The Human Bocavirus 1 NP1 Protein Is a Multifunctional Regulator of Viral RNA Processing

The Parvovirinae are small nonenveloped icosahedral viruses that are important pathogens in many animal species, including humans. The NP1 protein of human bocavirus 1 (HBoV1), similar to NP1 of the bocavirus minute virus of canine (MVC), regulates viral alternative RNA processing by both suppressing polyadenylation at an internal site, (pA)p, and facilitating splicing of an upstream adjacent intron. These effects allow both extension into the capsid gene and splicing of the viral pre-mRNA that correctly registers the capsid gene open reading frame. Characterization of HBoV1 NP1 generalizes this central mode of parvovirus gene regulation to another member of the bocavirus genus and uncovers both important similarities and differences in function compared to MVC NP1 that will be important for future comparative studies.

ABSTRACT

Human bocavirus 1 (HBoV1) encodes a genus-specific protein, NP1, which regulates viral alternative pre-mRNA processing. Similar to NP1 of the related bocavirus minute virus of canine (MVC), HBoV1 NP1 suppressed cleavage and polyadenylation of RNAs at the viral internal polyadenylation site (pA)p. HBoV1 (pA)p is a complex region. It contains 5 significant cleavage and polyadenylation sites, and NP1 was found to regulate only the three of these sites that are governed by canonical AAUAAA hexamer signals. HBoV1 NP1 also facilitated splicing of the upstream intron adjacent to (pA)p. Alternative polyadenylation and splicing of the upstream intron were independent of each other, functioned efficiently within an isolated transcription unit, and were responsive independent of NP1. Characterization of HBoV1 NP1 generalizes its function within the genus Bocaparvovirus, uncovers important differences, and provides important comparisons with MVC NP1 for mechanistic and evolutionary considerations.

IMPORTANCE The Parvovirinae are small nonenveloped icosahedral viruses that are important pathogens in many animal species, including humans. The NP1 protein of human bocavirus 1 (HBoV1), similar to NP1 of the bocavirus minute virus of canine (MVC), regulates viral alternative RNA processing by both suppressing polyadenylation at an internal site, (pA)p, and facilitating splicing of an upstream adjacent intron. These effects allow both extension into the capsid gene and splicing of the viral pre-mRNA that correctly registers the capsid gene open reading frame. Characterization of HBoV1 NP1 generalizes this central mode of parvovirus gene regulation to another member of the bocavirus genus and uncovers both important similarities and differences in function compared to MVC NP1 that will be important for future comparative studies.

INTRODUCTION

Human bocavirus 1 (HBoV1) is a member of the Bocaparvovirus genus, which also includes bovine parvovirus and minute virus of canine (MVC) (1, 2). HBoV1 can cause mild to severe respiratory tract infections, primarily in children (3–6). Until its recent cloning, and the development of a useful tissue culture system to grow HBoV1 (7–11), MVC was often used as a surrogate for the characterization of aspects of bocavirus gene expression and virus-host interactions (12–17).

Like MVC, HBoV1 generates a single pre-mRNA from a promoter at the left-hand end of the genome (P5) that is processed via alternative splicing and alternative polyadenylation into multiple nonstructural (NS) protein- and capsid-encoding transcripts (12, 18, 19). As with other parvoviruses, an open reading frame (ORF) in the left half of the genome encodes NS proteins, while an ORF in the right half encodes the capsid proteins VP1 and VP2 (20, 21). HBoV1 NS proteins have been shown to help initiate and sustain the replication of the viral genome and mediate a number of important virus-host cell interactions (20, 22). However, the bocaparvoviruses also encode a genus-specific protein, NP1, from a small ORF spanning the center of the genome (12, 19, 23). This protein contains an extensive disordered region in its amino terminus and multiple SR dipeptide repeats (14).

MVC NP1 has been shown to play a major role in regulating viral alternative RNA processing. MVC NP1 has been shown to suppress the potent internal polyadenylation signal (pA)p located within the capsid-coding region in the middle of the genome (13, 19). The NP1 proteins of both MVC and HBoV1 also facilitate splicing of the 3D/3A intron that lies immediately upstream of (pA)p (14, 19). Both of these processes are necessary to gain proper access to the capsid gene ORF (13, 14, 19). Additionally, the C-terminal region of three of the MVC NS proteins are generated from mRNAs spliced at the third intron; thus, their expression is also facilitated by MVC NP1 (17). RNAs which encode the HBoV1 NS proteins are not spliced at the analogous intron (20).

HBoV1 NP1 is less well characterized than its MVC counterpart. It is 219 amino acids in length and shares only 46% identity and 62% similarity with the 185-amino-acid MVC protein (7). HBoV1 NP1 has been shown to be required for virus replication and is localized to viral replication centers (24). In certain contexts, it has been shown to interact with interferon regulatory factor 3 (IRF3) and thereby suppress interferon beta (25). HBoV1 NP1 was shown to enhance expression of the HBoV1 capsid proteins, and when the viral P5 promoter was replaced with the cytomegalovirus (CMV) immediate early (IE) promoter, subsequent knockout of the internal polyadenylation site (pA)p abrogated the need for NP1 (19). Interestingly, HBoV1 NP1 has also been shown to complement some early functions provided by the NS2 protein of minute virus of mice (MVM) (26).

In this report, we have defined the role that HBoV1 plays in the alternative processing of viral pre-mRNAs more thoroughly. We have defined the cleavage sites and polyadenylation elements that comprise the internal polyadenylation site (pA)p. Interestingly, while there are five cleavage sites, only three of them were governed by AAUAAA cleavage and polyadenylation specificity factor (CPSF)-binding motifs. We show that like the NP1 of MVC, HBoV1 NP1 also was required to suppress polyadenylation at (pA)p, allowing extension of RNA into the capsid gene; however, only the canonical cleavage sites were affected by NP1. HBoV1 NP1 was required for splicing of the third intron, immediately upstream of (pA)p, which brings the capsid gene mRNAs into register with the capsid-coding ORF, and so, similar to MVC, HBoV1 NP1 plays an essential role in the alternative RNA processing that governs access to the capsid-coding gene. However, the interaction between (pA)p and the upstream adjacent third intron of HBoV1, and how HBoV1 NP1 governs the interaction between the two motifs, was found to be quite different from what was previously seen for MVC and in at least one case may be due to the nature of the promoter driving expression. These results revealed important species-specific similarities and differences in this central aspect of the bocavirus life cycle.

RESULTS

Identification and characterization of the internal cleavage and polyadenylation sites of HBoV1.

Because the internal polyadenylation site of HBoV1 (pA)p contains five individual AAUAAA hexamer sequences (diagrammed in Fig. 1A, left), we suspected that the cleavage-and-polyadenylation profile of HBoV1 nonstructural mRNAs would be more complex than previously suggested. 3′ rapid amplification of cDNA ends (RACE) experiments, using HBoV1 RNA generated following transfection of a pHBoV1 wild-type (WT) Rep-Cap plasmid into 293T cells, identified 6 sites of cleavage within this region (sites A to F) (diagrammed in Fig. 1A, left). It is important to note that we (data not shown) and others (9, 20, 23, 27) have shown that the overall transcription profile of the nonreplicating HBoV plasmid, the infectious clone, and viral infection are similar. To determine the steady-state levels of RNAs using each cleavage site, we performed RNase protection assays (RPAs) using a probe spanning the region (diagrammed in Fig. 1A, left [projected protected fragments from each cleavage are also indicated]). As can be seen in Fig. 1B, 5 of these sites (sites B to F) were used at a significant and reproducible frequency (Fig. 1B, lane 1). Quantification of these results indicated that following transfection, RNAs cleaved at nucleotide (nt) 3341 (site B) represented, on a molar basis, approximately 24% of the total (pA)p-cleaved RNAs, and RNAs cleaved at nt 3350 (site C), nt 3360 (site D), nt 3406 (site E), and nt 3505 (site F) represented approximately 28%, 19%, 14%, and 15%, respectively (Fig. 1C, top). The cleavage site determined by 3′ RACE at nt 3316 (site A) was not reproducibly detected by an RPA. These results indicated a surprising heterogeneity in the cleavage of HBoV1 pre-mRNAs within (pA)p. Previous studies have suggested that RNAs cleaved and polyadenylated internally at HBoV1 (pA)p primarily used cleavage site A and either site C or D (8) or, alternatively, site F (23).

FIG 1.

Identification and characterization of the internal cleavage and polyadenylation sites in HBoV1. (A, left) Maps of proximal polyadenylation sites [(pA)p] in HBoV1 showing the locations of five AAUAAA hexamers (marked 1, 2, 3, 4, and 5) and six cleavage sites (marked A, B, C, D, E, and F, located at nt 3316, 3341, 3350, 3360, 3406, and 3505 of the HBoV1 genome, respectively [GenBank accession number JQ923422]). The position of the RNase protection probe (pA)p (nt 3250 to 3536) is indicated. The expected sizes of HBoV1 transcripts protected by the (pA)p probe are shown: the protected readthrough product [RT or (pA)d] (286 nt) and products cleaved at (pA)p (nt 3505 [site F, 252 nt], nt 3406 [site E, 156 nt], nt 3360 [site D,110 nt], nt 3350 [site C, 100 nt], and nt 3341 [site B, 91 nt]). (Right) Nucleotide locations and sequences of the indicated pHBoV1 mutants. Different nucleotide locations of AAUAAA core element mutations or T-rich-region mutations are shown in constructs m1, m2, m3, m4, m5, and mTT, respectively. The double mutant (m1/2) includes first and second indicated mutations shown in m1 and m2, while the joint mutant (mALL) includes all five AAUAAA core element mutations. (B) RPAs of total RNA, extracted 48 h following transfection of 293T cells with pHBoV1 WT and AAUAAA hexamer mutants, as indicated, using the (pA)p probe diagrammed in panel A. The RNA marker and the probe (323 nt) are shown on the left. Protected bands representing RNA species extending through (pA)p to (pA)d, or cleaved at the various cleavage sites as described, are indicated to the right. The protected band representing the readthrough product from m5 [(pA)p] (lane 6) is shorter due to a mismatch in the m5 probe at nt 3520 generated during cloning. (C, top) Quantifications of the data from RPAs in panel B showing the percentage of total RNA cleaved at (pA)p for each individual cleavage site (sites B, C, D, E, and F) following transfection of pHBoV1 WT (WT) (lane 1 in panel B). (Bottom) Quantifications of data from RPAs in panel B showing the ratios [(pA)p/(pA)d] for each of major cleavage sites (sites B, C, D, E, and F) following transfection of the AAUAAA core motif mutants 1 to 5, as indicated. Error bars indicate standard deviations from at least three individual experiments.

To identify which AAUAAA hexamer signals governed which individual cleavage events in this complex site, we undertook a mutational analysis of these motifs within the (pA)p region. The results presented in Fig. 1B and C show the levels of steady-state RNAs that used each of the major cleavage sites, relative to the readthrough RNAs using the distal (pA)d site, following mutation of each of the hexamer sites individually or in combination (mutations are diagrammed in Fig. 1A, right). As can be seen, mutation of hexamer site 1 had little effect on the levels of any of the RNAs using (pA)p (Fig. 1B, lane 2, and C, bottom). Mutation of site 2 had little effect on levels of RNAs cleaved at site F or E but significantly reduced levels of RNAs using the downstream cleavage sites B, C, and D (Fig. 1B, lane 3, and C, bottom). Mutation of hexamer sites 3, 4, and 5 led to only slight, nonsignificant reductions in the levels of any of the RNAs using (pA)p (Fig. 1B, lanes 4 to 6, and C, bottom). Mutations of all 5 sites (Fig. 1B, lane 8, and C, bottom) or only sites 1 and 2 (Fig. 1B, lane 7, and C, bottom) yielded results similar to those for the mutation in site 2 alone; i.e., cleavage at sites E and F was relatively unchanged, while cleavage at sites B, C, and D was significantly reduced. These results indicated that AAUAAA hexamer 2, spanning nucleotides 3329 to 3334, was the dominant signal within (pA)p, affecting the cleavage of sites immediately downstream. Interestingly, cleavage at sites E and F, which together represented approximately 30% of the total RNAs cleaved at (pA)p, was resistant to mutations of all the AAUAAA hexamers in the region. The (pA)p region also contains a U-rich region downstream of AAUAAA hexamer 2 (nt 3409 to 3414) (Fig. 1A). Such sites are often binding sites for the polyadenylation cleavage stimulation factor CstF and can aid in upstream cleavage and polyadenylation at canonical sites. Surprisingly, mutation of this site did not affect cleavage of the sites governed by AAUAAA hexamer 2 (sites B, C, and D); however, it dramatically reduced cleavage of site E, which was independent of the AAUAAA hexamers in the region (Fig. 1B, lane 9, and C, bottom). Hexamer-independent cleavage and potential polyadenylation of the HBoV1 RNA using site F are discussed further, below.

Analysis of a strong cleavage site within the (pA)p region that was cleaved independently of an AAUAAA hexamer.

Cleavage site F, at nt 3505, was independent of the five AAUAAA hexamers within the (pA)p region and was not affected by mutation of the U-rich region at nt 3400 to 3406. However, it represented approximately 15% of RNAs cleaved in the (pA)p region and has been reported to be the primary (pA)p cleavage site in other studies. Therefore, we decided to characterize this site further and investigate its importance.

Cleavage site F is embedded in an A-rich region (Fig. 2A). We chose to first confirm the cleavage of this site in RNA generated by both the wild type and a mutant of the A-rich region (mF) (diagrammed in Fig. 2A), using Northern analyses, which do not depend upon nuclease-digested protection and so are resistant to artifacts that such assays can produce. These experiments demonstrated that a mutation of the A-rich site led to a relative decrease in the amount of RNAs using (pA)p, consistent with the ∼15% cleavage of RNA at the A-rich site quantified in Fig. 1C (Fig. 2B, top, compare lane 3 to lane 1, and bottom). In addition, adding the A-rich region mutation to a mutation specifically disabling the 5 AAUAAA hexamers (mALLmF and mALL, respectively) led to both a further decrease in the amount of RNAs using (pA)p as well as a concomitant increase in the amount of those that utilized downstream (pA)d. The decrease observed was consistent with the ∼15% expected, based on the levels of A-rich cleaved RNAs described above (Fig. 2B, top, compare lane 5 [mALLmF] to lane 4 [mALL], and bottom). Data from RNase protection assays were consistent with these results (Fig. 2C, top, compare lane 3 to lane 1 and lane 4 to lane 2, and bottom). To probe the upstream requirements for cleavage at the A-rich region, the complete (pA)p region upstream of site F was replaced with the heterologous sequence from the capsid region of MVM (MVM_in) (diagrammed in Fig. 2A). Following transfection of this mutant, the majority of (pA)p cleavage products were lost, as expected; however, when the A-rich region was further mutated in this background (MVM_inmF), an additional decrease in the amount of RNAs cleaved in this region, consistent with expected levels, was observed (Fig. 2B, top, compare lane 7 to lane 6, and bottom, and C, top, compare lane 6 to lane 5, and bottom).

FIG 2.

Analysis of a strong cleavage site within the (pA)p region that was cleaved independently of an AAUAAA hexamer. (A) Diagram of HBoV1 proximal polyadenylation site mutation constructs (mALL, mALLmF, mF, MVM_in, and MVM_inmF), as described in the text, and pHBoV1 WT. The locations of ³²P-labeled forward primers (3250-F and 3387-F) are shown. The MVM_in fragment (nt 3501 to 3711), inserted in HBoV1 (nt 3280 to 3490), is shown as a gray line. Gray ovals indicate the locations of the five AAUAAA hexamer core motif signals, while black ovals indicate mutations of those signals as diagrammed in Fig. 1A, right. At the bottom is an alignment of nt 3494 to 3513 from the wild type (WT) and the A-rich mutant (mF) showing the locations of mutant nucleotide substitutions (underlined). For the 10A(pA)p probe, 10 adenines (underlined) were inserted at nt 3513 of the wild-type (pA)p probe. The arrows indicate the locations of the readthrough-protected bands [(pA)d] and the band cleaved at site F (nt 3505), using the 10A(pA)p probe for the RPA. (B) Northern blot assays of total RNA extracted from 293T cells following transfection with wild-type HBoV1 (lane 1) and the indicated mutant constructs using a ³²P-labeled probe as described in Materials and Methods. Bands representing mRNA species polyadenylated at (pA)p (np1/ns4 mRNA) and polyadenylated at (pA)d (vp mRNA and various ns mRNAs) are shown on the top. Lighter (left) and darker (right) exposures are shown. Quantification of data from the Northern assay showing changes of relative np1/ns4 mRNA [(pA)p] and vp mRNA levels for the various mutants shown at the bottom was performed as follows. The accumulated levels of vp mRNAs and np1/ns4 mRNAs were quantified relative to the total RNA for each sample. Next, the percentage of np1/ns4 RNAs relative to the total and the percentage of vp mRNAs relative to the total for the mF mutant compared to wild type, the percentage of np1/ns4 RNAs relative to the total and and the percentage of vp RNAs relative to the total for the mALL mutant compared to the mALLmF mutant, and the percentage of np1/ns4 RNAs relative to the total and the percentage of vp mRNAs relative to the total for MVM_in compared to MVM_inmF were calculated and are displayed as explained in the text. Error bars indicate the standard deviations from at least three individual experiments. (C, top) RPAs of total RNA extracted 48 h following transfection of 293T cells with the wild type or mutants, as indicated, using homologous wild-type or mutant (pA)p probes or the homologous MVM (pA)p probe, as described in Materials and Methods. Bands representing RNAs that either read through or are polyadenylated at the various cleavage sites of (pA)p are designated (pA)d, (F)3505, (E)3406, (D)3360, (C)3350, and (B)3341, to the left. P, probe alone. (Bottom) Quantification of data from RPAs showing ratios of (pA)p or (pA)d RNA in each lane relative to (pA)p plus (pA)d for each mutant. Errors bars indicate the standard deviations from at least three individual experiments. (D) Extracts from wild-type and mutant HBoV1 constructs (as indicated) bearing Flag-tagged NP1 genes were transfected into 293T cells, and extracts were subjected to immunoblotting with anti-Flag and antitubulin antibodies. The ratios for the WT (lane 1) were arbitrarily set as 100%, and the relative ratios expressed by the other mutants were normalized to the value for the WT. Error bars indicate the standard deviations from at least three individual experiments.

Importantly, mALLmF, in which all (pA)p signals were debilitated and consequently generated primarily readthrough products polyadenylating at (pA)d, expressed only ∼36% of the levels of NP1 of the wild type (Fig. 2D, compare lane 4 to lane 1). Additionally, the NP1 proline insertion mutant, which is deficient for (pA)p suppression and accumulates the internally polyadenylated product (as shown in Fig. 2B, lane 2; see also Fig. 4B, lane 4), generated approximately twice the level of NP1 as the wild type (Fig. 2D, compare lane 2 to lane 1). Together, these results indicated that RNAs internally cleaved at (pA)p, rather than those that read through to (pA)d, are the primary source for NP1 expression.

FIG 4.

NP1 affects alternative polyadenylation of only canonical cleavage/poly(A) sites as well as third-intron splicing. (A). Diagram of various HBoV1 NP1 mutant constructs and (pA)p replacement mutants (HB/MVCpAp and MVC/HBpAp). The locations of nucleotide substitutions, the 5× proline mutagenic insertion, and the locations of the (pA)p replacements are indicated. (B, top) RPAs of total RNA extracted following transfection of 293T cells with HBoV1 wild-type and mutant constructs, as indicated, using the (pA)p probe. The RNA species polyadenylated at (pA)d and cleaved at the various (pA)p cleavage sites are shown on the right. (Bottom) Quantifications of data from RPAs showing the ratios [(pA)p/(pA)d] for each of the major cleavage sites (sites B, C, D, E, and F). Errors bars indicate the standard deviations from at least three individual experiments. (C, top) RPAs of total RNA extracted following transfection of 293T cells with HBoV1 wild-type or mutant constructs, as indicated, using the 3A probe. The 3A probe (237 nt) is indicated to the left, while the corresponding RNA species (unspliced across the third-intron acceptor [3Aunspl] [200 nt] and spliced at the third-intron acceptor [3Aspl] [150 nt]) are indicated on the right. (Bottom) Quantifications of data from RPAs at the top. The RNA ratio (3Aunspl/3Aspl) of the WT was arbitrarily set as 100%, and the relative ratios for the mutants were normalized to the value for the WT. Error bars indicate the standard deviations from at least three individual experiments. (D) Extracts from wild-type and mutant HBoV1 constructs (as indicated) bearing Flag-tagged NP1 genes were transfected into 293T cells, and extracts were subjected to immunoblotting with anti-Flag and antitubulin (Tub) antibodies. (E, left) RPAs of total RNA extracted from 293T cells transfected with either wild-type HBoV1 or the mutants indicated, as explained in the text. The RNA species polyadenylated at (pA)d and (pA)p in HBoV1 or MVC are shown on the left or right, respectively. (Right) Quantifications of data from RPAs showing the ratios [(pA)p/(pA)d] for noncanonical cleavage sites (sites E and F) and canonical cleavage sites (sites B, C, and D) generated by wild-type HBoV1 or the mutants indicated. Error bars indicate the standard deviations from at least three individual experiments.

To characterize the unusual cleavage site F in greater detail, we next determined whether it could be captured on an oligo(dT) column. As the predicted cleavage site lies within the A-rich region, and this site was identified using 3′ RACE, this seemed likely. Figure 3A demonstrates that RNAs cleaved at site F, generated following transfection of either the pHBoV1 WT Rep-Cap clone (Fig. 3A, lane 4) or the clone with only the nt 3505 cleavage site present (Fig. 3A, lane 8), could be captured with oligo(dT).

FIG 3.

Analysis of a strong cleavage site within the (pA)p region that was cleaved independently of an AAUAAA hexamer. (A) Poly(A) RNA (lanes 4 and 8) selected from 100 μg total RNA extracted 48 h following transfection of 293T cells with pHBoV1 WT or MVM_in was used for RPAs. A total of 10 μg or 20 μg total RNA (lanes 1, 2, 5, and 6) or the nonbound supernatant fraction RNA (lanes 3 and 7) from 100 μg total RNA was used as control. The RNA species polyadenylated at (pA)d and the various cleavage sites of (pA)p are shown on the right. (B) RPAs of 20 μg total RNA extracted following transfection with pHBoV1 WT (lanes 1 and 2), detected with the wild-type or 10A(pA)p probe, respectively. Probe species are shown in the lanes to the left. The RNA species polyadenylated at (pA)d and (pA)p are shown on the right. (C) Poly(A) tail length assay, as described in the text, with ³²P-labeled primer F-3250 or F-3387. A darker exposure of lanes 5 to 8 (primer F-3387) is shown to the right. The RNA marker is shown on the left, and the distinct band (∼128 nt) is shown on the right. On the left, a band of approximately 262 nt (nt 3250 to 3515), identified by a star, is likely due to spurious binding of the oligo(dT) primer/adapter to the A-rich region. The diamond symbols indicate likely nonspecific amplified products from the forward primer.

To analyze the 3′ end of the RNA cleaved at site F in more detail, we performed two experiments. In the first experiment, we constructed an RPA probe that included an additional 10 adenine residues downstream of the proposed cleavage site [10A(pA)p probe] (Fig. 2A). If additional A residues had been added to the 3′ end of the RNA cleaved at site F during processing, the band representing protection of that RNA would be 10 nt larger. As can be seen in Fig. 3B, lane 3, the band obtained was actually slightly shorter, likely representing stuttering of cleavage among the multiple A residues. These results indicated that this RNA acquired no significant additional polyadenylation following cleavage. This result was confirmed by direct analysis of the size of the A-residue sequence on the 3′ end of this RNA, employing reverse transcription-PCR (RT-PCR) using an oligo(dT) primer/adapter and either of two labeled upstream primers indicated in Fig. 2A. Primer 3250-F was predicted to detect additional A residues on all (pA)p RNAs, while primer 3387-F was predicted to detect only additional A residues added to the RNA cleaved at nt 3505. As can be seen in Fig. 3C, when primer 3250-F was used, a smear, indicating heterogeneous levels of poly(A) addition, was seen added to (pA)p RNA generated following transfection of either the wild-type pHBoV1 Rep-Cap clone (WT) (Fig. 3C, lane 1) or a mutant with a disabled cleavage site F (mF) (Fig. 3C, lane 2). No poly(A) addition was detected following transfection of F-site-containing mutants either lacking the 5 AAUAAA hexamers (MVM_in) (Fig. 3C, lane 3) or with the hexamer sites mutated (mALL) (Fig. 3C, lane 4). In contrast, when the 3387-F primer was used, a discrete band of approximately 128 nt was observed with wild-type-generated RNA as the target (Fig. 3C, lane 5), which was dramatically reduced when site F was mutated (mF) (Fig. 3C, lane 6). The mutant in which the hexamer region was replaced with a heterologous sequence (MVM_in) generated no band because the binding site for the primer was lacking (Fig. 3C, lane 7); however, when only the 5 AAUAAA hexamers were specifically mutated (mALL), the 128-nt band was regained (Fig. 3C, lane 8). Together, these results indicated that the RNA cleaved at site F, which was not governed by a CPSF-binding AAUAAA signal, acquired no significant additional poly(A) tail following cleavage. The role that this RNA might play in HBoV1 infection is currently under further investigation.

NP1 affects alternative polyadenylation of only canonical cleavage/poly(A) sites as well as splicing of the upstream proximal third intron.

The NP1 protein of the bocavirus MVC suppresses polyadenylation at (pA)p and facilitates splicing of the upstream adjacent third intron to allow proper access to the capsid gene (13, 14). To determine whether the NP1 protein of HBoV1 performs similar functions, we tested a series of HBoV1 NP1 mutations and chimeric constructs (diagrammed in Fig. 4A) for their effects on HBoV1 alternative RNA processing.

Analysis of four separate NP1 mutations were tested: mAUG, which mutated the initiating AUG; 5×Pm′ and 5×Pm, in which 5 proline residues were inserted at nt 2745 and 2820, respectively; and mTAG, in which a termination signal was inserted at NP1 residue 2590 (Fig. 4A). These analyses demonstrated that in the absence of wild-type NP1, (pA)p cleavage sites governed by an AAUAAA hexamer (cleavage sites B, C, and D) accumulated significantly (Fig. 4B, top, compare lanes 2 to 5 to lane 1, and bottom). This indicated that, similar to MVC, HBoV1 NP1 was required to suppress these sites in HBoV1 pre-mRNAs and allow readthrough into the capsid gene. Surprisingly, however, levels of RNA that used hexamer-independent cleavage sites E and F were not significantly affected by the mutations of NP1 (quantification of these results is shown in Fig. 4B, bottom). Interestingly, the mutagenic addition of five proline residues at nucleotide 2745 (5×Pm′) had a much more modest effect on (pA)p suppression than a similar mutation inserted at nt 2820 (5×Pm) (Fig. 4B, top, compare lane 3 to lane 4, and bottom), even though the Flag-tagged mutant NP1 expression levels were similar (Fig. 4D, compare lanes 3 and 4). Whether the increase in total RNA apparent in the NP1 mutant samples shown in Fig. 3 was due to the lack of a specific NP1 function is under investigation. The less deleterious, first 5-proline insertion mutation lies in a highly disordered region in the NH₂-terminal portion of HBoV1 NP1 (as determined by PrDOS analysis [http://prdos.hgc.jp/cgi-bin/top.cgi]) (data not shown). Figure 4D also shows that, as expected, neither the mAUG nor the mTAG construct expresses an NP1 protein detectable by anti-Flag antibody.

Effects of HBoV1 on third-intron splicing showed concordant results. A mutant in the HBoV1 NP1 initiating AUG codon, which prevented its expression (mAUG) (Fig. 4D, lane 2), reduced third-intron splicing by approximately 50% (Fig. 4C, top, compare lane 2 to lane 1, and bottom). The proline insertion in the disordered region of NP1 (5×Pm′ at nt 2745) was less disruptive on third-intron splicing, reducing it to approximately 70% of the wild-type level (Fig. 4C, top, lane 3, and bottom), while the proline insertion 79 nt downstream (5×Pm at nt 2820) was similar to the initiating AUG mutation, mAUG, debilitating this function by nearly 50% (Fig. 4C, top, lane 4, and bottom). Interestingly, the mutation that caused a premature termination of NP1 28 amino acids from its start (mTAG) was not similarly debilitating; it still allowed third-intron splicing at levels approximately 75% of those of the wild type, suggesting, surprisingly, that this disordered amino terminus retained substantial function in this regard (Fig. 4C, top, lane 5, and bottom). Taken together, these results indicate that although the magnitude of its effect was not as great, similarly to MVC NP1, NP1 of HBoV1 was important for both suppression of internal polyadenylation at (pA)p and splicing of the third intron, two functions necessary to access the viral capsid open reading frame.

Since HBoV1 NP1 affected HBoV1 alternative RNA processing similarly to MVC NP1, we tested whether HBoV1 NP1 could affect polyadenylation of the MVC (pA)p signal and vice versa. As can be seen in Fig. 4E, when substituted in place into HBoV1 (as diagrammed in Fig. 4A), suppression of MVC (pA)p was indeed dependent upon HBoV1 NP1 (Fig. 4E, left, compare lane 3 [HB/MVCpAp] to lane 4 [mAUG/MVCpAp], and right). Similarly, when the HBoV1 (pA)p region was substituted in place into MVC, as shown (as diagrammed in Fig. 4A), its use was also regulated by MVC NP1 (Fig. 4E, left, compare lane 1 [MVC/HBpAp] to lane 2 [5×Pm/HBpAp], and right). These results indicate some similarity in both cis-acting signals and NP1 function for the regulation of the internal polyadenylation site for these two bocaviruses.

The role of NP1 in splicing is independent of (pA)p.

The effects of MVC NP1 on alternative polyadenylation and alternative splicing are independent of one another: when MVC (pA)p was debilitated, MVC NP1 was still required for third-intron splicing, and when the MVC third intron was debilitated, NP1 remained necessary for efficient accumulation of the readthrough RNA polyadenylated at (pA)d (13). HBoV1 NP1 function showed a similar independence. Mutation of the 5 AAUAAA hexamers plus the A-rich site in HBoV1 (pA)p did not substantially decrease splicing at the upstream third intron in the presence of wild-type NP1 (Fig. 5B, compare lane 2 [mALLmF′] to lane 1 [WT′]), and under these conditions, NP1 was still required for third-intron splicing (Fig. 5B, compare lane 4 [mAUGmALLmF′] to lane 1 [WT′] and lane 3 [mAUG′]). This indicated that in the full-length HBoV1 transcription unit driven by the HBoV1 left-end P5 promoter, NP1's effect on third-intron splicing remained primarily (pA)p independent. Data from Northern analysis (Fig. 5C) were consistent with these results. (pA)p mutations, in the presence of wild-type NP1, were seen to prevent internal polyadenylation of hexamer-dependent cleavage sites, as expected (Fig. 5C, lane 2 [mALLmF′]), and, as expected, Northern analysis also showed that NP1 mutations inhibited wild-type (pA)p suppression and reduced the accumulation of (pA)d-polyadenylated RNA (Fig. 5C, lane 3 [mAUG′]). However, the double mutant generated neither RNAs polyadenylated at (pA)p nor capsid-encoding RNAs spliced at the third intron (Fig. 5C, lane 4 [mAUGmALLmF′]). Thus, similar to the results seen in Fig. 5B, Northern analysis showed that even in the absence of (pA)p, NP1 remained necessary for splicing of capsid-encoding RNA generated from the authentic HBoV1 left-end promoter. This implied that in this context, NP1 is necessary for efficient third-intron splicing and also that it likely can act directly at the third intron itself. This is addressed further below. As expected, the role of NP1 in governing the accumulation of readthrough (pA)d-polyadenylated RNA (Fig. 5D, compare lane 2 [mAUG′] to lane 1 [WT′]) and the requirement of the (pA)p hexamers for internal polyadenylation were confirmed by an RPA (Fig. 5D, compare lane 3 [mALLmF′] to lane 4 [mAUGmALLmF′]). The concomitant lack of capsid protein production in both NP1 and the double mutant can be seen in Fig. 5E, lanes 3 and 4. In a reciprocal experiment, mutation of the third-intron acceptor, which dramatically inhibited splicing (Fig. 5F, compare lane 4 to lane 3), had little effect on polyadenylation choice (Fig. 5F, compare lane 2 to lane 1).

FIG 5.

The role of NP1 in splicing is independent of (pA)p. (A) Diagram of the HBoV1 (pA)p region showing the locations of wild-type and mutant constructs (pHBoV1 WT′ and mALLmF′) and (pA)p replacement mutant WT/bGH. mAUGmALLmF′ and mAUG/bGH have, in addition, a mutation in the initiating AUG codon of NP1 (not shown). All mutants are in a background in which the capsid gene is HA tagged, as described in Materials and Methods. Landmarks are as described in the legend of Fig. 1A, except that the bGH polyadenylation sequence (nt 1021 to 1263) is shown as a gray line. (B) RPAs of total RNA extracted following transfection of 293T cells with HBoV1 wild-type and indicated mutant constructs using the 3A probe. The probe (237 nt) is shown at the left, and the RNA species corresponding to RNAs unspliced or spliced at the third intron (3Aunspl [200 nt] and 3Aspl [150 nt], respectively) are indicated on the right. Quantifications below show the ratios of spliced/unspliced (Spl/un) RNAs compared to the ratio for the wild type, which has been set to 1. Standard errors are derived from at least three independent experiments. (C) Northern blot assays of total RNA from the same experiment shown in panel B. Bands representing mRNA species polyadenylated at (pA)p (np1/ns4 mRNA) or polyadenylated at (pA)d (vp mRNA and various ns mRNAs) are indicated on the right. Quantification of data from the Northern assay shows changes of vp mRNA levels relative to the total mRNA for the wild type (lane 1) and the indicated mutants, as described in the text. The ratios for WT′ (lane 1) are arbitrarily set as 100%, and the relative ratios in other lanes were normalized to the value for WT′. VP mRNAs were not detectable for mutants shown in lanes 3 and 4. Errors bars indicate the standard deviations from at least three individual experiments. (D) RPAs of total RNA extracted following transfection with HBoV1 wild-type and mutant constructs, as indicated, using the wild-type or hexamer mutant (mALL) (pA)p probe. The RNA species polyadenylated at (pA)d and using various cleavage sites in (pA)p are shown on the right. Quantifications below show the ratios of (pA)p to (pA)d. Standard errors are derived from at least three independent experiments. (E) 293T cells transfected with constructs described above for panels B to D were harvested and analyzed by immunoblotting with anti-HA and antitubulin antibodies. (F) RPAs of total RNA extracted following transfection with the HBoV1 wild-type or m3A mutant constructs using either the (pA)p probe (lanes 1 and 2), the wild-type 3A probe (lane 3), or the homologous mutant 3A probe (lane 4). The RNA species polyadenylated at (pA)d and (pA)p are shown on the left. For lanes 1 and 2, quantifications below show the ratios of (pA)p to (pA)d. Standard errors are derived from at least three independent experiments. For lanes 3 and 4, the corresponding RNA species unspliced and spliced at the third intron (3Aun and 3Aspl, respectively) are indicated on the right. Quantifications below show the ratios of spliced/unspliced RNAs compared to the ratio for the wild type, which has been set to 1. Standard errors were derived from at least three independent experiments. (G) RPAs of total RNA extracted following transfection with HBoV1 wild-type or mutant constructs, as indicated, using either the (pA)p probe (lanes 1 and 2) or the bGH probe (lanes 3 and 4). The RNA species polyadenylated at (pA)d and using the various cleavage sites at (pA)p are shown on the right. Quantifications below show the ratios of (pA)p to (pA)d. Standard errors are derived from at least three independent experiments. (H) RPAs of total RNA extracted following transfection with HBoV1 wild-type or mutant constructs, as indicated, using the 3A probe. Quantifications below show the ratios of spliced/unspliced RNAs compared to the ratio for the wild type, which has been set to 1. Standard errors were derived from at least three independent experiments.

As previously seen for MVC (14), replacement of (pA)p with the strong bGH polyadenylation site essentially prevented accumulation of RNAs polyadenylated at downstream (pA)d in either the presence or absence of NP1 (Fig. 5G, compare lane 3 [WT/bGH] and lane 4 [mAUG/bGH] to lane 1 [WT′] and lane 2 [mAUG′]). However, in contrast to what was seen for MVC, this substitution in HBoV1 significantly reduced upstream third-intron splicing in either the presence (WT/bGH) or absence (mAUG/bGH) of NP1 (Fig. 5H, compare lanes 3 and 4 to lanes 1 and 2). [Similar levels of hemagglutinin (HA)-tagged NP1 were generated from the constructs containing (pA)p or replaced with bGH (data not shown).] Thus, while debilitation of (pA)p did not abrogate the role of NP1 in upstream intron splicing, replacement of (pA)p with bGH inhibited splicing in an NP1-independent manner. These results suggested that there is a complex interaction between the HBoV1 (pA)p and third-intron signals, which differs substantially from MVC and which is not present when (pA)p is replaced with bGH.

Alternative polyadenylation of HBoV1 pre-mRNAs chooses either (pA)p or (pA)d. Does the choice of polyadenylation site depend upon the proximity of (pA)d to (pA)p? To test this, we made deletions in the intervening region to bring (pA)d closer to (pA)p (constructs diagrammed in Fig. 6A). As can be seen in Fig. 6, not only relative alternative polyadenylation (Fig. 6B, compare lanes 3 to 5 to lanes 1 and 2) but also splicing of the third intron (Fig. 6C, compare lanes 3 to 5 to lanes 1 and 2) was only slightly changed when (pA)d was brought closer to (pA)p. Interestingly, the Δ1 mutation, which joined nt 4851 and 3375 (which also deleted AAUAAA hexamers 4 and 5), showed no additional defects in this context.

FIG 6.

The (pA)p transcription unit can function independently of (pA)d. (A) Diagram of various HBoV1 deletion mutant constructs designed to bring (pA)d closer to (pA)p. Locations of the hexamer core motifs are shown as black ellipses. LEH, left-end hairpin; REH, right-end hairpin. (B) RPAs of total RNA extracted following transfection of 293T cells with the indicated HBoV1 wild-type and mutant constructs using the (pA)p probe. The RNA species polyadenylated at (pA)d and (pA)p are shown on the right [NC (pA)p and CON (pA)p], using the nonconsensus and consensus cleavage sites, respectively. Quantifications, shown below, display the ratios of RNAs cleaved at the canonical sites relative to those that read through. Standard errors were derived from at least three independent experiments. (C) RPAs of total RNA extracted following transfection of 293T cells with the indicated HBoV1 wild-type and mutant constructs using the 3A probe. The probe (237 nt) and the corresponding unspliced and spliced RNA species (3Aunspl [200 nt] and 3Aspl [150 nt], respectively) are indicated on the right. Quantifications below show the ratios of spliced/unspliced RNAs compared to the ratio for the wild type, which has been set to 1. Standard errors were derived from at least three independent experiments. (D) Diagram of various HBoV1 simple transcription unit constructs showing their start and end nucleotide positions. Insertion of an HA tag sequence at the C-terminal region of NP1 is shown as black rectangles, while the black ellipses depict the AAUAAA hexamer motifs, and the black circles designate positions of the 5-proline insertion mutants. (E) RPAs of total RNA extracted following transfection of HBoV1 wild-type and mutant constructs, as indicated, using the (pA)p (lanes 1, 2, and 5) or bGH (lanes 3 and 4) probe. The RNA species polyadenylated at (pA)d and (pA)p are shown on the right; RT indicates RNAs that read through (pA)p, and bGH indicates RNAs cleaved at the bGH polyadenylation site. For lanes 1, 2, and 5, quantifications below show the ratios of (pA)p to (pA)d. Standard errors are derived from at least three independent experiments. For lanes 3 and 4, all detectable RNAs are polyadenylated at bGH. (F) RPAs of total RNA extracted following transfection of 293T cells with HBoV1 wild-type and mutant constructs using the 3A-HA probe. The probe (264 nt) and the corresponding unspliced and spliced RNA species (3Aunspl [227 nt] and 3Aspl [177 nt], respectively) are indicated on the right. Lanes 3 and 4 have lower 3Aunspl (187 nt) and 3Aspl (137 nt) bands due to the mismatch at nt 3201 in constructs skWT/bGH and sk5Pm/bGH generated during the cloning procedure. Quantifications below show the ratios of spliced/unspliced RNAs compared to the ratio for the wild type, which has been set to 1. Standard errors were derived from at least three independent experiments. (G) 293T cells transfected with constructs described in panel F were harvested and analyzed by immunoblotting with anti-HA and anti-β-actin antibodies.

Since the position of (pA)d had little effect on alternative polyadenylation or splicing of the third intron, we chose to examine expression of the (pA)p transcription unit in isolation (from nt 96 to 3536) (diagrammed in Fig. 6D). The NP1s in these constructs were tagged at their carboxyl termini with HA. 3′ cleavage of RNAs generated by the parent construct (skWT) was efficient, and the profile, and accumulated level, of the various products, including the readthrough product (presumably terminating within vector sequences), was similar to that seen with the Rep-Cap construct (Fig. 6E, lane 1). Like the Rep-Cap constructs described above, mutation of NP1 (sk5Pm) increased the level of products accumulated at (pA)p relative to the readthrough product (Fig. 6E, compare lane 2 to lane 1). The isolated (pA)p transcription unit also exhibited efficient splicing of the third intron (Fig. 6F, lane 1 [skWT]) in the presence of NP1, and splicing was dependent upon NP1 (Fig. 6F, lane 2 [sk5Pm]). These results suggested that the isolated transcription unit itself could undergo NP1-dependent alternative RNA processing similarly to the wild-type construct. Similarly to expression from HBoV1 Rep-Cap, replacement of (pA)p with the strong bGH polyadenylation motif (skWT/bGH) led to full cleavage and polyadenylation at this site (Fig. 6E, lanes 3 and 4) as well as inhibition of splicing at the third intron, in the presence or absence of NP1 (Fig. 6G, compare lane 3 [skWT/bGH] to lane 4 [sk5Pm/bGH]). Mutant skWT′, which lacks hexamers 4 and 5, was significantly deficient in polyadenylation at (pA)p, suggesting that the remaining hexamers (hexamers 1 to 3) were deficient in the isolated transcription unit context (Fig. 6E, lane 5). Also, in contrast to the Δ1 mutation described above, mutant skWT′, which also lacked hexamers 4 and 5, was deficient in splicing of the upstream intron in the presence of NP1 (Fig. 6F, lane 5). Thus, (pA)d sequences in Δ1 could overcome a deficiency in NP1-dependent splicing due to the lack of a region within the (pA)p region. This suggested that there was an important interaction between the two motifs in this regard. The levels of HA-tagged NP1 protein generated from the individual constructs are shown in Fig. 6G.

The R/S domain of HBoV1 NP1 is not required for its role in HBoV1 splicing but facilitates accumulation of RNAs extending to the right-hand end of the genome.

Arginine/serine dipeptides (often also found as SR dipeptides) are a signature of the SR and SR-related groups of proteins, many of which are involved in cellular and viral RNA processing (28–30). The arginine-serine (SR) dipeptide motifs, which lie within the intrinsically disordered amino terminus of MVC NP1, are required for splicing of the MVC capsid transcript but not suppression of polyadenylation at MVC (pA)p (14). HBoV1 NP1 also contains a series of SR dipeptides in its disordered amino terminus; however, their number and relative location are somewhat different (diagrammed in Fig. 7A). To examine the importance of the SR motifs for HBoV1 NP1 function, we constructed a series of mutations of these residues (diagrammed in Fig. 7A). As can be seen in Fig. 7B, mutations at residues 11 and 12 (RSm1) or residues 34 to 39 (RSm2), either separately or together (RSm1/2), had no discernible effect on splicing of the proximal upstream third intron (Fig. 7B, compare lanes 3 to 5 to lane 1). This was in contrast to the results for MVC. Also in contrast to results with MVC, the full SR mutant, RSm1/2, but not either individual mutant, was seen to have a modest, but significant, effect on the ability to enhance internal polyadenylation at (pA)p, when assayed by Northern blotting (Fig. 7C, top, compare lanes 3 to 5 to lane 1, and bottom) or an RPA (Fig. 7D, top, compare lanes 3 to 5 to lane 1, and bottom). These results were also evident as a deficiency in capsid protein production, as expected (Fig. 7E, lanes 4 to 6). Together, the NP1 full SR mutant, RSm1/2, displayed separable, albeit subtle, effects on alternative RNA processing and, hence, subsequent capsid mRNA production. However, as seen for MVC NP1 (also diagrammed in Fig. 7F), the HBoV1 SR mutations had no significant effect on NP1's role in suppressing the polyadenylation of MVC (pA)p when the MVC (pA)p region was inserted into the HBoV1 background (RSm/MVCpAp series) (Fig. 7G, compare lanes 3 to 5 to lane 1). Additionally, splicing of the HBoV1 third intron was not significantly affected by HBoV1 NP1 SR mutations in the presence of the MVC (pA)p insertion (data not shown). Although the results presented in Fig. 3 demonstrated that HBoV1 NP1 could affect MVC (pA)p and vice versa, results with HBoV1 NP1 SR mutants revealed additional potential differences between NP1's effects on the HBoV1 and MVC (pA)p regions in alternative RNA processing of their respective pre-mRNAs.

FIG 7.

The R/S domain of HBoV1 NP1 is not required for its role in HBoV1 splicing but facilitates accumulation of RNAs extending to the right-hand end of the genome. (A) Alignment of NP1 amino acid sequences of wild-type pHBoV1 VPHA (WT′), NP1 RS1m VPHA (RS1m), NP1 RS2m VPHA (RS2m), NP1 RS1/2m VPHA (RS1/2m), and MVC NP1 showing the RS dipeptide repeats or their alanine (A) or nonalanine (A) mutations (underlined). The positions of RS dipeptide repeats in MVC are shown in italics. (B) RPAs of total RNA extracted following transfection with HBoV1 wild-type and mutant constructs using the 3A probe. The probe (237 nt) and the corresponding unspliced and spliced RNA species (3Aunspl [200 nt] and 3Aspl [150 nt], respectively) are indicated on the right. Quantifications below show the ratios of spliced/unspliced RNAs compared to the ratio for the wild type, which has been set to 1. Standard errors were derived from at least three independent experiments. (C) Northern blot assays of total RNA extracted following transfection of 293T cells with HBoV1 wild-type and mutant constructs, as indicated, processed as described in Materials and Methods. RNA species polyadenylated at (pA)p (np1/ns4 mRNA) or polyadenylated at (pA)d (vp mRNA and various ns mRNAs) are indicated to the right. Quantification below shows changes of vp mRNA levels relative to the total mRNA level for the wild type (lane 1) and the mutants, as indicated and as described in the text. The ratios for the WT (lane 1) are arbitrarily set as 100%, and the relative ratios in other lanes were normalized to the value for the WT. Errors bars indicate the standard deviations from at least three individual experiments. (D) RPAs of total RNA extracted following transfection with HBoV1 wild-type and mutant constructs, as indicated, using the (pA)p probe. The RNA species polyadenylated at (pA)d and (pA)p are shown on the right. Quantifications, shown below, display the ratios of RNAs cleaved at either the canonical sites [Con/(pA)d] (left) or noncanonical sites [NC/(pA)d] (right) relative to those that read through, for each of the mutants. Standard errors were derived from at least three independent experiments. (E) 293T cells transfected with constructs described in panels B to D were harvested and analyzed by immunoblotting with anti-HA and anti-β-actin antibodies. (F) Diagram of various HBoV1 (pA)p replacement mutant constructs, as indicated and as described in the text. (G) RPAs of total RNA extracted following transfection of HBoV1 replacement mutant constructs, as indicated, using the MVC/(pA)p probe. The RNA species polyadenylated at (pA)d and (pA)p are shown on the right. Quantifications below show the ratios of (pA)p to (pA)d. Standard errors are derived from at least three independent experiments.

DISCUSSION

All parvoviruses have subtle early-to-late expression shifts; however, they regulate access to their capsid genes in different ways. Adeno-associated virus type 2 (AAV2) and MVM use transactivation of a capsid gene promoter to generate capsid protein-encoding mRNAs (31, 32). Other parvoviruses, like AAV5, B19, and goose parvovirus, have an internal poly(A) site, which typically lies within an intron, so excision of this intron allows extension of the spliced RNA into the capsid gene (33–38). For the bocaviruses, a potent internal polyadenylation motif, (pA)p, is retained in the capsid protein-encoding cytoplasmic mRNA (8, 13, 23). Thus, this signal must be suppressed to allow the accumulation of mRNAs that extend into the capsid gene and allow production of the capsid proteins.

NP1, unique to the Bocaparvovirus genus of the Parvovirinae, has been shown to be essential for the accumulation of MVC capsid mRNAs and capsid protein by both allowing for suppression of the internal polyadenylation site (pA)p and facilitating splicing of the adjacent upstream third intron (14). These effects allow both extension into the capsid gene and splicing of the viral pre-mRNA that correctly registers the capsid gene open reading frame. The role that NP1 proteins of other members of the Bocavirus genus play in alternative RNA processing had not been clearly established, and so it was important to determine how HBoV1 NP1 functioned in this regard. HBoV1 NP1 is 219 amino acids in length and shares only 46% identity and 62% similarity with the 185-amino-acid MVC NP1 protein (7). In addition, fine characterization of the HBoV1 internal polyadenylation site, and its interaction with the adjacent upstream third intron, was also necessary to complete this analysis. Our analysis has both identified similarities and uncovered important differences in the two systems.

Two groups have previously investigated the cleavage and polyadenylation sites of HBoV1 (pA)p. These studies have suggested that RNAs cleaved and polyadenylated internally at HBoV1 (pA)p primarily used cleavage site A and either site C or D (8) or, alternatively, site F (23). Our analysis indicates that cleavage and polyadenylation at this region are more complex than previously appreciated. We have identified a total of 6 cleavage sites in the region, 5 of which (sites B to F) make up the bulk of internally polyadenylated HBoV1 RNA. Interestingly, two of these sites (sites E and F) are not governed by CPSF-binding AAUAAA hexamer motifs. The two noncanonical sites made up close to 30% of the mRNA cleaved within (pA)p. The noncanonical sites were first identified by oligo(dT)-primed 3′ RACE, verified by Northern assays not susceptible to nuclease artifacts, and also confirmed by RNase protection assays. Site F, cleaved within an A-rich region at nt 3505, contains a significant run of A residues at its 3′ end; however, it is not further polyadenylated following cleavage. How this RNA is generated and whether or not these RNAs encode an HBoV1 protein are under investigation. Preliminary data have shown that these RNAs were transported efficiently to the cytoplasm.

Importantly, we have demonstrated that HBoV1 NP1 was indeed necessary for the accumulation of the viral capsid mRNAs. Although the magnitude of the effect was smaller than what was seen for MVC NP1, HBoV1 NP1 allowed both suppression of the internal (pA)p and splicing of the adjacent upstream third intron, both of which are necessary to properly express the capsid proteins. Interestingly, NP1 affected suppression of only the (pA)p-cleaved RNAs that were governed by canonical AAUAAA hexamers; the noncanonically cleaved RNAs were not affected. Whether this implicates an interaction between NP1 and factors associated with cellular factors that associate with the AAUAAA motif, like CPSF, is under investigation. It is interesting that proline insertion mutagenesis within the ordered region of NP1 (5×Pm) affected both polyadenylation and upstream splicing, while similar insertion mutagenesis within the N-terminal disordered region (5×Pm′) had an effect (albeit modest) on splicing but not polyadenylation. Together with the observation that the NP1 termination mutant, which expressed only the N-terminal 28 amino acids of NP1 (mTAG), retained the ability to significantly enhance splicing of the third intron, this suggested that this region is likely involved in this aspect of NP1 function. Additionally, little role for the downstream polyadenylation site (pA)d was observed in regulation at (pA)p or third-intron splicing.

Although we found that HBoV1 NP1 could suppress MVC (pA)p when it was substituted in place into HBoV1, and vice versa, distinct differences between the two systems were uncovered, which suggested that the complex interaction between alternative polyadenylation at (pA)p and splicing of the adjacent upstream third intron is not conserved. We have found that within a Rep-Cap-based construct, mutation of the 5 AAUAAA hexamers and the A-rich site in HBoV1 (pA)p did not substantially decrease splicing at the upstream third intron in the presence of NP1, and under these conditions, NP1 was still required for third-intron splicing. This indicated that in the full-length HBoV1 transcription unit driven by the authentic HBoV1 left-end P5 promoter, NP1's effect on third-intron splicing remained primarily (pA)p independent. This was in contrast to what has been seen by others (19). In those experiments, HBoV1 NP1 was shown to enhance expression of the HBoV1 capsid proteins when the viral P5 promoter was replaced with the CMV IE promoter, and subsequent knockout of the internal polyadenylation site (pA)p abrogated the need for NP1 (19), implying that the two events were not independent. It is not clear why these particular results differ; however, RNA processing can be dependent upon the promoter driving the pre-mRNA being processed (39, 40), and in these experiments, the transcription units were driven by different promoters. We have previously shown that processing of the parvovirus adeno-associated virus pre-mRNA is dependent upon the promoter that directs its transcription (32).

Although the HBoV1 third intron splicing signals are not fully consensus, their sequence predicts it to be strong. Therefore, it is unclear why it is not functional in the absence of NP1. In the wild-type context, it is not likely that (pA)p alone governs third-intron splicing because in either the presence or absence of (pA)p, in Rep-Cap-based or minimal constructs, the third intron was functional only in the presence of NP1. However, the observation that third-intron splicing was reduced for HBoV1 when bGH replaced (pA)p, in both the presence and absence of NP1, suggested that for HBoV1, downstream signals could play an inhibitory role in upstream splicing. Clearly, the interaction between the two signals is complex and warrants further investigation. In preliminary experiments, we have found that when the third-intron acceptor sequence was made fully consensus, it still could not overcome the need for NP1 (data not shown). Our experiments in this regard highlight another basic difference between alternative RNA processing of MVC and HBoV1 pre-mRNAs: when bGH was substituted for (pA)p in the MVC background, upstream splicing was not affected and still facilitated by MVC NP1 (14).

Interestingly, although mutants of (pA)p still allowed efficient NP1-dependent third-intron splicing, significantly less NP1 was made under these conditions. This suggested that either only modest levels of NP1 are sufficient to facilitate third-intron splicing or splicing factors are attracted to the third intron better in the absence of (pA)p. This observation may also prove important in understanding the role of NP1 in alternative RNA processing.

Finally, the SR dipeptide repeats in HBoV1 NP1 seem to function differently than they do in MVC NP1. The MVC NP1 SR motifs, which lie within the intrinsically disordered amino terminus of MVC NP1, are required for splicing of the MVC capsid transcript but not suppression of polyadenylation at MVC (pA)p (14). HBoV1 NP1 also contains a series of SR dipeptide in its disordered amino terminus; however, their number and relative location are somewhat different. Mutation of the HBoV1 NP1 SR motifs together had no discernible effects on third-intron splicing in our assays but had a modest but reproducible effect on its ability to suppress (pA)p. SR motifs typically mediate interactions between RNA processing factors (28–30). This highlighted another difference between the two NP1 proteins that should allow comparative experimentation into their mechanism of action.

Together, these experiments revealed similarities and also subtle, but significant, species-specific differences in how NP1 orchestrated alternative RNA processing and negotiated the complex connection between the (pA)p motifs and the upstream introns.

It has recently been reported that the C-terminal regions of three of the MVC NS proteins are generated from mRNAs spliced at the third intron (17). Therefore, their expression is facilitated by MVC NP1 (17). For HBoV1, the C terminus of the viral NS proteins is determined by splicing of the second intron of their encoding mRNAs, which is independent of NP1 (20). Thus, in contrast to MVC, HBoV1 NP1 is not predicted to govern their expression. The requirement of NP1 for expression of MVC NS proteins may impose a constraint on the interaction between MVC alternative polyadenylation and splicing not found for HBoV1, perhaps consistent with the differences in the interactions between the signals that we observe.

Our analysis of expression from mutants within the (pA)p region also indicated that, as previously seen for MVC, mRNAs cleaved internally at (pA)p were the primary source for NP1 expression. Thus, similar to the suggested model for MVC, it is most likely that HBoV1 RNAs initially use the internal (pA)p site, and as these RNAs accumulate and program the expression of NP1, NP1 can facilitate suppression of (pA)p and splicing of the third intron to allow proper access to the viral capsid gene. However, as described above, while MVC NP1 also participates in the expression of three MVC NS proteins, the role of HBoV1 NP1 seems restricted to the generation of the viral capsid proteins.

MATERIALS AND METHODS

Cells and transfections.

Human 293T cells were propagated, as previously described (41), in Dulbecco's modified Eagle medium (DMEM) with 10% fetal calf serum. Transfections were performed with 293T cells using LipoD293 transfection reagent (SignaGen Laboratories, MD), and cells were maintained at 37°C in 5% CO₂ for 36 to 48 h.

RNase protection assays.

Total RNA was extracted using the TRIzol reagent (Invitrogen), and RNase protection assays (RPAs), using the indicated probes, were performed on 20 μg total RNA as previously described (42). RPA signals were analyzed with the Typhoon FLA9000 system and quantified with Multi Gauge software (GE). Relative molar ratios of individual RNA species were calculated after adjustment for the number of ³²P-labeled uridines in each protected fragment as previously described (41). Probes utilized in this study are as follows. The HBoV1 (pA)p probe spanned HBoV1 nucleotides (nt) 3250 to 3536 around the proximal polyadenylation sites. Appropriate homologous probes were used when the HBoV1 proximal polyadenylation region was replaced by MVM heterologous sequences or when the AAUAAA hexamers were individually or jointly mutated. The 10A(pA)p probe included an additional 10 adenines downstream of nt 3512 in the HBoV1 proximal polyadenylation region and was used to analyze the poly(A) tail. The wild-type 3A or m3A probe (nt 3041 to 3241) was used to analyze splicing across the HBoV1 third intron. The 3A-HA probe (nt 3041 to 3241) contains an HA tag sequence (27 nt) at nt 3162 and was also used to analyze splicing across the HBoV1 third intron. The MVC/(pA)p probe, which spanned MVC nt 3107 to 3333, was designed as previously described (14). The bGH probe (bovine growth hormone polyadenylation sequence) (nt 1021 to 1263) was amplified from pcDNA3.1(+) (Life Technologies, CA). All the indicated probe fragments were cloned into pGEM-3Z (Promega, WI) with restriction enzymes EcoRI and XbaI.

Selection of poly(A)⁺ RNA by oligo(dT)-cellulose chromatography.

Dynabeads oligo(dT)₂₅ (catalog number 610.11; Invitrogen) was used for isolation of highly purified poly(A)⁺ RNA from total RNA, and the procedures were performed according to the manufacturer's instructions. Briefly, 100 μg total RNA extracted from cells at 48 h posttransfection was incubated with the treated beads. After bead binding and washing, mRNA was eluted with heated 10 mM Tris-HCl elution buffer and used for an RNase protection assay.

Immunoblot analysis.

Transfections were conducted with the indicated constructs described below and in the figure legends. 293T cells were harvested and lysed at 48 h posttransfection with WL lysis buffer (50 mM Tris [pH 8.0], 400mM NaCl, 0.5% NP-40, 10% glycerol, 1 mM EDTA), which contained protease and phosphatase inhibitors. SDS-PAGE and Western blotting were performed as previously described (13). Anti-HA, anti-Flag, anti-β-actin, or antitubulin antibodies were used to detect the HA-tagged NP1, HA-tagged VP1/2/3, Flag-tagged NP1, actin, and tubulin proteins where indicated. Anti-β-actin, antitubulin, anti-HA, and anti-Flag antibodies were purchased from Sigma-Aldrich.

Northern blotting.

293T cells were transfected with pHBoV1 WT (WT) and its derivatives as described below and in the figure legends. Total RNA were extracted at 48 h posttransfection and assayed by Northern blotting as described previously (21). Briefly, 10 μg total RNA of each sample was resolved in formaldehyde–1.4% agarose gels and run for 16 h at 35 mA. After staining, washing, and soaking, the gel was transferred to nitrocellulose overnight. Blots were baked and hybridized with a randomly primed radiolabeled BstBI/EcoRI-digested genomic fragment (nt 2704 to 4398) of pHBoV1 WT to detect all the viral mRNAs.

Poly(A) tail length assay.

293T cells were grown in 60-mm dishes and transfected, as indicated, with the HBoV1 Rep-Cap construct (WT) or various HBoV1 mutant constructs described below and in the figure legends. Forty-eight hours later, total RNA was extracted using TRIzol reagent and treated with DNase I (Ambion). A poly(A) tail length assay was performed as previously described (43). Briefly, cDNA was synthesized from the indicated total RNA sample with an oligo(dT) primer/adapter. The cDNA was used as the template for PCR with two different 5′-end-labeled target-specific primers (Primer1-F [HBoV1 nt 3250 to 3268] [5′-ATTAGAATTCCGCTGATCGCGCTGCTCAA-3′] and Primer2-F [HBoV1 nt 3387 to 3410] [5′-ATTAGAATTCTTATTGGATCCAGTTTTTTTAAAA-3′]) and an unlabeled oligo(dT) primer/adapter (5′-GGGGATCCGCGGTTTTTTTTTT-3′). After PCR, samples were resolved in an 8% polyacrylamide gel with 1× Tris-borate-EDTA (TBE) at 50 W for 2.5 h and exposed to a phosphorimaging screen overnight.

3′ RACE assay.

Rapid amplification of cDNA ends (RACE) was conducted using the 3′-RACE system (3′-Full Race core set; TaKaRa) according to the manufacturer's protocol. The forward primer Primer1-F (HBoV1 nt 3251 to 3268) (5′-ATTAGAATTCCGCTGATCGCGCTGCTCAA-3′) was used for the PCR amplification reaction.

Plasmid constructs.

The generation of the multiple plasmid constructs used in this study was done as follows.

(i) HBoV1 wild-type or mutant constructs.

pHBoV1 WT (WT), the wild-type HBoV1-nonreplicating construct pHBoV1 m630, was kindly provided by Jianming Qiu, University of Kansas Medical Center (GenBank accession number DQ000496).

All subsequent numbering is positioned to align with pIHBoV1 (see reference 9 and references therein) (GenBank accession number JQ923422).

pHBoV1 NP1 5×Pro 2820 (5×Pm) and NP1 5×Pro 2745 (5×Pm′) are two structurally defective mutants of HBoV1 NP1, generated by introducing five consecutive proline substitutions into the NP1 ORF at nt 2820 and 2649.

pHBoV1 NP1 mTAG (mTAG) has an NP1 translation termination mutant generated at nt 2588 as previously described (14).

pHBoV1 NP1 mAUG (mAUG) has the NP1 initiating codon AUG at nt 2521 mutated to TAA.

pHBoV1 m1 (m1), pHBoV1 m2 (m2), pHBoV1 m3 (m3), pHBoV1 m4 (m4), pHBoV1 m5 (m5), pHBoV1 m1/2 (m1/2), and pHBoV1 m12345J (mALL) have the five AAUAAA core elements within HBoV1 (pA)p (nt 3250 to 3536) disrupted by a GCA substitution in these constructs (see Fig. 1A for diagram).

pHBoV1 mAAA (mF) and pHBoV1 mTT (mTT) have the A-rich motif (nt 3494 to 3519) or the T-rich element (nt 3400 to 3406) within (pA)p disrupted, as shown in Fig. 1A and 2A.

pHBoV1 m12345J+mAAA (mALLmF) has the five AAUAAA core elements within HBoV1 (pA)p (nt 3250 to 3536) disrupted by a GCA substitution in the pHBoV1 mAAA (mF) background.

pHBoV1 MVM_in (MVM_in) has the MVMp heterologous sequences (nt 3501 to 3711) amplified from the MVMp infectious clone (GenBank accession number NC001510) cloned into (pA)p (HBoV1 nt 3184 to 3394) in the wild-type HBoV1 background.

pHBoV1 MVM_in mAAA (MVM_inmF) has the MVMp heterologous sequence at nt 3501 to 3711 cloned into (pA)p (HBoV1 nt 3184 to 3394) within the pHBoV1 mAAA (mF) mutant background.

HB/MVCpAp (WT mAUG) has the MVC (pA)p region (nt 3107 to 3316) amplified from pIMVC (13) and cloned into the HBoV1 (pA)p region (nt 3280 to 3517) in the pHBoV1 WT or pHBoV1 NP1 mAUG background using restriction enzymes BclI and EcoNI.

pHBoV1 m3A (m3A) has the splice acceptor site of the third intron of HBoV1 (nt 3090) changed as an in-frame G-to-A substitution.

pHBoV1 Δ3375/4851 (Δ1) was generated by deletion of HBoV1 genomic fragments from nt 3375 to 4851 in pHBoV1 WT, digested with MfeI, treated with Klenow fragment (NEB), and ligated with T4 DNA ligase.

pHBoV1 Δ3815/4049 (Δ2) was generated by deletion of HBoV1 genomic fragments from nt 3815 to 4049 in pHBoV1 WT, digested with NcoI, treated with Klenow fragment (NEB), and ligated with T4 DNA ligase.

pHBoV1 Δ3815/4618 (Δ3) was generated by deletion of HBoV1 genomic fragments nt 3815 to 4618 in pHBoV1 WT, digested with NcoI/KpnI, treated with Klenow fragment (NEB), and ligated with T4 DNA ligase.

(ii) HBoV1 wild type or mutants with VP-HA tagged constructs.

pHBoV1 VPHA (WT′) was created in the wild-type HBoV1-nonreplicating construct backgrounds by insertion of an HA tag sequence at nt 4128 in the capsid ORFs.

pHBoV1 NP1 mAUG VPHA (mAUG′) was created in the AUG translation initiation mutant background by insertion of the HA tag sequence at nt 4128 in capsid ORFs.

pHBoV1 m12345J+mAAA VPHA (mALLmF′) was created in the HBoV1 hexamer and AAA region mutant (mALLmF) backgrounds by insertion of the HA tag sequence at nt 4128 in capsid ORFs.

pHBoV1 NP1 mAUG m12345J+mAAA VPHA (mAUGmALLmF′) was constructed by inserting NP1 mAUG initiation mutations in the pHBoV1 m12345J+mAAA VPHA (mALLmF′) backgrounds.

To make pHBoV1 NP1 RSm1 VPHA (RSm1), NP1 RS dipeptides 11 and 12 were mutated to AA (nt 2536 to 2541; CGC TCC to GCA GCT); to make pHBoV1 NP1 RSm2 VPHA (RSm2), RS dipeptides 34 to 39, RSRSRS, were mutated to KFANVN (nt 2605 to 2622; AGG AGC AGG AGC CGC AGC to AAA TTT GCT AAT GTT AAC); and pHBoV1 NP1 RSm1/2 VPHA (RSm1/2) contains both RS1 and RS2 mutations. Mutations are diagrammed in Fig. 7A.

The RSm/MVCpAp series constructs pHBoV1 NP1 RSm1/MVCpAp VPHA (RSm1/MVCpAp), pHBoV1 NP1 RSm2/MVCpAp VPHA (RSm2/MVCpAp), and pHBoV1 NP1 RSm1/2/MVCpAp VPHA (RSm1/2/MVCpAp) were made by inserting the MVC (pA)p region (nt 3107 to 3316) amplified from pIMVC (13) and cloned into the HBoV1 (pA)p region (nt 3280 to 3517) in the pHBoV1 NP1 RSm1 VPHA, pHBoV1 NP1 RSm2 VPHA, or pHBoV1 NP1 RSm1/2 VPHA background using restriction enzymes BclI and EcoNI.

pHBoV1 WT/bGH VPHA (WT/bGH) and pHBoV1 mAUG/bGH VPHA (mAUG/bGH) were made by inserting the bovine growth hormone polyadenylation sequence (nt 1021 to 1263), synthesized by IDT, into the pHBoV1 VPHA or pHBoV1 NP1 mAUG VPHA construct in place of HBoV1 proximal polyadenylation elements using restriction enzymes BclI and EcoNI.

(iii) HBoV1 wild type or mutants with NP1-HA/Flag-tagged constructs.

pHBoV1 NP1HA (WTNP1HA) and pHBoV1 5×Pro 2820 NP1HA (5×PmNPHA) were created in the pHBoV1 WT or pHBoV1 5×Pro 2820 background by insertion of an HA tag sequence (5′-TACCCATACGATGTTCCAGATTACGCT-3′) into the NP1 C terminus at nt 3162.

skWTNPHA/pAp (skWT) and sk5×PmNPHA/pAp (sk5Pm) were generated first by amplification of HBoV1 nt 96 to 3536 from pHBoV1 NP1HA and pHBoV1 5×Pro 2820 NP1HA, followed by cloning into pBlueScript II SK(+) (Life Technologies, CA).

skWTNPHA/bGH (skWT/bGH) and sk5×PmNPHA/bGH (sk5Pm/bGH) were generated by insertion of the bovine growth hormone polyadenylation element (nt 1021 to 1263) into the skWTNPHA/pAp or sk5×PmNPHA/pAp construct background.

skWTNPHA3375/pAp NPHA (skWT′) was generated by deletion of the HBoV1 genomic fragment from nt 3375 to 5395 in the pHBoV1 NP1HA construct, using enzymes MfeI and XbaI, followed by treatment with Klenow fragment (NEB) and ligation with T4 DNA ligase.

pHBoV1 NP1 Flag (WT, mAUG, 5×Pm, 5×Pm′, mTAG, mALL, and mALLmF) was created in the indicated pHBoV1 WT, pHBoV1 NP1 5×Pro 2820, pHBoV1 NP1 5×Pro 2745, pHBoV1 NP1 mTAG, pHBoV1 m12345J, and pHBoV1 m12345J+mAAA construct backgrounds, followed by insertion of the Flag tag sequence (5′-GACTACAAAGACGATGACGACAAG-3′) into nt 2704 (BstBI) of the HBoV1 genome.

(iv) Other constructs.

pIMVC and pIMVC NP1 5×Pm (5×PmMVC) were generated as previously described (13).

To generate MVC/HBpAp (WT or 5×), the HBoV1 proximal polyadenylation region (nt 3077 to 3536) was amplified from pHBoV1 WT and cloned into pIMVC SK in place of the MVC proximal polyadenylation region from nt 3111 to 3353 as previously described (14).