The Unstructured Paramyxovirus Nucleocapsid Protein Tail Domain Modulates Viral Pathogenesis through Regulation of Transcriptase Activity

ABSTRACT

The paramyxovirus replication machinery comprises the viral large (L) protein and phosphoprotein (P-protein) in addition to the nucleocapsid (N) protein, which encapsidates the single-stranded RNA genome. Common to paramyxovirus N proteins is a C-terminal tail (Ntail). The mechanistic role and relevance for virus replication of the structurally disordered central Ntail section are unknown. Focusing initially on members of the Morbillivirus genus, a series of measles virus (MeV) and canine distemper virus (CDV) N proteins were generated with internal deletions in the unstructured tail section. N proteins with large tail truncations remained bioactive in mono- and polycistronic minireplicon assays and supported efficient replication of recombinant viruses. Bioactivity of Ntail mutants extended to N proteins derived from highly pathogenic Nipah virus. To probe an effect of Ntail truncations on viral pathogenesis, recombinant CDVs were analyzed in a lethal CDV/ferret model of morbillivirus disease. The recombinant viruses displayed different stages of attenuation ranging from ameliorated clinical symptoms to complete survival of infected animals, depending on the molecular nature of the Ntail truncation. Reinfection of surviving animals with pathogenic CDV revealed robust protection against a lethal challenge. The highly attenuated virus was genetically stable after ex vivo passaging and recovery from infected animals. Mechanistically, gradual viral attenuation coincided with stepwise altered viral transcriptase activity in infected cells. These results identify the central Ntail section as a determinant for viral pathogenesis and establish a novel platform to engineer gradual virus attenuation for next-generation paramyxovirus vaccine design.

IMPORTANCE Investigating the role of the paramyxovirus N protein tail domain (Ntail) in virus replication, we demonstrated in this study that the structurally disordered central Ntail region is a determinant for viral pathogenesis. We show that internal deletions in this Ntail region of up to 55 amino acids in length are compatible with efficient replication of recombinant viruses in cell culture but result in gradual viral attenuation in a lethal canine distemper virus (CDV)/ferret model. Mechanistically, we demonstrate a role of the intact Ntail region in the regulation of viral transcriptase activity. Recombinant viruses with Ntail truncations induce protective immunity against lethal challenge of ferrets with pathogenic CDV. This identification of the unstructured central Ntail domain as a nonessential paramyxovirus pathogenesis factor establishes a foundation for harnessing Ntail truncations for vaccine engineering against emerging and reemerging members of the paramyxovirus family.

INTRODUCTION

The paramyxovirus family comprises a multitude of major human and animal respiratory pathogens, such as measles virus (MeV), canine distemper virus (CDV), the parainfluenzaviruses, mumps virus (MuV), and the recently emerged highly pathogenic Hendra and Nipah (NiV) viruses (1). Together with the rhabdo-, borna-, filo-, and pneumoviruses, the paramyxoviruses belong to the order Mononegavirales, which is characterized by nonsegmented negative-polarity RNA genomes that are encapsidated by viral nucleocapsid (N) proteins into helical ribonucleoprotein (RNP) complexes (2–5). Only encapsidated viral RNA serves as a template for the viral RNA-dependent RNA polymerase (RdRp) complex that functions as both the replicase and transcriptase of the viral genome.

Virus-encoded components forming the RdRp are the large (L) protein and its obligatory cofactor, the tetrameric phosphoprotein (P protein). Of these, the L protein harbors all catalytic centers required for RNA polymerization, mRNA capping and methylation, and mRNA polyadenylation, while the P protein binds to both the L and N proteins. This interaction is thought to tether the P-L polymerase to the RNP and trigger transient local release of the encapsidated RNA that gives L protein access to the template (6–13). In addition, the P protein prevents nonproductive aggregation of newly synthesized N proteins through the formation of cytosolic N⁰-P complexes and chaperoning of N protein to nascent genomic and antigenomic RNA emerging from the polymerase complex for cotranscriptional encapsidation (14–17).

Paramyxovirus N proteins feature an N-terminal core domain of approximately 400 residues that is responsible for RNA encapsidation and a structurally largely disordered C-terminal tail of approximately 120 to 150 amino acids. Functionally analogous Ncore domains are present in all Mononegavirales. However, only members of the bornavirus, filovirus, and paramyxovirus families contain unstructured Ntail extensions, whereas N proteins of the closely related rhabdo- and pneumoviruses are composed of Ncore only (18). Crystal structure data of N⁰-P complexes of MeV and NiV have revealed that the protein-protein interface consists of a short stretch of residues positioned at the N terminus of the P protein, the P peptide, binding to Ncore (16, 17). However, the initial recruitment of paramyxovirus P-L polymerase to the RNP genome was thought to require an interaction between a C-terminal binding domain in the P protein, designated P-XD, and Ntail (5, 19, 20).

In the case of MeV and CDV, two well-characterized paramyxoviruses of the Morbillivirus genus, Ntails harbor three short conserved microdomains, designated box 1 to box 3, that are positioned at its N-terminal (box 1) and C-terminal (box 2 and box 3) ends, respectively (Fig. 1A). Of these domains, box 2 contains a molecular recognition element (MoRE) that mediates interaction with the P-XD through an induced-fit protein-protein interaction (6, 8, 9, 21, 22). The adjacent box 3 of Ntail binds to the viral matrix (M) protein, aiding proper particle assembly (23, 24). In addition, Ntail is reportedly recognized by host cell interferon regulator factor 3 (IRF-3) (25) and box 2 and box 3 residues can interact with the major inducible heat shock protein HSP72 (26).

FIG 1.

MeV and CDV N protein mutants with different length deletions in the disordered central tail section. (A) Morbillivirus N protein organization. The Ntail domain missing from the cryo-electron microscopic reconstruction of Ncore (blue-gray; PDB code 4UFT) was added using Coot for length illustration, assuming a perpendicular orientation of Ntail to the axis of the helical RNP assembly (right). Conserved box regions are highlighted in yellow, orange, and green. Heat maps (red) represented the predicted degree of structural disorder. The lower portion shows a model of an Ntail mutant after removal of the disordered central Ntail section. The truncation posits box 2 and 3 regions in close proximity to the trunk of the RNP assembly. (B) Sequence alignment of the MeV and CDV Ntail domains. Box 1 to 3 domains are color-coded as described for panel A. Truncation donor (red) and acceptor (green) residues explored in this study are highlighted; residues in the central structurally disordered region are underlined. Alignments were generated using T-Coffee (69). (C and E) Steady-state levels of MeV (C) and CDV N (E) protein mutants in transiently transfected BSR-T7/5 cells. Immunoblots were developed with specific antibodies directed against MeV and CDV N protein, respectively, or cellular GAPDH. Numbers represent means of densitometry quantitations of three biological repeats ± SEM. MW, molecular weight, in thousands. (D and F) Monocistronic minigenome assays with N protein mutants specified in panels C and E. Symbols represent relative luciferase units of each biological repeat, determined each in nine technical replicates and normalized for standard N protein. Columns show sample means ± SEM; one-way ANOVA with Tukey's post hoc test was used. (G and H) Tricistronic minigenome assays. Symbols represent mean relative reporter activity ratios of each biological replicate, with error bars showing SEM; one-way ANOVA with Tukey's post hoc test was used. NS, not significant.

In contrast, the molecular role of the N-terminal box 1 is less defined (27, 28). Cryo-electron microscopy-based reconstructions of the MeV RNP assembly positioned the Ntail origin at the internal surface of the RNP (29), thus requiring tail residues to protrude toward the outer surface between the rungs of the RNP helix. In fact, not only box 1 residues but also the first approximately 50 amino acids of Ntail were proposed to be buried based on nuclear magnetic resonance (NMR) predictions, leaving only MeV N protein residues 450 to 525 to extend away from the RNP core structure (29, 30). Consistent with this model, proteolytic removal of all of Ntail triggered a major conformational change in the RNP assembly, resulting in a rigidified helix with decreased pitch and diameter compared to those of the native complex. This rigidified helix is considered to be bioinactive for replication (31).

The three conserved box domains are separated by the highly variable central Ntail section of approximately 60 amino acids located in MeV from positions 420 to 480, which is considered to be structurally disordered (8, 20, 29, 30, 32). Since none of the other paramyxovirus proteins are known to harbor domains of comparable size that are dispensable for virus viability, it was assumed that the central Ntail section also serves a critical role in viral polymerase function. Representing a long-standing view, for instance, a fly-casting model entailed that the unstructured section extends spatial flexibility to MoRE, which was suggested to be required for interaction with P-L polymerase and efficient recruitment of the polymerase complex to the RNP template (33, 34).

When testing this model experimentally, however, we recently demonstrated that MoRE can be physically relocated into Ncore in a recombinant MeV (18), revealing efficient virus replication in the absence of Ntail-mediated spatial flexibility of MoRE. In contrast, combining MoRE relocation with a partial deletion of the central Ntail section resulted in a severely temperature-sensitive virus that was replication defective under physiological growth conditions. Cells infected with this recombinant contained a significantly larger amount of nonproductive polycistronic viral mRNAs. These results may highlight a previously unappreciated role of the unstructured Ntail section in proper mRNA polyadenylation and release, or they could simply be due to unrelated but synergistic adverse effects of MoRE relocation combined with partial central Ntail truncation.

To better define the role of the unstructured Ntail section in virus replication, we explored in this study the effect of internal Ntail truncations on RdRp activity in minireplicon assays and recombinant virus replication, and we employed the lethal CDV/ferret model of morbillivirus disease to query a relation between Ntail-controlled viral polymerase activity and pathogenesis. Our experiments identified the unstructured Ntail section as a nonessential paramyxovirus pathogenesis factor that contributes to viral fitness through ensuring properly balanced viral transcriptase activity. We provide proof of concept that Ntail modifications can be harnessed to engineer genetically stable attenuated paramyxovirus recombinants for next-generation vaccine design.

RESULTS

The functional importance of the structurally disordered central Ntail sections for paramyxovirus polymerase activity is currently mechanistically poorly understood (33, 34). To test whether the morbillivirus unstructured central Ntail section is required for virus replication, we designed a series of progressively larger internal tail deletions in MeV N protein, ultimately extending to elimination of most of the structurally disordered residues between box1 and MoRE (Fig. 1A). Figure 1B provides an overview of the morbillivirus (MeV and CDV) Ntail truncations targeted. The largest deletion in MeV N protein eliminated all of Ntail from residues 399 to 482, which also includes the conserved box1 region at the N-terminal origin of the tail. Naturally, gradual removal of the central Ntail residues positions the C-terminal end of Ntail harboring MoRE and box3 in immediate proximity of the trunk of the helical RNP assembly.

Bioactivity of transiently expressed MeV and CDV Ntail mutant proteins.

Western blot analyses of cell lysates after transient expression of the different Ntail mutants confirmed stable expression of all modified N proteins and revealed the anticipated gradual increase in electrophoretic mobility with expanding truncation size (Fig. 1C). The two constructs harboring partial or complete deletions of box 1 (NΔ399–482 and NΔ409–482) showed substantially reduced and abrogated bioactivity, respectively, in minireplicon reporter assays. All other mutants supported efficient RdRp activity, and none of the small variations from activity observed in the presence of standard N protein was statistically significant.

Encouraged by these results, we generated a comparable series of CDV N deletion mutants informed by the Ntail linear sequence alignments (Fig. 1B). Resembling the corresponding MeV N protein mutants, expression levels and bioactivity of CDV NΔ441–479 and NΔ425–479 were indistinguishable from those of standard CDV N protein in Western blotting and minireplicon assays (Fig. 1E and F). However, larger truncations (CDV NΔ423–479 and NΔ421–479, respectively) resulted in significantly enhanced (approximately 2-fold) minireplicon activities. As noted for MeV, truncation encroaching into the box 1 section abolished CDV N protein bioactivity.

To address whether the viral RdRp successfully negotiates intergenic junctions in the presence of the mutant N proteins, we employed a recently described tricistronic minireplicon construct (Fig. 3D in reference 18 shows a schematic of the construct) that contains two intergenic junctions and distinct firefly luciferase and nanoluciferase reporter genes in the first and third reading frame positions, respectively. In this assay, the relative ratio of downstream versus upstream reporter activity represents the relative efficiency with which the RdRp complex advances through intergenic junctions and reinitiates mRNA synthesis. When tested against a subset of the MeV (Fig. 1G) and CDV (Fig. 1H) Ntail mutants, only the MeV NΔ420–482 construct returned a slight (approximately 25%), but statistically significant, relative reduction of third open reading frame (ORF) expression compared to the relative reporter ratios observed with standard N protein. These results suggest successful negotiation of intergenic junctions by the RdRp complex in the absence of the unstructured Ntail section.

FIG 3.

Recombinant MeV and CDV expressing Ntail mutants. (A and B) Electrophoretic mobility profiles of N proteins in lysates of cells infected with the different recMeV (A) or recCDV (B) strains. (C and E) Multistep growth curves of the different recMeV (C) and recCDV (E) mutants. Symbols represent cell-associated virus titers of individual biological experiments; lines connect sample means. (D and F) Regression modeling of growth profiles shown in panels C and E. Bindslev's population growth four-parameter variable slope model was applied. PDT_max, maximal population doubling time; Titer_max, titer corresponding to the top plateau of the regression models. Values in parentheses specify 95% confidence intervals; NS, overlapping confidence intervals. *, nonoverlapping confidence intervals relative to standard recMeV. (G) Relative levels of type I IFN and selected ISG message in HeLa cells infected with the different CDV recombinants. Symbols represent relative fold change in mRNA levels normalized to mock-infected control cells for each biological replicate. Horizontal lines and error bars show means and SEM, respectively; one-way ANOVA with Tukey's post hoc test was applied to each message set targeted.

Effect of Ntail truncations on NiV bioactivity.

To evaluate whether continued RdRp activity in the absence of the central Ntail section extends to paramyxoviruses outside the morbillivirus genus, we applied an equivalent truncation strategy to the Ntail of highly pathogenic NiV, a member of the henipavirus genus (Fig. 2A and B). Although the organization of the NiV tail is more complex than that of the morbilliviruses, featuring an additional box4 in the central Ntail section (11), NiV RdRp readily accepted template RNA encapsidated by tail-truncated N proteins (Fig. 2C). The NiV Ntail modifications did not substantially affect N protein steady-state levels in immunoblots (Fig. 2D). Whereas the NiV NΔ423–471 mutant protein with the larger truncation showed standard NiV N-like activity in minireplicon assays, bioactivity of the NiV N construct harboring a shorter (NΔ442–471) truncation was significantly increased.

FIG 2.

Bioactivity of henipavirus Ntail mutants. (A) Schematic of the NiV Ntail organization and disorder prediction, color-coded as in Fig. 1A. (B) Sequence alignment of the MeV and NiV Ntail domains, color-coded as in Fig. 1A. Truncation donor (red) and acceptor (green) sites are highlighted. (C) Monocistronic minigenome assays. Symbols represent means of individual biological replicates, with error bars showing SEM; one-way ANOVA with Tukey's post hoc test was used. (D) Steady-state levels of NiV N proteins expressed in transiently transfected BSR-T7/5 cells. Immunoblots were developed with polyclonal anti-NiV antiserum or specific antibodies directed against cellular GAPDH. *, cross-reaction with a host cell protein with electrophoretic mobility similar to that of standard NiV N protein.

Recovery of recombinant MeV and CDV encoding tail-truncated N proteins.

RdRp activity in minireplicon assays is necessary but not sufficient for support of a full viral replication cycle. To test a role of the unstructured Ntail section in virus replication, we exchanged the N protein-encoding ORFs in cDNA copies of the genomes of two pathogenic viral isolates, MeV-IC-B (35) and CDV-5804PeH (36). Two different Ntail truncations each were inserted, MeV NΔ439–482 and NΔ420–482 and CDV NΔ441–479 and NΔ425–479, respectively, which in each case represent the shortest and largest internal tail truncation that did not significantly alter N protein bioactivity in the monocistronic minireplicon assays. The corresponding recombinant viruses were recovered readily on receptor-positive Vero-human SLAM (Vero-hSLAM) and Vero-canine SLAM (Vero-cSLAM) cells, respectively. DNA sequencing after reverse transcription-PCR (RT-PCR) amplification of recovered virus genomes and Western blot analyses of infected cell lysates (Fig. 3A and B) confirmed the presence of the respective Ntail truncations in the recombinant viruses.

Multiple-step growth curves revealed parent virus-like replication of recMeV-IC-B NΔ439–482 (Fig. 3C), whereas recMeV-IC-B NΔ420–482 with the larger Ntail truncation showed an initial 12-h growth delay. Modeling of growth profiles confirmed that the maximal population doubling times of recMeV-IC-B NΔ420–482 were significantly longer than those of standard recMeV-IC-B, but differences between recMeV-IC-B NΔ439–482 and the parent strain remained not significant (Fig. 3D).

This growth pattern of the MeV mutants was inversed in the recombinant CDV strains (Fig. 3E). recCDV NΔ425–479 carrying the larger Ntail truncation showed significantly shorter maximal doubling times than recCDV NΔ441–479 (Fig. 3F). While peak growth rates of both mutant strains lagged behind that of the parental recCDV-5804PeH, growth profile modeling did not reveal significant differences in final progeny titers reached. These data demonstrate that the unstructured central Ntail section is not associated with an essential RdRp function required for virus replication, establish distinct effects of different length truncations on viral fitness in cell culture, and indicate that the actual impact of internal Ntail truncations on virus growth is not proportional to the length of the deletion but must be individually determined.

In preparation of in vivo pathogenesis testing in the lethal CDV/ferret model, which recapitulates key features of human measles (37), we determined stimulation of the type I interferon response by the different CDV mutants in cultured cells. While recCDV NΔ425–479 did not significantly change levels of induction of alpha interferon (IFN-α), IFN-β, or a subset of selected interferon-stimulated genes (ISGs) compared to those in cells infected with standard recCDV-5804PeH, significantly higher levels of induction of IFN-β, IFIT1, and ISG-15 were found after infection with recCDV NΔ441–479 (Fig. 3G). These data suggest stronger stimulation or impaired downregulation of the host innate antiviral response by recCDV NΔ441–479.

Pathogenesis of Ntail-mutated CDV in ferrets.

Ferrets were infected intranasally with 2 × 10⁵ 50% tissue culture infectious dose (TCID₅₀) units of standard recCDV-5804PeH or the two recCDV Ntail mutants, and clinical signs, peripheral blood mononuclear cell (PBMC)-associated viremia titers, white blood cell counts, and lymphocyte proliferation response were monitored in regular intervals. Consistent with our previous experiences with the model (38), peak viremia titers were reached 7 days postinfection, followed by a rapid decline in viral load in animals infected with either of the N mutant viruses (Fig. 4A). All animals infected with standard recCDV-5804PeH succumbed to the disease by day 14. In contrast, the group that had received recCDV NΔ441–479 and 75% of animals in the recCDV NΔ425–479 group survived the infection (Fig. 4B). In recovering animals, viremia fully subsided 21 (recCDV NΔ441–479) and 35 (recCDV NΔ425–479) days after infection.

FIG 4.

Pathogenesis in the CDV ferret model. (A) Cell-associated viremia titers after intranasal infection of animals. Symbols represent mean TCID₅₀ units for each biological replicate; lines connect sample means (recCDV NΔ425–479-infected animals, n = 4; recCDV NΔ441–479-infected animals, n = 7; recCDV-5804PeH-infected animals, n = 4); two-way ANOVA with Tukey's post hoc test was applied. (B) Survival curves. Mantel-Cox test was applied; predicted median survival is shown. (C) Virulence index for standard and mutant CDV. Each box represents one animal, clinical score index ranges from high (black) to intermediate (gray) and absent (white). (D) Animal body weight changes. Weight is expressed as the percentage of the initial weight at the day of infection. Symbols represent individual biological replicates, and lines connect sample means; two-way ANOVA with Sidak's post hoc test was used. (E) Fever curves. Symbols represent individual biological replicates, and lines connect sample means; two-way ANOVA with Sidak's post hoc test was used. (F) Lymphopenia. Symbols represent mean leukocyte counts for each biological replicate, and lines connect sample means; two-way ANOVA with Tukey's post hoc test was used. (G) Nonspecific PBMC proliferation capacity. Symbols represent mean BrdU incorporation relative to nonstimulated PBMCs for each biological replicate, and lines connect sample means; two-way ANOVA with Tukey's post hoc test was used.

Animals infected with the parental virus experienced severe disease with extensive rash, substantial weight loss, and high fever (Fig. 4C to E). In comparison, disease progression was less aggressive in animals infected with the recCDV NΔ425–479 mutant virus and mild in animals of the recCDV NΔ441–479 group. Specifically, recCDV NΔ441–479-infected ferrets showed a benign, localized rash and only transiently lost a moderate (<10%) amount of body weight before making a full recovery (Fig. 4D). Fever peaked in these animals 2 to 3 days earlier and at a lower level than in recCDV-5804PeH and recCDV NΔ425–479-infected animals and resolved within the second week after infection (Fig. 4E). In contrast, recCDV NΔ425–479-infected ferrets presented with a longer weight loss period and fever fully resolved only in the third week after infection.

Acute lymphopenia and temporary lack of lymphocyte responsiveness to stimulation are hallmarks of morbillivirus infections (39). When assessing immune competence of animals in the different groups, we noted significantly milder lymphopenia early after infection in the recCDV NΔ441–479 group than in ferrets that had received standard recCDV-5804PeH or recCDV NΔ425–479 (Fig. 4F). However, lymphocytes derived from animals of all groups showed similar declines in proliferation responsiveness during the first 2 weeks after infection (Fig. 4G). Proliferation response improved in all surviving animals only at 35 days postinfection, although we noted a temporary rebound in cells extracted from recCDV NΔ441–479-infected ferrets at the 21-day time point.

Ferret immune response to Ntail-truncated CDVs.

Ferrets infected with the two mutant viruses mounted overall comparable type I interferon responses, reaching approximately 10-fold induction levels in IFN-β and Mx-1 message, the latter representing one of the major ISGs in the ferret response to CDV infection (Fig. 5A). Resembling our results obtained in cultured cells, however, PBMCs derived from recCDV NΔ441–479-infected animals showed significantly higher induction levels in IFN-β and ISG message at day 7 postinoculation than at day 3.

FIG 5.

Ferret immune responses to mutant CDVs. (A) Relative levels of type I IFN and Mx-1 message in PBMCs isolated at days 3 and 7 after infection of animals. Symbols represent relative fold change in mRNA level normalized to uninfected naive controls (day 0) for each biological replicate. Horizontal lines and error bars show means and SEM, respectively; t tests were used for pairwise comparisons, and one-way ANOVA with Sidak's post hoc test was applied to Mx-1 day 7 fold change analysis. (B) Neutralizing antibody responses. Titers are shown as the reciprocal of the highest dilution in which CPE was observed. Symbols represent individual biological replicates, and lines connect sample means; two-way ANOVA with Sidak's post hoc test (comparison of the recCDV NΔ425–479- and recCDV NΔ441–479-infected groups only) was used. (C) Challenge of surviving animals from Fig. 4 with 2 × 10⁵ TCID₅₀ units of standard recCDV-5804PeH at day 49 postinfection. Survival curves are shown for rechallenged and CDV-naive ferrets (before infection with recCDV NΔ425–479, n = 3; recCDV NΔ441–479, n = 4; naive, n = 4). (D) Cell-associated viremia titers in rechallenged animals. Symbols represent TCID₅₀ for each biological replicate, and lines connect sample means; two-way ANOVA with Tukey's post hoc test was used. (E) Lymphopenia in rechallenged animals. Symbols represent leukocyte counts for each biological replicate, and lines connect sample means; two-way ANOVA with Tukey's post hoc test was used.

Anti-CDV antibody responses were robust in animals of either group, although neutralizing antibody titers induced by the more attenuated recCDV NΔ441–479 peaked slightly higher than those found in animals of the recCDV NΔ425–479 group (Fig. 5B). Consistent with our past experience (38), low-level neutralizing antibodies were detectable in ferrets inoculated with standard recCDV-5804PeH at day 7, but titers did not reach robust levels by the time animals succumbed to the infection. To determine whether immune responses mounted by the surviving animals infected with the Ntail mutant viruses were protective, we rechallenged them with a lethal dose of standard recCDV-5804PeH at 49 days after the original infection. All challenged animals survived (Fig. 5C), and none developed appreciable viremia (Fig. 5D), showed clinical signs, or experienced severe lymphopenia (Fig. 5E).

These results identify the unstructured Ntail section as a determinant of paramyxovirus pathogenesis. Gradual shortening of the tail induces different degrees of viral attenuation, although not with direct proportionality. Importantly, all surviving animals in the CDV/ferret model were completely protected against a lethal challenge with standard CDV, underscoring efficient induction of functional immune responses by the Ntail-modified recombinants.

Genetic stability of Ntail truncation in cell culture and in vivo.

To assess the genetic stability of the Ntail-truncated CDV, we subjected viral RNA preparations to deep sequencing before and after 10 or 11 passages in cell culture and determined N ORF consensus sequences in viral RNA extracted from PBMCs harvested from ferrets 7 days after infection through Sanger sequencing (Table 1). Neither standard recCDV-5804PeH nor recCDV NΔ441–479 showed any consensus changes in the N or P protein ORF compared to the corresponding genomic cDNA plasmids. However, recCDV NΔ425–479 carried a glutamate-to-glutamine substitution at N residue 156 that was dominant in the viral population after four cell culture passages and acquired an alanine-to-aspartate exchange at N residue 410 that became increasingly fixed during passaging. All recCDV NΔ425–479 recovered from infected ferrets at the peak of viremia contained both substitutions. Comparison with N protein sequences representing a variety of different circulating CDV strains and isolates revealed that the N protein ORF is sequence conserved at position 156 and shows only conservative variations from the aspartate substitution at residue 410 (i.e., threonine or alanine residues in circulating strains).

TABLE 1.

Sequence analysis of recCDV strains after passage in cell culture and through ferrets^a

^aaa, amino acid; NA, not applicable; TC, tissue culture; NC, no change from previously published sequence (36); ND, not determined based on low allele variations after tissue culture passaging (exceptions are noted). Boldface indicates deviations from the input consensus sequence.

^b The input cDNA plasmid and ferret DNA sequences were determined through Sanger sequencing. “TC” indicates that the virus was passaged in tissue culture and that its sequence was determined through deep sequencing. P_n-x, the number (n) of consecutive passages (P) carried out in x independent parallel assessments.

^c Not determined because the sequence was not subject to in vitro amplification or PCR-based modification because it was the last sequence verified.

^d Virus was extracted after in vivo passage from harvested PBMCs. The input viruses for animal studies were TC P₂ (recCDV NΔ441–479) and TC P₄ (recCDV NΔ425–479).

^e Determined for each animal (n = 4) after Sanger sequencing of virus extracted from PBMCs.

None of the CDV recombinants harbored coding mutations in the P and L protein ORFs, with the exception of a single recCDV NΔ441–479 passaging line that carried a leucine-to-proline substitution at L protein residue 2175 with approximately 50% allele frequency after 15 passages in cell culture (Table 1). Since all other recCDV NΔ441–479 lines analyzed in parallel lacked any allele variation at this position, this mutation most likely represents a stochastic event that became partially fixed in the genome.

These observations demonstrate that the NΔ441–479 truncation is genetically stable over a number of generations in cell culture and after in vivo passage of the recombinant strain. Efficient growth of recCDV NΔ425–479 appears to be linked to the presence of two compensatory mutations, one located in Ncore (residue 156) and the second in box 1 of Ntail (residue 410). Localization of CDV N residue 156 in a structural model of the morbillivirus RNP assembly posits this substitution at the C-terminal end of a flexible loop in Ncore (18), orientated toward the interface between consecutive turns of the RNP helix (Fig. 6) where the N tails are thought to emerge from the RNP core (40). This substitution may thus highlight direct cross talk between Ntail residues located immediately downstream of position 425 and in the Ncore loop region harboring residue 156.

FIG 6.

Location of candidate compensatory mutation E156Q in a model of CDV RNP. (A) E156Q (red spheres) in Ncore is posited near the predicted site of Ntail (yellow circles) protrusion from the RNP. CDV RNP homology models were created with the SWISS model homology modeling server based on the MeV Ncore structure (PDB code 4UFT). (B) Side view of consecutive rungs of the helical CDV RNP. No structural information is available for the position of Ntail, but box 1 (yellow rectangles) is predicted to locate close to the outer surface of the RNP or between consecutive turns of the RNP helix. The A410D substitution near the center of box 1 is highlighted (red rectangles).

Quantitation of viral RNA populations in cells infected with Ntail mutant MeV or CDV.

To further elucidate the mechanistic basis for the altered CDV pathogenesis profiles, we analyzed viral RNA populations synthesized in cells infected with the different recombinant virus strains. Cotranscriptional paramyxovirus mRNA editing results in the expression of two additional proteins, V and W proteins, from the viral P protein ORF through the insertion of nontemplated G residues at an editing site (41–44). RNA editing is thought to result from backsliding of the RdRp complex on the RNP template (41, 45, 46), which requires structural flexibility that could be mediated by the flexible central Ntail section. Since impaired V protein expression causes viral attenuation (47), we employed a MiSeq assay to quantify the relative ratios of P, V, and W protein-encoding mRNAs in infected cells. Only recCDV NΔ441–479 and the corresponding recMeV-IC-B-NΔ439–482 were analyzed, based on higher attenuation of this shorter CDV truncation mutant in the ferret model. Relative mRNA distributions were comparable between MeV and CDV, but we noted only minor changes when mutant and the corresponding parent viruses were compared (Fig. 7A and B).

FIG 7.

RNA populations present in cells infected with Ntail mutant viruses. (A and B) MiSeq analysis of viral P mRNA editing by recCDV NΔ441–479 (A), recMeV NΔ439–482 (B), or the corresponding parent virus. Values represent a minimum of 91,741 reads each and are expressed as mean percentage of the differentially edited mRNAs relative to total P ORF transcripts ± SEM. (C) qRT-PCR quantitation of relative CDV genome copy numbers in cells infected with the recCDV Ntail mutants. First-strand synthesis was done with specific primers binding to the viral genome untranslated region (UTR). (D and E) qRT-PCR quantitation of relative CDV N mRNA (D) and L mRNA (E) copy numbers present in RNA preparations as in panel C. First-strand synthesis was done with oligo(dT) primers. (F to H) qRT-PCR quantitations of RNA preparations as in panel C of the relative ratios of L and N protein-encoding mRNAs (F) and of polycistronic mRNAs covering the N/P (G) and mKate/L (H) intergenic sequence (IGSs). First-strand synthesis was done with oligo(dT) primers. In panels C to H, symbols represent individual values of three biological repeats analyzed in two technical replicates each. Columns show means ± SEM; one-way ANOVA with Tukey's post hoc test was performed.

We therefore examined whether attenuation of the recCDV Ntail mutants in ferrets may alternatively result from deregulated viral RdRp activity. Using an RT-quantitative PCR (qPCR)-based approach, we quantified viral genome copies in infected cells, determined relative N and L protein-encoding mRNA levels, examined relative ratios of L to N protein-encoding mRNAs, and calculated the relative frequencies with which polycistronic viral mRNAs are synthesized by the different recombinants.

Viral genome copy numbers of both mutant recCDV strains were reduced by approximately 19 to 29% at the end of the replication cycle compared to standard virus (Fig. 7C). In contrast, relative N protein-encoding mRNA levels of either mutant strain were increased approximately 1.8-fold (Fig. 7D). When we examined mRNA levels of the downstream-most positioned L protein ORF, we noted that this 1.8-fold relative increase was maintained in recCDV NΔ425–479-infected cells but boosted to an approximately 3-fold relative excess in the case of recCDV NΔ441–479 (Fig. 7E). Replication of recCDV NΔ441–479 furthermore results in a significant increase in the amounts of L protein-encoding message relative to N protein mRNAs, while essentially identical ratios of L to N protein message were obtained for standard recCDV and the less attenuated recCDV NΔ425–479 (Fig. 7F).

This higher relative L protein mRNA level produced by recCDV NΔ441–479 may reflect an increase in bona fide L protein message due to a lowered premature termination rate of RdRp or a higher proportion of nonproductive polycistronic mRNAs. We therefore quantified the relative content of polycistronic message generated at the first (Fig. 7G) and last (Fig. 7H) intergenic sequences (IGSs) in the recCDV genomes. Depending on the IGS examined, replication of standard recCDV-5804PeH and recCDV NΔ425–479 produced 2 to 10% polycistronic message relative to total message synthesized for the preceding ORF. At each IGS examined, however, we noted a significantly higher ratio of polycistronic message present in cells infected with the recCDV NΔ441–479 mutant strain.

These results implicate the structurally disordered central Ntail section in affecting paramyxovirus transcriptase function on two levels. Both Ntail mutant strains show heightened transcriptase activity relative to that of the parental recCDV strain. In addition, the more severely attenuated recCDV NΔ441–479 further disturbs the relative ratio of viral message in infected cells by generating a higher proportion of nonproductive polycistronic mRNAs.

DISCUSSION

Our characterization of Ntail truncation mutants revealed three major effects of the unstructured Ntail region on RdRp activity, viral fitness in cell culture, and viral pathogenesis.

First, all Ntail mutants with truncations in the disordered central tail section remained bioactive in monocistronic and, tested for MeV and CDV constructs only, tricistronic minireplicon assays. Only MeV-derived Ntail mutants showed a slightly impaired ability to transcribe the downstream reporter ORF in the tricistronic assay, demonstrating that the central Ntail domain is not essential for negotiation of the IGS sections. In contrast, truncations eliminating the conserved box1 caused a decline in RdRp activity, which is most likely due to an impaired protein-protein interface between box1 and Ncore. The compensatory mutation found at Ncore position 156 in recCDV-NΔ425–479 reinforces this hypothesis of a direct cross talk between Ncore and tail residues. Altering the molecular nature of the interface between N-terminal Ntail residues and the RNP rungs could possibly affect local structural rearrangements that are required to transiently provide polymerase access to the genomic RNA.

Second, recombinant MeV and CDV replicate productively in the absence of the central Ntail section in cell culture. This surprising finding could not be extrapolated from the minireplicon data, since in our experience sustained RdRp activity in minireplicon assays is necessary but not sufficient for completion of a viral replication cycle (48). Intriguingly, the extent of the internal deletion affected viral growth kinetics, but no direct proportional correlation exists between the number of residues removed and fitness penalty. Our study demonstrates, however, that morbillivirus replication is tolerant toward changes in both central Ntail sequence and length, which is consistent with the notion that N proteins of ancestral Mononegavirales may have lacked an unstructured tail domain (18).

Third, consistent with this proposed role of the central Ntail section as a regulatory element for optimal viral polymerase activity, we found that Ntail is a determinant for paramyxovirus pathogenesis in vivo. Depending on the extent of the central Ntail truncation, we found different degrees of viral attenuation. Our quantitations of viral RNA populations present in infected cells linked this differential loss in pathogenesis to a single versus double hit on polymerase function. The first hit is experienced by both CDV mutants and entails a partial shift of polymerase activity from replicase to transcriptase mode, evidenced by a higher relative proportion of first ORF-encoding viral message and lower relative genome copy numbers in cells infected with either mutant compared to standard recCDV-5804PeH. The decision of whether the paramyxovirus RdRp complex proceeds as transcriptase or replicase after initialization of de novo RNA synthesis depends on the concurrent encapsidation of a short leader RNA strand with N protein (49). Ntail may directly participate in the structural reorganization required for transfer of N proteins from N⁰-P complexes to the nascent RNA for encapsidation (45). Tail shortening may affect the efficiency of this reaction, driving the RdRp complexes into transcriptase mode (Fig. 8A). Alternatively, Ntail could modulate the time window available for leader RNA encapsidation by slowing the progress of the advancing polymerase complex. Cryo-electron microscopy (12) and NMR spectroscopy studies (29) suggest that binding of the RdRp complex to the viral RNP involves local Ntail ordering. RNPs with shortened Ntails could mimic a permanently ordered state, potentially accelerating advance of the RdRp complex along the RNP and thus narrowing the window of opportunity for efficient leader RNA encapsidation (Fig. 8B). At later stages of the replication cycle, the ensuing relative excess of viral message may alleviate transcriptase bias of the polymerase complex to some degree, restoring sufficient replicase function in the presence of Ntail truncations for successful genome replication. It is noteworthy that the Ntail deletions did not simply reduce all RdRp activity. Our data thus show that the disordered central Ntail section is not required for initial polymerase loading onto the RNP template, eliminating a central prediction of the original model of Ntail attraction of P-L polymerase to the RNP (33, 34).

FIG 8.

Mechanistic models of possible roles of Ntail in RdRp transcriptase activity. Upon initialization at the promoter, RdRp synthesizes a short leader RNA (L in gray, P tetramers in blue, and MoRE and box3 as orange and yellow circles). Encapsidation of this leader triggers switch to replicase mode. Without encapsidation, RdRp releases leader and generates viral mRNAs. Advance of RdRp along the RNP involves release of existing N-MoRE/P-L interactions, Ntail reordering in front of the polymerase, and local deencapsidation of the viral RNA. (A) Removal of the central Ntail section may reduce the encapsidation efficiency of leader RNA, causing a bias toward transcriptase (thick horizontal arrow). (B) Alternatively or in addition, Ntail truncation may eliminate the need for local Ntail ordering ahead of RdRp. Advance of polymerase is accelerated, narrowing the time window available for leader encapsidation (staggered horizontal arrows) and promoting transcriptase mode.

As a second hit experienced only by recCDV NΔ441–479, we noted a larger amount of polycistronic viral message. Since only the first ORF of a polycistronic paramyxovirus mRNA can be translated (49), this hit directly affects the viral protein transcription gradient established in a recCDV NΔ441–479-infected cell. This finding implicates the unstructured central Ntail section in the proper recognition of viral gene end sequences located near the beginning of each intergenic junction. Paramyxovirus mRNA polyadenylation is not templated but is thought to be achieved through backsliding of the RdRp complex on the RNP template, which also entails slippage of the nascent mRNA relative to the template strand (45). Possibly, Ntail is involved in providing additional structural flexibility to optimize backsliding efficiency. However, the recCDV NΔ425–479 strain, with the larger truncation, was not subject to the second hit on transcriptase function. While providing a mechanistic explanation for the higher relative fitness of this mutant strain both in cell culture and in vivo compared to recCDV NΔ441–479, this finding suggests that the actual Ntail amino acid sequence rather than tail length determines how efficiently gene end signals are recognized by the transcriptase complex. The shorter NΔ425–479 tail may provide higher structural flexibility than the longer NΔ441–479 mutant.

Although rational viral attenuation for vaccine design has gained considerable traction in recent years (50, 51), we propose that three features in particular recommend the central paramyxovirus Ntail section as an attractive target for the engineering of next-generation recombinant vaccine strains: the promise of genetic stability, adjustability of attenuation, and application to related or emerging members of the family.

Engineering attenuation through individual point mutations or a small panel of mutations is straightforward but at risk of spontaneous reversion to the pathogenic form. A recent approach to mitigate the problem introduces a large number of changes, i.e., through codon deoptimization strategies (50), which generate sufficient redundancies to ensure that attenuation is maintained faithfully. However, calibrating the balance between residual pathogenicity and immunogenicity remains a challenge. Our identification of a role of the unstructured paramyxovirus Ntail section in modulating RdRp activities provides an opportunity to dial attenuation toward the desired balance. The CDV/ferret data provide proof of concept for the validity of the approach. Large internal deletions in coding sequences are furthermore at very low risk of spontaneous reversion, since the high error rates of viral RdRps result from base mismatches rather than de novo creation of code (52) and homologous recombination events are rare in negative-sense RNA viruses with encapsidated genomes (53). In addition, sequence analysis of recCDV NΔ441–479 mutants after virus passage did not reveal any additional mutation or larger genome rearrangements in the N protein or its binding partner, the P protein. Efficient growth of the recCDV NΔ425–479 recombinant harboring the larger Ntail truncation appeared to require, however, the presence of two candidate compensatory mutations. Future work will probe functional and mechanistic links between these substitutions and the NΔ425–479 truncation, but the available data suggest that placing engineered Ntail truncation junctions too close to the RNP core calls for additional mutations and should therefore be avoided in the interest of genetic stability of the recombinant virus.

We furthermore expect that attenuation by Ntail truncation will be applicable to different paramyxovirus targets. A safe and effective anti-MeV prophylaxis based on the live attenuated MeV Edmonston strain exists, but CDV vaccine safety and efficacy remain problematic in species with unknown or high sensitivity, making the development of a next-generation vaccine with a known attenuation mechanism desirable (54, 55). Importantly, the NiV minireplicon data imply that the approach may be extendable to the highly pathogenic henipaviruses, which are listed by the World Health Organization among the top emerging diseases likely to cause major epidemics (56, 57). Provided that acceptable safety and immunogenicity profiles can be established and protection demonstrated in surrogate models (58), a broadly applicable, genetically stable, and tunable attenuation strategy could shorten vaccine development response times to clinically significant emerging or reemerging paramyxoviruses.

MATERIALS AND METHODS

Cell culture.

Baby hamster kidney cells (C-13; ATCC) stably expressing T7 polymerase (BSR-T7/5 [59]), human cervix adenocarcinoma (HeLa, CCL-2; ATCC), and African green monkey kidney epithelial cells (CCK-81; ATCC) stably expressing human or canine signaling lymphocytic activation molecule (Vero-hSLAM or Vero-cSLAM, respectively [60]) were maintained at 37°C and 5% CO₂ in Dulbecco's modified Eagle's medium (DMEM) supplemented with 7.5% fetal bovine serum. All stable cell lines were incubated in the presence of G-418 (100 μg/ml) at every fifth passage. GeneJuice (Novagen) reagent was used for all transient transfections of cells.

Molecular biology.

Plasmids encoding expression constructs of MeV strain Edmonston N, P, and L proteins (61), MeV strain IC-B N, P, and L proteins (35, 48), CDV strain Onderstepoort N, P, and L proteins (62), and NiV N, P, and L proteins (63) were previously described. Likewise, plasmids harboring full-length cDNA copies of the MeV strain IC-B genome (35), CDV strain 5804PeH genome (36), and the different MeV minireplicons (18) and shuttle vectors harboring MeV strain IC-B- and CDV strain 5804PeH-derived N protein ORFs were previously reported (48). A cloning strategy developed in our earlier work (48) was applied to generate all MeV, CDV, and NiV N genes with internal Ntail truncations. Briefly, sets of PCR primers were engineered that flanked the specific nucleotides targeted for deletion and contained terminal AfeI restriction sites in frame with the N protein ORF. Religation of the AfeI-digested PCR products reconstituted the expression plasmid, now replacing the targeted Ntail section with Ser-Ala residues encoded by the AfeI site. All Ntail modifications were confirmed by DNA sequencing. In addition, all full-length genome plasmids were sequence confirmed prior to recovery transfection of recombinant virions. To generate an NiV nanoluciferase minireplicon reporter construct, the nanoluciferase gene was amplified using appropriate PCR primers and the resulting product cloned into an existing NiV replicon backbone (63) that was likewise PCR amplified using appropriate primers. The nanoluciferase amplicon was ligated to the replicon vector backbone using the NeBuilder kit in accordance with the manufacturer's protocols (New England BioLabs), and the resulting plasmid sequence was verified.

Immunoblotting, SDS-PAGE, and antibodies.

BSR-T7/5 cells transfected in a 12-well plate format (4 × 10⁵ per well) with 2 μg of MeV or CDV N protein-encoding expression plasmid DNA were washed once 40 h after transfection with phosphate-buffered saline (PBS) and lysed in radioimmunoprecipitation assay (RIPA) buffer (1% sodium deoxycholate, 1% NP-40, 150 mM NaCl, 50 mM Tris-Cl [pH 7.2], 10 mM EDTA, 50 mM NaF, 0.05% sodium dodecyl sulfate [SDS], protease inhibitors [Roche]). Cleared lysates (centrifugation at 20,000 × g and 4°C for 10 min) were mixed with 5× urea buffer (200 mM Tris [pH 6.8], 8 M urea, 5% SDS, 0.1 mM EDTA, 0.03% bromphenol blue, 1.5% dithiothreitol). Samples were denatured for 30 min at 50°C, fractionated on 10% SDS-PAGE gels, blotted on polyvinylidene difluoride (PVDF) membranes, and subjected to enhanced chemiluminescence detection using specific antibodies directed against MeV N protein (MAB8905; Millipore), CDV (DV2-12; Bio-Rad), CDV Ncore (clone 1214) (64), polyclonal whole NiV immune serum, or glyceraldehyde-3-phosphate dehydrogenase (GAPDH; 6C; Ambion). Immunoblots were developed using a ChemiDoc digital imaging system (Bio-Rad) and the Image Lab software package (Bio-Rad) for image visualization. Densitometry was carried out on nonsaturated images with global background correction. Positive and negative controls were included on each individual gel, and no normalization across different blots was carried out.

Minireplicon luciferase reporter assay.

BSR-T7/5 cells (5,000 in a 96-well plate format) were transfected with plasmids encoding IC-B-L (0.02 μg), IC-B-P (0.02 μg), and IC-B-N (0.016 μg) and the respective MeV luciferase replicon reporter (0.044 μg). CDV minireplicon assays were performed accordingly using CDV helper plasmids. For NiV minireplicon experiments, cells were transfected with plasmid DNA encoding NiV L protein (0.005 μg), NiV P protein (0.005 μg), NiV N protein (0.01 μg), and NiV nanoluciferase replicon reporter (0.06 μg). Firefly luciferase or nanoluciferase activities were determined 40 h posttransfection in a Synergy H1 microplate reader (BioTek), using Bright-Glo or Nano-Glo luciferase substrate (Promega), respectively. Substrates were directly added to the cells, and bioluminescence was quantified after a 3-min incubation for signal stabilization. Relative RdRp activities (relA), expressed as percentages, were determined on the basis of the formula relA = (signal_sample − signal_min)/(signal_max − signal_min) × 100, with signal_max corresponding to cells transfected with plasmids encoding the standard NiV proteins and signal_min representing cells that received equal amounts of empty vector (pUC19) in place of the N protein-encoding plasmid. All experiments were performed in at least 3 independent replicates, each measured in nine dependent repeats.

Virus recovery.

Recombinant MeV or CDV was recovered in BSR-T7/5 cells by transfecting 1.25 μg of the cDNA copy of the modified genome and IC-B-N (0.42 μg), IC-B-P (0.54 μg) and IC-B-L (0.55 μg). All recombinant CDV genomes harbored an additional transcription unit encoding the mKate fluorescent protein in pre-L protein ORF position, which does not affect viral pathogenicity (36). Transfected cells were overlaid 48 h after transfection onto Vero-hSLAM or Vero-cSLAM cells, and emerging infectious particles were passaged in Vero-hSLAM or Vero-cSLAM cells, respectively. All recombinant virus strains used in this study were recovered in the Plemper lab. The integrity of newly recovered virus strains was confirmed by extracting total RNA from infected cells (RNeasy minikit; Qiagen) and generating cDNA copies using random hexamer primers and SuperScript III reverse transcriptase (Invitrogen). Modified genome regions were amplified using appropriate primers and subjected to Sanger sequencing.

Preparation of virus stocks.

MeV and CDV virus stocks were prepared by infecting Vero-hSLAM or Vero-cSLAM cells at a multiplicity of infection (MOI) of 0.01 50% tissue culture infectious dose (TCID₅₀) units per cell, followed by incubation at 37°C. When microscopically observed virus-induced cytopathicity reached approximately 90%, cell-associated progeny particles were released through freeze-thaw cycles and titers determined by TCID₅₀ titration on Vero-hSLAM or Vero-cSLAM cells as described previously (65).

Multistep virus growth curves.

Prior to infection for multistep growth curves, viral stocks were diluted to approximately 1 × 10⁴ TCID₅₀ units/ml and exact titers determined in a separate aliquot by TCID₅₀ titration. Vero-hSLAM or Vero-cSLAM cells (1 × 10⁵ per well in a 12-well format) were infected with the different MeV or CDV strains at an MOI of 0.01 TCID₅₀ unit per cell for 1 h and the inoculum was replaced with DMEM. Individual wells were harvested at 12-h intervals and cell-associated progeny virus titers determined by TCID₅₀ titration. At least three independent growth curves were generated for each virus strain examined. Virus-induced cytopathicity in infected cells was documented using an inverted fluorescence microscope (Nikon) equipped with a digital imaging package.

CDV ferret studies.

Male and female adult European ferrets (Mustela putorious furo without immunity against CDV) were used in this study. Animals were infected intranasally with 2 × 10⁵ TCID₅₀ units of recCDV-5804PeH per animal, and blood samples were collected from the jugular vein at the desired time points. Three parameters of virulence—rash, body temperature, and weight loss—were measured and graded based on a scale of 0 to 2. Scoring for rash was as follows: 0, no rash; 1, localized rash; and 2, generalized rash. Scoring for body temperature was as follows: 0, no fever; 1, temperature reached >39°C; and 2, temperature was >40°C. Scoring for loss compared to weight on day 0 was as follows: 0, 0 to 5%; 1, 5 to 10%; and 2, >10%. For white blood cell counts, 10 μl of heparinized blood was diluted in 990 μl of 3% acetic acid and white blood cells were counted. Cell-associated viremia was quantified by first isolating total white blood cells, followed by addition to Vero-cSLAM cells in 10-fold dilution steps for TCID₅₀ titration. To assess proliferation activity of Ficoll-purified (GE Healthcare) PBMCs, cells were stimulated with 0.2 μg of phytohemagglutinin (PHA; Sigma) for 24 h, followed by addition of 10 μM 5-bromo-2′deoxyuridine (BrdU; Roche). After another 24-h incubation period, cells were fixed and BrdU incorporation was quantified using a peroxidase-linked anti-BrdU antibody in a chemiluminescence assay (Roche). Signals were detected in a microplate luminescence counter (Pherastar), and the extent of proliferation was expressed as the ratio of nonstimulated to stimulated cells.

Quantification of neutralizing antibodies.

Neutralizing antibody titers were determined by mixing serial dilutions of ferret plasma collected at different time points with 10² TCID₅₀ units of recCDV-5804PeH. Virus and plasma were incubated at 37°C for 20 min, followed by addition of Vero-cSLAM cells. Neutralizing titers are expressed as the reciprocal of the highest dilution at which no cytopathic effect (CPE) was observed after 4 days.

Cytokine mRNA induction analysis.

Relative IFN-α, IFN-β, MxA, IFIT1, and ISG-15 or IFN-α, IFN-β, and Mx-1 message levels present in cultured HeLa cells or ferret PBMCs, respectively, were determined by semiquantitative real-time PCR analysis as described previously (38). Briefly, HeLa cells were spin inoculated (2,000 × g for 30 min at 4°C) with an MOI of 1 and total RNA was isolated at 20 h postinfection. For PBMCs, RNA was isolated from cells collected on days 3 and 7 after infection, or from CDV-naive ferrets representing day 0. In all cases, RNA was reverse transcribed using the SuperScript III reverse transcriptase kit (Invitrogen), and 10 ng of the resulting cDNAs was subjected to real-time PCR using the Fast SYBR green master mix (Applied Biosystems) or QuantiFast SYBR green PCR kit (Qiagen). GAPDH mRNA served as an internal control, and mRNA induction levels were normalized to the average of mock-infected cells (HeLa) or average of the day 0 values (PBMCs), respectively. The relative changes in transcription levels were calculated according to fold change in threshold cycle (2^−ΔΔCT) (66).

Deep sequencing of viral genomes.

RNA was extracted from infected cells using the ZR viral RNA kit (Zymo Research) according to the manufacturer's instructions. Metagenomic next-generation sequencing libraries were constructed as described previously (67). Briefly, 20 μl of extracted RNA was reverse transcribed using SuperScript III reverse transcriptase (Thermo), and second-strand synthesis was performed using Sequenase v2.0 (Agilent). cDNA was purified using DNA Clean and Concentrator-5 (Zymo) and subjected to Nextera XT tagmentation (Illumina) followed by 19 cycles of PCR amplification and a 0.8× Ampure XP cleanup (Beckman Coulter). Libraries were sequenced on a 2 × 300-bp run on an Illumina MiSeq. Sequencing reads were adapter and quality trimmed (Q20) using cutadapt (68). Majority consensus genomes were called via mapping trimmed reads to the canine distemper virus reference genome (NC_001921) in Geneious v9.1, and allele frequencies were called via remapping trimmed reads to the majority consensus genome.

MiSeq-based quantitation of relative mRNA editing efficiency.

Next-generation sequencing using MiSeq (Illumina) was performed to determine the ratio of P/V/W transcripts in recMeV constructs. When CPE reached approximately 75 to 90%, total RNA was isolated from infected cells using the RNeasy kit (Qiagen) and cDNAs of mRNA transcripts were synthesized using oligo(dT) primers and SuperScript III reverse transcriptase (Invitrogen). MiSeq primers with Illumina overhangs targeting a region of ∼500 nucleotides surrounding the P gene editing site were used to PCR amplify cDNAs representing all P, V, and W protein transcripts. Sequences were queried for the relative ratios of P, V, and W protein transcripts for each sample.

Profiling of viral RNA populations in infected cells.

To monitor relative quantities of viral genomic and mRNA populations during virus replication, Vero-cSLAM cells were infected with the recCDV strains at an MOI of 0.1 TCID₅₀ unit per cell and total RNA was isolated 24 h posttransfection using the RNeasy kit (Qiagen). To quantify relative genome copy numbers, a specific primer located in the viral trailer sequence was used for first-strand cDNA synthesis with SuperScript III reverse transcriptase, while oligo(dT) primers were applied to determine relative amounts of message. Relative RNA ratios were determined in an Applied Biosystems 7500 real-time PCR system using Fast SYBR green master mix (Thermo Fisher Scientific) and appropriate primer pairs annealing in the N, L, and mKate protein and host GAPDH ORFs, respectively, or flanking the N/P or mKate/L protein ORF intergenic junctions as specified in the figure legends. To calculate relative differences between distinct viral RNA populations or relative to RNA present in standard recCDV-infected cells, threshold cycle values obtained for each sample were standardized to expression levels of cellular GAPDH as a reference, and then ΔΔC_T values were determined by normalization of the standardized values to the individual reference RNA population as specified in the figure legends.

Statistical analysis.

To assess experimental variation and the statistical significance of differences between sample means, one-way or two-way analysis of variance (ANOVA) was carried out in combination with Tukey's or Sidak's post hoc tests as specified in the figure legends, using the Prism (GraphPad) software package. For virus growth profiling, regression models were built using Bindslev's population growth four-parameter variable slope model. Results for individual biological replicates are shown wherever possible. When appropriate, experimental uncertainties are identified by error bars, representing standard deviations (SD) or standard errors of the means (SEM) as specified in the figure legends.

Ethics statement.

All animal experiments were approved by the responsible state authority, the Regierungspräsidium Darmstadt (approval no. F107/121), and carried out in compliance with the regulations of German animal protection law.

Accession number(s).

All deep sequencing files of CDV genomes are available from GenBank under accession numbers MG228429, MG228443, MG228444, MG228431, MG228437, MG228439, MG228440, MG228441, MG228432, MG228433, MG228434, MG228435, and MG228436. All other relevant data are within the paper and/or available from the authors.