A Mechanism Underlying Attenuation of Recombinant Influenza A Viruses Carrying Reporter Genes

Influenza A viruses (IAV) carrying reporter genes provide a powerful tool to study viral infection and pathogenesis in vivo, however, incorporating a non-essential gene into the IAV genome often results in virus attenuation and genetic instability. Very few studies have systematically compared different reporter IAVs, and most optimization attempts seem to lack authentic directions. In this study, we evaluated the ratio of genome copies to the number of infectious unit of two reporter IAVs, PR8-NS1-Gluc and PR8-PB2-Gluc. As a result, PR8-NS1-Gluc and PR8-PB2-Gluc produced 41.4 and 3.8 genomes containing noninfectious particles respectively for every such particle produced by parental PR8 virus. RdRp assay demonstrated that modification of segment NS by inserting reporter genes can interfere with the replication competitive property of the corresponding vRNAs, and the balance of the 8 segments of the reporter IAVs were drastically impaired in infected cells. As a consequence, large amounts of NS-null noninfectious particles were produced during the PR8-NS1-Gluc packaging. In summary, we unravel a mechanism underlying attenuation of reporter IAVs, which suggests a new approach to restore infectivity and virulence by introducing extra mutations compensating for the impaired replication property of corresponding segments.


Introduction
Influenza A virus (IAV) is a major cause of respiratory illness in humans, accounting for up to 500,000 annual deaths worldwide [1]. IAV belongs to the Orthomyxoviridae family of enveloped viruses, and its genome contain 8 single-stranded, negative-sense RNA segments (vRNAs), coding for at least 11 proteins [2]. The segments vary in length from 2341 to 891 nucleotides (nts) and are named
Reporter virus PR8-NS1-Gluc and parental virus PR8 were generated and stocked as previously described [29]. Reporter virus PR8-PB2-Gluc was rescued with pDZ-PB2-Gluc along with the other seven IAV rescue plasmids pDZ-PA, -PB1, -NP, -HA, -NA, -M, and -NS1 as previously described [29]. The plasmid pDZ-PB2-Gluc, pPolI-NS and pPolI-NS-Gluc were kindly provided by Dr. Balaji Manicassamy (the University of Iowa). The TCID50 values of viral stocks were determined by inoculation of serial 10-fold dilutions of stock virus onto MDCK cells, and the titer was calculated by the Reed-Muench method.

Single-Cycle and Multiple-Cycle Replication Assay
In order to perform a virus replication assay, MDCK cells grown in 6-well plates were infected by indicated viruses at multiplicities of infection (MOI) of 1 or 0.01 TCID 50 per cell respectively for single-cycle or multiple-cycle replication assay. After 1 h incubation at 37 • C, the virus was removed and the cells were washed, followed by adding fresh Opti-MEM containing 2 µg/mL TPCK-trypsin. Aliquots were removed at various time points and titrated by determining the TCID 50 values.

qPCR Analysis of Packaged vRNAs
MDCK cell derived stocks of indicated viruses were clarified by centrifugation at 8000× rpm for 10 min, and passed through a 0.45 µm filter. The vRNAs of virus stocks were extracted using Simply P Virus RNA Extraction Kit (Bioflux, Zhejiang, China) and reverse transcribed using a universal 3' primer 5'-GATCGCTCTTCTGGGAGCRAAAGCAGG-3' and PrimeScript RT reagent Kit (Takara, Beijing, China) according to the manufacturers' instructions. The RT product were then used as templates for quantitative PCR (qPCR) with segment specific primers (Table 1) separately. Table 1. Segment specific primers used for qPCR analysis.

Segments
Forward Primer Reverse Primer To determine the concentration of segment M, plasmid pDZ-M was serial log diluted and used to generate a standard curve range from 1 × 10 8 to 100 copies. The standard curve is the linear regression line through the data points on a plot of Ct (threshold cycle) versus logarithm of standard sample concentration. In order to examine the composition of the 8 vRNA segments of indicated viruses, the relative concentrations of vRNAs were normalized to that of parental PR8 and presented as the ratios of each vRNA segments versus M. The standard deviations were calculated based on three replicates.
To detect the replication efficacy of vRNAs of NS and M segments, reverse transcription was performed using the 3' primer 5'-GATCGCTCTTCTGGGAGCAAAAGCAGG-3' and PrimeScript RT reagent Kit (Takara, Beijing, China). RT product was then analyzed by qPCR as described above and the ratios of vRNA NS versus M were presented.
To examine the mRNA level of NS and M gene, the extracted RNAs were transcripted using oligo(dT) as the primer, and qPCR were performed using primers specific to NS, M and NP respectively.

Analysis of vRNA Replication in Infected Cells
MDCK cells were infected with indicated viruses at an MOI of 1 TCID 50 /cell. After incubation for 1 h at 37 • C, the cells were washed and fresh Opti-MEM containing 2 µg/mL TPCK-trypsin were added. Infected cells were harvested at indicated time points, and total RNAs were extracted using Simply P Total RNA Extraction Kit (Bioflux, Zhejiang, China). The vRNAs were transcripted and analyzed by qPCR as described above. The copies of M-segment vRNA derived from 50 ng total RNA were calculated. The

Comparison of the Replication Kinetics of Two Reporter IAVs PR8-NS1-Gluc and PR8-PB2-Gluc
Reporter IAVs PR8-NS1-Gluc and PR8-PB2-Gluc were generated by two different strategies. PR8-NS1-Gluc was constructed by fusion of the Gaussia luciferase gene to NS1 (Figure 1a) [29], while PR8-PB2-Gluc was modified by insertion the same reporter gene into PB2, the longest segment of IAV genome, which is believed to have higher tolerance with exogenous sequences (Figure 1b) [24]. Our previous data demonstrated that reporter IAV PR8-NS1-Gluc showed a delay in replication kinetics and >1 log reduction in virus titer when grown in MDCK cells [29], while PR8-PB2-Gluc showed about 1 log reduction in titer when grown in eggs, compared to the parental PR8 [24]. To better understand the difference in replication properties of the two reporter IAVs, both the multiple-cycle and single-cycle replication assays were performed in MDCK cells.
As shown in Figure 1c, for multiple-cycle replication assays, PR8-NS1-Gluc showed a delay in replication kinetics at as early as 12 h (h) post infection (p.i.), while PR8-PB2-Gluc did not show any delay until 24 h p.i. At 36 h p.i., titers of both PR8 and PR8-NS1-Gluc peaked, with PR8-NS1-Gluc having a 1.8 log reduction compared to PR8. Although PR8-PB2-Gluc also showed a 1 log reduction at 36 h p.i., it kept growing till 60 h p.i., and the titer almost reached the highest titer of PR8 eventually. Similar results were observed for single-cycle replication assays. PR8-NS1-Gluc showed significant reduction in titers all through the infection, compared to parental PR8, While PR8-PB2-Gluc displayed nearly identical kinetics as PR8 (Figure 1d).
These results suggest that both reporter IAVs are attenuated in vitro, however, PR8-PB2-Gluc is less attenuated.

The Genome Copies-to-TCID50 Ratio of PR8-NS1-Gluc is Greatly Increased
The IAV genome consists of 8 segments, which complicates the genome packaging of progeny virions. Since PR8-NS1-Gluc and PR8-PB2-Gluc respectively acquire a modified segment NS or PB2 of altered sizes, we speculated that the vRNA packaging might be affected.
To address this, the ratios of genome copies to the number of infectious unit (genome copies-to-TCID50 ratio) of parental PR8, PR8-NS1-Gluc and PR8-PB2-Gluc were determined. The number of copies of segment M was chosen to represent genome copies since vRNP M plays pivotal role during packaging, and it is the shortest segment besides NS [30]. We assumed that incorporation of vRNA M was less likely to be affected by modification of either PB2 or NS. As a result, the genome copiesto-TCID50 ratio of PR8-NS1-Gluc was drastically higher (41.4-fold, p < 0.05) compared to that of parental PR8, while that of PR8-PB2-Gluc was less affected, with a 3.8-fold increase (p > 0.05; Figure  2). These results suggest that PR8-NS1-Gluc produces ~41 folds more of the genome (M)-containing noninfectious particles for every such particle produced by the parental PR8 virus.

The Genome Copies-to-TCID 50 Ratio of PR8-NS1-Gluc is Greatly Increased
The IAV genome consists of 8 segments, which complicates the genome packaging of progeny virions. Since PR8-NS1-Gluc and PR8-PB2-Gluc respectively acquire a modified segment NS or PB2 of altered sizes, we speculated that the vRNA packaging might be affected.
To address this, the ratios of genome copies to the number of infectious unit (genome copies-to-TCID 50 ratio) of parental PR8, PR8-NS1-Gluc and PR8-PB2-Gluc were determined. The number of copies of segment M was chosen to represent genome copies since vRNP M plays pivotal role during packaging, and it is the shortest segment besides NS [30]. We assumed that incorporation of vRNA M was less likely to be affected by modification of either PB2 or NS. As a result, the genome copies-to-TCID 50 ratio of PR8-NS1-Gluc was drastically higher (41.4-fold, p < 0.05) compared to that of parental PR8, while that of PR8-PB2-Gluc was less affected, with a 3.8-fold increase (p > 0.05; Figure 2). These results suggest that PR8-NS1-Gluc produces~41 folds more of the genome (M)-containing noninfectious particles for every such particle produced by the parental PR8 virus. Relative genome copies-to-TCID50 ratio of reporter IAVs PR8-NS1-Gluc and PR8-PB2-Gluc. Indicated viruses derived from MDCK were titrated and extracted for vRNAs followed by quantification of segment M by qPCR. The genome copies-to-TCID50 ratio were evaluated. The standard deviations were calculated based on three replicates. * p < 0.05; ns, no significance, students' t test.

PR8-NS1-Gluc Produces Large Amounts of NS-Null Noninfectious Particles
In order to investigate whether the altered genome copies-to-TCID50 ratio of reporter IAVs was caused by impaired packaging efficacy of vRNAs NS-Gluc or PB2-Gluc, respectively, the vRNA compositions of PR8-NS1-Gluc and PR8-PB2-Gluc were examined. MDCK-derived virus stocks were extracted for vRNAs and relative amounts of eight segments were quantified by qPCR analysis. As shown in Figure 3a, the packaged vRNA NS-Gluc was drastically reduced compared to seven other segments. The NS-to-M ratio decreased for about 16-fold compared to that of PR8, suggesting large amounts of NS-null noninfectious particles were produced among PR8-NS1-Gluc virions.
In contrast, the vRNA composition in virions of PR8-PB2-Gluc was not significantly affected, and the ratio of the eight segments was almost equal to that of parental PR8 (Figure 3b), which is in accordance with our finding that genome copies-to-TCID50 ratio of PR8-PB2-Gluc is not significantly affected ( Figure 2).  Relative genome copies-to-TCID 50 ratio of reporter IAVs PR8-NS1-Gluc and PR8-PB2-Gluc. Indicated viruses derived from MDCK were titrated and extracted for vRNAs followed by quantification of segment M by qPCR. The genome copies-to-TCID50 ratio were evaluated. The standard deviations were calculated based on three replicates. * p < 0.05; ns, no significance, students' t test.

PR8-NS1-Gluc Produces Large Amounts of NS-Null Noninfectious Particles
In order to investigate whether the altered genome copies-to-TCID 50 ratio of reporter IAVs was caused by impaired packaging efficacy of vRNAs NS-Gluc or PB2-Gluc, respectively, the vRNA compositions of PR8-NS1-Gluc and PR8-PB2-Gluc were examined. MDCK-derived virus stocks were extracted for vRNAs and relative amounts of eight segments were quantified by qPCR analysis. As shown in Figure 3a, the packaged vRNA NS-Gluc was drastically reduced compared to seven other segments. The NS-to-M ratio decreased for about 16-fold compared to that of PR8, suggesting large amounts of NS-null noninfectious particles were produced among PR8-NS1-Gluc virions.

Figure 2.
Relative genome copies-to-TCID50 ratio of reporter IAVs PR8-NS1-Gluc and PR8-PB2-Gluc. Indicated viruses derived from MDCK were titrated and extracted for vRNAs followed by quantification of segment M by qPCR. The genome copies-to-TCID50 ratio were evaluated. The standard deviations were calculated based on three replicates. * p < 0.05; ns, no significance, students' t test.

PR8-NS1-Gluc Produces Large Amounts of NS-Null Noninfectious Particles
In order to investigate whether the altered genome copies-to-TCID50 ratio of reporter IAVs was caused by impaired packaging efficacy of vRNAs NS-Gluc or PB2-Gluc, respectively, the vRNA compositions of PR8-NS1-Gluc and PR8-PB2-Gluc were examined. MDCK-derived virus stocks were extracted for vRNAs and relative amounts of eight segments were quantified by qPCR analysis. As shown in Figure 3a, the packaged vRNA NS-Gluc was drastically reduced compared to seven other segments. The NS-to-M ratio decreased for about 16-fold compared to that of PR8, suggesting large amounts of NS-null noninfectious particles were produced among PR8-NS1-Gluc virions.
In contrast, the vRNA composition in virions of PR8-PB2-Gluc was not significantly affected, and the ratio of the eight segments was almost equal to that of parental PR8 (Figure 3b), which is in accordance with our finding that genome copies-to-TCID50 ratio of PR8-PB2-Gluc is not significantly affected ( Figure 2).  In contrast, the vRNA composition in virions of PR8-PB2-Gluc was not significantly affected, and the ratio of the eight segments was almost equal to that of parental PR8 (Figure 3b), which is in accordance with our finding that genome copies-to-TCID50 ratio of PR8-PB2-Gluc is not significantly affected ( Figure 2).

The vRNA NS-Gluc is Less Competitive in Replication
It is well known that the replication and transcription of the 8 segments are balanced in IAV by competing with each other for available polymerases. The competition can be affected by segment length, coding region, and UTRs [31]. Thus we asked whether the insertion of reporter Gluc negatively affected the competitive capacity of NS segment, which might cause the deficient incorporation. To address this, the replication efficacy of vRNA NS and NS-Gluc were respectively examined in the presence of natural vRNA M. Cell-based IAV RdRp assay was performed, and pPolI-NS and pPolI-NS-Gluc respectively were co-transfected with pDZ-M as well as plasmids expressing NP, PB1, PB2 and PA. As shown in Figure 4a, the relative replication of vRNA NS-Gluc was reduced by about 10-fold, suggesting that vRNA NS-Gluc is less competitive in replication than the parental NS.

The vRNA NS-Gluc is Less Competitive in Replication
It is well known that the replication and transcription of the 8 segments are balanced in IAV by competing with each other for available polymerases. The competition can be affected by segment length, coding region, and UTRs [31]. Thus we asked whether the insertion of reporter Gluc negatively affected the competitive capacity of NS segment, which might cause the deficient incorporation. To address this, the replication efficacy of vRNA NS and NS-Gluc were respectively examined in the presence of natural vRNA M. Cell-based IAV RdRp assay was performed, and pPolI-NS and pPolI-NS-Gluc respectively were co-transfected with pDZ-M as well as plasmids expressing NP, PB1, PB2 and PA. As shown in Figure 4a, the relative replication of vRNA NS-Gluc was reduced by about 10-fold, suggesting that vRNA NS-Gluc is less competitive in replication than the parental NS.
The mRNA levels of NS and M were also examined. The mRNA NP, which is thought to be transcribed constantly under the control of promoter CMV, was set as an internal reference. The mRNA level of NS-Gluc was significantly decreased when compared to the parental NS (Figure 4b). Interestingly, we observed obvious increase of mRNA M in the presence of vRNA NS-Gluc compared to that in the presence of the parental vRNA NS (Figure 4b), which suggests that NS-Gluc is less competitive.

The Balance of the 8 Segments of Reporter IAVs Is Impaired
To further determine whether the less competitive replication of vRNA NS-Gluc is the cause for the impaired packaging of vRNA, the replication kinetics of vRNA segments in infected cells were examined. One cycle infection was performed by infecting MDCK cells with an MOI of 1 of parental PR8, PR8-NS1-Gluc or PR8-PB2-Gluc, respectively, and the relative amounts of vRNAs in infected cells were monitored. As shown in Figure 5, the replication of vRNA NS-Gluc was impaired by about 1 log all through the infection (Figure 5a), while none of the other seven segments were obviously affected except for PA, which was slightly reduced by 2-3 fold at different phases (Figure 5b-h). Interestingly, we also observed a slight reduction of vRNA PB2-Gluc replication in PR8-PB2-Gluc infected cells (Figure 5h). The impaired replication greatly reduced the amounts of vRNAs NS-Gluc or PB2-Gluc available for packaging (Figure 5i,j).
These results clearly demonstrate that modification of vRNA segments alters their replication property in infected cells. The reduced amount of vRNA NS-Gluc may predominately contribute to the NS-null noninfectious particles. The mRNA levels of NS and M were also examined. The mRNA NP, which is thought to be transcribed constantly under the control of promoter CMV, was set as an internal reference. The mRNA level of NS-Gluc was significantly decreased when compared to the parental NS ( Figure 4b). Interestingly, we observed obvious increase of mRNA M in the presence of vRNA NS-Gluc compared to that in the presence of the parental vRNA NS (Figure 4b), which suggests that NS-Gluc is less competitive.

The Balance of the 8 Segments of Reporter IAVs Is Impaired
To further determine whether the less competitive replication of vRNA NS-Gluc is the cause for the impaired packaging of vRNA, the replication kinetics of vRNA segments in infected cells were examined. One cycle infection was performed by infecting MDCK cells with an MOI of 1 of parental PR8, PR8-NS1-Gluc or PR8-PB2-Gluc, respectively, and the relative amounts of vRNAs in infected cells were monitored. As shown in Figure 5, the replication of vRNA NS-Gluc was impaired by about 1 log all through the infection (Figure 5a), while none of the other seven segments were obviously affected except for PA, which was slightly reduced by 2-3 fold at different phases (Figure 5b-h). Interestingly, we also observed a slight reduction of vRNA PB2-Gluc replication in PR8-PB2-Gluc infected cells (Figure 5h). The impaired replication greatly reduced the amounts of vRNAs NS-Gluc or PB2-Gluc available for packaging (Figure 5i,j).
These results clearly demonstrate that modification of vRNA segments alters their replication property in infected cells. The reduced amount of vRNA NS-Gluc may predominately contribute to the NS-null noninfectious particles.

Discussion
Reporter viruses provide a powerful tool to study infections; however, their development and application usually encounter huge challenges. First, introduction of foreign genes or modification of existing viral genes might alter the virological properties, such as delay of replication kinetics or reduced virulence. Second, such alterations can create an evolutionary pressure, leading to the loss of reporter genes [32]. For example, a recombinant bunyavirus expressing GFP fused to the Nterminus of Gc is highly attenuated, which might be due to the fact that GFP directly inhibited the function of Gc, while replacing GFP with mCherry partially solved the problem [33]. As another example, expression of Renilla luciferase attenuated respiratory syncytial virus (RSV) in vivo, but expression of Katushka2 did not, suggesting that the observed attenuation was purely caused by the reporter protein itself rather than the changes to the virus genome [34]. However, for most reporter viruses, how reporter expression can cause attenuation remains unclear. It's more likely to be through multiple mechanisms rather than by a single one. This study investigated the mechanisms underlying attenuation of two reporter IAVs, PR8-NS1-Gluc and PR8-PB2-Gluc [24,29]. Our data clearly demonstrated that elongation of either vRNA segment NS or PB2 reduced their replication efficacy respectively (Figures 4 and 5). Subsequently, the incorporation of NS-Gluc into progeny virions was impaired, producing a large amount of NS-null noninfectious particles. Although we didn't observe

Discussion
Reporter viruses provide a powerful tool to study infections; however, their development and application usually encounter huge challenges. First, introduction of foreign genes or modification of existing viral genes might alter the virological properties, such as delay of replication kinetics or reduced virulence. Second, such alterations can create an evolutionary pressure, leading to the loss of reporter genes [32]. For example, a recombinant bunyavirus expressing GFP fused to the N-terminus of Gc is highly attenuated, which might be due to the fact that GFP directly inhibited the function of Gc, while replacing GFP with mCherry partially solved the problem [33]. As another example, expression of Renilla luciferase attenuated respiratory syncytial virus (RSV) in vivo, but expression of Katushka2 did not, suggesting that the observed attenuation was purely caused by the reporter protein itself rather than the changes to the virus genome [34]. However, for most reporter viruses, how reporter expression can cause attenuation remains unclear. It's more likely to be through multiple mechanisms rather than by a single one. This study investigated the mechanisms underlying attenuation of two reporter IAVs, PR8-NS1-Gluc and PR8-PB2-Gluc [24,29]. Our data clearly demonstrated that elongation of either vRNA segment NS or PB2 reduced their replication efficacy respectively (Figures 4 and 5). Subsequently, the incorporation of NS-Gluc into progeny virions was impaired, producing a large amount of NS-null noninfectious particles. Although we didn't observe equivalent reduction of Viruses 2018, 10, 679 9 of 11 packaged vRNA PB2-Gluc when compared to other segments, this might due to the critical role of vRNP PB2 during the genome-packaging process [30].
Due to the segmented nature, replication of IAV genome is much more complicated. It has been reported that IAV balances the replication and transcription of its multiple genome segments by competition between them. The competition could be affected by segment length, coding region, and UTRs, of which any change might result in balance disruption [31]. It could be hypothesized that the impaired competition property of one segment can cause reduced replication and transcription, subsequently interfering with the balance of the genome segments. In the case of PR8-NS1-Gluc, the replication of vRNA NS-Gluc showed a drastic reduction (>1 log compared to M). Although in the case of PR8-PB2-Gluc, the reduction in replication of vRNA PB2-Gluc was less drastic (about 5 fold), this is reasonable, since the sequence changes relative to the length of the segment are much more important in the NS-segment as compared to those in the PB2 segment. However, since the ratio of the eight segments vRNAs is quite dynamic in the infection (Supplementary Figure S1), caution should be taken in analyzing and interpreting the kinetics of vRNA replication of recombinant reporter IAVs using the parental PR8 as a control.
The packaging of IAV genome segments into progeny virions is also a very complicated process. Previous studies address a selective model in which a complete set of eight vRNA segments is selectively packaged into each progeny virion [35][36][37]. It was proposed that the IAV segments were usually packaged into virions as an interlinked complex of 8 vRNPs, however, the contribution of each vRNA segment to virion production is different. The vRNPs PB2, PA, NP and M play a more pivotal role during the packaging process than NS and others [30]. This may explain why we observed large amounts of NS-null noninfectious particles, but no PB2-null ones.
As we proposed that the mechanism underlying attenuation of reporter viruses is comprehensive, we could not exclude the possibility that elongation of IAV segments directly interfere with the packaging efficacy due to altered sizes. Also, the reduced replication of NS or PB2 may decrease the expression of NS1 or PB2 proteins. Therefore, the NS1-mediated evasion of cellular innate immune response and PB2 related RdRp activity might also be affected [38,39], resulting in impairment of virus growth.
The loss of competition of elongated segment in replication and transcription could be restored. Providing more RdRp's or introducing mutations that enhance the activity of RdRp's could alleviate the competition between segments [28,31], while stabilizing the pan-handle structure of a vRNA segment can increase its binding to RdRp's, promoting the competitiveness [31]. Based on our finding that impaired replication of corresponding vRNAs of modified segments fundamentally contribute to the attenuation, we propose that introduction of extra mutations that could compensate for the impairment might restore the viral infectivity and virulence.
In summary, this study unravels a mechanism underlying attenuation of two reporter IAVs, and provide feasible directions for their further optimization.