Analysis of cis and trans Requirements for DNA Replication at the Right-End Hairpin of the Human Bocavirus 1 Genome

ABSTRACT

Parvoviruses are single-stranded DNA viruses that use the palindromic structures at the ends of the viral genome for their replication. The mechanism of parvovirus replication has been studied mostly in the dependoparvovirus adeno-associated virus 2 (AAV2) and the protoparvovirus minute virus of mice (MVM). Here, we used human bocavirus 1 (HBoV1) to understand the replication mechanism of bocaparvovirus. HBoV1 is pathogenic to humans, causing acute respiratory tract infections, especially in young children under 2 years old. By using the duplex replicative form of the HBoV1 genome in human embryonic kidney 293 (HEK293) cells, we identified the HBoV1 minimal replication origin at the right-end hairpin (OriR). Mutagenesis analyses confirmed the putative NS1 binding and nicking sites within the OriR. Of note, unlike the large nonstructural protein (Rep78/68 or NS1) of other parvoviruses, HBoV1 NS1 did not specifically bind OriR in vitro, indicating that other viral and cellular components or the oligomerization of NS1 is required for NS1 binding to the OriR. In vivo studies demonstrated that residues responsible for NS1 binding and nicking are within the origin-binding domain. Further analysis identified that the small nonstructural protein NP1 is required for HBoV1 DNA replication at OriR. NP1 and other viral nonstructural proteins (NS1 to NS4) colocalized within the viral DNA replication centers in both OriR-transfected cells and virus-infected cells, highlighting a direct involvement of NP1 in viral DNA replication at OriR. Overall, our study revealed the characteristics of HBoV1 DNA replication at OriR, suggesting novel characteristics of autonomous parvovirus DNA replication.

IMPORTANCE Human bocavirus 1 (HBoV1) causes acute respiratory tract infections in young children. The duplex HBoV1 genome replicates in HEK293 cells and produces progeny virions that are infectious in well-differentiated airway epithelial cells. A recombinant AAV2 vector pseudotyped with an HBoV1 capsid has been developed to efficiently deliver the cystic fibrosis transmembrane conductance regulator gene to human airway epithelia. Here, we identified both cis-acting elements and trans-acting proteins that are required for HBoV1 DNA replication at the right-end hairpin in HEK293 cells. We localized the minimal replication origin, which contains both NS1 nicking and binding sites, to a 46-nucleotide sequence in the right-end hairpin. The identification of these essential elements of HBoV1 DNA replication acting both in cis and in trans will provide guidance to develop antiviral strategies targeting viral DNA replication at the right-end hairpin and to design next-generation recombinant HBoV1 vectors, a promising tool for gene therapy of lung diseases.

INTRODUCTION

Human bocavirus 1 (HBoV1) is a recently identified respiratory virus associated with acute respiratory tract infections in young children (1–9). HBoV1 belongs to the Bocaparvovirus genus within the Parvoviridae family (10, 11). Other species in the Bocaparvovirus genus include minute virus of canines (MVC), bovine parvovirus 1 (BPV1), porcine bocavirus, and gorilla bocavirus (11–15).

A unique feature of bocaparvoviruses that makes them different from other parvoviruses is that they express the small nuclear phosphoprotein NP1 from an open reading frame (ORF) located in the middle of the genome (12, 13, 16). NP1 is a nonstructural protein and is indispensable for viral DNA replication (12, 17). The role of the NP1 proteins in viral DNA replication is conserved within MVC BPV1 and HBoV1, and the HBoV1 and BPV1 NP1 proteins can replace the MVC NP1 to support MVC DNA replication (12). While the mechanism defining how NP1 facilitates bocaparvovirus DNA replication remains largely unknown, it has been revealed that NP1 plays an important role in processing viral precursor mRNA (pre-mRNA) to matured viral mRNA polyadenylated at the distal polyadenylation site and is therefore important for capsid protein expression (18–20). In addition, HBoV1 holds unique features in the genus Bocaparvovirus: it naturally replicates in nondividing/polarized human airway epithelial cells, and replication is dependent on the cellular DNA damage and repair machinery (21). Upon transfection, the duplex replicative-form (RF) genome (RF DNA) of HBoV1 replicates in human embryonic kidney 293 (HEK293) cells and produces infectious progeny virions (17). Aside from NP1, HBoV1 expresses two large nonstructural proteins, NS1 and NS1-70 (an isoform of NS1), and three other small nonstructural proteins, NS2, NS3, and NS4 (22). NS1, NS2, NS3, and NS4 share a C terminus of 144 amino acids (aa). NS2 is indispensable for viral replication during infection of polarized human airway epithelial cells (22), whereas the NS2 to NS4 proteins are dispensable for viral DNA replication in HEK293 cells. NS1-70 is expressed at a very low level in HEK293 cells transfected with the duplex HBoV1 genome compared to the level at which it is expressed during HBoV1 infection of differentiated human airway epithelial cells (22).

Parvovirus replication generates monomer and dimer replicative-form DNA (mRF and dRF DNA, respectively) intermediates via a unidirectional strand-displacement mechanism, in which the ends of progeny genomes are excised by viral replication initiator protein Rep78/68 or NS1 (23–25). For homotelomeric parvoviruses (e.g., adeno-associated virus [AAV]), in which the two genomic termini are inverted in sequence and identical in structure, the replication process is symmetrical. The tip of the T structure formed on the termini is critical for Rep78/68 to nick the replication origin (26). For heterotelomeric autonomous parvoviruses (e.g., minute virus of mice [MVM] and HBoV1), in which the two genomic termini are dissimilar, the 3′ end hairpin (left-end hairpin [LEH]) structure of the negative-sense single-stranded DNA (ssDNA) viral genome, which has a replication origin, is critical to progress the replication intermediates and to produce the ssDNA genome for progeny virion production, and the 5′ end hairpin (right-end hairpin [REH]) structure contains another replication origin that is required to replicate viral RF DNA (23). In MVM, the cruciform structure at REH is required for RF DNA replication (27). As bocaparvovirus packages predominantly negative-sense single strands (12), we also termed the 3′ end of the HBoV1 genome the LEH and the 5′ end the REH, similar to those in MVM. The HBoV1 LEH and REH also have unique sequences and secondary structures.

At either hairpin end of both homotelomeric and heterotelomeric viruses, the replication origin contains Rep78/68 or NS1 binding elements (RBEs and NSBEs, respectively) and a nicking site which is recognized and nicked by Rep78/68 or NS1 (26–29). Replication at the origins of both termini follows a so-called rolling hairpin replication model and undergoes a series of replicative intermediates (30, 31). Rep78/68 or NS1 binds to RBEs/NSBEs in the origin, executes its endonuclease activity, nicks the 5′-3′ single strand, and liberates at the 3′ end a hydroxyl (OH) group that functions as a primer for viral DNA synthesis by cellular DNA polymerases. The binding and nicking properties characterized by in vivo or in vitro studies suggest that the RBE or NSBE is several tetranucleotide repeats which are directly recognized by the origin-binding domain (OBD) of Rep78/68 or NS1. The nicking site is normally 7 to 17 nucleotides (nt) ahead of the RBE or NSBE at the 5′ end. The genome structure of HBoV1 is unique among these heterotelomeric parvoviruses. The LEH forms a rabbit ear structure of 140 nt with mismatched nucleotides (“bubbles”) inside, and the REH consists of a perfect palindromic sequence of 200 nt in length (17). Of note, the REHs of two other bocaparvoviruses, MVC and BPV1, are able to form a cruciform structure (12).

Because of the unique REH structure of the HBoV1 genome, we sought to define the minimal requirements for HBoV1 DNA replication at the REH both in cis and in trans using the duplex HBoV1 genome in HEK293 cells. We identified a 46-nt minimal replication origin at the REH of HBoV1 (OriR), which contains a nicking site and unconventional NSBEs. In addition, we uncovered new properties of nonstructural proteins NS1 and NP1 during viral DNA replication at the OriR.

MATERIALS AND METHODS

Cell culture.

HEK293 cells (CRL-1573; ATCC) were cultured in Dulbecco's modified Eagle medium (DMEM) with 10% fetal calf serum (FCS). Cells were incubated at 37°C with 5% CO₂. HEK293F cells (Life Technologies/Thermo Fisher Scientific Inc., Carlsbad, CA) were cultured in suspension in Freestyle 293 medium (Life Technologies) at 37°C with 8% CO₂.

Primary human airway epithelium (HAE) cultured at the air-liquid interface (ALI) was generated and cultured as previously described (17, 21). HAE-ALI cultures that had a transepithelial electrical resistance (TEER) of >1,000 Ω · cm² were selected for use in this study.

Plasmid DNA construction. (i) pIHBoV1-based constructs.

The parent HBoV1 infectious clone (pIHBoV1) has been described previously (17). A plasmid from which LEH was deleted, pIHBoV1(LEH−), was constructed by deleting nt 1 to 140 from the HBoV1 genome sequence of pIHBoV1. pIHBoV1(NS1−) and pIHBoV1(NP1−), which have early terminated NS1 and NP1 ORFs, respectively, and pIHBoV1(VP1/3−), which has both the early terminated capsid proteins VP1 and VP3 ORFs, were previously described (17). pIHBoV1(REH−), the plasmid with the REH deletion, was constructed by deleting nt 5303 to 5543 from the HBoV1 sequence of pIHBoV1. pIHBoV1(LEH-VP1/3−) was constructed by deleting LEH and the early terminating VP1 and VP3 ORFs. Based on the pIHBoV1(LEH-VP1/3−) sequence, an XbaI site was inserted at nt 5344 to construct REH-truncated mutants pΔREH1 to pΔREH24, which are diagrammed in Fig. 2. pIHBoV1(3′NCR−), a plasmid from which the 3′ noncoding region (3′ NCR) was deleted, was constructed by deleting the HBoV1 sequence from nt 5221 to 5291 in pIHBoV1. pIHBoV1(VP−) was constructed by deleting nt 3380 to 5119 of the HBoV1 sequence in pIHBoV1.

FIG 2.

Identification of the minimal replication origin at the REH (OriR). (A) Diagram of REH deletion mutants. All REH deletion mutants were constructed on the basis of plasmid pIHBoV1(LEH-VP1/3−), from which LEH was deleted and VP1/3 was knocked out. Group I mutants had deletions at the 3′ end from nt 5543 to 5383. Group II mutants shared the 3′ end deletion to nt 5430 and had a further deletion from nt 5354 to nt 5384 at the 5′ end. Group III mutants shared the 5′ end deletion to nt 5357, except for the ΔREH18 mutant, which had a 5′ end at nt 5360. The deletion at the 3′ end started from nt 5410 to nt 5390. (B to D) Southern blot analyses. HEK293 cells were transfected with linearized mutant HBoV1 DNAs, and Hirt DNA samples were extracted at 2 days posttransfection. DpnI-digested Hirt DNA samples were applied for Southern blotting. (E) Quantification. Three experiments were repeated for panel D, and the amount of RF DNA in each lane was quantified. The level of RF DNA from wild-type HBoV1 was set equal to 100% (lane 1) after normalization with the main DpnI-digested bands. N.S., not significant; *, P < 0.05; **, P < 0.01. (F) Minimal replication origin (OriR). The sequence of the identified OriR is shown in duplex DNA. WT, wild type.

(ii) pHBoV1Ori- and pHBoV1Ori-based constructs.

pΔREH21 was renamed pHBoV1Ori. pHBoV1Ori has the minimal replication origin at the REH (OriR) of nt 5357 to 5402.

pHBoV1λ200Ori, pHBoV1λ400Ori, and pHBoV1λ1000Ori were constructed by inserting bacteriophage lambda (λ) DNA sequences of 200 bp (nt 10,030 to 10,230), 400 bp (nt 10,030 to 10,430), and 1,000 bp (nt 10,030 to 11,030), respectively, in front of the OriR in pHBoV1Ori.

pHBoV1Ori(NS1−) and pHBoV1Ori(NP1−) were constructed by early termination of the NS1 and NP1 ORFs, respectively, in pHBoV1Ori using strategies described previously (17).

pNS1 m¹³QOri (Mut Q), pNS1 m⁵⁴PHP⁵⁶Ori (Mut P), pNS1 m¹²³EGL¹²⁵Ori (Mut E), pNS1 m¹¹⁵HCH¹¹⁷Ori (Mut Endo), pNS1 m¹²⁷KR¹²⁸Ori (Mut K), and pNS1 m¹⁹³RR¹⁹⁴Ori (Mut R) were constructed by mutating the indicated amino acids at the indicated positions to an alanine(s) in the context of pHBoV1Ori. The constructs are diagrammed in Fig. 9.

FIG 9.

Characterization of the OBD of NS1 in vivo. (A) Diagram of NS1 mutants. The NS1 OBD is diagrammed to scale with the putative Ori binding site, and endonuclease activity is shown. On the basis of the pHBoV1Ori sequence, OBD mutants that have NS1 amino acid mutations at the indicated positions are shown. (B) Superimposition of the structures of three NS1 OBD mutants with the wild-type NS1 OBD structure. The NS1 OBD structures of Mut K, Mut R, and Mut Endo were predicted on the basis of the wild-type structure by use of the web server I-TASSER (62). The predicted structures of the mutants were individually superimposed with the wild-type NS1 OBD structure using the web server Superpose (version 1.0) (63). The mutated amino acid residues in the putative DNA binding loops K and R and the endonuclease activity core are labeled with green boxes. (C) Southern blot analysis. Linearized HBoV1 DNAs, as indicated, were transfected into HEK293 cells. At 2 days posttransfection, Hirt DNA samples were extracted, digested with DpnI, and analyzed for DNA replication using Southern blotting. The control was 60 ng of linearized HBoV1Ori DNA digested with DpnI.

Plasmids pNSBEm1 to pNSBEm8 and pTRSm1 to pTRSm7 were constructed by introducing various mutations in the putative NSBEs and the nicking site of the OriR in pHBoV1Ori, respectively. These plasmids are diagrammed in Fig. 7A and 8A, respectively, where the mutations are shown.

FIG 7.

Characterization of the putative NS1 binding element in vivo. (A) Diagram of NSBE mutants. The wild-type OriR and mutated sequences of the putative NSBE are shown. The putative NSBE includes (TGT)₄, and mutants with various mutations of the TGT repeats were constructed. Dots represent identical nucleotides. (B) Southern blot analysis. HEK293 cells were transfected with linearized wild-type Ori and NSBE mutants, as indicated. Hirt DNA samples were extracted and applied for Southern blotting after DpnI digestion. (C) Quantification. After normalization to the amount of DpnI-digested bands, the RF DNA bands on the blots were quantified. Levels relative to the amount of RF DNA from HBoV1Ori (which was considered 100%) are shown as averages and standard deviations from three independent experiments.

FIG 8.

Identification of the nicking site within OriR in vivo. (A) Diagram of the nicking site mutants. The dynamic nicking sites identified by rapid amplification of the 3′ end of nicked DNA are indicated by arrows. The sequencing results for the polyadenosine starting sites derived from transfection of the HBoV1 full-length genome (IHBoV1) and HBoV1Ori are shown with green and black arrows, respectively. Mutations at seven of these poly(A) starting sites, as shown, were performed. (B) Southern blot analysis. Linearized HBoV1 DNAs were transfected into HEK293 cells. At 2 days posttransfection, Hirt DNA samples were extracted, DpnI digested, and analyzed by Southern blotting. Sixty nanograms of HBoV1Ori was digested by DpnI as a control.

(iii) HBoV1 NS1- and B19V NS1-expressing plasmids.

HBoV1 and parvovirus B19 (B19V) NS1-coding sequences were optimized for mammalian cell expression at Integrated DNA Technologies, Inc. (IDT; Coralville, IA), tagged with a Flag tag at the C terminus, and cloned in the pCI expression vector (Promega, Madison, WI), resulting in pOptiHBoV1NS1 and pOptiB19VNS1, respectively.

All nucleotide positions of HBoV1 and bacteriophage lambda (λ) DNA refer to those in the sequences with GenBank accession no. JQ923422 and NC_001416, respectively, unless otherwise specified.

In vivo DNA replication analysis.

HEK293 cells were seeded in 6-well plates or 60-mm dishes 1 day before transfection. When the cells reached 70% confluence, they were transfected by use of the LipoD293 reagent (SignaGen, Gaithersburg, MD) or the Lipofectamine and Plus reagent (Life Technologies) following the manufacturers' instructions.

Low-molecular weight DNA (Hirt DNA) was extracted and digested with DpnI, followed by Southern blotting. These steps were performed exactly as previously described (32). After hybridization, the membrane was exposed to a phosphor screen. The screen was then scanned on a phosphor imager (GE Typhoon FLA 9000; Fuji). The developed image was processed and analyzed using ImageQuant TL8.1 software (GE Healthcare, Marlborough, MA).

BrdU incorporation, IF assay, and PLA.

For virus-infected differentiated airway epithelial cells, we treated the infected HAE-ALI with 5 mM EDTA for 5 min and then trypsinized the infected cells off the insert. We resuspended ∼1 × 10⁵ cells in 1 ml of the ALI medium with bromodeoxyuridine (BrdU; Sigma, St. Louis, MO) at 30 μM for 20 min. For transfected HEK293 cells, at 2 days posttransfection, the cells were incubated with BrdU for 30 min by a previously published method (33). Then, the labeled cells were spun onto slides by use of a cytospin apparatus for immunofluorescence (IF) analysis.

IF analysis was performed following a method described previously (33, 34) with antibodies against the proteins or BrdU, as indicated in Fig. 5. Confocal images were taken with an Eclipse C1 Plus confocal microscope (Nikon) controlled by Nikon EZ-C1 software. DAPI (4′,6-diamidino-2-phenylindole) was used to stain the nucleus.

FIG 5.

NS and NP1 colocalized with the BrdU-chased viral DNA replication centers. (A to C) Localization of NS, NP1, and BrdU-incorporated HBoV1 DNA in transfected cells. HEK293 cells were transfected with pBB (vector backbone), pHBoV1Ori, or pHBoV1Ori. At 2 days posttransfection, BrdU was incorporated into the cells before they were harvested, and then immunofluorescence staining was applied. Cells were costained with anti-NS1C with anti-BrdU (A), anti-NP1 with anti-BrdU (B), and anti-NS1C with anti-NP1 (C). (D to F) Localization of NS1, NP1, and BrdU-incorporated HBoV1 DNA during infection. HAE-ALI cultures were infected by HBoV1 at a multiplicity of infection of 10 viral genome copies (vgc) per cell as previously described (22). At 7 days postinfection, BrdU was incorporated into the cells and immunofluorescence analysis was performed. NS, NP1, and BrdU were costained in different combinations. (G) Interaction of NP1 with BrdU-chased viral DNA. Transfected cells, as indicated, were pulse-chased with BrdU and were analyzed by PLA with anti-NP1 and anti-BrdU antibodies. DAPI was used to stain the nucleus. Confocal images were taken under a Nikon confocal microscope. Magnifications, ×100.

A Duolink proximity ligation assay (PLA) kit (Sigma) was used to perform PLA according to the manufacturer's instructions as previously described (21).

Antibody production and antibodies used.

HBoV1 anti-NS1C antibody, which reacts with all four NS proteins (NS1 to NS4) (22), was produced as previously described (16). All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Kansas Medical Center. Rabbit anti-HBoV1 NP1 antibody (35) was a kind gift from Peter Tattersall at Yale University. Rabbit anti-BrdU (catalog no. 600-401-C29; Rockland, Limerick, PA), mouse anti-Flag monoclonal antibody (Sigma, St. Louis, MO), and secondary antibodies for the IF assay (Jackson ImmunoResearch Inc., West Grove, PA) were purchased.

Rapid amplification of the 3′ end of the nicked DNA.

A DNA tailing reaction was performed using terminal transferase (NEB, Ipswich, MA). Basically, the reaction mixture was composed of 5.0 μl of 2.5 mM CoCl₂, 1 μl of Hirt DNA (from ∼500 cells) diluted 10 to 50 times, 0.5 μl of 10 mM dATP, 0.5 μl of terminal transferase (20 units/μl), and deionized H₂O (dH₂O) in a final volume of 50 μl. The reaction mixture was incubated at 37°C for 30 min. Two microliters of the product was used as the template for amplification by PCR using forward primer 5′-CTG TCT AGA ATG ATC AAT GTA TGC CAG-3′ (nt 5121 to nt 5138) and reverse primer 5′-CAC GGA TCC TTT TTT TTT TTT TTT T-3′, where the underlined nucleotides are XbaI and BamHI sites, respectively. The amplified fragments were cloned into pcDNA3 (Life Technologies) through the XbaI and BamHI sites. Twenty positive clones were sequenced.

Protein expression and purification.

One 500-ml suspension culture of HEK293F cells (10⁶ cells/ml) was transfected with pOptiHBoV1NS1 or pOptiB19VNS1, using a TransIT-PRO transfection kit following the manufacturer's instructions. At 3 days posttransfection, the cells were lysed in L buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1 mM dithiothreitol [DTT], 5 mM ATP, 5 mM MgCl₂) supplemented with protease inhibitors (catalog no. S8820; Sigma). Crude lysate was sonicated for 3 min at a frequency of 15 s on and 25 s off with a 70% pulse by a VCX130 sonicator (Sonics & Materials Inc., Newtown, CT). The lysate was then centrifuged at 10,000 rpm for 15 min and filtered through a 0.2-μm-pore-size filter. The filtered lysate was incubated with 1 ml of phosphate-buffered saline (PBS; pH 7.4)-prewashed anti-Flag G1 affinity resin (GenScript, Piscataway, NJ) at 4°C for at least 1 h. Then the beads were washed with 5 times the resin volume of washing buffer (50 mM Tris, pH 7.4, 500 mM NaCl, 0.05% Triton X-100, protease inhibitors) and were eluted with 3× Flag peptide (Apexbio, Houston, TX) at a concentration of 200 μg/ml. Finally, the eluted protein was dialyzed against PBS twice and against binding buffer (B buffer; 20 mM Tris-HCl, pH 8.0, 125 mM NaCl, 10% glycerol, 1% NP-40, 5 mM DTT, protease inhibitors) once and was concentrated 10 times using polyethylene glycol (PEG) 6000. The concentrated protein was quantified, aliquoted, and stored at −80°C.

Gel shift assay.

A gel shift assay was performed essentially following a published method (36). A duplex DNA probe was generated by annealing complementary oligonucleotides at a concentration of 45 μM in an annealing buffer (10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM EDTA) after boiling for 5 min. The double-stranded DNA (dsDNA) probe was cleaned using a G-50 column (GE Healthcare). Duplex DNA probes were HBoV1 OriR (see Fig. 2F), HBoV1 OriR-mut (5′-CGC GAA ACT CTA TAT CTT TTA ATG GCA GAA TTC AGC ACA TGC GCC A-3′), B19V Ori (5′-GCC GCC GGT CGC CGC CGG TAG GCG GGA CTT CCG GTA CA-3′) (37), and B19V Ori-mut (5′-AGC TAT TGG TCG CTA TTG GTA GGC GGG ACT-3′) (37).

One microliter of the duplex DNA probe was labeled with [γ-³²P]ATP using T4 polynucleotide (NEB) following the manufacturer's instructions. The binding reaction mixture consisted of 8 μl of 2.5× B buffer (see above), 2 μl of diluted duplex DNA probe (at 1:10,000) with or without 2 μl of purified protein (300 ng/μl), 2 μl of unlabeled (cold) probe at a concentration (ratio) specified in Fig. 10, and dH₂O to a total volume of 20 μl. Poly(dI-dC) was added to some reaction mixtures at a final concentration of 2 μg/ml. The reaction mixtures were incubated on ice for 20 min, loaded directly into a prerun 5% nondenaturing polyacrylamide gel, and electrophoresed for 45 min at 100 V. Finally, the acrylamide gel was dried with a vacuum at 70°C and exposed to a phosphor screen.

FIG 10.

HBoV1 NS1 alone did not specifically bind the OriR in an in vitro setting. (A) NS1 protein purification. Five microliters (15 pmol) of purified HBoV1 NS1 (lane 1) and B19V NS1 (lane 2) was analyzed by SDS-PAGE, and the gel was stained with Coomassie brilliant blue. Lane 3, 2 μg of bovine serum albumin (BSA); lane 4, a protein ladder. (B) B19V NS1 specifically binds B19V Ori in vitro. [γ-³²P]ATP-labeled B19V Ori (lanes 1 to 7) or Ori-mut (lanes 8 and 9) was incubated with (lanes 2 to 6 and 9) or without (lanes 1 and 8) B19V NS1 in the binding buffer with 2 μg/ml poly(dI-dC). The cold Ori probe at ratios of 20 times (lane 3) and 200 times (lane 4) or the cold Ori-mut probe at levels of 20 times (lane 5) and 200 times (lane 6) was added for competition. Free and shifted probes are indicated. GST protein was added as a negative control (lane 7). (C) HBoV1 NS1 did not bind HBoV1 OriR in vitro. [γ-³²P]ATP-labeled HBoV1 OriR was incubated with (lanes 2 to 8) or without (lane 1) HBoV1 NS1 in the binding buffer with 2 μg/ml poly(dI-dC). Cold OriR and OriR-mut probes at 10 times (lanes 3 and 6), 100 times (lanes 4 and 7), and 1,000 times (lanes 5 and 8) were included for competition. Free and shifted probes are indicated. (D) Nonspecific HBoV1 NS1 binding to OriR. [γ-³²P]ATP-labeled HBoV1 Ori was incubated with (lanes 2 to 8) or without (lane1) HBoV1 NS1 in the absence of poly(dI-dC). Cold OriR and OriR-mut probes at 50 times (lanes 3 and 6), 10 times (lanes 4 and 7), or 4 times (lanes 5 and 8) were included for competition. The shifted probes are indicated by arrows in panels B through D.

NE preparation.

Nuclear extraction was performed following a method described previously (38). HEK293 cells from one dish of 100 mm were transfected with pOptiHBoV1NS1 or pOptiB19VNS1. At 2 days posttransfection, cells were collected, washed with ice-cold PBS, and pelleted. The cell pellet was lysed in 5 volumes of L buffer (10 mM HEPES, pH 7.5, 10 mM KCl, 0.1 mM EDTA, 1 mM DTT, 0.5% NP-40, protease inhibitors). The lysate was vortexed and centrifuged at 500 × g for 5 min at 4°C. The pelleted nuclei were washed 3 times with 1 ml W buffer (10 mM HEPES, pH 7.5, 10 mM KCl, 0.1 mM EDTA, 1 mM DTT, protease inhibitors). The nuclei were resuspended in 0.25 ml of nuclear extract (NE) buffer (20 mM HEPES, pH 7.5, 420 mM NaCl, 1 mM EDTA, 1 mM DTT, 25% [vol/vol] glycerol, protease inhibitors) and were incubated on ice for 30 min. Finally, the nuclear extract was obtained from the supernatant by centrifuging the lysed nuclei at 12,000 × g for 10 min at 4°C, and the NaCl concentration was adjusted to 100 mM using B1 buffer (NE buffer without NaCl).

Pulldown assay.

Biotin-labeled probe was generated by annealing two complementary oligonucleotides, in which one oligonucleotide was biotinylated, at a concentration of 5 μM. The annealed dsDNA probe was desalted using a G-50 column. Streptavidin-conjugated beads (Gold Biotechnology, St. Louis, MO) were prewashed following the manufacturer's instructions. The binding reaction mixture consisted of 200 μl of 2.5× B buffer with poly(dI-dC), 100 μl of nuclear extract (∼5 μg/ml), 1 μl of biotinylated probe with or without unlabeled probe, and dH₂O in a total volume of 0.5 ml. The reaction mixtures were rotated at 4°C overnight and were pelleted by centrifugation at 1,000 × g for 3 min. The pellet was then washed 3 times with cold PBS before addition of loading buffer. The samples were boiled for 5 min and analyzed by SDS-polyacrylamide gel electrophoresis, followed by Western blotting.

Immunoprecipitation (IP) assay.

HEK293 cells cultured in one 60-mm dish were mock transfected or transfected with plasmids. At 2 days posttransfection, the cells were washed and lysed with 300 μl radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, protease inhibitors) with constant agitation for 30 min at 4°C. The cell lysates were centrifuged at 14,000 × g for 3 min at 4°C, and the supernatant was collected. Eighty microliters of the supernatant was reserved as samples for input loading; the rest of the supernatant (∼220 μl) was preincubated with 40 μl of normal rat serum, 30 μl of prewashed protein A/G beads (Thermo Fisher), and 2 μl of Benzonase (Thermo Fisher) for 3 h at 4°C with rotation. After it was precleared, the supernatant was equally divided for incubation with either 2 μl of normal rat IgG (0.4 mg/ml; Santa Cruz, Dallas, TX) or 20 μl of rat antiserum, 30 μl of protein A/G beads, and 300 μl of RIPA buffer at 4°C overnight. Finally, the protein A/G beads were pelleted down and washed 3 times with ice-cold PBS before mixing with loading dye for Western blotting.

RESULTS

Identification of viral elements required in cis or in trans for HBoV1 replication.

To understand the DNA replication mechanism of HBoV1, we first delineated the viral DNA signals and viral proteins that are required for viral DNA replication. As diagrammed in Fig. 1A, the HBoV1 genome contains four noncoding sequences (LEH, P5 promoter region, 5′ NCR, REH) and three ORFs that encode the NS1 to NS4, NP1, and VP1 to VP3 proteins, respectively (18, 22). The 5′ noncoding region (5′ NCR) contains a sequence of ∼200 nt between the polyadenylation signal of VP mRNA and the REH. To dissect the minimal requirements for viral DNA replication, we carried out sequential deletions of the noncoding regions and early termination of the ORFs in the context of pIHBoV1. Southern blotting results showed a representative pattern of HBoV1 DNA replication, including DpnI digestion-resistant bands: viral single-stranded DNA (ssDNA; genome size), viral RF DNA and double-RF (dRF) DNA, and plasmid-replicated DNA (Fig. 1B, lane 1). The results suggested that without the LEH sequence, pIHBoV1(LEH−) still replicated but failed to excise the HBoV1 RF DNA out of the plasmid (Fig. 1B, lane 2). VP1/3-knockout plasmid pIHBoV1(VP1/3−) replicated better than the parent, pIHBoV1 (Fig. 1B, lane 5 versus lane 1). As noted above, ssDNA was produced only from the transfections of pIHBoV1 and pIHBoV1(VP1/3−). However, pIHBoV1(NS1−), pIHBoV1(3′NCR−), and pIHBoV1(REH−) did not replicate (Fig. 1B, lanes 3, 6, and 7, respectively) and pIHBoV1(NP1−) replicated very poorly (Fig. 1B, lane 4), suggesting that the sequence at the 3′ end of the RF DNA, including REH and the 3′ NCR, and expression of the NS1 and NP1 proteins are apparently critical to HBoV1 DNA replication. pIHBoV1(VP−), from which the VP-encoding ORFs were largely deleted, replicated as well as pIHBoV1 (Fig. 1C, lane 2).

FIG 1.

Identification of the cis-acting sequences and trans-acting proteins required for HBoV1 replication. (A) Diagram of HBoV1 genome and mutants. The HBoV1 genome is shown as negative-sense ssDNA. Sequentially, from left to right, the functional elements are the left-end hairpin (LEH; 3′ end), P5 promoter, NS-coding region, NP1-coding region, VP-coding region, 3′ noncoding region (3′ NCR), and right-end hairpin (REH; 5′ end). The wild-type infectious clone is shown as duplex-form DNA (dsDNA). The noncoding sequences LEH, 3′ NCR, and REH were deleted as described in Materials and Methods. The ORF of NS, NP1, or VP was terminated as described in Materials and Methods. Dotted lines, sequence deletion; vertical bars, nucleotide mutations. (B and C) Southern blot analysis. The wild-type infectious clone (pIHBoV1) or the indicated mutants were transfected into HEK293 cells. At 2 days posttransfection, low-molecular-weight (Hirt) DNA was extracted and applied for Southern blotting after DpnI digestion. One hundred twenty nanograms of pIHBoV1 digested with DpnI under the same conditions used for the Hirt DNA samples was used as a DpnI digestion control (DpnI-Dig. Ctrl). RF, replicative-form DNA intermediate; dRF, double-RF DNA intermediate; ssDNA, single-stranded DNA. Arrow in panel B and asterisk in panel C, DNA replicated from the plasmid containing the entire HBoV1 DNA sequence. Lanes M, size markers.

Taken together, our results suggest that the REH and 3′ NCR are largely responsible for HBoV1 DNA replication as cis-acting elements, while NS1 and NP1 serve trans-acting functions (22). In the following experiments, we used both LEH deletion and the VP1/3-knockout HBoV1 plasmid, pIHBoV1(LEH-VP1/3−), for further characterization of HBoV1 DNA replication at REH.

Identification of an HBoV1 right-end minimal replication origin (OriR).

For DNA replication of either dependoparvovirus or autonomous parvovirus, RBEs or NSBEs and a nicking site at the end palindromic hairpin sequence are requisite (26–29). These two cis signals are necessary for Rep78/68 or NS1 to recognize the replication origin, to perform strand-specific nicking, and to initiate DNA replication. To identify the minimal replication-requisite sequence on REH, we constructed a serial of truncation mutants of the REH on the base of pIHBoV1(LEH-VP1/3−) (Fig. 2A). We used linearized HBoV1 DNA for in vivo DNA replication analysis, in order to avoid circular plasmid DNA replication like that seen from pIHBoV1(LEH−) (Fig. 1B, lane 2). The first group of truncations contained a progressive deletion from the 3′ end of the REH (Fig. 2A, group I). The results showed that the level of viral DNA replication decreased as 3′ end nucleotides of the REH were removed (Fig. 2B). As there was a clear decrease in the level of viral DNA replication from ΔREH2 to ΔREH3 HBoV1 DNA (Fig. 2B, lane 3 versus lane 4), we fixed the right end at nt 5430 of the ΔREH2 HBoV1 DNA and started progressive truncations from the 5′ end (nt 5344) (Fig. 2A, group II). The results of viral DNA replication analysis showed a large decrease in RF DNA from the ΔREH10 to ΔREH11 mutants (Fig. 2C, lane 3 versus lane 4). Thus, we determined the 5′ end of the OriR to be at nt 5357, the 5′ end of the ΔREH10 mutant.

To define the 3′ end of the OriR more carefully, we performed progressive deletions from the 3′ end of the ΔREH10 mutant (nt 5357). DNA replication analysis of the ΔREH17 and ΔREH19 to ΔREH24 mutants showed that further deletions of the 3′ end after nt 5402 (ΔREH21; Fig. 2D, lane 6) significantly decreased the level of RF DNA (Fig. 2E). This result defined the 3′ end of OriR to be at nt 5402. In addition, we created the ΔREH18 mutant, which consisted of the ΔREH17 mutant from which 3 nucleotides from the 5′ end were deleted, to confirm the 5′ end of OriR. The result showed that replication from the ΔREH17 mutant was significantly decreased by ∼3-fold (Fig. 2D, lane 3).

Collectively, we defined HBoV1 OriR to be a 46-nt DNA fragment from nt 5357 to 5402 (Fig. 2F). Although these deleted sequences which flank the OriR are dispensable for the replication at the OriR, they are required for the maximum DNA replication at the REH. Of note, HBoV1 has a long 3′ NCR in front of the OriR. To clarify whether the OriR replicates viral DNA independently of the 3′ NCR, we inserted various bacteriophage λ DNAs of 0.2, 0.4, and 1.0 kb between the 3′ NCR and the OriR in HBoV1Ori DNA (Fig. 3A). DNA replication analysis showed that the OriR conferred viral DNA replication regardless of the insertion sizes between the 3′ NCR and OriR, supporting the suggestion that the OriR functions as a template of DNA replication independently of the adjacent 3′ NCR (Fig. 3B).

FIG 3.

HBoV1 OriR replicated independently of the upstream 3′ NCR. (A) Diagram of bacteriophage λ DNA insertion mutants. On the basis of the pHBoV1Ori sequence, 200 bp, 400 bp, or 1,000 bp of λ DNA was inserted between the 3′ NCR and REH. (B) Southern blot analysis. HEK293 cells were transfected with linearized mutants. At 2 days posttransfection, Hirt DNA samples were extracted and digested with DpnI before they were applied for Southern blotting. Sixty nanograms of IHBoV1 was digested as a DpnI digestion control.

NP1 colocalizes with NS proteins within the viral DNA replication centers and is required for viral DNA replication at OriR.

We next checked the trans-acting factors that facilitate viral DNA replication at OriR. We constructed NS1- or NP1-knockout plasmids based on pHBoV1Ori and performed viral DNA replication analysis. We observed that, without NS1/2 or NP1 expression, replication at the OriR was abolished (Fig. 4, lanes 2 and 3), reflecting that NS1 and NP1 are required for HBoV1 DNA replication at the OriR, since NS2 is not required for HBoV1 DNA replication in HEK293 cells (22).

FIG 4.

Both NS1 and NP1 are required for DNA replication at OriR. HEK293 cells were transfected with HBoV1Ori or its mutants HBoV1(NS1−)Ori and HBoV1(NP1−)Ori. Hirt DNA samples were extracted at 2 days posttransfection and were digested with DpnI for Southern blotting. Sixty nanograms of IHBoV1 digested by DpnI was used as a digestion control.

Replication of parvovirus MVM and H-1 takes place in discrete subnuclear compartments, termed autonomous parvovirus-associated replication (APAR) bodies (39, 40). APAR bodies are active sites of viral DNA replication and contain cellular DNA replication factors and parvovirus NS1 (39). We located the APAR bodies using BrdU to pulse-chase active replicating viral ssDNA (33, 34). Similar to other parvoviruses (39, 40), the APAR bodies of HBoV1 showed various patterns in different cells, from focus-like dots to areas with a broad distribution in the nucleus, and colocalized with NS proteins stained with anti-NS1C in HEK293 cells transfected with pHBoV1Ori or pIHBoV1 (Fig. 5A, NS and BrdU). Notably, patterns of NP1 and BrdU colocalization similar to those for NS and BrdU were observed, suggesting that both NP1 and NS localized within APAR bodies (Fig. 5B, NP1 and BrdU). In support of this hypothesis, NS and NP1 proteins colocalized in the nucleus of pHBoV1Ori- or pIHBoV1-transfected cells (Fig. 5C, NP1 and NS). More importantly, in HBoV1-infected HAE cells, NS and NP1 colocalized well within the BrdU-chased APAR bodies (Fig. 5D to F). To confirm the localization of NP1, we used a proximity ligation assay (PLA) to visualize the interactions of NP1 with BrdU-labeled viral ssDNA or dsDNA/ssDNA intermediates. We observed clearly positive fluorescent foci in the nucleus of pHBoV1Ori- and pIHBoV1-transfected or HBoV1-infected cells stained with anti-NP1 and anti-BrdU antibodies (Fig. 5G). The PLA showed bright signals only if the two molecules localized proximately at a distance of ∼20 nm (41).

To explore whether a direct interaction exists between NS1 and NP1, we performed immunoprecipitation assays. Cells transfected with pHBoV1Ori were lysed and precleared with normal rat serum. Precleared lysates were immunoprecipitated with control rat IgG or anti-NP1 serum in the presence of nuclease treatment and were then evaluated for NS1 by Western blotting. Neither NS1, NS2, NS3, nor NS4 was coimmunoprecipitated with NP1 (Fig. 6A, lane 6 versus lane 4). As a control, using an anti-NS1C antibody, all NS proteins could be immunoprecipitated (Fig. 6B, lane 3 versus lane 1). These results suggest that the NS1 to NS4 and NP1 proteins do not directly interact.

FIG 6.

NS and NP1 did not interact directly. HEK293 cells were transfected with pHBoV1Ori or mock transfected. At 2 days posttransfection, the cells were washed, lysed, and precleared with normal rat serum as described in Materials and Methods. Precleared lysates were immunoprecipitated with either rat IgG and rat anti-NP1 rat serum (A) or rat IgG and rat anti-NS1C serum (B). Immunoprecipitated proteins were analyzed by Western blotting (WB) using an anti-NS1C antibody. Ten microliters of the precleared lysates was loaded as input controls. IgG Ctrl, IgG control.

Taken together, these results suggest that either by transfection of pIHBoV1/pHBoV1Ori or during viral infection, the HBoV1 NS1 to NS4 and NP1 proteins function synergistically in the viral DNA replication centers (APAR bodies) but without a direct interaction. More importantly, these results confirm a direct involvement of the HBoV1 NP1 during viral DNA replication at OriR.

Identification of NSBEs and nicking site within OriR.

The replication origins of parvoviruses harbor multiple binding elements of Rep78/68 or NS1 and a specific nicking site (26–29). Therefore, we next aimed to define the NS1 binding elements (NSBEs) and nicking site by means of mutagenesis. The NSBEs characterized in AAV and MVM by both in vivo and in vitro studies are several tetranucleotide repeats, which are directly recognized by the origin-binding domain (OBD) of AAV Rep68/78 or MVM NS1. Notably, there are no identical NSBEs for HBoV1 NS1 binding that can be located in both the REH and LEH (17). In contrast, we found that, similar to the 4 repeats of trinucleotides in densonucleosis virus of Galleria mellonella (GmDNV), there are 4 repeats of TGT trinucleotides (5′-TGT TGT TGT TGT-3′) in OriR (Fig. 7A) (42).

We therefore made serial mutations in the TGT repeat region in OriR and performed in vivo DNA replication analysis. The results showed that all these mutations of the TGT trinucleotides significantly decreased DNA replication to various levels that were less than 40% of the activity of the wild type (Fig. 7B and C); moreover, the mutation of multiple TGT trinucleotides progressively disabled DNA replication to a greater extent. For example, with mutation of two TGT trinucleotides, DNA replication remained better than that with mutation of three TGT trinucleotides (Fig. 7B and C, lanes 1 and 2 versus lanes 4 to 7). Mutation of all four TGT sequences decreased DNA replication the most to a level that was only 5% of that of the wild type (NSBEm3; Fig. 7B, lane 3, and C). Thus, our results suggest that this TGT repeating sequence is likely the HBoV1 NSBE.

The nicking sites of parvoviruses are not conserved (29). In principle, after nicking, the Ori reveals a transient nicked intermediate (ssDNA breaks) before the free 3′ OH is extended by DNA polymerase (23). We employed a strategy of rapid amplification of the nicked 3′ end to characterize the intermediates of the resolved OriR after NS1 nicking. Using Hirt DNA extracted from IHBoV1- or HBoV1Ori-transfected cells, we added adenosines to extend the 3′ end with poly(A) residues. With subsequent PCR amplification and cloning, we mapped a few transient ending sites in the OriR (Fig. 8A). To identify the nicking among these ending sites, we mutated each site located outside the NSBE (Fig. 8A). Analysis of in vivo DNA replication showed that a mutation at either the A or T nucleotide at nt 5368 or 5369 but not mutations at other locations completely abolished viral DNA replication (Fig. 8B, lanes 5 and 6 versus lanes 1 to 4). A mutation at nt 5373, within the putative nicking motif (5′-CTA TAT CG-3′; the mutated site is underlined) also reduced the level of RF DNA (Fig. 8A, TRSm7, and B, lane 7). This result defined the T at nt 5369 to be the nicking site [5′-CT(A/T)ATCT-3′].

Identification of the key OriR-binding residues in the OBD of HBoV1 NS1.

We have previously resolved the structure of the HBoV1 OBD and predicted that NS1 utilizes two nonstructured positively charged loop regions to bind Ori (43). We thus examined the binding residues of NS1 using a mutagenesis approach (Fig. 9A). To this end, we performed mutations of either amino acids at the positively charged loops (loop K, ¹²⁷KR¹²⁸; loop R, ¹⁹³RR¹⁹⁴) or the predicted endonuclease activity motif of the NS1 (¹¹⁵HCH¹¹⁷), together with three control mutations (Mut Q, P, and E), in the context of HBoV1Ori. As controls, Mut Q and Mut P mutated the first and the second loops, respectively, of the HBoV1 NS1 OBD structure. The Mut E control mutated the nearby amino acids (¹²³EGL¹²⁵) of putative loop K (Fig. 9B). The structures of the mutated NS1 OBD were predicted on the basis of the sequence of the wild-type NS1 OBD (43). Superimposition of the mutant NS1 OBD with wild-type NS1 OBD showed that mutagenesis did not significantly change the structure (Fig. 9B). In vivo DNA replication analysis showed that mutation of either the basic charged loop region (Mut K and Mut R) or the endonuclease core (Mut Endo) abolished replication, whereas the three control mutations did not alter DNA replication very much (Fig. 9C). These data suggest that the two positively charged loops of the OBD, as well as the two histidine residues at the predicted endonuclease motif, are required for HBoV1 DNA replication and that the two positively charged loops likely play a significant role in NS1 and OriR binding.

HBoV1 NS1 did not specifically bind OriR in vitro.

To understand the interaction between HBoV1 NS1 and OriR (Fig. 2F), we purified HBoV1 NS1 (Fig. 10A, lane 2) and studied its binding with the OriR in vitro. As a positive control for NS1 binding to OriR, B19V NS1 was purified (Fig. 10A, lane 1) and carried along in parallel. We used B19V NS1 binding to B19V Ori as a positive control because B19V also has a short Ori of 67 nt (32). An electrophoresis gel shift assay showed that B19V NS1 shifted the B19V Ori (Fig. 10B, lane 2). This binding was specific because it was competed by excess cold B19V Ori but not by excess cold mutated B19V Ori (Ori-mut) (Fig. 10B, lane 4 versus lane 6). As a negative control, the glutathione S-transferase (GST) protein did not shift the B19V Ori (Fig. 10B, lane 7), and the B19V Ori-mut was not shifted by B19V NS1 (Fig. 10B, lane 8 versus lane 9). In contrast, however, under the same experimental condition of the electrophoresis gel shift assay, HBoV1 NS1 did not specifically bind its own OriR (Fig. 10C, lane 2). Addition of either cold OriR (Fig. 10C, lanes 3 to 5) or cold Ori-mut (Fig. 10C, lanes 6 to 8) showed similar negative signals. Of note, we found a weak nonspecific interaction between HBoV1 NS1 and nonspecific DNA in the binding buffer without poly(dI-dC) (Fig. 10D, lane 2). This nonspecific binding was competed either by excess cold HBoV1 OriR (Fig. 10D, lanes 3 to 5) or by cold HBoV1 OriR-mut (Fig. 10D, lanes 6 to 8). Thus, the results of the gel shift assays suggest that HBoV1 NS1 alone does not bind HBoV1 OriR specifically in vitro.

To further examine the binding property between HBoV1 NS1 and OriR in the presence of cellular proteins, we designed a biotin pulldown assay. Nuclear extract prepared from NS1-transfected HEK293 cells was used for pulldown by biotin-labeled OriR (Bio-OriR) and streptavidin-conjugated beads and then analyzed by Western blotting. In the experiment with B19V Ori, B19V Bio-Ori pulled down B19V NS1, which was competed by excess B19V Bio-Ori but not by excess mutant Ori (Ori-mut) (Fig. 11A, lane 4 versus lane 5, and B). As a control, non-biotin-labeled B19V Ori did not pull down any B19V NS1 (Fig. 11A, lane 2, and B), and the Bio-Ori-mut pulled down only a small amount of NS1 (Fig. 11A, lane 6, and B). In contrast, HBoV1 NS1 was not pulled down by either biotinylated Ori or Ori-mut (Fig. 11C, lanes 2 and 3), despite the high level expression of NS1 (Fig. 11C, lane 1). As NP1 colocalized with BrdU-labeled viral replication centers (Fig. 5), we tested the hypothesis that NP1 is critical for NS1 binding with OriR. We prepared nuclear extracts from cells coexpressing NP1 and NS1 and performed pulldown assays. The results demonstrated that biotinylated HBoV1 OriR did not pull down NS1, even in the presence of NP1 (Fig. 11D, lane 2). Collectively, the pulldown assays suggest that HBoV1 NS1 does not bind HBoV1 OriR in vitro in the presence of cellular factors and NP1.

FIG 11.

HBoV1 NS1 alone or with NP1 did not specifically bind OriR in a pulldown assay. (A and B) Detection of B19V NS1 and B19V Ori binding. (A) NEs prepared from OptiB19VNS1-transfected HEK293 cells were incubated with biotin-labeled B19V Ori (Bio-Ori; lanes 3 to 5) and B19V Ori-mut (lane 6), followed by streptavidin-conjugated beads. Additionally, some reaction mixtures were additionally incubated with B19V Ori-mut at 30 times (lane 4) and B19V Ori at 30 times (lane 5). The NS1 bands pulled down by Bio-Ori are indicated. (B) The NS1 bands pulled down by Bio-Ori were quantified. Levels relative to those of NS1 (which were considered 100%) are shown as averages and standard deviations from three independent experiments. (C and D) Detection of HBoV1 NS1 and HBoV1 OriR binding. (C) NEs prepared from OptiHBoV1NS1-transfected HEK293 cells were incubated with HBoV1 OriR (lane 2) and HBoV1 OriR-mut (lane 3), followed by streptavidin-conjugated beads. The NS1 detected in the input is indicated. (D) NEs prepared from OptiHBoV1NS1- and OptiHBoV1NP1-cotransfected HEK293 cells were incubated with HBoV1 OriR or OriR-mut, followed by streptavidin-conjugated beads. The NS1 and NP1 detected in the input are indicated.

DISCUSSION

In the study described in this report, we studied the viral components that are required both in cis and in trans for HBoV1 terminal resolution at the REH. We defined a 46-nt sequence at nt 5357 to 5402 to be the HBoV1 OriR. It contains both the NSBEs and the nicking site and is used as a template for HBoV1 DNA replication at the REH. Notably, a sequence in the 3′ NCR was critical in cis for viral DNA replication at the OriR, while the large NS protein NS1 and the small NP1 played a pivotal role in trans for viral DNA replication at the OriR. These basic findings of HBoV1 DNA replication at the OriR lay a foundation for further understanding of the mechanism underlying NS1 binding to and nicking at the OriR, which are the key steps in HBoV1 RF DNA replication. Additionally, in this study, we show, for the first time, the specific binding of the B19V NS1 with B19V Ori in vitro.

Functions of viral proteins in HBoV1 DNA replication at OriR. (i) NS1 function.

The large nonstructural protein of parvovirus, Rep78/68 or NS1, is composed of a putative DNA OBD/endonuclease domain, a helicase activity domain, and a transactivation domain (TAD) at the N terminus, middle, and C terminus, respectively (22). We have previously resolved the crystal structure of the HBoV1 OBD at a 2.7-Å resolution and showed that it is similar to the structures of canonical histidine-hydrophobic residue-histidine (HUH) superfamily of nucleases. The OBD structure combines two distinct functions: (i) a positively charged region formed by a surface β-hairpin (aa 190 to 198) and an α5 helix (aa 127 to 132), which is responsible for recognizing the viral DNA replication origin, and (ii) the endonuclease active site, which contains the signature motif HUH and performs strand-specific cleavage at Ori (43). The HUH motif of the HBoV1 OBD contains two histidine residues (H115 and H117) separated by cysteine C116, followed by the three hydrophobic residues I118, L119, and V120 (43). However, these active sites are predicted only from superposition of the HBoV1 OBD structure onto the AAV5 OBD structure (44). Here, we confirmed that R193 and R194 in the surface hairpin of the OBD and K127 and R128 in the loop region are critical to terminal resolution and, in contrast, the neighboring ¹²³EGL¹²⁵ residues are not. Mutation of the two histidine residues (H115 and H117) confirmed their nicking function. A study of the structure of the MVM NS1 OBD also revealed conserved residues with DNA binding and nicking activities (45), highlighting the importance of the confirmed DNA binding and nicking residues of the HBoV1 NS1 OBD.

(ii) NP1 function.

We have previously shown that bocaparvovirus NP1 plays an important role in the replication of viral duplex DNA (12, 17). HBoV1 NP1 or BPV1 NP1 can complement the replication of a mutant MVC infectious clone that does not express MVC NP1 (12). Three other newly identified HBoV1 small NS proteins, NS2 to NS4, are not required for the replication of HBoV1 duplex DNA genome in HEK293 cells (22). Since NS2 to NA4 all contain the NS1C terminus, the anti-NS1C antibody reacts with all these isoforms (22). Thus, we demonstrated that all NS1 to NS4 proteins and NP1 colocalize within APAR bodies. Nevertheless, it is unlikely that NS2 to NS4 recruit NP1 to the APAR bodies. Direct interactions between NS1 to NS4 and NP1 were not confirmed (Fig. 6). NP1 contains a nonclassical nuclear localization signal (ncNLS) at aa 7 to 50 (46) and is able to complement the functions of the MVM NS2 in viral DNA replication during an early phase of infection (35). In fact, during MVM infection of NP1-expressing A9 cells, NP1 was progressively lost from its nucleolus localization and began to be colocalized with MVM NS1 in APAR bodies. Moreover, NP1 expression rescues APAR body maturation in cells infected with an NS2-null mutant of MVM, MVMp (35). Additionally, NP1 is involved in viral pre-mRNA processing (18–20). HBoV1 NP1 is required for viral mRNA splicing at the A3 splice site and read-through of the viral mRNA through the (pA)p site (18). Taken together, we hypothesize that during infection or viral DNA replication, NP1, together with NS1, in which cellular DNA replication factors are enriched, is critical in the development of viral DNA replication centers (APAR bodies); however, after the formation of the APAR bodies, NP1 could be relocated to the cellular compartment for viral RNA processing. On the other hand, NP1 could possibly be recruited by cellular DNA replication factors to the APAR bodies, in response to the efforts of NS1 in interacting with the viral Ori and cellular DNA replication factors (47–49).

Identification of the HBoV1 OriR.

For a homotelomeric parvovirus whose replication depends on a helper virus, a 43-nt DNA sequence containing the Rep binding element (RBE) and nicking site was identified to be the AAV Ori (26). For homotelomeric parvoviruses that replicate autonomously, a specific 38-nt DNA sequence has been identified to be the Ori of goose parvovirus (GPV) (50), and a 67-nt DNA was identified to be the B19V Ori that contains a nicking site and four GC-rich NSBEs that are required for optimal virus replication (32). The B19V inverted terminal repeat (ITR) resembles that of GPV, in that both have an arrow-like hairpin structure (29, 51). For heterotelomeric parvoviruses that replicate autonomously, there are two replication origins located at the LEH and REH, respectively. The active form of the MVM LEH Ori (OriL_TC) is ∼50 nt in length and is composed of a parvovirus initiation factor (PIF) binding site, the (ACCA)₂ NSBEs, and the nicking site (23, 28). It functions as a template for junction resolution that generates the ssDNA genome. In contrast, the MVM REH Ori (OriR) is about 125 nt in length and contains a region composed of a nicking site and two closely contacted (ACCA)₂ NSBEs, a degenerate NSBE (CGGT) at the tip of the hairpin, and a cis sequence that is nonspecifically bound by HMG1/2 family DNA binding proteins (27). Therefore, the MVM OriR includes almost the entire sequence of the REH (23, 27).

In this study, we identified that a 46-nt sequence (OriR) at the REH of the HBoV1 genome is responsible for the replication of the duplex HBoV1 genome in HEK293 cells. This OriR represents the first Ori in members of heterotelomeric parvoviruses that functions as a template of terminal resolution in a short closely contacted DNA sequence (46 nt) containing the NSBEs and nicking site. We speculate that cis-acting sequences surrounding the nicking site and NSBEs are required for interacting with cellular factors, e.g., HMG1/2 with MVM OriR (27) and PIF with MVM OriL (52).

Characterization of the nicking site and NSBEs.

The site at which Rep78/68 or NS1 nicks is specific to each parvovirus (29). The nicking site 5′-GAG(T/T)GG-3′ is conserved only in AAV1 to AAV4 and AAV6, but AAV5 uses 5′-AGT(G/T)GGC-3′ (53). For autonomous parvoviruses, GPV uses 5′-TGA(G/T)CT-3′ (50), B19V Ori uses a unique nicking site [5′-GAC(A/C)CA-3′] (32), and MVM uses the nicking site 5′-CTW(W/T)CA-3′, where W is A or T (50). Thus, the nicking sites of autonomous parvoviruses differ from each other. The HBoV1 nicking site [5′-CT(A/T)ATCT-3′] identified in this study closely resembles the MVM terminal resolution site (TRS) with an A/T-rich sequence in the center. Of note, such a similar nicking site is not found at the LEH of the HBoV1 genome; however, one unique nicking site of AAV2 Rep78/68 [5′-CTCC(A/T)TT-3′] has been identified in the minimal replication origin present within the AAV2 P5 promoter (54, 55). The nicking of AAV2 Rep78/68 at the nicking site in the P5 promoter involves the TATA box in cis and the TATA-binding protein in trans (54). We hypothesize that HBoV1 NS1 must employ a different nicking site to perform junction resolution at the OriL of the LEH and that cellular transcriptional factors or DNA binding proteins should facilitate nicking of the NS1 at the nicking site at the OriL.

Several Rep78/68 and NS1 binding elements have been characterized and confirmed by an in vitro binding assay. The AAV RBE consists of three tetramer repeats [(GCTC)₃] plus a degenerate GCGC (26). Similarly, three tetramer repeats [(GTTC)₃] plus GAAC were found in the ITR of GPV (50). Two hexamer repeats [(GCCGCCGG)₂] were confirmed to bind to the B19V NS1 OBD in an in vitro binding assay (37). The MVM NSBE in either OriL or OriR comprises 2 to 3 tandem copies of the tetranucleotide TGGT [(TGGT)₂ and (TGGT)₃, respectively]. The densoparvovirus GmDNV NS1 binds a (GAC)₄ trimer repeat sequence in its ITR (42). A consensus NSBE for HBoV1 NS1 binding cannot be found in the LEH and REH of the HBoV1 genome (17). In the HBoV1 OriR identified in this study, at ∼12 nt downstream of the nicking site, a tetramer trinucleotide [(TGT)₄] proved critical to viral duplex DNA replication at the OriR. However, we were not able to confirm specific binding between HBoV1 NS1 and OriR in vitro. We believe that our in vitro binding assay is capable of revealing the specific binding. We adapted an in vitro binding buffer which has been successfully used to confirm MVM NS1 binding to its NSBE (36, 47). We also demonstrated B19V NS1 binding to the B19V Ori in side-by-side studies assessing HBoV1 NS1 binding of the HBoV1 OriR. Although a strong and specific binding between B19V NS1 and its Ori was observed and confirmed, we did not observe any specific binding between HBoV1 NS1 and its OriR. We also used a more sensitive in vitro binding assay, in which an anti-NS1 antibody was used to pull down NS1-associated OriR (47), but the result of this assay was also negative (data not shown). Therefore, we speculate that HBoV1 NS1 and OriR binding may require the involvement of viral and cellular proteins. However, in a subsequent in vitro pulldown assay, the biotinylated HBoV1 OriR did not pull down any HBoV1 NS1 from the lysate of cells expressing HBoV1 NS1. As a control, the B19V Ori pulled down NS1 at a significantly higher level than the mutant Ori, which can be competed by the wild-type B19V Ori but not the mutant Ori. A further experiment using a nuclear extract prepared from NS1- and NP1-coexpressing cells also did not show any binding between HBoV1 NS1 and OriR. We hypothesize that in vivo oligomerization of NS1 may be necessary for recognition of the NSBEs, as Rep78/68 must be oligomerized for recognition of the RBEs (56).

Nevertheless, on the basis of the model of rolling circle replication in which Rep78/68 or NS1 has to bind the origin, melt the duplex viral DNA, and perform nicking of the ssDNA at the nicking site ∼20 nt upstream of the NSBEs (23), we believe that the (TGT)₄ repeat in HBoV1 OriR should be the NSBEs. The (TGT)₄ repeat closely resembles the (GAC)₄ repeat of the GmDNV NSBE (42). Since the (TGT)₄ repeat contains the (TGTT)₂ repeat, the HBoV1 NSBEs also resemble the MVM NSBEs, (TGGT)₂ and (TGGT)₃ (47). Considering that no cognate binding sequences can be found in the origins of either the LEH or the REH of the HBoV1 genome, we hypothesize that, contrary to what has been observed, HBoV1 NS1 must bind the origins at a low affinity and requires the help of other viral components and cellular proteins to do so.

Function of the 3′ NCR of HBoV1.

Identification of the role of the 3′ NCR between the VP-coding region and the REH in HBoV1 DNA replication is important. In other parvoviruses, various cis sequences that are outside the terminal hairpins, e.g., an additional AAV2 minimal DNA replication origin at the P5 promoter (5′ NCR), have been identified to be important for DNA replication (55, 57, 58). In MVM, it has been shown that specific elements within the REH between nt 4489 to 4636 and nt 4636 to 4695 are necessary for efficient replication of MVM duplex DNA (59). During the development of the recombinant MVMp vector (rMVMp), a large portion of cis element remained at the 3′ end (nt 4631 to 5149) of the rMVMp genome (60). However, how these cis elements outside the hairpins facilitate viral DNA replication has not been studied.

We have previously developed a recombinant HBoV1 vector, in which both the recombinant AAV2 genome and recombinant HBoV1 (rHBoV1) genomes were used. However, large portions were retained at the 3′ and 5′ ends in the rHBoV1 genome, in order to ensure efficient replication in the presence of a packaging plasmid (61). Unfortunately, the cis sequences that remained resulted in a high rate of recombination that generated wild-type virus in the rHBoV1 preparations (61). Therefore, it is important to define the minimal requirement for cis sequences for HBoV1 DNA replication to develop a better rHBoV1 vector that may hold benefits for gene targeting in human airways, since 95% of the HBoV1 genome is negative sense, while the AAV genome has equal polarity (12, 24). We plan to further define the cis element at the left end in order to construct an rHBoV1 genome that has a minimal sequence of HBoV1 (to avoid homologous recombination with the HBoV1 packaging plasmid).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.