Capsid Proteins Are Necessary for Replication of a Parvovirus

Abstract

Despite tight genetic compression, viral genomes are often organized into functional gene clusters, a modular structure that might favor their evolvability. This has greatly facilitated biotechnological developments such as the recombinant adeno-associated virus (AAV) systems for gene therapy. Following this lead, we endeavored to engineer the related insect parvovirus Junonia coenia densovirus (JcDV) to create addressable vectors for insect pest biocontrol. To enable safer manipulation of capsid mutants, we translocated the nonstructural (ns) gene cluster outside the viral genome. To our dismay, this yielded a virtually nonreplicable clone. We linked the replication defect to an unexpected modularity breach, as ns translocation truncated the overlapping 3' untranslated region (UTR) of the capsid transcript (vp). We found that the native vp 3' UTR is necessary for high-level VP production but that decreased expression does not adversely impact the expression of NS proteins, which are known replication effectors. As nonsense vp mutations recapitulate the replication defect, VP proteins appear to be directly implicated in the replication process. Our findings suggest intricate replication-encapsidation couplings that favor the maintenance of genetic integrity. We discuss possible connections with an intriguing cis-packaging phenomenon previously observed in parvoviruses whereby capsids preferentially package the genome from which they were expressed. IMPORTANCE Densoviruses could be used as biological control agents to manage insect pests. Such applications require an in-depth biological understanding and associated molecular tools. However, the genomes of these viruses remain difficult to manipulate due to poorly tractable secondary structures at their extremities. We devised a construction strategy that enables precise and efficient molecular modifications. Using this approach, we endeavored to create a split clone of Junonia coenia densovirus (JcDV) that can be used to safely study the impact of capsid mutations on host specificity. Our original construct proved to be nonfunctional. Fixing this defect led us to uncover that capsid proteins and their correct expression are essential for continued rolling-hairpin replication. This points to an intriguing link between replication and packaging, which might be shared with related viruses. This serendipitous discovery illustrates the power of synthetic biology approaches to advance our knowledge of biological systems.

Keywords: capsid; densovirus; replication; synthetic biology.

FIG 1

Initial design for the construction of a nonpropagatable JcDV clone. (A) Functional architecture of the infectious JcDV clone pWT. ns (blue) and vp (orange) gene blocks are in a tail-to-tail orientation. Each produces a single transcript (blue and orange lines) from a promoter partly located in the inverted terminal repeats (ITRs) (red). The ns transcript is alternatively spliced, as shown. Due to the locations of ns and vp polyadenylation signals and sites (blue and orange small rectangles and triangles, respectively), the two transcripts overlap over a complementary region of 60 nucleotides (top, gray arrow). (B) Architecture of the nonpropagatable clone pDef-nf. The ns block (from the start of ns3 to the end of the ns transcript) was cloned out of the genome under the control of a strong baculovirus promoter and replaced by a reporter block of the same size comprising the mVenus fluorescent protein gene and a neomycin resistance gene. In this process, VP’s 3′ UTR was truncated by 44 nt to avoid recombination-prone sequence duplication. (C) Continued infection upon transfection of the pWT clone. Upon packaging of the WT genome, the produced virions can propagate in naive cells, leading to the uncontrolled spread of the infection. (D) Transfection of the nonpropagatable pDef-nf clone should not permit downstream infection. trans complementation by NS function enables initial genome replication in transfected cells. Upon packaging, genomes devoid of NS functions cannot replicate in naive cells, containing the spread of the virus.

FIG 2

Amended versions of the nonpropagatable clone function as intended. The GFP signal from various Ld652Y cell populations was evaluated by epifluorescence microscopy. The first column shows Ld652Y cells at 3 days p.t. The second column shows naive cells 3 days after transduction with 100 μl of the clarified supernatants obtained after lysing transfected cells at 7 days p.t. The third column shows naive cells 3 days after transduction with 100 μl of the clarified supernatants obtained from lysing first-round-infected cells at 7 days p.t. The top diagram recalls the expected behavior of the nonpropagatable clone (Fig. 1D). No fluorescence could be observed at the first round of infection with the original pDef-nf construct, suggesting no or defective virion production from this construct (first row). Based on ulterior results (Fig. 3), the amended versions pDef-3p91 and pDef-3p115 were constructed, which exhibit infection commensurate with observed viral replication levels (middle column, middle and bottom rows) (Fig. 3). Second rounds of infection from these virions show no signs of transduction, as expected from nonpropagatable clones.

FIG 3

Impaired replication of the nonfunctional clone is rescued by vp complementation. (A) Schematic diagram of genetic constructs. ITRs are shown in red, vp is in orange, ns is in blue, and the reporter cassette is in green. Sizes are not to scale. (B) Global qPCR measurement of viral replication pinpoints the region necessary and sufficient for rescue. Shown are fold changes in genome quantities between 16 h and 72 h after transfection of Ld652Y cells with constructs depicted in panel A (the qPCR target site is marked by a gray arrowhead). The original design of the nonpropagatable genome (pDef-nf) (top) shows an almost complete absence of replication. Cotransfections with both the WT genome (pWT) and an ITR-less derivative (pWT-noITR) complement this phenotype, while a fragment restricted to the ns gene block does not (pNS). A large vp block comprising 115 nt downstream of the vp coding sequence (pVP-3p115) effects a high rescue level, while blocks restricted to 49 nt (pVP-3p49) or 75 nt (pVP-3p75) of downstream context do not rescue at all. Extension of vp’s native downstream context from 16 nt (pDef-nf) to 91 nt (pDef-3p91) or 115 nt (pDef-3p115) yields increasingly more functional nonpropagatable clones that function as originally intended (Fig. 2). The introduction of punctual nonsense (*, TAT 542 TAG) (pDef-3p115-S) and frameshift (**, AAT 687 AA–) (pDef-3p115-F) mutations (boldface type) that result in the production of truncated VP proteins (Fig. 6D) abolishes viral replication. (C) Apparent stalling of rolling-hairpin replication. Shown is a Southern blot probing vp sequences on DpnI-treated DNA extracted 96 h after transfection of Ld652Y cells with the indicated plasmids. Monomer-length (M) and dimer-length (D) species that are typical rolling-hairpin intermediate products are indicated by arrowheads. pDef-nf produces faint but detectable levels of intermediates. The functional derivatives pDef-3p115 and pDef-3p91 support elevated replication levels but do not restore wild-type performance. The discrepancy in the relative performances of pWT between panels B and C is apparent: panel B reports the rescue of pDef-nf replication by pWT, while panel C reports the replication of the pWT clone itself. The vp point mutants pDef-3p115-S and pDef-3p115-F also produce small quantities of replication intermediates. As the introduced mutations involve minimal DNA sequence modification and are separated by 435 nt, these data strongly support the link between VP proteins and replication.

FIG 4

Antisense transcript interactions do not explain vp’s impact on replication. (A) Alignments of WT and mutated transcripts in the overlapping region between ns and vp. WT ns and vp (reversed) transcript sequences are shown in the middle, with ambisense complementarity marked by vertical bars. ns and vp coding sequences are in blue and orange, respectively. Complementary mutations (7 or 12) were independently introduced into distinct ns and vp constructs. Mutations (highlighted in red) correspond to synonymous ns1 substitutions and avoid putative polyadenylation signals (green). Dots indicate sequence identity. (B) Mutations in ns and vp transcripts show moderate but complex effects on replication. Shown are fold changes in genome abundances as measured by qPCR between 16 and 72 h after transfection of Ld652Y cells with the indicated plasmid pairs and the replication reporter pITR-R. Error bars show standard errors across 3 biological replicates. While complementary mutations have little net impact on replication, mismatches show largely different effects on ns or vp transcripts. (C) Contrasting effects of complementary mutations on ns and vp. Shown are the marginal mean fold changes in genome abundances over the 3 assays involving each construct. Error bars show standard errors over 9 biological replicates. In line with the antisense regulation of ns, mutations affecting the ns transcript tend to increase replication. Symmetrical mutations in the vp transcript show opposite effects, hinting at a distinct mechanism. (D) Transcript mismatches modulate mutational effects. Shown are absolute values of interactions between constructs against the number of mismatches between ns and vp transcripts. Interactions represent the differences between measured fold changes and linear predictions based on the main mutation effects plotted in panel C. Modulation by transcript mismatches confirms a role for antisense regulation.

FIG 5

The full vp 3′ UTR is required for replication and specified by its immediate downstream context. (A) Impaired replication is associated with the truncated 3′ UTR. Shown are representative chromatograms of the sequence downstream of vp’s stop codon as obtained by 3′ RACE. The sequence corresponding to each construct is shown at the top. Mismatches (red background) correspond to the sequence context of the plasmid backbone. A construct comprising 115 nt of native context (pVP-3p115) produces the same transcript as the WT clone. A construct truncated just after the poly(A) site (pVP-3p75) produces shorter transcripts that lack the last 17 nt (bold) of the native 3′ UTR and do not support replication (data not shown). (B) Identification of a potential downstream sequence element (DSE). The inset shows the %GT content calculated for sliding windows of 15 nt over JcDV’s genome. Blue arrowheads mark the 18 peaks with a %GT content of ≥80%. The red region corresponds to the sequence shown in panel A and is enlarged in the main plot, where points show %GT contents of sliding windows centered at these positions. The %GT peak located immediately downstream of the poly(A) site represents a potential DSE (solid black line) that is largely attenuated in pVP-3p75 (dashed gray line).

FIG 6

Alteration of vp’s 3′ UTR impacts VP but not NS expression. (A) The shorter 3′ UTR decreases the vp transcript abundance. Shown are relative transcript abundances as quantified by RT-qPCR 12 h after transfection of Ld652Y cells with the indicated plasmids. A functional version of the nonpropagatable clone (pDef-3p115) produces the native vp transcript that accumulates to a 100-fold-higher level than truncated transcripts from the nonfunctional clone (pDef-nf). Abundances of ns transcripts are identical for both constructs. (B and C) The shorter 3′ UTR decreases VP protein abundance but has little impact on NS expression. A cotransfection strategy between nonreplicable ns and vp blocks was adopted to quantify potential expression feedbacks in the absence of variable gene dosages due to differential genome replication. (B) Distributions of cellular fluorescence intensities (left), as quantified by immunofluorescence microscopy for VP (top), NS1 (middle), and NS2 (bottom) proteins 3 days after transfection of Ld652Y cells with different combinations of vp and ns blocks, as color-coded in the key (top). Data points represent cell frequencies for each percentile of the fluorescence range. For visual clarity, points are underlaid with locally estimated scatterplot smoothing (LOESS) regression smoothers. The 99th percentile of control cells (gray background) is used as a threshold to quantify the total fluorescence signals (insets). (C) Corresponding Western blot marked by primary antibodies against VP (top), NS1 (middle top), NS2 (middle bottom), and α-tubulin (bottom), as shown. Protein species are indicated by arrows on the right, and apparent molecular weights are shown on the left. Samples were diluted 20-fold less in the last two lanes for comparison purposes. Consistent with the lower abundances in panel A, truncated vp transcripts led to strongly reduced VP protein levels in both the absence and presence of ns genes. Neither NS1 nor NS2 (bottom) abundances are dramatically impacted by differential VP production, suggesting that VP is not involved in upregulating NS expression. (D) Protein productions from potentially replicable viral genome clones. Compared to the WT, the functional nonpropagatable clones (pDef-3p115 and pDef-3p91) show decreased VP production that is consistent with lower replication levels (Fig. 3C) but higher NS production that might reflect expression from a strong heterologous promoter (Fig. 1B). vp coding sequence point mutants (pDef-3p115-S and pDef-3p115-F) produce truncated VP proteins that are unable to support replication (Fig. 1B and C), leading to lower viral protein production. Western blots were obtained from denatured total protein extracted from Ld652Y cells 3 days after transfection with different plasmids, as indicated at the top of each lane. Unless stated otherwise, all samples were diluted to the same extent prior to loading. Images were acquired using a chemiluminescence imager and have been cropped to the signal of interest and individually adjusted for brightness and contrast. afu, arbitrary fluorescence units.

FIG 7

Potential structural binding site for regulatory proteins in the 3′ UTR of the vp transcript. (A) Secondary structure prediction for the 3′ UTR of vp’s native transcript. The 75-nt-long 3′ UTR was appended with a 15-nt poly(A) tail and folded. Bases are colored according to structural status. The sequence absent from early-terminating transcripts that are nonfunctional for replication is highlighted with a dotted line: 8 nt are involved in the secondary structure. (B) Secondary structure prediction for the 3′ UTR of vp’s truncated transcript (from, e.g., p118) (Fig. 4). The lower stem of the structure can no longer be formed. (C and D) Secondary structure prediction from vp 3′-UTR mutants originally constructed to test antisense regulation. The 3′ UTR is assumed to have the WT length because the cryptic poly(A) site is mutated, although this was not verified experimentally. Mutations (7 and 12 in pVP-3p115-7m [C] and pVP-3p115-12m [D], respectively) result in increasingly altered secondary structures in both strength and position, which may explain the increasingly negative impact on replication associated with these variants (Fig. 4). Secondary structures were predicted using the Vienna package RNAfold Web server and rendered with FORNA (49). Black triangles mark the WT poly(A) site, and red triangles mark the cryptic poly(A) site, which is further annotated with an asterisk if mutated.