Evaluation of the potential defensive strategy against Influenza A in cell line models

Background: Influenza virus can cause both seasonal infections and unpredictable pandemics. Rapidly evolving avian H5N1 and H7N9 viruses have a potential pandemic threat for humans. Since avian Influenza can be transmitted by domestic birds, serving as a key link between wild birds and humans, an effective measure to control the influenza transmission would be eradication of the infection in poultry. It is known that the virus penetrates into the cell through binding with the terminal oligosaccharides - sialic acids (SA) - on the cell surfaces. Removal of SA might be a potential antiviral strategy. An approach to developing chicken lines that are resistant to influenza viruses could be the creation of genetically modified birds. Thus it is necessary to select a gene that provides defense to influenza. Here we have expressed in cells a range of exogenous sialidases and estimated their activity and specificity towards SA residues. Methods: Several bacterial, viral and human sialidases were tested. We adopted bacterial sialidases from Salmonella and Actinomyces for expression on the cell surface by fusing catalytic domains with transmembrane domains. We also selected Influenza A/PuertoRico/8/34/H1N1 neuraminidase and human membrane sialidase ( hNeu3) genes. Lectin binding assay was used for estimation of a α (2,3)-sialylation level by fluorescent microscopy and FACS. Results: We compared sialidases from bacteria, Influenza virus and human. Sialidases from Salmonella and Influenza A neuraminidase effectively cleaved α (2-3)-SA receptors. Viral neuraminidase demonstrated a higher activity. Sialidases from Actinomyces and hNeu3 did not show any activity against α (2-3) SA under physiological conditions. Conclusion: Our results demonstrated that sialidases with different specificity and activity can be selected as genes providing antiviral defence. Combining chosen sialidases with different activity together with tissue-specific promoters would provide an optimal level of desialylation. Tissue specific expression of the sialidases could protect domestic birds from infection.


Introduction
The global influenza pandemic remains a real threat. Avian influenza viruses H5N1 and H7N9 have killed millions of birds, including domestic poultry, causing enormous financial losses. 860 cases of disease in human population has been detected since 2003 and 454 of them with fatal outcome (see http://www.who. int/influenza/human_animal_interface/2018_03_02_tableH5N1. pdf?ua=1). The risk of a new pandemic of viral infection remains high. Seasonal flu vaccination is used as a traditional way to protect individuals in an outbreak nearby location, however it has a lot of limitations. Also it is difficult to reduce the infection in bird population that are natural transmitters of the Influenza. Therefore, it is critical to develop alternative approaches to prevent influenza infection. An alternative way to reduce the infection could be the creation of genetically modified domestic birds. It will decrease the risk of the infection spread, because domestic birds are known to transmit the infection acting as an intermediary between wild ducks to humans (Kim et al., 2009). It is necessary to select gene-candidates for developing the approach.
The best way to prevent the infection is to impede the virus from entering the cell. Influenza virus hemagglutinin (HA) provides attachment to the host cell that leads to fusion between the virion envelope and the host cell membrane (Skehel & Wiley, 2000). Surface sialic acid (SA) residues are host cell epitopes that are recognized by influenza virus A and B (Ito, 2000). SA are a family of nine-carbon acidic monosaccharides that naturally terminate sugar chains attached to the proteins on the cell surface (Varki et al., 2009). Neuraminidase (NA) -the second major surface antigen -is an exoglycosidase, or saildase, that cleaves SA from cell membrane glycolipids and glycoproteins, thus destroying the recognition epitopes on the surface of the host cell for the viral receptor HA. NA activity helps viral particles to penetrate through mucous secretions that are rich in sialic acids, to reach the target cells in the airway epithelium (Palese et al., 1974). The role of the enzyme in facilitating release of newly formed viral particles from the infected cell surface and preventing aggregation of the viral particles was experimentally confirmed (Matrosovich et al., 2004;Palese et al., 1974). There are known antiviral agents -oseltamivir and zanamivir -that inhibit NA, blocking the release of virus particles from infected cells (Moscona, 2005;Varghese, 1999). Sialidases have also demonstrated to be effective inhibitors of Influenza virus infection in vitro. It has been shown that cells treated with Vibrio cholerae or Micromonospora viridifaciens bacterial sialidase are resistant to influenza virus infection (Air & Laver, 1995;Bergelson et al., 1982;Griffin et al., 1983;Stray et al., 2000).
We focused on exogenous expression of sialidases as a defense strategy against influenza infection ( Figure 1). Here we show the protective effect of exogenous expression of different sialidases. The range of exogenous sialidases has diverse activity and specificity towards SA residues. Tissue-specific expression of sialidases in transgenic poultry might protect domestic birds against Influenza virus.

Methods
Cloning of sialidases Viral neuraminidase. Viral neuraminidase from human Influenza A virus was amplified from the plasmid pNAHA (Plasmid # 44169; Addgene, Cambridge, MA, USA) that encodes the neuraminidase gene of Influenza A / Puerto Rico / 8/34, subtype N1. The amplified product was inserted by restriction cloning in the lentiviral transfer plasmid based on a pHAGE backbone (#24526, Addgene) after the CMV promoter. The RFP reporter was cloned in the same reading frame and the resulted construction was pHAGE-CMV-infNA-P2A-RFP (Supplementary Figure 1A). Additionally the construction RFP-P2A-infNA was cloned in the same vector having RFP and infNA in reverse order (Supplementary Figure 1A).

Bacterial sialidases Sialidases from Salmonella typhimurium and Actinomyces viscosus.
Synthesis of the codon-optimized bacterial sialidase from Salmonella typhimurium (Uniprot / P29768, 2-382aa) and the codon-optimized bacterial sialidase from Actinomyces viscosus (uniprot / Q59164, 347-631aa) fused with the transmembrane domain of viral neuraminidase (1 -75 aa) were done in BioCat. The transmembrane domain was added to provide membrane localisation of the enzymes, since sialidase S.typhimurium and A. viscosus are cytoplasmic. The codon-optimization was done with the IDT tool with the choice of the Gallus gallus genome. The genes were ordered in the standard vector pUC57 and were flanked by the restriction sites NheI and SphI for S. typhimurium

Amendments from Version 1
Changes with the text in general: the misprints pointed by referees were rectified; figures were supplemented with additional information; and several new references were added.
Title: This has been changed to "Evaluation of the potential defensive strategy against Influenza A in cell line models" Introduction: The information regarding cases of diseases among human population was specified and the previous reference was replaced.
Methods: Some information about RFP-P2A-infNA and d_infNA (pHAGE-CMV-d_infNA-P2A-RFP) genetic constructs was added. The description of lentivirus titre calculation using FACS was added in the part 'Lentiviral production, determination of viral titre and viral transduction of MDCK cell line'.
Results: Additional information was added in the part 'Transient expression of genetic constructs in HEK293 cell lines'.
Discussion: The related approach such as usage of a recombinant sialidase DAS181 as an antiviral defence was mentioned. The point that de-sialylation could not completely abrogate influenza infectivity was considered in more details. The probable physiological consequences of sialidases expression for respiratory and digestive tract were discussed. Speculations regarding how to reduce a probable negative impact of constitutive sialidase expression for animal physiology were supplemented with additional information. The practicalities of using the Dox-inducible system for our approach was added. and XhoI and SphI sites for A. viscosus. The genes were excised from the vector pUC57 using the sites and inserted into the pHAGE backbone after the the CMV promoter. As a result, the plasmids pHAGE-CMV-Sal.t.Sia-P2A-RFP and pHAGE-CMV-Act.v.Sia-P2A-RFP have been cloned (Supplementary Figure 1A).

Human neuraminidase
The genes coding sialidase 3 (membrane sialidase) in G.gallus and H.sapiens are homologous based on the analysis in Homolo-Gene (NCBI). We selected human membrane sialidase (Gene ID: 10825) for cloning because there is more information about it compared with other orthologs. The coding exons were amplified from human genomic DNA and overlapped. The resulted product was inserted by restriction cloning in the lentiviral transfer plasmid based on the pHAGE backbone (#24526, Addgene) after CMV promoter. The RFP reporter was cloned in the same reading frame and the resulted construction was pHAGE-CMV-hNeu3-P2A-RFP (Supplementary Figure 1A).

Catalytically inactive sialidase
Enzymatically inactive neuraminidase (d_infNA) was taken as a negative control. Two amino acids at positions 262-263 (EE) of the Influenza A virus strain A/Puerto Rico/8/1934 H1N1 are responsible for the substrate binding. The inactive variant has a mutation -E262D (Huang et al., 2008). The mutation was made by two overlapping primers: 5'-GCACCTAATTCTCACTATGAt-GAATGTTCCTGTTAC-3', 5'-CATTCaTCATAGTGAGAATT-AGGTGC-3' that change the codon from GAG to GAT (The changed nucleotides for the overlapping primers are marked as lowercase letters). Primers were made by Evrogen (Moscow, Russia). The changed PCR product was then cloned into the same backbone. The resulting construction was pHAGE-CMV-d_ infNA-P2A-RFP Tet-on system, design and cloning Tet-on system was set up to make transcription of the gene of interest to be inducible (Gossen & Bujard, 1992;Gossen et al., 1995). rtTA (tetracycline transactivator), CMV promoter, TRE (Tet Response Element) promoter P2A and RFP were amplified from different plasmids. Influenza A neuraminidase and sialidase from S. typhimurium were inserted under control of TRE-promoter. Neuraminidase was amplified from the plasmid pNAHA (Plasmid # 44169, Addgene) and sialidase S. typhimurium was amplified from pHAGE-CMV-Sal.t.Sia-RFP. All amplified products were cloned in the pHAGE backbone in the following order: TRE-infNA/Sal.t.Sia-P2A-RFP-CMV-rtta by restriction cloning (Supplementary Figure 1B). The plasmid pTagBFP encoding BFP (blue fluorescent protein) was used for cotransfection to approximately estimate the effectiveness of transfection before addition of the doxycycline. In order to determine a viral titre, different aliquots of supernatant were added to HEK293 cells in the presence of 4 µg/ml polybrene (Hexadimethrine bromide, H9268, Sigma). Infection of six 10-fold serial dilutions was performed in 96well plates. The medium was replaced 12 hours after infection. Infection with the same aliquot was used in three repeats. At 48 hours after infection, transduction effectiveness was calculated by FACS analysis. FACS analysis was performed using noninfected cells as a negative control. The cells with 10%-40% RFP-positive were selected for titer calculation. The titer was calculated using the equation: Titer (Transduction Unit/ml) = (cell number in each well on the day of infection × percentage of GFP/RFP positive cells (should be 10%~40%))/(the virus volume used for infection in each well × dilution fold).

Doxycycline induction
The supernatant of the known titre was used for viral transduction of the MDCK cell line. On the first day about 1.0 × 10 4 MDCK cells were seeded into 6-well plates. The cells were incubated 18-20 hours at 37°C in a humidified incubator in an atmosphere of 5-7% CO2. Upon transduction the confluency around 30-50% was estimated. Next, the appropriate volumes of unconcentrated virus and polybrene with concentration 5ug/ml were added to each well. After 48 hours, cells were analyzed using Flow cytometry to define the percentage of infected/fluorescent cells. Lentiviral transduction efficiency was calculated as (the number of RFP-positive cells/the total number of cells counted × 100) and the number of transduced cells (transduction efficiency (%) × the total number of RFP recovered/100).

Genomic DNA extraction and PCR analysis
Genomic DNA was isolated using Wizard® Genomic DNA Purification Kit (A1120, Promega, Madison, WI, USA) according the manufacturer protocol.
Lectin binding assay Lectin binding assay was used for the cell surface SA detection. For lectin binding assay we used Maackia amurensis (MA) lectins that are specific for α (2-3)-bound SA. The lectin were conjugated with the FITC fluorophore (21761036-1 (510183), bio-WORLD; Dublin, OH, USA). The assay was made for HEK293 and MDCK cells. The experiment was setup in 48-well plates. The cells expressing a desired construct were washed three times with PBS. Then cells were incubated with MA lectins (1:100) about 1 hour and three times washed with PBS. Hoechst (RRID:AB_2651133, Thermo Fisher) was added for visualization of the nuclei for 5 min at recommended concentration with the subsequent washing with PBS.

Transient expression of genetic constructs in HEK293 cell lines
Sialidases vary in enzyme kinetic and substrate specificity to the type of the SA linkage. Several sialidases from bacteria, Influenza virus and vertebrate were selected for our study. We were mostly interested in α(2-3) specialized sialidases, because avian Influenza prefers this type of the linkage (Ito, 2000). However the broad substrate specificity was also an area of interest. Regarding enzyme kinetics, sialidases with various levels of activity were useful for us in order to combine them with tissue-specific promoters and find an optimal level of desialylation of the cell surface. This optimal level prevents penetration of the virus into the cell and does not affect cell function.
Substrate specificity of bacterial sialidases is very diverse, in terms of type of SA linkages and enzyme kinetics. We selected two different bacterial sialidases: Salmonella typhimurium sialidase, which specializes in cleavage of α(2-3) SA residues (Hoyer et al., 1991;Rogerieux et al., 1993) and Actinomyces viscosus sialidase, which has a substrate specificity to both α (2-3) and α (2-6), but preferentially cleaves the α (2-6) linkage (Teufel et al., 1989). We used sequence of the catalytic domains of the sialidases and added to their N-terminal sequences the transmembrane domain with the stem loop of Influenza NA transmembrane protein in order to target the proteins to the membrane and demonstrate its catalytic activity outside of the cell.
Exogenous overexpression of a vertebrate membrane sialidase also would be interesting to analyze. The sialidases are poorly investigated, but some experimental data exists for human membrane neuraminidase hNeu3 (Monti et al., 2000;Zhang et al., 2010) and the enzyme was selected for cloning.
All cloned genetic constructs were firstly tested by transient expression in the HEK293 cell line that has glycosylation patterns including α (2-3) and α (2-6) SA linkages (Picanco-Castro et al., 2013). All plasmids had the RFP reporter to mark the cells that express a sialidase. The construction coding catalytically inactive neuraminidase d_infNA-RFP was taken as the negative control. Using lectin binding assays it has been shown that the cells expressing the sialidase from S. typhimurium have decreased level of SA. Expression of the sialidase from A. viscosus did not affect the α (2-3) sialylation level, probably because it is more specific to the α (2,6) SA linkages. However, it has previously been shown that this sialidase effectively removes both α (2,3)-and α (2,6)-linked SA from the cell surface (Malakhov et al., 2006). Human neuraminidase hNEU3 also has not shown activity towards α (2-3)-linked SA. Probably, it might be related to unsuitable pH conditions. Cultivation in DMEM in 5% CO 2 provides pH in range of 7,6-7,7. Earlier, hNeu3 activity was demonstrated in the pH range of 2.8-6.6 (Monti et al., 2000). However we do not demonstrate here the membrane localisation of the protein, thus we can not be sure in its proper localisation on the cell surface. In the case of influenza A neuraminidase, sialylation level decreased not only in RFP-positive cells, but also in the RFP-negative cells, which are not supposed to express the enzyme ( Figure 2A). According to the flow cytometry analysis, the most significant α (2-3)-linked SA cleavage was in the case of Influenza A neuraminidase expression ( Figure 2B). On the contrary, it is known that the sialidase from S.typhimurium is more active than viral neuraminidase N1. It could be explained by slightly different protein folding of sialidase from Salmonella when transmembrane domain and stem loop are added. In the case of viral neuraminidase we used unchanged sequence of natural protein.
Representative histograms from three independent repeats are shown. Lectin binding assay was quantified as fluorescence and calculated as a mean fluorescence intensity -MFI (

Lentiviral transduction of MDCK cell lines
The MDCK cell line is used for propagation of influenza viruses. We transduced MDCK cells with lentiviruses. The expressing constructs were integrated into the genome by lentiviral transduction to obtain cell lines that stably express exogenous genes coding for different sialidases. were absent not only in the RFP-positive cells but in the RFPnegative cells as well in the case of infNA-RFP expression. A possible explanation of the evidence could be a 'slipstream' translocation in a polycistronic vector of the P2A downstream protein without a signal of localisation that is RFP in our case (de Felipe et al., 2010). An alternatively explanation being when a ribosome encounters 2A within an open reading frame the synthesis of a specific peptide bond could be "skipped". This results in termination of translation at the end of 2A peptide (Donnelly et al., 2001). It means that RFP is not expressed and can explain the absence of SA in some RFP-negative cells.
In the case of improper translation of RFP the lentivirus titre determined by FACS could be underestimated. In result, the cells were infected by larger quantity of viral particles and desialylation was more prominent. The effects were not observed in the case of other sialidases. Usage of a longer sequence for P2A (with a favourable upstream sequence composition) or modifying the order of proteins could solve the problem. When we changed the order of the RFP and the infNA sequences (RFP-P2A-infNA), the result of lectin binding assay was as expected: RFPpositive cells showed decreased level of α (2-3) SA ( Figure 4A). Nevertheless, in the case of inverted position of the neuraminidase the enzyme activity was less pronounced. It can be explained    by the fact that a protein located at the second position can be slightly less expressed in a polycistronic construct with a 2A peptide (Liu et al., 2017). In the case of the S. typhimurium sialidase expression (Sal.t.Sia-RFP) the 'slipstream' effect was not observed.

Discussion
Sialic acid is the receptor of Influenza A virus (Matrosovich et al., 2013). Therefore the removal of SA from the cell surface might be a powerful defence strategy against Influenza virus. Previously, a recombinant sialidase (DAS181) has already shown its antiviral effectivity in the cell culture (Malakhov et al., 2006) and in vivo, using a mouse model (intranasal injection of the recombinant fusion protein -DAS181) (Belser et al., 2007). Thus, the principal possibility of the usage of membrane-anchored sialidase as antiviral defence has been demonstrated. In the current study we reviewed a variety of sialidases from different sources and compared their activity. We showed that expression of Salmonella typhimurium sialidase and human Influenza A neuraminidase on the cellular membrane effectively removed α (2-3)-sialic residues on the cell surface under physiological conditions. The viral neuraminidase had higher cleavage activity against α (2-3)-linked SA than sialidase from S. typhimurium.
We are planning to use a sialidase gene as a defensive strategy to create genetically modified domestic birds resistant to Influenza infection. However there are several concerns to use the strategy. The first is that influenza virus receptors may be different from SA, thus, sialidase expression may be ineffective. It has been reported that the virus binding could still occur in the MDCK cell line desialylated with addition in the medium Micromonospora viridifaciens sialidase (Stray et al., 2000). The authors suggested that influenza virus infection can result from SAindependent receptors, either directly or in a multistage process. The presence of SA may enhance virus binding to the cell surface to increase interaction with secondary receptors to mediate entry. High multiplicity of infection has an increased requirement for SA thus desialylation will inhibit the virus amplification. Although the HA of a human H1N1 strain was shown to bind other glycoconjugates (Rapoport et al., 2006), the exact receptors to enter into the cell have not been found. Stray et al. used NA-deficient virus in the research which was obtained in a laboratory. It is unlikely that such virus would exist in nature. Probably, the barrier to efficient infection and transmission between organisms is due to inefficient use or expression of a yet unidentified entry mediator.
Another concern of the stable expression of sialidases is disruption of physiological functions required for proper glycosylation (Gutierrez et al., 1987;Michalek et al., 1988;Pangburn et al., 2000;Varki, 1992). The probable consequences of constitutive expression of membrane sialidase in vivo are not known. It has been shown that overexpression of the human ortholog NEU3 membrane sialidase under the β-actin promoter in transgenic mice resulted insulin-resistant diabetes mellitus (Sasaki et al., 2003). The authors focused on insulin signaling only.
However at least the data shows that the constitutive membrane sialidase expression was not lethal for mice.
We are going to reduce a probable negative impact on organism using tissue-specific expression of the selected gene. It is necessary to predict a potential impact on the epithelium of intestine and trachea in vivo. There are no similar genetic experiments. However the consequences of tissue-specific expression could be considered from the researches where local desialylation was done. It has been shown that intranasal sialidase treatment does not affect the properties of respiratory mucus, nor did it affect the normal mucus transport activity on ciliated epithelium (King et al., 1974;Meyer et al., 1975). Some authors showed that Influenza infection is associated with an increased risk of secondary infection by Streptococcus pneumoniae (van der Sluijs et al., 2006). There are concerns that, probably, tissuespecific expression of sialidases would result in a greater chance of bacterial infection in the respiratory tract due to exposure of its binding sites that were masked by terminal SA. At the same time it has been demonstrated that the binding sites for the bacteria already exist on the normal respiratory surface and sialidase treatment does not introduce de novo bacterial adhesion sites. Also the risk of infection arises due to damage of mucus and epithelial cells. But it is not clear whether the secondary S. pneumoniae infection happens due to viral NA activity or from a secondary effects of influenza virus intracellular amplification. Influenza infection in the respiratory tract causes much deeper changes to the epithelium than treatment of the surface of the epithelial cells with a sialidase, because Influenza infection in humans and in mice results in vast exposure of the ciliated epithelial cells down to the basal cells and the basement membrane (Plotkowski et al.,1986;Walsh et al., 1961). Thus, the most likely the increased S. pneumoniae infection is the result of airway epithelial damage caused by the virus infection but not just desialylation. In the gastrointestinal tract SA is an abundant sugar residue in mucin that is a key target of intestinal bacteria. Expression of sailidases will release free SA from mucins that could drives intestinal inflammation and infection. For example, elevated levels of free SA in the gut, during and post antibiotic treatments, promoted the expansion of Clostridium difficile, that utilises the free monosaccharide. Also sialidase activity can promote the outgrowth of Escherichia coli and result in imbalanced microbiota, inflammation, colitis development due to α (2-3)-sialic residues promote expansion of E. coli (Huang et al., 2015). Nevertheless, we do not know the probable negative impact on the physiology of avian respiratory and digestive tract. Thus, an optimal level of desialylation can be established in the tissues when combining genetic constructs with varying sialidase activity and tissuespecific promoters in order to avoid disruption of physiological functions in our approach.
Additionally, a gene expression with time limit could prevent a probable negative impact on a living organism. Michael P. Malakhov et al. showed that temporary desialylation of respiratory tract during 7 days using recombinant sialidase protein did not cause any symptoms of toxicity or inflammation in ferrets and mice that were not infected by the virus (Malakhov et al., 2006). This in vivo data support our idea that temporary inducible expression of a sialidase will not have serious negative physiological effects for transgenic animals. We are planning to use the Tet-on system as a potential model for creation of genetically modified organism that allows to induce a sialidase expression temporary. Transgenic transcription factor rtTA will be expressed under control of a tissue-specific promoter and a sialidase will be expressed under control of TRE promoter. This enables tissue-specific inducible control of the transgene expression and will provide a powerful and flexible resource for studies of influence of sialylation on chicken physiology and on viral infection. The level of sialylation will return to the previous level during a period of time after withdrawal of an inducing agent. The required time depends on the SA turnover rate. The inducing agent can be treated prophylactically to genetically modified birds in an outbreak that can protect livestock and locally stop infection spread. It is known that the inducible system has some limitations. Firstly, it has basal transgene 'leak'. However the 'leak' is weak and in our case it will be restricted by the specific tissues. The potential consequences of the 'leak' will be studied. Secondly, the feeding of domestic birds with the doxycycline antibiotic can compromise its safety as biological products. Nonetheless it is a temporary measure that could help to save the stock. The efficacy of our strategy needs to be evaluated to resolve all the concerns.

Supplementary Figure 1A & 1B. Genetic constructs.
Click here to access the data.

Supplementary Figure 2. Confirmation of inserted genes in MDCK genome after lentiviral transduction.
Click here to access the data.

Supplementary Figure 3. Schematic of integrated constructs with primers for PCR analysis.
Click here to access the data.
In conclusion, our current preliminary in vitro data indicates that sialidases from Salmonella typhimurium and neuraminidase from Influenza A virus could be the potential candidates that provide antiviral defence against avian Influenza virus. Since sialidases target cellular receptors but not a viral gene product, the chance of influenza viruses developing resistance is low.
14. The consistent apparent activity of the N1 neuraminidase acting on non-transduced cells, could more simply be an indication that the protein is being secreted. The lack of data on protein localisation does not help with the interpretation. The "slip-stream" explanation hypothesising that many more cells are transduced than were apparent in the photos is possible, but not very credible as this would also need the lentivirus titre/transduction efficiency to be much higher for the infNA vector. 15. Little consideration regarding the topology of their constructs has been given. Discussion: 16. The discussion is quite superficial. The authors might like to mention related approaches such as DAS181 and the discussions regarding this general approach 17. There are reports that de-sialylation does not completely abrogate influenza infectivity. A point perhaps worth mentioning briefly.
18. They may wish to mention more about the possible physiological consequences e.g. the potential impact on the MCE of constitutively expressing a sialidase 19. The choice of appropriate promoter is not discussed in any detail. Presumably the intention is to restrict expression to relevant tissues i.e. the respiratory and GI tract in the case of poultry. Do they have a particular promoter in mind?
20. The practicalities of using a Dox-inducible promoter are not considered. Is the intention to treat flocks continuously with Dox to prevent infection, or to respond prophylactically to an outbreak in a nearby location? The use of Dox in this way is undesirable for several reasons.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? No Abstract 1-2. Agreed. The statement regarding H5N1 was modified to 'Avian H5N1 and recently H7N9 virus has a potential pandemic threat.' in the version 2. Introduction 3.The sentence has been corrected from "Several thousands of cases of human infection were recorded and many of them with a fatal outcome" to "860 cases of disease in human population have been detected since 2003 and 454 of them with fatal outcome (see http://www.who.int/influenza/human_animal_interface/2018_03_02_tableH5N1.pdf?ua=1 4. The reference has been changed in the version 2. 5. Agreed. The sentence has been corrected Methods 6. The information has been added in the Methods part: "Additionally the reverse construction RFP-P2A-infNA was cloned in the same vector. (Supplementary Figure 1A)". 7. The sentence "The viral titre corresponded to the number of colonies developed at the highest dilution" was deleted due to it is related to the titre determination with drug selection. However lentiviral constructions with resistant genes were not considered in the publication. Instead, the following text has been added in the version 2: "FACS analysis was performed using non-infected cells as a negative control. The cells with 10%-40% RFP-positive were selected for titer calculation. The titer was calculated using the equation: Titer (Transduction Unit /ml) = (Cell Number in each well used for infection on Day 2 × percentage of GFP/RFP positive cells (should be 10%-40%))/(virus volume used for infection in each well × dilution fold)" 8. The three repeats were mentioned. The phrase "three serial infections" was corrected in the text to avoid confusion: "Infection with the same aliquot was used in three repeats." 9. The typo has been corrected in the version 2. Please check the supplementary availability now. Supplementary Figure 1A & 1B. Genetic constructs. Click here to access the data. Supplementary Figure 2. Confirmation of inserted genes in MDCK genome after lentiviral transduction. Click here to access the data. Supplementary Figure 3. Schematic of integrated constructs with primers for PCR analysis. Click here to access the data. Results 10.The mention has been added in the part Methods/Catalytically inactive sialidaseю 11. Corrected: 4.725(±0.983) 11. Corrected: 4.725(±0.983) 12. We used well-known transmembrane domain of Influenza neuraminidase for proper localization of both Act.v.Sia and Sal.t.Sia sialidases on cytoplasmic membrane. The sequence of the transmembrane domain and stem loop was the same for both bacterial sialidases, however only one was active. It is known that the sialidase hNeu3 is localised on the plasma membrane. We cloned the gene and checked the sequence. Lectin binding assay was used to confirm its functionality on the cell surface hence the correct localisation (in the case if activity was demonstrated). However we agree with your point and we have noticed it in the text (Results/Transient expression of genetic constructs in HEK293 cell lines) 13. Corrected: 1.468 (±0.439) 14. We have considered the hypothesis regarding secreted neuraminidase and rejected it. First, because we used neuraminidase from Influenza Puerto Rico strain (Addgene). And it is well-known that the neuraminidase is located on the plasma membrane surface. We sequenced the neuraminidase gene after cloning. Second, the reversion of positions of neuraminidase and RFP would not influenced on neuraminidase secretion. But in our case only RFP-positive cells were desialylated (CMV-RFP-P2A-infNA expression) that exclude secretion. Regarding the lentivirus titre/transduction efficiency. The lentivirus titre was defined by RFP fluorescence. However not all RFP sequences could be translated properly due to the effect of termination of translation at the end of 2A peptide that resulted in lower RFP fluorescence. Hence the lentivirus titre could be underestimated. In result the cells were infected by larger quantity of viral particles. FACS data was normalised to RFP fluorescence level when analysing the level of lectin binding. 15. We used well-characterised transmembrane domain to target all the enzymes (except human Neu3) on the cell surface and made functional estimation of catalytic activity using lectin binding assay. Enzymatic activity of a protein in the cell expressing reporter confirms membrane location of the protein on the cell membrane. Discussion 16. The mention regarding DAS181 has been added in the discussion part: "Previously, a recombinant sialidase (DAS181) has already demonstrated its antiviral effectivity in the cell culture (Malakhov MP et al., 2006)) and in vivo, using a mouse model (intranasal injection of the recombinant fusion protein -DAS181) (Belser JA et al., 2007). Thus, the principal possibility of the usage of membrane-anchored sialidase as antiviral defence has been demonstrated." 17. We added in the discussion points regarding the possible risk: "There are several concerns to use the strategy. The first is that influenza virus receptors may be different from SA, thus, sialidase expression may be ineffective. It has been reported that the virus binding could still occur in the MDCK cell line desialylated with addition in the medium sialidase ( Micromonospora viridifaciens Stray , 2000). The authors suggested that influenza virus infection can result from sialic et al. acid-independent receptors, either directly or in a multistage process. The presence of sialic acid may enhance virus binding to the cell surface to increase interaction with secondary receptors to mediate entry. Also the authors mentioned that high multiplicity of infection has an increased requirement for sialic acids and desialylation will inhibit the virus amplification. Although the HA of a human H1N1 strain was shown to bind other glycoconjugates (Rapoport , 2006), the exact et al. receptors to enter into the cell have not been found. Stray et al., used NA-deficient virus in the research which was obtained in a laboratory. It is unlikely that such virus would exist in nature. Probably, the barrier to efficient infection and transmission between organisms is due to inefficient use or expression of a yet unidentified entry mediator." 18. Added: "Another concern of the stable expression of sialidases is disruption of physiological functions required for proper glycosylation ( Gutierrez , 1987;Michalek , 1988;Pangburn et al. et al. , 2000;Varki, 1992). The probable consequences of constitutive expression of membrane et al. sialidase are not known. It has been shown that overexpression of the human ortholog in vivo membrane sialidase under the β-actin promoter in transgenic mice resulted insulin-resistant NEU3 membrane sialidase under the β-actin promoter in transgenic mice resulted insulin-resistant NEU3 diabetes mellitus (Sasaki ., 2003). The authors focused on insulin signaling only. However at et al least the data shows that the constitutive membrane sialidase expression was not lethal for mice. We are going to reduce a probable negative impact on organism using tissue-specific expression of the selected gene. It is necessary to predict a potential impact on the epithelium of intestine and trachea in vivo. There are no similar genetic experiments. However the consequences of tissue-specific expression could be considered from the researches where local desialylation was done. It has been shown that intranasal sialidase treatment does not affect the properties of respiratory mucus, nor did it affect the normal mucus transport activity on ciliated epithelium (King , 1974;Meyer , 1975). Some authors showed that Influenza infection is associated with et al. et al. an increased risk of secondary infection by (Sluijs et al., 2006). There Streptococcus pneumoniae are concerns that, probably, tissue-specific expression of sialidases would result in a greater chance of bacterial infection in the respiratory tract due to exposure of its binding sites that were masked by terminal sialic acid. It has been shown that the binding sites for the bacteria already exist on the normal respiratory surface and sialidase treatment does not introduce de novo bacterial adhesion sites. Also the risk of infection arises due to damage of mucus and epithelial cells. But it is not clear whether the secondary infection happens due to viral NA S. pneumoniae activity or from a secondary effects of influenza virus intracellular amplification. Influenza infection in the respiratory tract causes much deeper changes to the epithelium than treatment of the surface of the epithelial cells with a sialidase, because Influenza infection in humans and in mice results in vast exposure of the ciliated epithelial cells down to the basal cells and the basement membrane (Walsh , 1961;Plotkowski ,1986 In the gastrointestinal tract SA is an abundant sugar residue in mucin that is a key target of intestinal bacteria. Expression of sailidases will release free SA from mucins that could drives intestinal inflammation and infection. For example, elevated levels of free sialic acid in the gut, during and post antibiotic treatments, promote the expansion of Clostridium difficile, that utilises free SA. Also it has been shown that sialidase activity can promote the outgrowth of Escherichia and result in imbalanced microbiota, inflammation, colitis development due to α (2-3)-sialic coli residues promote expansion of (Huang ., 2015). Nevertheless, we do not know the E. coli et al probable negative impact on the physiology of avian respiratory and digestive tract. Thus, an optimal level of desialylation can be established in the tissue when combining genetic constructs with varying sialidase activity and tissue-specific promoters to avoid disruption of physiological functions in our approach. Additionally, a gene expression with time limit could prevent a probable negative impact on a living organism. Michael P. Malakhov showed that temporary et al. desialylation of respiratory tract during 7 days using recombinant sialidase protein did not cause any symptoms of toxicity or inflammation in ferrets and mice that were not infected by the virus (Malakhov ., 2006). This data support our idea that temporary inducible expression of a et al in vivo sialidase will not have serious negative physiological effects for transgenic animals." 19. Yes, we have. We have analysed several promoters specific for intestine and trachea. We have measured expression in scraping of the epithelium from the chicken intestine and trachea by RT-PCR and selected several tissue specific genes that have different expression level. Respiratory tract: SFTPA2, SLC34A2. Related expression normalized to GAPDH: SFTPA2 (3,5±0,7); SLC34A2 (0,002). p<0,001 Digestive tract: RBP2, GUCA2A, MUC13, OLFM4. Related expression normalized to GAPDH: RBP2 (1,1±0,01); GUCA2A (0,4±0,005); MUC13 (0,18±0,05); OLFM4 (0,09±0,015), p<0,01. 20. It is planned to use Dox prophylactically to an outbreak. The practicalities of using Tet-on system were added: "We are planning to use the Tet-on system as a potential model for creation of 1.

2.
system were added: "We are planning to use the Tet-on system as a potential model for creation of genetically modified organism that allows to induce a sialidase expression temporary. Transgenic transcription factor rtTA will be expressed under control of a tissue-specific promoter and a sialidase will be expressed under control of TRE promoter. This enables tissue-specific inducible control of the transgene expression and will provide a powerful and flexible resource for studies of influence of sialylation on chicken physiology and on viral infection. The level of sialylation will return to the previous level during a period of time after withdrawal of doxycycline. The required time depends on the sialic acid turnover rate. The inducing agent can be treated prophylactically to genetically modified birds in an outbreak that can protect livestock and locally stop infection spread. It is known that the inducible system has some limitations. Firstly, it has basal transgene 'leak'. However in our case at least the leak will be in the specific tissue. The probable consequences of the leak will be studied. Secondly, the feeding of domestic birds with the doxycycline antibiotic can compromise its safety as biological products. However it is a temporary measure that could help to save the stock. The efficacy of our strategy needs to be evaluated in to resolve all the concerns.' The authors have expressed membrane-anchored constructs of influenza, human and bacterial sialidases in cells and measured the effect on cell surface sialylated glycans. The study is preliminary but well executed as far as it goes.

General comments:
The manuscript does not include evaluation of influenza virus infection so the title is exaggerated. Maybe "Evaluation of defense strategy…". would be less misleading. potential The Figure legends are very terse and a reader needs to make some guesses at what is being shown. For example, what are the boxes on the panels in Figures 3 and 4? The manuscript needs to be checked for spelling. "sialilation" appears several times and there are other errors such as S. typhimyrium on p.7.

Specific comments:
Abstract: I have not seen data that shows the H5N1 virus is increasingly infective for humans. Either reference this statement or modify it.
Abstract and text: the increased activity seen with influenza neuraminidase is likely due to the version 2.
Respond to the specific comments: 1.The statement was modified to 'Avian H5N1 and H7N9 virus has a potential pandemic threat' in the version 2. 2.The information was added in the Results/Transient expression of genetic constructs in HEK293 cell lines. 3.Maakia amurensis lectins from bioWORLD that have declared alpha(2;3) specificity were used as analogous to an avian hemagglutinin due to they share similar structural fold. The lectins were used for preliminary study and it is planned to continue the research with an avian virus. 4. Both statements make sense, however "the optimal sialic acids level" is the result of sialidases expression and it undermines not only the level of sialidases on the cell surface but their catalytic activity. 5.The statement «Unlike human HA that preferentially recognises α (2-6), the viral NA of H1N1 has cleavage activity to both (2-6) and α(2-3) types of linkages» was supplemented with the information that NA also cleaves alpha2-6 linked sialic acids at lower rate than alpha2-3 . 6.The aim of the project was to compare created genetic constructs between each other and use some of them in future researches. We didn't try to achieve similar activity of the constructs to their recombinant counterparts. However it would be interesting to check how folding and localisation influence on the enzymes activity.
No competing interests were disclosed.

Competing Interests:
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com