A method for the unbiased quantification of reassortment in segmented viruses

Abstract

Reassortment of segmented viruses can be an important source of genetic diversity underlying viral evolution and emergence. Methods for the quantification of reassortment have been described but are often cumbersome and best suited for the analysis of reassortment between highly divergent parental strains. While it is useful to understand the potential of divergent parents to reassort, outcomes of such heterologous reassortment are driven by differential selection acting on the progeny and are typically strain specific. To quantify reassortment robustly, a system free of differential selection is needed. We have generated such a system for influenza A virus and for mammalian orthoreovirus by constructing well-matched parental viruses carrying small genetic tags. The method utilizes high-resolution melt technology for the identification of reassortant viruses. Ease of sample preparation and data analysis enables streamlined genotyping of a large number of virus clones. The method described here thereby allows quantification of the efficiency of reassortment and can be applied to diverse segmented viruses.

Introduction

Virus genome organization varies widely across different families. Segmented genomes, which are comprised of distinct RNA molecules, have been documented in eleven virus families to date. Segment numbers vary, as do replication strategies. Each genome segment encodes a different protein or proteins which are involved in establishing a productive infection. During infection, cells can be co-infected by multiple parent viruses. In segmented viruses, co-infection events have the potential to give rise to many different progeny genotypes, fueling evolution.

Viruses with segmented genomes undergo a type of genetic recombination termed reassortment. Reassortment occurs when two or more parent viruses co-infecting a single cell exchange whole gene segments, which are then packaged together to yield progeny viruses with novel genotypes. This mode of generating diversity is unique to segmented viruses and contrasts with classical recombination, in which a chimeric genome is formed intramolecularly, through the combination of nucleic acid sequences derived from two viral genomes. Reassortment events most often yield attenuated progeny due to incompatibilities between nucleic acids and/or proteins derived from heterologous parents^1-5. There is, however, the potential for the coupling of compatible, beneficial alleles and the subsequent emergence of novel pathogens, as was observed in the 2009 influenza A virus (IAV) pandemic^{6, 7}.

To date, reassortment has been investigated in several viruses, including IAV and mammalian orthoreovirus (reovirus), using a variety of methods. Measurement of reassortment is often complicated by fitness differences among progeny viruses or a lack of sensitive quantification methods. In the former case, difficulties arise due to selection. When reassortment yields progeny of variable fitness, those with the most beneficial segment combinations will be amplified more rapidly. Preferential propagation of parental genotypes results in underestimation of the frequency of reassortment. Similarly, preferential amplification of certain reassortant viruses results in underestimation of population diversity. To avoid these issues, it is preferable to quantify reassortment between viruses of more similar genotypes. This can be challenging, however, as most detection methods rely on parental viruses being significantly different genetically. Detection of reassortment in genetically similar viruses requires sensitive molecular technologies, which were not available until relatively recently⁸.

The earliest method to identify reassortants utilized polyacrylamide gel electrophoresis^9-11. This method depends on all segments of each parental genome having different electrophoretic mobility. Thus, each segment must differ significantly in length and/or sequence. This requirement for significant differences for detection necessitates that parent viruses are divergent, which – as discussed above – can impact the fitness of reassortant progeny and reduce observed rates of reassortment. Additionally, there is a practical limitation on the number of samples that can feasibly be analyzed using this approach.

As an alternative approach, temperature-sensitive (ts) mutants have been used to quantify reassortment^{12, 13}. In this system, a single segment confers temperature-sensitivity in each parental virus and pairs of viruses used for co-infection carry ts mutations in differing segments. Temperature sensitivity is abrogated if these segments are exchanged for wild type segments from the opposite parent in a reassortment event. Culture of progeny viruses at the nonpermissive temperature results in selection of reassortants. Titration at the nonpermissive temperature therefore allows quantification of the frequency of exchange of the ts segments. This approach makes it possible to detect reassortment between identical parents (with the exception of the ts lesions), thus avoiding the confounding effects of protein or segment incompatibilities. However, it does not yield the frequency of reassortment between all virus segments but rather only the two ts segments^{12, 14}.

PCR-based methods can also be used to differentiate parental segments that differ sufficiently to allow the design of specific primers^15-17. Amplicons can be detected using gel electrophoresis with ethidium bromide staining or by determining C_t values in quantitative PCR. These methods may be limited when parental genomes are too similar to allow unique primer design for all segments.

Alternatively, whole or partial genome sequencing of clonal virus populations offers a very flexible approach to identify reassortants¹⁸. With sequencing as a read-out, the need for primer design does not constrain the parental viruses that can be examined in combination. Traditionally, the costs of sequencing approaches have prohibited their use on a larger scale. In recent years, however, the price of sequencing has decreased and the technology has improved.

To address the shortcomings of prior methodologies and enable robust quantification of reassortment, we present here two methodological innovations. First, to eliminate selection bias, we generated well-matched pairs of parental viruses that differ only by one or a handful of synonymous mutations in each genome segment. These introduced synonymous changes act as genetic markers to indicate the parental origin of each segment. Second, to streamline the detection of reassortant viruses, we applied high resolution melt analysis, a post-PCR method originally developed to differentiate single nucleotide polymorphisms in eukaryotic genomes⁸. Because this method is sensitive enough to detect single nucleotide differences, it allows for the quantification of reassortment between highly similar parental viruses. We have applied these approaches to both IAV and reovirus.

Materials and Methods

Cell lines and cell culture media

293T cells from the American Type Culture Collective (ATCC) were maintained in Dulbecco’s Modified Eagle Medium (DMEM; Gibco) supplemented with 10% fetal bovine serum (FBS; Atlanta Biologicals). A549 cells (ATCC CCL-185) were maintained in F-12K nutrient mixture with L-glutamine (Corning) supplemented with 10% FBS and penicillin (100 IU), and streptomycin (100 ug/mL) (PS; Corning). Baby hamster kidney cells stably expressing T7 RNA polymerase (BFIK-T7) cells¹⁹ were maintained in DMEM supplemented with 5% FBS, 2 mM L-Glutamine (Corning), PS and 1 mg/mL G418 (Gibco). Spinner-adapted L929 cells (gift from Bernardo Mainou) were grown in Joklik’s modified MEM supplemented with 5% FBS, 2 mM L-Glutamine, PS and 0.25 mg/mL amphotericin B (Sigma), termed SMEM.

Design of wild-type and variant viruses

To allow quantitative analysis of reassortment, we designed parental viruses that are i) highly homologous, so that reassortment does not give rise to genetic or protein incompatibilities, and ii) genetically distinct in all segments to allow tracking of genetic exchange. For both IAV and reovirus, a variant virus was generated from the wild-type strain using reverse genetics. The wild type (wt) virus strains used were influenza A/Panama/2007/99 (H3N2) (Pan/99) virus and Type 3 Dearing (T3D) reovirus. To generate variant (var) viruses, silent mutations were added to the first 1kb of each wild-type virus coding sequence, as shown in Table 1. We have made multiple versions of the Pan/99var virus^{20, 21}; the Pan/99var virus described here is Pan/99var15. The mutations made were A to G/G to A or C to T/T to C, which have the greatest impact on melting properties⁸. The melting properties of a short amplicon (65-110 base pairs) containing a mutation of this nature is typically altered sufficiently to allow robust detection by high resolution melt analysis^{8, 22}. The change in melting properties of the short amplicon is dependent upon the amplicon sequence and the nature of the nucleotide change, with multiple changes in the same direction (ex. A → G and T → C) typically resulting in a greater T_m difference. To avoid introducing attenuating mutations, sites of natural variation were targeted where sequence data was available. For IAV, isolates from the same lineage within a 10-year time frame were selected from the NCBI GenBank database, and 20-30 sequences were aligned. Sites within the first 1000 nucleotides relative to the 3’ end of the vRNA that displayed high nucleotide diversity were targeted for introduction of variant mutations. Point mutations were introduced into plasmid-encoded viral cDNAs using QuikChange mutagenesis (Agilent) according to the manufacturer’s protocol.

Table 1:

Point mutations used to generate variant viruses

T3D Mammalian Orthoreovirus		Pan/99 Influenza A Virus
Segment	Polymorphism	Segment	Polymorphism(s)*
L1	C612T	PB2	C354T C360T
L2	C853T	PB1	A540G
L3	G481A	PA	A342G G333A
M1	C919T	HA	T308C C311A C314T A646T C467G T470A
M2	A650G	NP	C537T T538A C539G
M3	T702C	NA	C418G T421A A424C
S1	G312A	M	G586A
S2	A438G	NS	C329T C335T A341G
S3	T318C
S4	C383T

^*Note, although a single nucleotide tide change is sufficient to support HRM genotyping, the Pan/99var15 virus contains multiple changes in a number of segments owing to use of alternative genotyping approaches during development of the IAV system.

To distinguish infected cells by flow cytometry, sequences encoding a 6xHIS or an HA epitope tag were added to the N-terminus of the IAV hemagglutinin protein, connected by a GGGGS linker sequence²³. The 6xHIS tag was added to the Pan/99wt virus and the HA tag was added to the Pan/99var15 virus. The linker sequence provides flexibility so that the epitope tags do not interfere with HA protein folding. To ensure the tags were retained on the mature HA proteins, they were inserted after the signal sequence. It was not possible to add an epitope tag to reovirus, as this has been demonstrated in the literature and in our hands to cause growth defects (data not shown)²⁴.

Generation of virus stocks

IAVs were generated by reverse genetics from viral cDNA^{25, 26}. Eight pPOL1 reverse genetics plasmids encoding the eight viral cDNAs were combined with four pCAGGS protein expression vectors encoding PB2, PB1, PA, and NP proteins. These plasmids were co-transfected into 293T cells using XtremeGene transfection reagent (Sigma-Aldrich) according to the manufacturer’s recommended procedure. At 16-24 h post transfection, the 293T cells were resuspended in growth medium and injected into the allantoic cavity of 9-11 day old embryonated chicken eggs (Hy-Line). Eggs were incubated 40 h in a humidified 33°C incubator. Incubation at 33°C was used because we have observed improved growth of the Pan/99 strain at this temperature, compared to 37°C²⁷. After chilling eggs overnight at 4°C, allantoic fluid was harvested, clarified of cell debris and aliquoted for storage at −80°C.

Reoviruses were generated from viral cDNA cloned into the pT7 plasmid in BHK-T7 cells²⁸. Plasmids containing each of the 10 viral cDNA’s and pCAG FAST P10 plasmid were transfected into BHK-T7 cells using TransIT-LT1 (Mirus)²⁹. Cells were fed with an additional 500 μL SMEM on day 2. BHK-T7 cells were incubated at 37°C for a total of 5 days, after which virus was harvested with 3 freeze-thaw cycles at −80°C. Plaque assays were performed using viral lysates as previously described³⁰. Reoviruses were amplified for two passages in L929 cells³⁰. Purified virus stocks were made from second passage L929 lysates. Purification was performed as described previously using Vertrel XF extraction and a CsCI density gradient³¹. The band at 1.36 g/cm³ was collected and dialyzed exhaustively against virus storage buffer (15mM NaCl, 15mM MgCl₂, 10mM Tris-HCl [pH 7.4]). The resultant purified material was stored in glass vials at 4°C for up to six months.

Analysis of single-cycle growth

To synchronize the infections, inoculation and virus attachment were performed on ice prior to warming dishes for viral entry. To limit viral growth to a single cycle, NH₄Cl was added to the medium for IAV experiments and either NH₄Cl or the protease inhibitor E64D was added for reovirus experiments. The details of these treatments are given below.

IAV

A549 cells were seeded at a density of 4 x 10⁵ cells per well in 6-well dishes 24 h prior to infection, with three wells prepared per virus. Stocks of Pan/99var15 and Pan/99wt viruses were individually diluted to achieve an MOI of 5 PFU/cell in 200 μL 1X PBS per well. Growth medium was aspirated from the cells, and cells were washed three times with 1X PBS, with the third wash performed on ice. The dishes were kept on ice for inoculation and three wells per virus were inoculated with diluted virus. Cells were incubated at 4°C for 45 minutes to permit viral attachment, and dishes were rocked every 10 minutes during incubation. Cells were then placed back on ice, and inocula were removed before washing the cells three times with cold 1X PBS. Dishes were then removed from ice, and warm virus medium (2 mL) was added to each well, followed by incubation at 33°C for 2 h. Virus medium was removed, and acid inactivation to remove residual extracellular virus particles was performed by applying 0.5 mL PBS-HCl (pH 3.0) to each well. Cells were incubated at 33°C for 5 minutes, then PBS-HCl was removed. Virus medium (2 mL per well) was again added to each well. After incubation at 33°C for 1 h, virus medium was removed and replaced with 2 mL virus medium containing NH₄Cl (20 mM) and HEPES buffer (50 mM) to ensure single cycle growth. Cells were incubated at 33°C for the duration of the growth curve, and a sample of cell supernatant (120 μL) was taken from each well at 2, 6, 12, 16, 24, 36, and 48 h post-infection. Samples were stored at −80°C and thawed once prior to plaque assay in MDCK cells for titer determination.

Reovirus

L929 cells were seeded at a density of 2.5 x 10⁵ cells per well in 24 well dishes 24 h prior to infection. One plate was prepared for each time point, with 3 wells per plate dedicated to each virus. Inocula of T3Dwt and T3Dvar were prepared separately in OPTI-MEM, for an MOI of 10 PFU/cell in each well. Cells were placed on ice, washed once with 1X PBS, and 100 μL of virus inoculum was added to each well. Virus was allowed to attach for 1 h at 4°C, and plates were rocked every 10 minutes during incubation. Cells were placed back on ice, inoculum was removed, and cells were washed 3 times with cold 1X PBS. Next, 500 μL of warm SMEM was added to each well, and cells were placed at 37°C. Starting a 0 h post infection, one plate was placed at −80°C every 3 h. At 4 h, SMEM was removed and replaced with SMEM containing 20 mM NH₄Cl. Samples were freeze-thawed 3 times at −80°C, and titers at each time point were determined using a plaque assay in L929 cells.

Primers

Reverse Transcription

For IAV, reverse transcription was performed using universal primers that anneal to the 3’ end of all eight IAV vRNA’s (GCGCGCAGC[A/G]AAAGCAGG) ³². Due to a lack of universally homologous sequences in reovirus, random hexamer primers (Thermo) were used in place of virus-specific primers.

High-resolution melt

Primers for quantitative PCR followed by high-resolution melt analysis were designed to flank the polymorphisms introduced into variant viruses. These primers were designed such that the amplicon size would be 65–110 base pairs and annealing temperatures were 58–62°C. Primer sequences for IAV and reovirus are listed in Table 2. Primer mixes were made by combining 5 μL of the 100 μM forward primer stock and 5μL of the 100 μM reverse primer stock with 240 μL molecular biology grade water for a final concentration of 4 μM.

Table 2:

Primers used to generate amplicons for high-resolution melt analysis

T3D Mammalian Orthoreovirus
Segment Forward Primer		Reverse Primer
L1	570F GCATAATTGCCCTTTATGGTG	624R AAGGTGCCCGATCTGGTAAT
L2	832F GCAACCCGTTACACGCTTAG	906R TAACACCCCCAACCGATATG
L3	451F TCAGAAGCCGATGTCTACCA	539R TGATACCCATGACCACTGCT
M1	839F TTGATGCATTTGCCTTACCA	923R CATCGGCCACATCCACTAC
M2	606F AGAGTGGCTCAAACGTTGCT	659R TCACTACCGACTGCATTGGA
M3	643F GGGATAATGAAGGCTGCTGA	720R ACCGCCCCTCGTTATAGATT
S1	278F GAGCCCTCCAAACAGTTGTC	329R AAGTTGTCCCACTCGAGCAC
S2	415F CTAGCGCGTGATCCAAGATT	488R GTAGGAAATCGGGCCAAAAC
S3	266F GGGATATCCTTCAGACTCGTG	334R CTCATGGTGGATGCTTGATG
S4	323F GGGTATGCTGTCCTTCGTTG	391R ACCTCCCTCAGTACGCACAC
Pan99 IAV
Segment Forward Primer		Reverse Primer
PB2	322F TGGAATAGAAATGGACCTGTGA	414R GGTTCCATGTTTTAACCTTTCG
PB1	508F AGGCTAATAGATTTCCTCAAGGATG	596R ACTCTCCTTTTTCTTTGAAAGTGTG
PA	307F TGCAACACTACTGGAGCTGAG	398R CTCCTTGTCACTCCAATTTCG
HA	251F CCTTGATGGAGAAAACTGCAC	313R CAACAAAAAGGTCCCATTCC
NP	482F CAACATACCAGAGGACAAGAGC	571R ACCTTCTAGGGAGGGTCGAG
NA	386F TCATGCGATCCTGACAAGTG	461R TGTCATTTGAATGCCTGTTG
M	563F GTTTTGGCCAGCACTACAGC	662R CCATTTGCCTGGCCTGACTA
NS	252F ACCTGCTTCGCGATACATAAC	342R AGGGGTCCTTCCACTTTTTG

Co-infection

IAV

Co-infections were performed with wild-type and variant (wt/var) viruses mixed at a 1:1 ratio and then diluted in PBS such that the total PFU/mL would give the desired multiplicity of infection. A549 cells seeded at a density of 4x10⁵ cells/well in 6-well dishes 24 h prior to infection were infected under single-cycle, synchronized conditions as detailed above for the growth analyses. These conditions include inoculation on ice (for synchronization), acid treatment (for inactivation of residual inoculum virus) and addition of NH₄Cl-containing medium at 3 h post infection (to prevent acidification of the endocytic compartment and thus multiple cycles of infection). At 12 h post infection, supernatants were collected and stored at −80°C. Cells were harvested and prepared for analysis by flow cytometry.

Reovirus

Co-infections followed the above procedure for synchronized, single-cycle infections. L929 cells were seeded in 12-well dishes at a density of 1.8 x 10⁵ cells/well 24 h prior to infection. The virus inoculum containing equal parts of wt/var virus as in IAV (above) was prepared in OPTI-MEM (Gibco), and cells were incubated in SMEM at 37°C. Rather than ammonium chloride, E64D protease inhibitor (Sigma) was added at a final concentration of 4 μM at 4 h post infection to block secondary infection. At 24 h post infection, three replicate wells of reovirus-infected cells at each MOI were harvested for flow cytometry, and the remaining three replicates were freeze-thawed 3 times at −80°C to release virus, and lysates stored at −80°C for future analysis.

Flow cytometry to quantify infected cells

IAV

Cells were harvested by the addition of 200 μL trypsin (Corning) and, once cells were detached, 800 μL FACS buffer (1X PBS with 2% FBS). Cells were transferred to 1.5 mL tubes on ice and pelleted by spinning at 1500 rpm for 3 minutes in a Beckman Coulter Microfuge 22R tabletop centrifuge. Supernatant was removed, and cells were washed two more times with 1 mL FACS buffer and 200 μL FACS buffer, respectively, pelleting and removing supernatant between washes. After washes, cells were resuspended in 50 μL stain buffer (FACS buffer containing Qiagen Penta-HIS Alexa Fluor 647 #35370 at a final concentration of 5 μg/mL and Sigma-Aldrich Monoclonal Anti-HA-FITC, Clone HA-7 #H7411 at a final concentration of 7 μg/mL) on ice in the dark for 35-45 minutes. Cells were washed twice with 200 μL of FACS buffer and resuspended in FACS buffer for analysis.

Reovirus

Cells were trypsinized and washed two times with FACS buffer, as above. Fixation, permeabilization, and staining were performed according to the BD Cytofix/Cytoperm protocol including a 15-minute block step with BD rat anti-mouse CD16/CD32 Fc block. To stain infected cells, a mouse monoclonal anti-σ3 antibody (clone 10C1) at a concentration of 1 μg/mL was added for 30 minutes at 4°C. After two washes, an AlexaFluor-647 conjugated donkey anti-mouse secondary antibody (Invitrogen) was added at a 1:1000 dilution.

In both virus systems, data was collected on a BD LSR II Flow cytometer running FacsDiva software. A minimum of 50,000 events was collected for each sample. Subsequent data analysis was performed using FlowJo (v10.1), gating for single cells. The threshold for positivity was determined based on a mock-infected control population stained with the relevant antibodies.

Viral genotyping

Collection of clonal isolates

For both reovirus and IAV, samples stored at −80°C were thawed and plaque assays were performed as previously described³⁰. Individual, well-isolated plaques were picked by aspirating the agar plug using a 1 mL pipette and deposited into 160 μL PBS in a 96-well assay block with 1 ml capacity wells (Costar 3958). From each sample, 21 plaques were picked for IAV, while 32 were picked for reovirus. Additionally, a single wild-type and a single variant plaque were included as controls for each series of 21 or 32 plaque picks. Assay blocks containing plaque isolates can be sealed and stored at −20°C or used directly for RNA extraction.

RNA extraction

RNA was extracted directly from agar plugs. Frozen assay blocks were thawed in a 37°C water bath and spun down at 2000 rpm for 2 min in a Heraeus Megafuge 16 tabletop centrifuge equipped with Thermo M-20 swinging bucket plate rotor. The Zymo Quick-RNA Viral Kit extraction protocol was followed using 96-well plates. Filter and collection plates were provided with the kit. No DNA/RNA Shield was used. Samples were eluted in 40 μL nuclease-free water into MicroAmp Optical 96-well reaction plates (Applied Biosystems). RNA can be covered and stored at −80°C or used directly for reverse transcription.

Reverse Transcription

Working on ice, a 12.8 μL volume of each viral RNA sample was combined with Maxima RT buffer at a final concentration of 1X, dNTP’s at a final concentration of 0.5 mM, either IAV primer at a final concentration of 0.3 μM or random hexamers (Thermo SO142) at a final concentration of 5 μM for reovirus, 100 U Maxima RT (Thermo), and 28 U RiboLock RNAse inhibitor (Thermo). Total reaction volume was 20 μl. Samples were capped, mixed by vortexing, and spun down briefly. Reactions were incubated at 55°C for 30 minutes and 85°C for 10 minutes in a BioRad T100 thermocycler. cDNA can then be stored at −20°C or used directly for qPCR and high-resolution melt analysis.

qPCR and high-resolution melt analysis

Viral cDNA was used as a template in qPCR reactions. Separate reactions were set up with primers targeting each of the viral gene segments. First, master mixes were made by combining appropriate primer mixes (see Table 2) with BioRad Precision Melt Supermix at volumes sufficient for the number of samples plus 15% extra. For each well, 0.5 μL of a 4 μM primer mixture containing both the forward and reverse primers was added to 2.5 μL Supermix. A 3 μl volume of this master mix was loaded into a 384 well plate (BioRad HSP3805) using a multichannel pipette according to the layouts in Figure 2. A 2 μl volume of cDNA diluted 1:4 (IAV) or 1:5 (reovirus) in molecular biology grade water (Invitrogen) was added to the 384 well plate. Plates were centrifuged at 2600 rpm in a Heraeus Megafuge 16 tabletop centrifuge equipped with Thermo M-20 swinging bucket plate rotor for 3 minutes to collect liquid in the bottom of wells and remove bubbles. qPCR and melt analysis were performed using a BioRad CFX384 Real-Time PCR Detection System. Amplicons were generated by initial denaturation at 95°C for 2 min, then 40 cycles of 95°C for 10 s and 60°C for 30 s. Melting properties of PCR amplicons were examined by heating from 67°C to 90°C in 0.2°C increments. Successful amplification of targets was verified in CFX Manager software (BioRad). Melt curves were analyzed using Precision Melt Analysis software (BioRad) to determine viral genotypes.

Figure 2. Example 384-well plate layouts for high-resolution melt analysis.

A) Example plate layout for the analysis of IAV reassortment. Each plate holds 21 unknown samples (indicated by numbers at the top and bottom of each schematic plate), wt and var positive controls (bottom right) and a negative control in which water was loaded in place of cDNA (bottom right). Each of the 8 segments are analyzed in duplicate for each sample, in rows as indicated at the left with segment designations. B) Example plate layout for the analysis of reovirus reassortment. Each plate holds 16 samples (indicated by numbers to the left), and wt and var positive controls (right side of the diagram). Each of the 10 segments are analyzed in duplicate, indicated by segment identification across the top and in wells for positive controls. Since 32 isolates are analyzed for reovirus reassortment, two plates must be used.

Results

Marker mutations in variant viruses do not cause fitness defects

To quantify reassortment in the absence of selection bias, it is important to ensure that the parental viruses do not differ in fitness. If one virus is fitter, that parental genotype and segments from it are likely to predominate in the progeny virus population due to uneven amplification. This outcome may be of interest in some contexts but obscures quantitative analysis of reassortment itself. For IAV, single-cycle growth of wild-type and variant viruses was assessed in A549 cells. The two viruses showed similar growth properties, and a two-way ANOVA yielded a nonsignificant P-value of 0.94 when comparing the two curves (Figure 3). For reovirus, single-cycle growth in L929 cells was evaluated to compare variant and wild type viruses. Again in this system, the variant strain was not attenuated relative to the wild-type counterpart, with a two-way ANOVA yielding a P-value of 0.95 (Figure 3). Additionally, there was no noted difference in plaque phenotype for either virus (data not shown). The highly homologous genotypes and comparable fitness of parental viruses are expected to result in similar fitness of reassortant progeny, allowing for an unbiased assessment of reassortment frequency.

Figure 3. Single nucleotide changes do not detectably alter variant virus growth.

(A) Pan/99 IAV wild-type and variant multi-cycle growth were analyzed under single-cycle conditions in A549 cells over the course of 48 h (N=3 for both viruses). (B) T3D mammalian orthoreovirus wild-type and variant virus growth were analyzed under single-cycle conditions in L929 cells over the course of 36 h (N=3 for both viruses). Means are plotted and error bars represent standard deviation.

Flow cytometry allows quantification of the proportion of cells infected

Owing to the presence of non-plaque forming particles in virus populations and routine experimental error, calculation of MOI based on plaque forming units does not always allow an accurate estimation of the proportion of a cell population that is infected. Flow cytometry analysis targeting viral antigens allows a more direct method to monitor infection levels.

In the case of IAV, the addition of different epitope tags to wt and var HA proteins allowed quantification of single- and co-infected cells within the population. Four distinct populations were observed representing uninfected, wild-type virus infected (expressing 6x HIS tag), variant virus infected (expressing HA tag), or wild type plus variant co-infected cells (Figure 4). Assessment of the proportion of cells in a population that are co-infected is useful, as co-infected cells are the only ones capable of producing reassortant progeny.

Figure 4. Flow cytometry allows for the quantification of infected and co-infected cells in a population.

(A) Uninfected cells, (B) Pan/99var15 infected cells, and (C) Pan/99wt infected cells were used to determine the appropriate gate locations for (D) IAV co-infected cells. Populations of cells infected by a single virus are indicated by the top left- and bottom rightmost gates. Co-infected cells expressing both the HA and 6xHIS epitope tags are shown in the top rightmost gate. (E) Uninfected cells were used to determine the appropriate gate location for (F) reovirus infected cells. The population shift indicates 98.6% infected cells divided between two populations, representing high and low levels of viral antigen expression.

Since insertion of epitope tags was not feasible in the reovirus system, flow cytometry of reovirus-infected cells was used to evaluate the proportion of cells that were infected, but no direct measurement of co-infected cells was made. Infected cells were detected using a primary antibody (clone 10C1) against viral structural protein σ3 and compared to an uninfected, stained control. Infected cells cluster within two groups with low and high antigen expression, respectively (Figure 4). To estimate the percentage of infected cells that are co-infected, Poisson statistics can be used, assuming an equal proportion of wild-type and variant viruses were present in the initial infection. We used the following to determine the expected fraction of infected cells that is co-infected with λ = −ln(1 – % infected cells):

$$\sum _{v=1}^{\infty }\frac{e^{-\lambda }{\lambda }^v}{v!}(1-2\ast {0.5}^v)$$

An example of expected % co-infection based on % infected cells is given in Table 3.

Table 3:

Expected % co-infection from Poisson statistics

% infection1	Expected % co-infection2
1	0.002
5	0.06
10	0.26
25	1.79
50	8.57
75	25
100	99.99

¹The percent of infected cells as determined by flow cytometry analysis

²The predicted percentage of cells infected with both wt and var viruses

High-resolution melt analysis allows determination of viral genotypes

High-resolution melt analysis allows detection of the nucleotide changes that differentiate wt and var viruses and can therefore be used for rapid assignment of wt or var genotypes to all segments present in clonal virus isolates. To this end, qPCR was performed with the cDNA of each viral isolate split into eight (IAV) orten (reovirus) separate duplicate reactions, each containing primers targeting a different segment. Following qPCR, samples with C_t values below 35 were used for melt analysis. The included wt and var controls were used as references and clusters based on similarity of T_m and melt curve shape were generated within BioRad High Precision Melt software (Figure 5). The distinct melt curves of wt and var amplicons indicate that the silent mutations introduced were sufficient for identification of the parental origins of each segment (Figure 5). In practice, we find that a minimum T_m difference of 0.15°C is needed to consistently differentiate between wt and var amplicons. Applying this approach to each segment in turn allows the full genotype of each clonal plaque pick to be determined.

Figure 5. Melt curves allow the determination of parental segment origin.

Melt curves on the left indicating relative fluorescent units (RFU) as a function of temperature were used to generate the difference curves on the right, which enabled differentiation of wt and var segments in the clusters. Wild-type (blue) and variant (red) controls were used to determine the parental origins of each segment. Melt curves of the PB2 and NP segments from IAV (A) and the L1 and S3 segments from reovirus (B) are shown.

Full genotypes are depicted in reassortment tables where each column is a separate genome segment, and rows represent clonal isolates (Figure 6). Occasionally, high resolution melt analysis yielded unclear results, with a given amplicon clustering neither with wt nor var controls. While exact causes of indeterminate results are not clear, possibilities include isolation of RNA from a mixed plaque pick (which would result in a T_m intermediate between that of the wt and var amplicons), co-packaging of a segment from both parental viruses (although we consider this to be unlikely), low cDNA quantities or loading error during plate preparation. Such results were recorded as indeterminate (white boxes in Figure 6). Viral isolates were excluded from analysis if there were more than two segments omitted due to unclear melt curves. If more than 20% of replicates were omitted from analysis, data collection was repeated starting from plaque picks.

Figure 6. Reassortment tables provide visual representation of progeny virus genotypes.

Each co-infection was performed in triplicate, and 21 (IAV) or 32 (reovirus) plaques were genotyped from each replicate. Blocks of genotypes numbered 1, 2 and 3 represent each replicate co-infection. Each row corresponds to a separate plaque isolate, and each column represents a gene segment (IAV segment order: PB2, PB1, PA, HA, NP, NA, M, NS; Reovirus segment order: L1, L2, L3, M1, M2, M3, S1, S2, S3, S4). Representative data from co-infections performed at a single MOI of both IAV (A) and reovirus (B) are shown. An MOI of 0.6 PFU/cell is shown for IAV, and an MOI of 3.16 PFU/cell is shown for reovirus. The calculated percent reassortment (%R) and the effective diversity (ED) as measured by Simpson’s index are indicated for each replicate. Blue boxes indicate wild-type parental segment origin, red boxes indicate variant parental segment origin, and white boxes indicate indeterminate results.

Following assembly of genotype tables, the percent reassortment observed in each sample is calculated as 100 times the number of reassortant clones identified divided by the total number of clones genotyped. To visualize the relationship between infection and reassortment, the calculated percent reassortment can be plotted as a function of the percent infected cells as calculated by flow cytometry (Figure 7). Additionally, due to the introduction of differing epitope tags in wt and var viruses, IAV reassortment can be plotted as a function of the percent co-infected cells. This is a useful measure, as only co-infected cells are capable of yielding reassortant progeny. For this quantitative assessment of reassortment to be meaningful, it is important that co-infections be performed under single cycle conditions. As noted in the Materials and Methods section, this was achieved for IAV using addition of ammonium chloride to cell culture medium at 3 h post-infection and for reovirus using E64D protease inhibitor added at 4 h post-infection. Blocking secondary spread of progeny virus ensures that detected frequencies of reassortant viruses reflect the efficiency of reassortment, rather than the efficiency of amplification. In contexts where infection cannot be limited to a single cycle, such as in vivo, analysis of genotypic diversity (as described below) is more appropriate than a simple readout of percent reassortment.

Figure 7. Comparison of IAV infection levels and reassortment.

Quantification of infected cells by flow cytometry can be combined with reassortment data to give insight into the dependence of reassortment on effective viral dose. (A) Percent reassortment as a function of the percentage of cells infected with Pan/99 wt and/or var viruses. (B) The effective diversity of Pan/99 isolates after co-infections in A549 cells as a function of the percentage of cells infected with Pan/99 wt and/or var viruses. (C) Percent reassortment as a function of the percentage of cells co-infected with Pan/99 wt and var viruses. (D) The effective diversity of Pan/99 isolates after co-infections in A549 cells as a function of the percentage of cells co-infected with Pan/99 wt and var viruses. Diversity increases more than five-fold from the lowest % infection to the highest in both (B) and (D).

Diversity analysis quantifies richness and evenness of reassortant population

A sample in which a single reassortant genotype is detected repeatedly would have high percent reassortment despite having low genotypic diversity. When using a wt/var co-infection system, this situation is unlikely to arise due to selection but can nevertheless occur under conditions where stochastic effects are strong (e.g. owing to within-host bottlenecks in vivo). Here, the percent reassortment readout is not highly relevant and a more sophisticated analysis of genotypic diversity is needed.

To quantify the diversity of genotypes present, Simpson’s index (given by D = sum(p_i²), where p_i is the proportional abundance of each genotype) was used. This approach accounts for both the raw number of species (richness) and variation in the abundance of each (evenness) and is sensitive to the abundance of dominant species. To determine effective diversity, the Simpson index value of each sample was converted to a corresponding Hill number, N₂ = 1/D. The Hill number N₂ is equivalent to the number of equally abundant species needed to generate the observed diversity in a sample community and is particularly useful because it scales linearly (i.e., a virus population with N₂ = 10 is twice as diverse as one with N₂ = 5). Hill’s N₂ therefore allows a more intuitive comparison between populations and is suitable for statistical analysis by basic linear regression methods³³. Diversity can be determined for each replicate and plotted as a function of percent infection (Figure 7). While the % reassortment and diversity plots shown here are similar, in cases where selection or drift have shaped the viral population, very different trends may be seen.

Discussion

Here we outline a conceptually simple approach to accurately quantify reassortment between co-infecting segmented viruses in the absence of selection bias. Our strategy utilizes reverse genetics derived parental viruses designed to allow both unbiased reassortment and streamlined genotyping. This approach overcomes limitations of previous methods in which quantitative analysis of reassortment was impeded by fitness differences among progeny viruses. The genotyping technology employed furthermore improves upon more cumbersome procedures involving gel electrophoresis or temperature-sensitive mutants. This method is useful for fundamental studies of reassortment and other interactions within virus populations.

We have used this method to evaluate reassortment of Pan/99 IAV and T3D reovirus. At high MOI, both viruses showed abundant reassortment. Further studies are needed with reovirus to assess the impact of MOI on reassortment levels and thereby gain more quantitative insight into the efficiency of segment exchange in this system. For IAV, the data included herein allow analysis of the frequency of reassortment in A549 cells as a function of both infected and co-infected cells. In line with our previous observations in other cell types³⁴, IAV reassortment was high even at co-infection levels below 25%. High reassortment at low % co-infection goes against the expectation of models that assume an equivalent burst size for all infected cells³⁴. The results suggest that co-infected cells produce more progeny than singly infected cells, as a consequence of beneficial virus-virus interactions within the cell. We previously showed that complementation of incomplete viral genomes is one such interaction^{35, 36}. Whether a similar effect occurs for reovirus has yet to be determined.

In addition to its use for differentiating wt and var viruses herein, high resolution melt analysis can also be applied to viral genotyping in systems where highly divergent parental viruses undergo reassortment. Although fitness differences among progeny viruses will obscure the quantitative assessment of reassortment in such an experiment, monitoring the combined outcomes of reassortment and selection is often highly relevant for assessing the public health risks posed by reassortment of parental strains of interest^{22, 37-39}. Highly divergent parental sequences are likely to exhibit detectable differences in melting properties, conducive to HRM analysis. It is important to note, however, that reciprocal changes within the region targeted for amplification and melt analysis will nullify differences in melting properties and therefore prohibit detection. Short regions with fewer single nucleotide polymorphisms are less likely to contain reciprocal changes and should be selected for amplification. In addition, regions of high homology must border the target region, as the same HRM primer set must be used for each parental virus. In practice for IAV, when considering reassortment between strains of differing subtypes, we found that HRM can be successfully applied to genotype the six non-HA, non-NA segments. The low sequence identity across HA and NA subtypes precluded common primer design and an alternative genotyping approach was used for these segments²².

Whole genome sequencing of clonal viral isolates is an alternative to the HRM approach that has been used recently to identify reassortant viruses⁴⁰. Next generation sequencing (NGS) allows parallel sequencing of all viral gene segments, which greatly reduces effort compared to classical Sanger sequencing. In addition, because viral genomes are typically small, the costs of NGS can be reduced by combining the barcoded cDNA derived from many isolates into a single sequencing lane. NGS can be applied to any pairing of parental viruses and may be more feasible than HRM where parental viruses are highly divergent and identical HRM primers cannot be generated for all segments. However, in experiments using matched parental viruses, such as the wt/var system, the HRM approach simplifies data collection. Whole genome sequencing requires additional preparation steps including the pre-amplification of cDNA, fragmentation and library generation. Additionally, customized bioinformatics approaches are necessary for NGS data analysis but not required in the HRM approach.

Segmented viruses utilize a variety of replication strategies which may impact their potential to undergo reassortment. For this reason, quantification of reassortment not only informs studies of viral diversification and evolution but can also offer insight into fundamental aspects of the virus life cycle. Reovirus, for example, replicates within cytoplasmic inclusions. These inclusions may prohibit exchange of gene segments and thus limit reassortment. High reassortment levels in this system would indicate an unknown role of inclusion dynamics or viral RNA trafficking in the life cycle. Other segmented viruses which are not known to form structured cytoplasmic inclusions may nevertheless possess organizational mechanisms which impact reassortment frequency. Deviation from the expectation that reassortment occurs freely at levels corresponding to the number of co-infected cells could indicate additional features of the virus’ replication cycle that are important, such as the prevalence of incomplete viral genomes in IAV^{34-36, 41, 42}. The role of compartmentalization and incomplete genomes in other virus families, such as the Reoviridae, remains incompletely understood, as does the frequency of reassortment in these systems. Utilization of the strategy outlined herein for monitoring reassortment may open up further avenues of study with respect to segmented virus replication mechanisms and population dynamics.

Figure 1. Design of variant mutations.

Genome segments are indicated by blue (wt) and red (var) bars. The single nucleotide polymorphism is indicated by a star in the variant segment. Vertical lines indicate the 1 kb region (beginning at the start codon) in which the polymorphism was introduced, and the borders of the amplicon used in qPCR and subsequent high resolution melt analysis. Arrows indicate the directionality of the qPCR primers. Criteria used in the selection of the variant mutation position are indicated in the box on the lower right.

Highlights.

• A system for quantifying reassortment of segmented RNA viruses in the absence of selection is described

• A single nucleotide difference in each gene segment is sufficient to allow differentiation of co-infecting viruses

• High-resolution melt analysis allows efficient genotyping of viral clones