the equivalent sequence from MVM, resulting in a substitution of the complementary outboard sequence in the palindrome after viral DNA replication (Fig. 3D). The mutant virus was readily recovered and replicated to the same titers in CRFK cells as wild-type FPV (results not shown).

DISCUSSION

We show that an hnRNP A/B protein binds to several positions in the genome of FPV and CPV, that the protein can attenuate DNA polymerase fill-in of the viral DNA in vitro, and that overexpression of the protein in cells significantly reduces the infection or replication of the virus. The initial binding site identified on full capsids could be removed by treatment with nucleases, indicating that it was likely to be part of the 5′ end of the genome extended to the outside the viral capsid (17). We were not able to identify any function for that binding, and a three-base alteration of that binding sequence in an infectious FPV clone did not reduce the infectivity of the virus or the viral yields obtained in tissue culture. However, the binding site was very close to the 5′ nick site of the genome predicted by comparison with the well-defined sequences within the MVM 5′-terminal palindrome (Fig. 7A), suggesting that there may still be some association between binding of the protein and replication at the 5′ end (9, 16, 18, 43, 50).

Expression of DBP40 in cells from a plasmid vector resulted in the cells becoming largely resistant to infection or replication (Fig. 8). This effect was due to some direct or indirect effect of DBP40 on the virus but was not due to arrest of the cell cycle, as the DBP40-expressing cells showed the same proportion of cells in S phase as the wild-type cells (results not shown). It is possible that DBP40 binds directly to the viral genomic DNA during replication, and in vitro the protein was seen to bind to at least three regions within the genome, either by coimmunoprecipitation of DBP40 DNA fragments along with the protein (Fig. 4) or by blocking the fill-in of the viral genomic DNA by the Klenow fragment of DNA polymerase I (Fig. 5). The block to fill-in appeared to occur at least in part between the SnaI site at nt 289 and the BseRI site at nt 469. Surveying the positive and negative strand sequences within that region for protein binding using DNase I footprinting showed protection of a 20- to 30-base region around nt 300 and 315 on the negative strand (Fig. 6). That region had been previously shown in studies of MVM to be responsible for attenuation of transcription of mRNA synthesis from the P4 promoter in a reaction that involved the binding of unidentified cellular proteins (4, 32). That region of the MVM genome was predicted to form an extended stem-loop structure in the ssDNA, and this was also seen for FPV (Fig. 7) (4). Some features of that predicted folded DNA structure were confirmed by the DNase I digestion profile, as most of the predicted stem structure between nt 295 and 330 was not digested by the DNase I under the conditions used (which were optimized for digestion of the ssDNA), and cleavage occurred mostly at predicted mismatches around nt 315, 316, and 300 (Fig. 7B).

DBP40 protein belongs to the hnRNP A/B subfamily, which have two RNA or DNA binding domains in the central part of the protein and an auxiliary glycine-rich region with one RGG box in the C terminus (10, 20). These two nucleic acid binding domains are very conserved, while the N- and C-terminal domains are more diverged. A variety of hnRNP A/B proteins that associate with pre-mRNA, forming the hnRNP complex have been found, and many family members also bind to ssDNA or dsDNA in sequence-specific manner (20, 21, 24, 46). Several of the proteins, including the CBF-A and the VLDL D-box binding factor, bind to DNA and regulate gene expression (30, 46), while others bind to and regulate the synthesis of telomeric regions of the genome (21, 33, 45). There are also similarities between some hnRNP proteins and the adenovirus 72-kDa DNA binding protein, which has an essential role in replication of AAV (29, 35, 54). The most closely related proteins were the CBF-A and the chicken VLDL D-box binding factor (Fig. 2), and the binding motif of DBP40 that we identified was TAAGG, similar to that of the CArG box (TTAGG) and the VLDL D box (TATGG), which both bound well in EMSA, although DBP40 clearly bound less efficiently to the sequence TAAGA (Fig. 3C).

Although no essential role(s) for this protein in enhancing or regulating viral replication has been defined, some possibilities are suggested by the data obtained on viral DNA binding and on the effect on replication and from the properties of similar proteins in this family. The protein may bind the viral genome and be involved in regulation of transcription or replication. The effect on replication or fill-in of the viral DNA may be similar to that seen for a cellular protein which binds the AAV genome just inside the 3′ hairpin and blocks viral DNA fill-in during infection (22, 23, 41), where binding is regulated by tyrosine phosphorylation of the protein (34, 39). Some hnRNP proteins are also phosphorylated on tyrosine by tyrosine kinases, or on serine or threonine by protein kinase C or casein kinase, and those modifications can regulate the binding or other functions of the proteins (6, 28, 38). We have not determined whether this has any effect of DBP40. The association of the protein with the 5′ end of the encapsidated genome may indicate a role in DNA packaging or in the location or transport of the full capsid within the cell. DBP40 may also play a role in the replication or resolution of the genomic 5′ end. Similar to the case for MVM, it is likely that FPV NS1 binds to the ACCA sites and introduces a nick at specific site in the genome adjacent to the DBP40 binding site, but we have not yet tested whether it can bind that sequence in the double-stranded form or whether there is any direct effect on that process.