Development and Characterization of a Reverse-Genetics System for Influenza D Virus

Influenza D virus (IDV) is a new type of influenza virus that uses cattle as the primary reservoir and infects multiple agricultural animals. Increased outbreaks in pigs and serological and genetic evidence of human infection have raised concerns about potential IDV adaptation in humans. Here, we have developed a plasmid-based IDV reverse-genetics system that can generate infectious viruses with replication kinetics similar to those of wild-type viruses following transfection of cultured cells. Further characterization demonstrated that viruses rescued from the described RGS resembled the parental viruses in biological and receptor-binding properties. We also developed and validated an IDV minireplicon reporter system that specifically measures viral RNA polymerase activity. In summary, the reverse-genetics system and minireplicon reporter assay described in this study should be of value in identifying viral determinants of cross-species transmission and pathogenicity of novel influenza D viruses.

ABSTRACT

Influenza D virus (IDV) of the Orthomyxoviridae family has a wide host range and a broad geographical distribution. Recent IDV outbreaks in swine along with serological and genetic evidence of IDV infection in humans have raised concerns regarding the zoonotic potential of this virus. To better study IDV at the molecular level, a reverse-genetics system (RGS) is urgently needed, but to date, no RGS had been described for IDV. In this study, we rescued the recombinant influenza D/swine/Oklahoma/1314/2011 (D/OK) virus by using a bidirectional seven-plasmid-based system and further characterized rescued viruses in terms of growth kinetics, replication stability, and receptor-binding capacity. Our results collectively demonstrated that RGS-derived viruses resembled the parental viruses for these properties, thereby supporting the utility of this RGS to study IDV infection biology. In addition, we developed an IDV minigenome replication assay and identified the E697K mutation in PB1 and the L462F mutation in PB2 that directly affected the activity of the IDV ribonucleoprotein (RNP) complex, resulting in either attenuated or replication-incompetent viruses. Finally, by using the minigenome replication assay, we demonstrated that a single nucleotide polymorphism at position 5 of the 3′ conserved noncoding region in IDV and influenza C virus (ICV) resulted in the inefficient cross-recognition of the heterotypic promoter by the viral RNP complex. In conclusion, we successfully developed a minigenome replication assay and a robust reverse-genetics system that can be used to further study replication, tropism, and pathogenesis of IDV.

IMPORTANCE Influenza D virus (IDV) is a new type of influenza virus that uses cattle as the primary reservoir and infects multiple agricultural animals. Increased outbreaks in pigs and serological and genetic evidence of human infection have raised concerns about potential IDV adaptation in humans. Here, we have developed a plasmid-based IDV reverse-genetics system that can generate infectious viruses with replication kinetics similar to those of wild-type viruses following transfection of cultured cells. Further characterization demonstrated that viruses rescued from the described RGS resembled the parental viruses in biological and receptor-binding properties. We also developed and validated an IDV minireplicon reporter system that specifically measures viral RNA polymerase activity. In summary, the reverse-genetics system and minireplicon reporter assay described in this study should be of value in identifying viral determinants of cross-species transmission and pathogenicity of novel influenza D viruses.

INTRODUCTION

Influenza viruses are enveloped, segmented, single-stranded, negative-sense RNA viruses that belong to the Orthomyxoviridae family. Four influenza types, A, B, C, and D, are classified on the basis of antigenic differences in the nucleoprotein (NP) and the matrix protein (M). Among these four types, influenza A virus (IAV) is the most common and the most pathogenic, causing seasonal epidemics every year in the Northern and Southern hemispheres and periodic pandemics (1). Despite lacking the ability to trigger pandemics, influenza B virus (IBV) also causes annual epidemics frequently associated with deaths in people (2). Influenza C virus (ICV) is generally not associated with annual influenza epidemics and gives rise to only mild respiratory infections in humans (3). Influenza D virus (IDV) was first isolated in 2011 (4) and officially named in 2016 (https://www.cdc.gov/flu/about/viruses/types.htm). Soon after its discovery, similar viruses were successively identified from swine and/or cattle in North and Central America, Asia, Europe, and Africa (5–16). It was shown that IDV utilizes cattle as a primary reservoir and amplification host, with periodical spillover to other mammalian hosts (17, 18). In addition to swine and bovines, antibodies against IDV were detected in sera from small ruminants (goats and sheep), horses, camels, and humans (especially cattle workers) (15, 16, 18–21). Significantly, a more recent study showed that the IDV genome was detected in nasal wash samples of a swine farm worker in Malaysia, Asia (22). Furthermore, molecular surveillance of respiratory viruses with bioaerosol sampling in the Raleigh-Durham International Airport found that among 4 (17%) of the 24 samples positive for known respiratory pathogens, 1 was positive specifically for IDV (23). It should be noted that none of the 24 samples tested positive for influenza A, B, and C viruses. Using a similar approach, one study detected IDV in hospital visitors in North Carolina (24). In addition to these epidemiological studies, several clinical studies were performed in cattle, showing that IDV replication can damage the structure of epithelial cells lining the respiratory tract and cause mild respiratory disease in infected animals (4, 25, 26). Considering the worldwide distribution and broad host range of IDV and its potential to adapt to humans, it is necessary to further characterize this novel influenza virus at the molecular level.

Unlike the genomes of IAV and IBV that consist of eight segments, ICV and IDV contain only seven genome segments. The three longest segments of IDV encode the polymerase subunits PB2, PB1, and P3, which form a heterotrimer to catalyze the transcription and replication of the viral genome RNA (vRNA). Transcription of vRNA to mRNA starts with the “cap-snatching” reaction, a process during which capped RNAs are bound by the cap-binding domain of the PB2 subunit and cleaved by the endonuclease of the P3 subunit; actual RNA synthesis is performed by the PB1 subunit that forms the core structure of the heterotrimeric RNA polymerase complex (18, 27). The fourth segment produces a major surface glycoprotein, the hemagglutinin-esterase fusion (HEF), which harbors viral receptor-binding, receptor-destroying, and membrane fusion activities (4, 28). The fifth segment encodes the nucleoprotein (NP), which, together with the vRNA and the polymerase complex, forms the ribonucleoprotein (RNP) complex. Each vRNA segment forms an RNP complex, and the RNP is the fundamental unit for vRNA transcription and replication (29). The sixth segment encodes the matrix proteins DM1 and DM2, and the DM2 protein exhibits the ion channel activity (30). The last segment produces two nonstructural proteins, NS1 and NS2. To date, the biological function of the individual proteins of IAV has been well studied, while very limited work has been performed to characterize functions of IDV proteins.

Reverse genetics is one of the most powerful tools to study the biological properties and molecular characteristics of influenza viruses. But until now, no system for the rescue of IDV has been reported. Over the past 20 years and more, several plasmid-based systems for the rescue of recombinant IAV, IBV, and ICV have been successfully developed (31–41). To rescue an influenza virus in the unidirectional system, plasmids with the human RNA polymerase I (Pol I) promoter and the hepatitis delta virus ribozyme sequence or the mouse Pol I terminator are used to generate vRNA-like transcripts and are cotransfected with four other helper plasmids expressing PB2, PB1, PA/P3, and NP (38, 39, 41). In the bidirectional system, viral cDNA is inserted between the human Pol I promoter and the murine Pol I terminator, and the entire Pol I transcription unit is flanked by the bovine growth hormone (BGH) polyadenylation site or the simian virus 40 (SV40) late mRNA polyadenylation signal and the cytomegalovirus (CMV) Pol II promoter (31, 32, 37). Plasmids with the Pol I-Pol II transcription units allow the generation of both vRNA and mRNA from one viral cDNA template via the two different orientations. A modified system has been made to reduce the number of plasmids required for virus generation. This system combines multiple Pol I transcription units on one plasmid and allows the efficient generation of IAV in Vero cells (33). Finally, to overcome the limitation of species specificity of the Pol I transcription system, a T7 RNA polymerase-driven reverse-genetics system (RGS) was developed for efficient rescue of influenza viruses in human, avian, and canine cells (40).

Here, we describe the first RGS for IDV. By using a bidirectional seven-plasmid-based system, we generated recombinant influenza D/swine/Oklahoma/1314/2011 (D/OK) viruses. Their biological characteristics were evaluated in detail, and our analyses showed that RGS-rescued viruses had replication and receptor-binding properties similar to those of wild-type viruses in cell culture. We also developed a minigenome replication assay for IDV. This assay was successfully used to identify determinants of the activity of the IDV RNP complex essential for viral replication. In summary, we have successfully developed a minigenome replication assay and reverse-genetics system, and these tools can be used to further investigate the biology, tissue tropism, and transmission of IDV.

RESULTS

Generation of infectious influenza D/swine/Oklahoma/1314/2011 virus by the reverse-genetics system.

The cDNAs of the seven genomic segments (PB2, PB1, P3, HEF, NP, M, and NS) of influenza virus D/swine/Oklahoma/1314/2011 (D/OK) were amplified by reverse transcription-PCR (RT-PCR) and cloned individually into the bidirectional vector pHW2000 (Fig. 1A). The pHW2000 vector contains the human Pol I promoter and the murine Pol I terminator that direct the precise synthesis of vRNA from one strand, flanked by the truncated CMV promoter and the BGH poly(A) signal that direct viral mRNA transcription from the opposite strand (31) (Fig. 1A). The seven plasmids encoding each of the genomic segments of D/OK virus were cotransfected into cocultured HEK-293T (293T)–Madin-Darby canine kidney (MDCK) cells. Five days after transfection, supernatants were collected and inoculated onto fresh MDCK cells. Three to five days after inoculation, infectious virus was detected only when all seven plasmids were present (data not shown). No virus was detected in the control group in which the plasmid encoding the PB2 segment was excluded. These results indicated that the infectious virus was generated from cloned plasmids following transfection.

FIG 1.

Generation of infectious influenza D/swine/Oklahoma/1314/2011 virus by the reverse-genetics system. (A) Schematic representation of the bidirectional seven-plasmid-based reverse-genetics system for IDV. The seven IDV cDNA segments were cloned into the pHW2000 vector (31). Each of the cDNA segments was flanked by the human Pol I promoter and the murine terminator. The RNA Pol I transcription unit was flanked by the truncated CMV promoter and the bovine growth hormone (BGH) poly(A) signal. (B and C) Growth kinetics of the wild-type D/OK and the rescued D/OK-RGS viruses. MDCK cells were infected with the D/OK or D/OK-RGS virus at an MOI of 0.1 (B) or an MOI of 0.01 (C). Samples of supernatants were collected at the indicated times, and virus titers were then determined in MDCK cells by a TCID₅₀ assay. The data presented in panels B and C are representative of results from three independent experiments, with each experiment analyzing samples in duplicate.

To determine whether there is any difference in replication between wild-type D/OK virus and rescued D/OK-RGS virus, we studied the growth kinetics of these viruses in MDCK cells. The cells were infected with the viruses at multiplicities of infection (MOIs) of 0.1 and 0.01. Samples of supernatants were collected every 24 h from day 0 to day 6, and virus titers were then determined by a 50% tissue culture infective dose (TCID₅₀) assay. The wild-type D/OK virus and the rescued D/OK-RGS virus showed similar growth kinetics in MDCK cells infected at an MOI of 0.1 (Fig. 1B) or an MOI of 0.01 (Fig. 1C). Taken together, these data led us to conclude that the influenza D/OK virus was successfully rescued by the reverse-genetics system presented here.

Replication kinetics of continuously passaged influenza D/OK-RGS viruses at different temperatures.

To ensure that rescued viruses from the transient transfection of 293T-MDCK cell cocultures with IDV RGS plasmids are stable in terms of infectivity during the course of serial passages, we conducted five more passages in MDCK cells using the RGS-derived virus and then analyzed their infectivity using the TCID₅₀ assay. As shown in Fig. 2A, similar virus stock titers were observed for D/OK-RGS viruses from passage 1 (P1) to P5, ranging from 5.62 × 10⁴ TCID₅₀/ml to 3.80 × 10⁵ TCID₅₀/ml. This result indicates that viruses rescued from our IDV RGS are stable and that the developed genetic system can be further exploited to study IDV infection biology.

FIG 2.

Growth kinetics of continuously passaged influenza D/OK-RGS viruses at different temperatures. (A) MDCK cells were infected with D/OK-RGS virus at an MOI of 0.01. After 5 days of infection, culture supernatants were collected and used for the next passage in MDCK cells. Virus titers at the indicated passages were determined by a TCID₅₀ assay. (B and C) MDCK cells were infected with different passages of D/OK-RGS virus at an MOI of 0.01, followed by further incubation at 37°C (B) or 33°C (C). Samples of supernatants were collected at the indicated days and then titrated by a TCID₅₀ assay. The data presented are representative of results from three independent experiments, with each experiment analyzing samples in duplicate.

Influenza D virus differs substantially from the related human influenza C virus (ICV) in temperature-restricted replication. Specifically, IDV replication is permissive at both 33°C and 37°C, while ICV replication is largely restricted to 33°C. Next, we determined whether RGS-derived viruses replicated in the same manner as the parental virus at both temperatures. As summarized in Fig. 2B and C, different passages of D/OK-RGS viruses (P2 to P5) replicated efficiently at both 33°C and 37°C, although the overall replication efficiency of these viruses was slightly lower at 33°C than at 37°C (Fig. 2B and C). Furthermore, the serially passaged D/OK-RGS viruses showed similar growth kinetics at either 33°C or 37°C (Fig. 2B and C). These data collectively show that IDV RGS-derived viruses, similar to the parental virus, replicate efficiently at both 33°C and 37°C, which further supports the utility of IDV RGS in future mechanistic studies toward elucidating a genetic basis of the temperature-mediated restriction that separates IDV from its related ICV.

Receptor binding of the wild-type D/OK and rescued D/OK-RGS viruses.

Viral attachment to receptors on the host cell surface is the initial step of the virus life cycle. Sialic acids (SAs) have been found to serve as receptors for influenza viruses. N-acetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc) are the most common SAs found on mammalian cell surfaces, both of which are ligands for influenza viruses. SAs are generally attached to glycan chains on glycoproteins or glycolipids via different glycosidic linkages. The most frequent linkage types are α2,3 and α2,6 linked to a galactose residue. IAV and IBV recognize α2,3- or α2,6-linked sialic acid moieties as receptors (42). However, ICV utilizes N-acetyl-9-O-acetylneuraminic acid (Neu5,9Ac₂), which has an additional O-acetyl group at the C-9 position as the receptor; this is independent of the α2,3- or α2,6-linkage to the following galactosyl residue (3). A recent study shows that the IDV HEF protein can bind both Neu5,9Ac₂ and 9-O-acetyl-N-glycolylneuraminic acid (Neu5Gc9Ac), regardless the α2,3- or α2,6-linkage, which indicates that IDV uses 9-O-Ac-SAs as its receptor (28). The chemical structures of these SAs are shown in Fig. 3A.

FIG 3.

Receptor-binding properties of the wild-type D/OK virus and the recombinant D/OK-RGS virus. (A) Chemical structures of sialic acids that serve as receptors for influenza viruses. (B and C) Inhibition of viral hemagglutination by receptor analogs. The receptor analog Neu5,9Ac₂ (B) or Neu5Gc9Ac (C) at the indicated concentrations was added to the virus containing 4 HA units. Mixtures were incubated for 30 min at room temperature. Aliquots of turkey RBCs were then added to the mixtures, and results were read after 30 min at room temperature. PBS was used as a negative control. Note that “−” indicates no hemagglutination, while “+” denotes evident hemagglutination. The data presented in panels B and C are representative of results from four independent experiments, with each experiment analyzing samples in duplicate.

To further investigate whether the RGS-derived IDV resembles the parental virus in its receptor-binding profile, we used the traditional hemagglutination (HA) assay-based competitive inhibition approach to determine their receptor-binding properties, replacing antibody with synthetic receptor analogs. The viruses were incubated with the receptor analogs at various concentrations ranging from 0 to 1,024 ng/μl. Both D/OK and D/OK-RGS viruses were capable of agglutinating turkey red blood cells (RBCs) in the presence of lower concentrations of receptor analogs (Fig. 3B and C). However, the HA activity of both viruses was completely inhibited in the presence of Neu5,9Ac₂ at concentrations of 128 ng/μl and above (Fig. 3B) or in the presence of Neu5Gc9Ac at a concentration of 512 ng/μl (Fig. 3C). Because the nonspecific reaction (agglutination) occurred between RBCs and higher concentrations of Neu5Gc9Ac, no HA inhibition was observed for both viruses in the presence of 1,204 ng/μl Neu5Gc9Ac (Fig. 3C). In summary, the results suggested that the wild-type D/OK and rescued D/OK-RGS viruses had the same capacity to interact with the sialic acid receptors Neu5,9Ac₂ and Neu5Gc9Ac. The data presented here are important, as they further confirm the utility of our reported IDV RGS in future studies toward illustrating the molecular and chemical details of IDV receptor binding and entry.

Activity of the IDV RNP complex with PB1-E697K or PB2-L462F mutation.

During the course of IDV RGS development, we identified two key mutations (PB1-E697K and PB2-L462F) that prevented us from successfully rescuing infectious virus particles. These two mutations were likely introduced by reverse transcription or PCR. These mutations have not been observed in natural IDV isolates. Here, we took advantage of these two coincident mutations and used them to assess the robustness of the newly developed IDV RGS. As demonstrated in Fig. 4B, we found that the PB1-E697K mutation caused attenuation of viral replication, while the PB2-L462F mutation rendered the virus replication incompetent. To find out the underlying mechanisms, we developed a minigenome replication assay to investigate whether the PB1-E697K or PB2-L462F mutation directly affects the activity of the IDV RNP complex. We synthesized a reporter plasmid, pUC57_mini-D/OK-HEF-Reporter, that contains the green fluorescent protein (GFP) reporter gene located between the 5′- and 3′-end sequences of the viral cRNA of the D/OK HEF segment and flanked by the Pol I terminator and the human RNA Pol I promoter (Fig. 4A). This reporter plasmid was cotransfected with plasmids encoding PB2, PB1, P3, and NP into 293T cells. The cellular RNA Pol I bound to the Pol I promoter to produce a reporter transcript that mimics the HEF vRNA, and the incoming viral polymerase complex then recognized the HEF terminal regions (viral promoter sequence) on the reporter transcript, initiating mRNA production for the expression of the GFP reporter. The GFP reporter was qualitatively detected by fluorescence microscopy (Fig. 4C), while its expression level was quantitatively analyzed by both Western blotting (Fig. 4D) and fluorescence-activated cell sorter (FACS) analysis (Fig. 4E).

FIG 4.

Activity of the IDV RNP complex with the PB1-E697K or PB2-L462F mutation. (A) Schematic diagram of the pUC57_mini-D/OK-HEF-Reporter plasmid. This reporter plasmid contains the GFP reporter gene inserted between the D/OK-HEF cRNA 5′ and 3′ ends and then flanked by the Pol I terminator and the human RNA Pol I promoter. All ATG codons before the GFP translation initiation codon in the construct were mutated to CTGs; thus, the translation of GFP utilized its own start codon. (B) Virus titers for PB1-E697K and PB2-L462F mutants were determined by a TCID₅₀ assay. Note that IDV OK-RGS^660-PB1 is identical to IDV OK-RGS except for the PB1 segment derived from IDV D/660. (C to E) Activity of the IDV wild-type RNP complex or the RNP complex with the PB1-E697K or PB2-L462F mutation was measured by a minigenome replication assay. In this assay, the GFP reporter plasmid and the wild-type or mutant plasmids encoding IDV RNP complex components were cotransfected into HEK-293T cells. At 48 h posttransfection, cells were collected, and the GFP reporter was detected and analyzed by fluorescence microscopy (C), Western blotting (D), and FACS analysis (E). NP or β-actin was also detected by Western blotting, which was set up as a transfection or loading control, respectively. Note that densitometry of Western blot bands was quantified by using ImageJ software (https://imagej.nih.gov/ij/). Specifically, the density of the GFP band (i.e., surrogate of viral RNP activity) was first normalized by the value obtained with the input transfected NP control (i.e., to gauge the transfection variability). Relative viral RNP activity was determined by setting the level of the wild-type “OK-RNP” group to 1.00. (F) Mutations PB1-E697K and PB2-L462F are localized on the complex structure of RNA polymerase from ICV (PDB accession number 5D98). The structure of RNA polymerase was downloaded from the PDB (https://www.rcsb.org/) and was shown in PyMOL. The data presented in panels B to E are representative of results from three independent experiments, with each experiment analyzing samples in duplicate. eGFP, enhanced GFP.

We analyzed the activity of the IDV RNP complex with the PB2-L462F or PB1-E697K mutation by using the minigenome replication assay. Interestingly, no green fluorescence was observed when wild-type PB2 was replaced by the mutated PB2-L462F segment (Fig. 4C). In support of this observation, the expression of GFP was below the detection level as either measured by Western blotting (Fig. 4D) or quantified by FACS analysis (Fig. 4E). These results indicated that the L462F mutation on PB2 severely abolished the activity of the IDV RNP complex. For the PB1-E697K mutation that was observed only in the influenza virus D/bovine/Oklahoma/660/2013 (D/660) PB1 segment during our initial effort in the development of the IDV RGS, we first determined whether the wild-type D/660 PB1 segment could replace its counterpart (D/OK PB1) in the minigenome replicon assay. As demonstrated in Fig. 4C to E, the substitution had no detectable effect on D/OK RNP activity. Interestingly, the RNP activity, as indicated by GFP reporter levels, was significantly decreased when wild-type D/OK PB1 was replaced with the mutant D/660 PB1-E697K segment (Fig. 4C to E), indicating an essential role of PB1 E697 in viral genome replication of IDV. In summary, the PB2-L462F and PB1-E697K mutations, encountered during initial IDV RGS development, can lead to either replication-incompetent or replication-attenuated virus, likely caused by their direct effect on the activity of the IDV RNP complex.

To further investigate the underlying mechanisms by which PB1-E697K and PB2-L462F mutations impair RNP complex activity, we took advantage of the recently resolved structure of the RNA polymerase complex from ICV (PDB accession number 5D98) in the absence of promoter RNA (43) to structurally visualize the location of these two mutations. Since the PB1 and PB2 proteins show high sequence similarity between ICV and IDV (4), and the corresponding positions are absolutely conserved in ICV, we assumed that the IDV RNA polymerase complex could adopt a structure similar to that of ICV and that structural modeling could offer mechanistic insight into mutation-associated replication defects. Our analysis showed that PB1 E697 most likely binds PB2 through hydrogen bonding to PB2 K207 (3.1 Å between E697 OE1 and K207 NZ atoms) and also forms an anion-pi interaction with PB2 F175 (∼3.6 Å between E697 OE1 and F175 CZ atoms) (Fig. 4F). The PB1-E697K mutation likely disrupts such interactions. PB2 L462 is located at the hydrophobic region of the cap-binding domain of PB2; thus, the PB2-L462F mutation may alter the stability or the conformation of the domain (Fig. 4F). However, such interpretations should be taken cautiously because the structural resolution of the ICV RNP complex is relatively suboptimal (around 3.9 Å), and the side-chain locations may be inaccurate. Further structure-function studies of IDV RNP will test these hypotheses.

Recognition and transcription of the heterotypic promoter sequence by the IDV polymerase complex.

Our previous work showed that the 1-nucleotide difference observed in the conserved 3′ noncoding ends of the vRNA segments between IDV and ICV could contribute to the failure of functional reassortment between these two viruses (4). It should be noted that in addition to this polymorphism at position 5 within the 3′ noncoding sequences, the noncoding regions (NCRs) across the seven segments between IDV and ICV are relatively variable in both sequence and length, especially after the first 11 to 12 nucleotides (Fig. 5A). These sequence variations as a whole may have a negative impact on the efficient cross-recognition of a heterotypic promoter by the viral RNP complex, resulting in the inhibition of vRNA replication and limiting functional reassortment between IDV and ICV (17). As a first step to address this hypothesis, we sought to determine whether the single polymorphism at position 5 between IDV and ICV affected the inefficient cross-recognition of the heterotypic promoter by the viral RNP complex. In this regard, we analyzed the activities of viral RNP complexes when cognate and noncognate promoters were provided in the minigenome replication assay.

FIG 5.

Activities of IDV and ICV RNP complexes with the conserved cognate or noncognate noncoding ends of vRNA segments. (A) Diagram showing nucleotide sequences of both the 3′ and 5′ noncoding regions in the M segment of influenza C and D viruses within the context of the GFP-based reporter. Note that the fifth nucleotide of the 3′ noncoding end is different between the M segment sequences of influenza C and D viruses and is in red with boldface type. “D(A→C)” indicates a mutated IDV reporter in which the fifth nucleotide, A, on the 3′ noncoding end of the D/OK M segment is altered to C, which makes it like the C/Vic M segment 3′ NCR in the first 11 nucleotides. (B to D) Activities of IDV and ICV RNP complexes with the cognate or noncognate promoter were determined by the minigenome replication assay. HEK-293T cells were cotransfected with the ICV or IDV M segment promoter sequence-based GFP reporter plasmid and the plasmids expressing RNP complex components from either D/OK or C/Vic. At 48 h posttransfection, cells were collected, and GFP reporter protein expression was analyzed by fluorescence microscopy (B), Western blotting (C), and FACS analysis (D). Renilla luciferase or β-actin detected by Western blotting was set up as a transfection or loading control. Note that densitometry of Western blot bands was quantified by using ImageJ software (https://imagej.nih.gov/ij/) (C). Specifically, the density of the GFP band (i.e., surrogate of viral RNP activity) was first normalized by the value obtained with the input transfected Renilla luciferase control (i.e., to gauge the transfection variability). Relative viral RNP activity was determined by setting the level of the wild-type group involving C-RNP/C-reporter or D-RNP/D-reporter (containing the respective cognate promoter) to 1.00. The data presented in panels B to D are representative of results from three independent experiments, with each experiment analyzing samples in duplicate.

Similar to the generation of the HEF segment-based GFP reporter plasmid, we constructed D/OK and influenza virus C/Victoria/2/2012 (C/Vic) M segment-based GFP reporter plasmids in which the full-length 5′ and 3′ noncoding regions of both M segments (containing both conserved and variable sequences) were used to direct the expression of the GFP reporter (Fig. 5A). HEK-293T cells were cotransfected with plasmids expressing the RNP complex components from either D/OK or C/Vic and a reporter plasmid that contains the 5′ and 3′ noncoding regions of the M segment from one of the two viruses. According to the fluorescence microscopy and Western blot results, the D/OK RNP complex (D-RNP) or the C/Vic RNP complex (C-RNP) initiated the expression of GFP not only from its cognate reporter but also from the noncognate reporter (Fig. 5B and C). However, the expression levels were visibly different between the cognate and noncognate groups, which showed that D-RNP with the cognate reporter or C-RNP with the cognate reporter produced higher levels of GFP than D-RNP with the ICV reporter or C-RNP with the IDV reporter, respectively (Fig. 5B and C). To further confirm these results, we quantified GFP expression by FACS analysis. There were 39.4% GFP-positive cells in the D-RNP-plus-IDV reporter group but only 22.4% GFP-positive cells in the D-RNP-plus-ICV reporter group (Fig. 5D). Similarly, C-RNP working with the ICV reporter generated 34.9% of the positive GFP expression, while C-RNP working with the IDV reporter generated only 19.7% of the positive GFP expression (Fig. 5D). These results indicated that IDV RNP or ICV RNP preferred using its cognate promoter to express proteins.

Since there is only one difference in the first 11 nucleotides of the 3′ NCR between IDV and ICV (4, 17), it is very likely that this nucleotide difference affects recognition and transcription of the heterotypic model GFP vRNA by the IDV or ICV polymerase complex. To address this hypothesis, we constructed a mutated IDV reporter plasmid in which the fifth nucleotide, A, on the 3′ NCR of the D/OK M segment was altered to C, which made it like the ICV M segment 3′ NCR in the first 11 nucleotides (Fig. 5A). It should be noted that the first 11 nucleotides are absolutely conserved in the 5′ NCR of the M segment between IDV and ICV. Interestingly, this substitution gave rise to significant increases in the expression and replication of the mutant IDV reporter by C-RNP compared to those observed for the wild-type IDV reporter with heterotypic C-RNP in the minireplicon assay (27% versus 19% GFP-positive cells) (Fig. 5B and D). Conversely, the substitution reduced substantially the activity of the IDV reporter by its cognate D-RNP in comparison to that observed in the minireplicon assay involving the wild-type IDV reporter and D-RNP (22% versus 39.4% GFP-positive cells) (Fig. 5B and D). Quantitative analysis of GFP expression levels in Western blot assays using ImageJ (https://imagej.nih.gov/ij/) also confirmed the differential activity observed in the cell-based minireplicon assay (Fig. 5C). For example, the GFP expression level from C-RNP/mutant D reporter containing an A-to-C substitution increased by more than 2-fold compared to that shown from the combination of C-RNP/D reporter. In contrast, the GFP expression level decreased by approximately 3-fold for the D-RNP group when the same mutant D reporter was provided in transfected cells. Altogether, these results revealed that the 1-nucleotide change in the highly conserved noncoding region negatively influences the heterotypic recognition and transcription of vRNAs by the viral RNP complex. The observed heterotypic promoter incompatibility together with other viral factors, such as packaging sequence variation, may contribute to the failure of viable reassortment between IDV and the closely related ICV, which warrants further investigation.

DISCUSSION

Reverse-genetics systems (RGSs) have already been described for IAV, IBV, and ICV but not yet for IDV. In this study, the first RGS of IDV was successfully developed using a bidirectional seven-plasmid-based system, and the resultant viruses were studied in comparison with the wild-type viruses. We also developed a minigenome replication assay for IDV. Biological characterizations involving mutations collectively demonstrated the robustness and versatility of these two systems, indicating their utility as powerful tools to study structure-function relationships of IDV proteins. The IDV RGS can also be used to elucidate critical determinants that drive IDV to use this agricultural animal as a primary reservoir and amplification host.

Many different ways have been established in the past 2 decades to generate recombinant influenza viruses, including earlier helper virus-dependent methods and later helper virus-independent plasmid-based systems. In helper virus-dependent methods, selection strategies are required because additional helper viruses are used and therefore have to be depleted through drug selection. Owing to this disadvantage, helper virus-dependent methods have been largely replaced by later helper virus-independent plasmid-based methods (36). The unidirectional and bidirectional plasmid-based systems have emerged as two predominant helper virus-independent RGSs, which have been previously reported for IAV, IBV, and ICV (31, 32, 35, 37–39, 41). In the present study, we succeeded in rescuing recombinant IDVs by using a bidirectional seven-plasmid-based system (Fig. 1), which is the first RGS reported for IDV. Generally, it is difficult to compare the generation efficiencies and titers of different influenza virus RGSs. Many factors, such as the cell line selected, transfection efficiency, technical protocol, and virus replication fitness, may determine which strategy should be used. Initially, we failed to generate the recombinant D/OK virus when we directly followed protocols (31) that are commonly used for the rescue of influenza A viruses. We realized that the D/OK virus cannot induce significant cytopathic effects and has a much lower replication capacity than influenza A virus in MDCK cells. Moreover, the 293T cell line has weak adhesion to the surfaces of cell culture vessels and is extremely sensitive to tosylsulfonyl phenylalanyl chloromethyl ketone (TPCK)-trypsin, which is required for the cleavage of the HEF protein and probably activation of viral fusion. Therefore, we added a lower concentration of TPCK-trypsin at a later time point, which was optimal to maintain 293T cell viability and, as a result, provided sufficient time for rescue of the recombinant D/OK virus. After making these subtle but critical changes, we could detect the rescued D/OK virus in the supernatant of transfected 293T-MDCK cells at 4 to 5 days posttransfection, which was slower than the rescue of influenza A virus, which could be detected at 2 to 3 days posttransfection (31). These findings suggest that rescue strategies for different types or strains of influenza virus should be adjusted according to their replication property and fitness, which will be vital for the successful rescue of IDV or IDV-similar influenza viruses.

Our comparative in vitro growth kinetics studies showed indistinguishable replication fitness of both the parental and RGS-derived viruses, and the RGS viruses can be stably cultured over multiple passages, at both 33°C and 37°C (Fig. 1 and 2). In addition, the RGS viruses displayed a receptor-binding capacity indistinguishable from that displayed by the authentic virus (Fig. 3). These studies clearly suggested that an efficient reverse-genetics system was successfully established and allowed for the genetic manipulation of viral genomes and the generation of mutant viruses, which opens the door for studying the functional importance of viral proteins and molecular aspects of viral replication and pathogenicity.

It has been known that both IDV and ICV contain only one major surface glycoprotein, HEF, which plays multiple critical roles during the virus life cycle (3, 4). First, HEF binds to the receptor to initiate virus entry. In agreement with previous reports (28), our results showed that IDV HEF could recognize both Neu5,9Ac₂ and Neu5Gc9Ac (Fig. 3), which is different from ICV HEF, which prefers Neu5,9Ac₂ over Neu5Gc9Ac as a viral receptor (3). HEFs of ICV and IDV have about 53% homology and possess almost identical overall structures (28). One interesting difference between them is that the IDV HEF protein has an open channel in the receptor-binding region, while in the structurally identical region of ICV HEF, there is a salt bridge (28). This subtle difference may indicate that IDV has different receptor-binding properties compared to ICV. With the availability of the RGSs, we can directly investigate the diverse receptor-binding properties arising from the subtle genetic differences between IDV and ICV. Second, HEF was a critical determinant of IDV’s antigenic diversity. Amino acid substitutions near the HEF receptor-binding region are involved in antigenic variation of influenza viruses (44). To understand the molecular basis for IDV lineage-dependent antigenicity, we can use the RGS to introduce select amino acid changes individually or in combination into the IDV HEF protein and analyze the antigenic properties of these recombinant viruses to identify amino acids determining the antigenicity of IDV. Third, HEF exhibits unique thermodynamic properties. It is known that ICV grows more efficiently at 33°C than at 37°C (3, 45). HEF is considered to be an important restriction factor for the temperature sensitivity of ICV (45). Interestingly, IDV replicates well at both 33°C and 37°C (4) (Fig. 2). Our previous experiments showed that IDV was able to tolerate a high-temperature environment, and HEF is the primary determinant for the high thermal stability of IDV (46). It will be very interesting to use the RGS to determine the amino acids or protein motifs that account for the different thermodynamic properties between ICV HEF and IDV HEF.

During the development of the IDV RGS, we identified several mutations in the viral polymerase complex proteins that rendered the RGS less productive and led to the failed rescue of viable viruses. To investigate the biological significance of these mutations and gain mechanistic insight, we developed a minigenome replication assay for IDV. This system has been widely used to study influenza virus RNP activity by cotransfecting cells with four plasmids encoding PB2, PB1, PA/P3, and NP and a viral promoter-driven reporter plasmid (41, 47). Our results of the minigenome replication assay showed that the RNP activity was lost due to the introduction of the L462F mutation into D/OK PB2 (Fig. 4), which explained why the PB2-L462F mutation caused the failure of RGS in rescuing viruses. The E697K change in PB1 resulted in significantly lower RNP activity than for wild-type PB1 (Fig. 4). This finding was consistent with the observation that the recombinant IDV containing the PB1-E697K mutation had an extremely low virus titer (Fig. 4B). These two mutations on the polymerase complex proteins showed a significant negative impact on RNP activity and, as a result, posed a challenge in rescuing viable IDVs. These data also provide additional justification for exploring conserved polymerase function as a target for anti-influenza drug development.

By using the minigenome replication assay, we further demonstrated that a single nucleotide difference at position 5 of the 3′ NCR within the first 11 nucleotides between IDV and ICV resulted in the inefficient cross-recognition of the heterotypic promoter by the viral RNP complex. This functional incompatibility in vRNA replication between IDV and ICV may contribute to the failure of viable reassortment between IDV and ICV, as demonstrated in our previous study. In addition to this single polymorphism, the 3′ NCR also shows considerable sequence variation between IDV and ICV in the region toward the segment-specific protein-coding sequence. Their impact on constraining IDV and ICV reassortment should be further investigated. Because productive packaging of vRNAs and successful reassortment between influenza viruses require multiple cis-acting elements and trans-acting factors (48–50), we suppose that multiple factors are probably involved in collectively restricting productive reassortment between IDV and the closely related ICV in animal hosts. This hypothesis can be addressed using the robust IDV RGS presented in this study.

MATERIALS AND METHODS

Cells, viruses, and antibodies.

Human embryonic kidney HEK-293T cells (ATCC) and Madin-Darby canine kidney (MDCK) cells (ATCC) were maintained in Dulbecco’s modified Eagle medium (DMEM; Gibco, Invitrogen, USA) supplemented with 10% (vol/vol) fetal bovine serum (FBS; PAA Laboratories, Dartmouth, MA, USA) and 100 U/ml penicillin-streptomycin (Life Technologies, Carlsbad, CA, USA). All cells were cultured at 37°C in the presence of 5% CO₂. Influenza viruses D/swine/Oklahoma/1314/2011 (D/OK) and D/bovine/Oklahoma/660/2013 (D/660) were propagated in MDCK cells at 37°C. Influenza virus C/Victoria/2/2012 (C/Vic) was grown in MDCK cells at 33°C. Polyclonal rabbit antibody against IDV D/OK was generated in-house by immunizing rabbits with concentrated IDV particles that were inactivated by UV light treatment. Polyclonal rabbit anti-GFP antibody was obtained from Santa Cruz Biotechnology, while monoclonal anti-β-actin antibody was purchased from Sigma-Aldrich. Secondary antibodies such as IRDye 680RD donkey anti-rabbit IgG and IRDye 680RD goat anti-mouse IgG were purchased from Li-Cor.

Construction of plasmids.

Viral RNAs were isolated from infectious virus particles with the TRIzol LS reagent (Invitrogen, CA, USA) according to the manufacturer’s instructions. RT-PCR was performed with a high-capacity cDNA reverse transcription kit (Thermo Fisher Scientific) according to the protocol provided by the manufacturer. The primers used to amplify each individual segment in the RT-PCRs are summarized in Table 1. After digestion of the PCR products with BsmBI or BsaI, the fragments were purified and cloned into the pHW2000 vector (31) (Fig. 1A). All inserted cDNAs were sequenced and confirmed (GenScript, NJ, USA) to ensure that there were no unwanted mutations in the constructs.

TABLE 1.

Primers used in the RT-PCR and cloning experiments

Primer	Sequence (5′–3′)
PB2_IDV_BsaI_For	TATTGGTCTCAGGGAAGCATAAGCAGAGGATGTCACTACTATTAAC
PB2_IDV_BsaI_Rev	ATATGGTCTCGTATTAGCAGTAGCAAGAGGATTTTTTCAATG
PB1_IDV_BsmBI_For	TATTCGTCTCAGGGAAGCATAAGCAGAGGATTTTATAAAATGG
PB1_IDV_BsmBI_Rev	ATATCGTCTCGTATTAGCAGTAGCAAGAGGATTTTTCTGTT
P3_IDV_BsaI_For	TATTGGTCTCAGGGAAGCATAAGCAGGAGATTTAGAAATGTCTAG
P3_IDV_BsaI_Rev	ATATGGTCTCGTATTAGCAGTAGCAAGGAGATTTTTAACATTACAAG
HEF_IDV_BsmBI_For	TATTCGTCTCAGGGAAGCATAAGCAGGAGATTTTCAAAGATG
HEF_IDV_BsmBI_Rev	ATATCGTCTCGTATTAGCAGTAGCAAGGAGATTTTTTCTAAGATTC
NP_IDV_BsmBI_For	TATTCGTCTCAGGGAAGCATAAGCAGGAGATTATTAAGCAATATGG
NP_IDV_BsmBI_Rev	ATATCGTCTCGTATTAGCAGTAGCAAGGAGATTTTTTG
M_IDV_BsmBI_For	TATTCGTCTCAGGGAAGCATAAGCAGAGGATATTTTTGACGCAATG
M_IDV_BsmBI_Rev	ATATCGTCTCGTATTAGCAGTAGCAAGAGGATTTTTTCGC
NS_IDV_BsmBI_For	TATTCGTCTCAGGGAAGCATAAGCAGGGGTGTACAATTTCAATATG
NS_IDV_BsmBI_Rev	ATATCGTCTCGTATTAGCAGTAGCAAGGGGTTTTTTCATACTAAAG

Transfection and rescue of recombinant viruses.

Approximately 20 to 24 h before transfection, cells (∼6 × 10⁵ HEK-293T cells and ∼1.5 × 10⁵ MDCK cells) were plated in 2 ml of complete growth medium per well in a 6-well plate. Prior to transfection, cells were 50% to 70% confluent, and the old growth medium was replaced with fresh medium containing 10% FBS without antibiotics. The reverse-genetics plasmids of IDV (1 μg for each segment) were mixed and added into 200 μl of Opti-MEM. Next, 18 μl of polyethylenimine (PEI) was dropped into the premixed diluted plasmids. Mixtures were incubated for 20 min at room temperature and added into different areas of the wells. The culture vessel was gently rocked back and forth and from side to side to evenly distribute the DNA-PEI complexes. At 6 to 9 h posttransfection, the old medium was replaced with 2 ml of fresh DMEM without FBS but containing antibiotics. After another 36 to 42 h of incubation at 37°C, 0.5 ml of DMEM containing 1 μg/ml TPCK-trypsin (Thermo Fisher Scientific) was added to each well. Virus HA titers of ∼32 could be detected 3 to 5 days after transfection.

Virus replication assay.

Viruses were cultured and measured in MDCK cells in DMEM supplemented with 1 μg/ml TPCK-trypsin and 100 U/ml penicillin-streptomycin. Viral growth kinetics studies were performed on a monolayer of MDCK cells using an inoculum at a multiplicity of infection (MOI) of 0.1 or 0.01. One hour after incubation at 37°C, cells were washed twice with phosphate-buffered saline (PBS), and fresh DMEM with 1 μg/ml TPCK-trypsin and 100 U/ml penicillin-streptomycin was added. Samples were collected at 0, 24, 48, 72, 96, 120, and 144 h postinfection and titrated by a TCID₅₀ assay (51).

Hemagglutination-based competitive inhibition assay.

Twenty-five microliters of virus suspensions containing 4 HA units of virus was added to 25 μl of receptor analogs (Neu5,9Ac₂ or Neu5Gc9Ac) (Glycohub). The virus-receptor analog mixtures were incubated for 30 min at room temperature. Fifty microliters of 1% turkey red blood cells (RBCs) (Lampire Biological Laboratories, Pipersville, PA, USA) was then added to the mixtures, and results were read after 30 min. PBS was used as a negative control.

Minigenome replication assay.

A reporter plasmid, pUC57_mini-D/OK-HEF-Reporter, was synthesized by GenScript. It contains the GFP reporter gene inserted between the 5′- and 3′-end sequences of the viral cRNA of the D/OK HEF segment, and it is flanked by the Pol I terminator and the human RNA Pol I promoter (Fig. 4A). All ATG codons before the GFP translation initiation codon in the construct were mutated to CTGs; thus, the translation of GFP used its own start codon (Fig. 4A). The M segment promoter sequence-based GFP reporter plasmids for both ICV and IDV were generated by conventional PCR and mutagenesis. The reporter plasmid was cotransfected with plasmids expressing RNP complex components (PB2, PB1, P3, and NP) into HEK-293T cells. At 48 h posttransfection, cells were collected, and the GFP reporter protein was detected by fluorescence microscopy and Western blotting. Percentages of GFP-positive cells were quantified by flow cytometry.

Statistical analysis.

The data presented in this paper are representative of results from three independent experiments, with each experiment assaying samples in duplicate. Standard-deviation bars indicate the variations among experiments. To analyze the difference between groups, statistical analysis by one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test was performed using GraphPad Prism 5.0. Statistically significant differences are indicated in the figures (**, P < 0.01; ***, P < 0.001).