We generated a replication-deficient, single-infectious cycle, recombinant influenza virus, expressing the firefly luciferase (Flux) reporter protein construct in the well-studied H1N1 PR8 influenza strain background. This was achieved by making specific modifications ( Figures 1a, b ) to the S-HA Flu genome (15) using methods described previously (15, 17). The resulting S-Lux strain retains all PR8 elements except that the hemagglutinin (HA) coding sequence has been replaced with the flux coding sequence for firefly luciferase. Viable PR8 S-Lux virus particles were rescued by supplying the A/PR/8/1934 HA gene in trans by transient transfection.

The ability of PR8 S-Lux to infect and express flux in cultured MDCK cells was confirmed at 24 h with detection of bioluminescence upon addition of D-Luciferin substrate and the signal was inversely proportional to dilutions of PR8 S-Lux ( Figure 1c ). The PR8 S-Lux infection was shown to be blocked by the known H1-neutralising mAb T1-3B (18) in a dose-dependent manner, but was not blocked by a non-relevant mAb (BJ-8C) ( Figure 1d ).

Four alternate S-Lux configurations were subsequently generated based on: (i) the potential pandemic avian influenza H5N1 strain (A/Vietnam/1203/2004), (ii) a 2013 H7N9 isolate [H7 A/Anhui/1/2013 (19)], (iii) a more recent 2017 H7N9 isolate (A/Taiwan/1/2017), and (iv) the 2017 H7N9 isolate (A/Guangdong/TH005/2017). For biosafety reasons, the polybasic HA cleavage site found in each of these four influenza isolates was converted to a monobasic trypsin sensitive site (20) prior to S-Lux pseudotyping. All H5 and H7 S-Lux viruses produced were able to infect cultured MDCK cells and express flux in a dose-dependent manner ( Supplementary Figure 1 ). Collectively, the generation of these five S-Lux variants supports the notion that inclusion of flux transgene in the HA vRNA is well tolerated.

To evaluate whether S-Lux could lead to bioluminescence which was detectable after in vivo infection, mice received an intranasally (i.n) dose of 5e4, 5e5 or 5e6 CID₅₀ (median cell infectious dose) of PR8 S-Lux and the level and duration of luciferase bioluminescence in the airways was monitored. At 24 hours post-delivery, bioluminescence was detected in the airways of dosed mice in a dose-dependent fashion ( Figures 2a, b ; p<0.0001; ANOVA with post-test for linear trend). For each of the active dosing groups, bioluminescence peaked at 24 h post-delivery and gradually decreased to background by approximately 8 days post-delivery. Crucially, these results confirm that S-Lux infection can lead to detectable bioluminescence in mice, which in turn can be used as a marker of infection to follow real-time infection kinetics within an animal treatment group. Importantly, the observed gradual decrease of the luciferase signal to background levels confirms that the S-Lux configuration does not support viral replication in the murine lung (15).

As proof of principle for the use of PR8 S-Lux as a surrogate to study anti-influenza treatments, we evaluated the in vivo efficacy of well characterised HA neutralising antibodies to limit S-Lux infection. Mice received an intraperitoneal (i.p) dose of 10 mg/kg of T1-3B, a known H1 HA stem neutralising antibody (18) or, as a negative control, a non-relevant antibody, BJ-8C, 24 hours prior to i.n infection with 3e5 CID₅₀ PR8 S-Lux. As shown in Figures 3a, b , mice treated prophylactically with T1-3B had ~2-orders of magnitude lower bioluminescence in the airways compared with mice receiving BJ-8C (p<0.001; ANOVA with Dunnett’s post-test) or those that had not received any antibody pre-treatment. Importantly, the findings with PR8 S-Lux were mirrored with PR8 influenza, where an identical dose of T1-3B given prophylactically, gave complete protection against a lethal challenge (10,000 TCID₅₀ (median tissue culture infectious dose); 1,000 LD₅₀ (median lethal dose)) ( Figure 3c ). Together, these results show that bioluminescence analysis following PR8 S-Lux infection can be used to predict the prophylactic efficacy of neutralising antibodies in vivo without the restrictions imposed on wild-type influenza studies, that commonly need to be terminated to avoid undue suffering to the animals in the study.

We subsequently evaluated the in vivo susceptibility of the panel of H5 and H7 S-Lux vectors to neutralising antibodies. To investigate H5 S-Lux, mice first received 10 mg/kg i.p dose of either the known pan-HA neutralising mAb MEDI8852 (21), an irrelevant BJ-8C mAb, or PBS as a control. Subsequently, mice were infected with 1e6 CID₅₀ i.n of H5 A/Vietnam/1203/2004 S-Lux. At 24 hours post-infection, mouse respiratory bioluminescence was significantly lower in the group treated with MEDI8852 than when treated with BJ-8C or PBS ( Figures 4a, b ; p<0.0001, ANOVA with Dunnett’s post-test).

Similarly, mouse respiratory infection with H7 A/Anhui/1/2013 and H7 A/Taiwan/1/2017 S-Lux configurations was evaluated with 10 mg/kg i.p of MEDI8852, a novel H7 cross-reactive mAb L4A-14 (22), the irrelevant mAb BJ-8C or PBS. In both instances, S-Lux infection was efficiently inhibited with L4A-14 ( Figures 4c–f ; p<0.0001, ANOVA with Dunnett’s post-test). Interestingly, MEDI8852 had only a modest neutralising ability against H7 A/Anhui/1/2013 S-Lux ( Figures 4c, d , p<0.05, ANOVA with Dunnett’s post-test) and no significant neutralising activity against H7 A/Taiwan/1/2017 S-Lux ( Figures 4e, f , not significant, ANOVA with Dunnett’s post-test). For H7 A/Guangdong/TH005/2017 S-Lux, we evaluated two dosing levels with L4A-14 and showed that while both 10 mg/kg and 20 mg/kg significantly reduced H7 A/Guangdong/TH005/2017 S-Lux infection, ( Figures 4g, h , p<0.0001, ANOVA with Dunnett’s post-test) the higher dose was superior ( Figures 4g, h , p<0.0001, ANOVA with Dunnett’s post-test). Importantly, 20 mg/kg of L4A-14 has also been previously shown to protect mice against weight loss and conferred full protection against ~100 LD₅₀ of the H7N9 (A/Guangdong/TH005/2017) influenza (22). Together, these results confirmed that H5 and H7 S-Lux can be used as surrogates for the respective wild-type influenza equivalents to aid prediction of prophylactic outcome of treatment with neutralising mAb in vivo.

We further evaluated the use of S-Lux to study and monitor clearance of infected lung epithelia mediated by CTL (Cytotoxic T lymphocytes) response following immunisation. To investigate that, a previously published live attenuated influenza vaccine which was known to induce strong protective heterotypic T-cell responses, S-FLU (15) was used. Mice were primed and boosted via i.n (1e6 CID₅₀) or i.p (1e7 CID₅₀) dosing with H5N1 S-FLU (H5 (A/Vietnam/1203/2004); N1 (PR8) or H7N2 S-FLU (H7 (A/Taiwan/1/201; N2 (A/Victoria/316/2011)), 2 weeks apart ( Figure 5a ). At 2 weeks post-boosting, all mice received i.n challenge with H5 (A/Vietnam/1203/2004) S-Lux (1e6 CID₅₀) and daily flux expression in the airways was measured. The results show that in both i.n and i.p routes where the mice were immunised with the “matched” S-FLU (H5N1) as the S-Lux, mice receiving i.n immunisation showed no expression of flux in the airways whereas mice receiving i.p immunisation showed flux expression at a log higher at 24h compared to i.n immunisation ( Figure 5b ). The results show that i.n immunisation of “matched” S-FLU was more effective in preventing influenza infection than i.p immunisation despite having lower serum neutralising antibody against H5 prior to challenge (EC₅₀; i.n: <40; i.p: 540, Supplementary Figure 2 ).

In mice immunised with “unmatched” S-FLU (H7N2), both routes of administration showed no detectable serum neutralising antibody against H5 ( Supplementary Figure 2 ) and did not block H5 (A/Vietnam/1203/2004) S-Lux infection in the airways. However, on day 1, both groups showed lower flux expression in the airways compared with the VGM (viral growth media) control group, suggesting that clearance of infected cells had happened before the first measurement (day 1). In general, both groups have more rapid clearance of flux signal compared with VGM control mice. Intranasal immunisation seems to exert a stronger (reduction by half a log on day 1) and more rapid clearance of signal in the lung (by a day) compared with i.p immunisation.

Following the successful use of S-Lux with influenza pseudotypes, we tested if glycoproteins from other infectious diseases could also be used to pseudotype S-Lux generating a reagent to facilitate the evaluation of anti-viral therapies targeted at the cognate glycoprotein and/or glycoprotein/receptor complex. As a pertinent exemplar, we investigated the utility of the S-Lux system with the Ebola virus (EBOV) glycoprotein. Ebola virus is a highly virulent and lethal human pathogen, which caused 11,316 deaths out of the 28,639 documented cases (~40% fatality) during the 2014–2016 Ebola outbreak in West Africa (23). Importantly, EBOV pseudotyped influenza has been shown to use the same entry pathway as the wild type Ebola virus (17), and can be useful as a surrogate to study EBOV entry. To facilitate the study of prophylaxis for EBOV in accessible low containment laboratories, we pseudotyped S-Lux with the glycoprotein from the Zaire EBOV (Makona variant). To evaluate Zaire EBOV S-Lux (ES-Lux), we dosed mice with 15 mg/kg of the EBOV neutralising mAb KZ52, which is perhaps one of the most extensively studied anti-Ebola antibody (24), or with PBS as a mock control 24 h prior to intravenous (i.v) infection with ~3.6e5 CID₅₀ of the ES-Lux. The result showed that at 24 h post-infection, mice that received PBS displayed high bioluminescence signal, which as expected was localised to the liver, whereas mice receiving the KZ52 mAb showed approximately three orders of magnitude lower bioluminescence, which was just slightly above background signal ( Figure 6 ; P<0.0001, t-test). This demonstrates that Zaire ES-Lux, similar to EBOV (25), has a high tropism for the liver after i.v. delivery. In addition, the results show that the relevant antibody against Ebola GP is also capable of blocking infection of Zaire ES-Lux when given prophylactically.

S-Lux was produced based on the described method (15, 17) with modifications highlighted in Figures 1a, b . The codon optimised flux sequence [GenBank accession no. CAA59282.1] was synthesised between appropriate NotI and EcoRI restriction sites and ligated into the S-FLU expression cassette. In brief, recombinant S-Lux viruses on the A/PR/8/1934 background were produced by transfection of HEK 293T cells as described (15, 37) with 1µg of each of the following plasmids pPol plasmids encoding the vRNA segments (pPol_PB1-PR8, pPol_PB2-PR8, pPol_PA-PR8, pPol_NP-PR8, pPol_NS-PR8, pPol_M-PR8, pPol_ vRNA), 4 core initiators to supply essential viral proteins in trans for vRNA replication and viral packaging (pCDNA3.1_PB1-PR8, pCDNA3.1_PB2-PR8, pCDNA3.1_PA-PR8, pCDNA3.1_NP-PR8), pCDNA3.1_H1_HA-PR8 (to supply HA in trans on the producer HEK293T cells) and pPol/S-US-HA-flux), and cloned twice by limiting dilution in MDCK-SIAT1 cells, transduced to express coating haemagglutinin from PR8 (Cambridge Strain) [GenBank accession no. CAA24272.1] to provide the pseudotyping haemagglutinin in trans. Viruses were then propagated in appropriate transduced MDCK-SIAT1 cells or MDCK-E-SIAT1 coated in H5 from A/Vietnam/1203/2004 [GenBank accession no. EF541403.1, (20)], H7 (A/Taiwan/1/2017 GISAID EPI917065; A/Anhui/1/2013 GISAID EPI439507; A/Guangdong/TH005/2017 GISAID EPI926825) or Ebola GP (Zaire ebolavirus Makona wt/GIN/2014/KissidougouC-15) [GenBank accession no. KJ660346.1; (17)], respectively, to generate S-Lux or ES-Lux. All HA sequences originally containing a polybasic cleavage site were converted to dependence on trypsin; for H5: PQRETR/GLFGAIA and H7: PEIPKGR/GLFGAIA (See OFFLU at http://www.offlu.net/: Influenza A Cleavage Sites, 31st Jan 2018).

Recombinant human IgG used in this study was produced by transient transfection of the IgG heavy and light chain expression plasmids (AbVec) in HEK293T grown in suspension in serum free media. Several of the IgG (BJ-8C and L4A-14) were produced using the ExpiCHO expression system (Thermo Fisher Scientific) according to the manufacturer’s protocol. IgG was then purified using the HiTrap MabSelect SuRe column (GE Healthcare) according to the manufacturer’s protocol. The VH of the MEDI8852 used in our study differs from the original sequence by a single amino acid (PKLLIYA to PKLLLYA) due to a misprint in the figure in (21).

S-Lux and ES-Lux were titrated as CID₅₀ as previously described (17) with slight modifications. Briefly, harvested supernatants containing S-Lux were titrated in a 2-fold serial dilution in viral growth media (VGM; DMEM-penicillin-streptomycin-0.1% BSA without trypsin) across a black flat-bottom 96-well plate seeded with 3e4 MDCK-SIAT1 cells. The plate was incubated at 37°C overnight and imaged for bioluminescence as described below. The dilution of virus giving 50% of the maximum plateau bioluminescence signal (EC₅₀) was calculated by linear interpolation. The CID₅₀/ml was calculated from the EC₅₀ dilution and the number of cells seeded per well (3e4 cells).

Microneutralisation assays were performed according to (15). Briefly, influenza viruses were diluted in VGM and titrated to give plateau expression of firefly luciferase in 3e4 MDCK-SIAT1 cells after overnight infection in 96-well flat-bottomed plates. Serial (2-fold) dilutions of mAb were incubated with S-Lux for 2 h at 37°C. A total of 3e4 MDCK-SIAT1 cells were then added in 100 µl of VGM and incubated overnight at 37°C and flux expression determined (as below). Titres were reported as fold-dilution of antibody that resulted in 50% reduction in flux expression (EC₅₀).

Animals used in this study were purchased from Envigo Ltd. (Shaw Farm, Bicester, United Kingdom) and procedures carried out at the Biomedical Services Unit (BMS) (University of Oxford, United Kingdom) under the terms of the Animal (Scientific Procedures) Act 1986. Mice were housed in accordance with the UK Home Office ethical and welfare guidelines and fed on standard chow and water ad libitum. Female BALB/c mice were used at 4 to 6 weeks of age. Antibody or vehicle (PBS) was delivered via intraperitoneal (i.p) delivery in a total volume of 500 µl. For delivery of S-Lux, mice were anaesthetised with isoflurane and 50 or 100 µl of S-Lux was delivered via the intranasal (i.n) route. For delivery of ES-Lux, mice were restrained and intravenous (i.v) delivery was performed via tail vein injection. For wild type influenza challenge study, mice were infected i.n with 10,000 TCID₅₀ of wild type A/PR/8/1934 (Cambridge) virus and a humane endpoint of weight loss and clinical score was used for mice that would otherwise have succumbed to infection. Animals were assessed for clinical score in terms of mobility, appearance, and breathing intensity. Mice reaching 20% weight loss from the pre-challenge body weight and/or a morbid clinical score were euthanised. Mice were euthanised by gradual fill of the chamber using compressed CO₂ (displacement rate of 30-70% chamber volume per minute).

For imaging of cell culture plates, supernatant was removed from all wells and washed with PBS. A working stock of 1x D-Luciferin (1.5 mg/ml) diluted in 100 µl D-PBS was then added to each well and incubated for 5 min before images were captured. For live animal imaging, mice were anaesthetised with isoflurane before i.n delivery of 100 µl of 100x (15 mg/mL) D-Luciferin for S-Lux studies and i.p delivery of 300 µl of 100x (15 mg/mL) D-Luciferin for ES-Lux studies. Images were captured after 5 min incubation. During imaging, anaesthesia was maintained via nasal delivery of 2.5% isoflurane. All images were acquired using the Xenogen IVIS Lumina LT Series III using the automated settings and analysed with LivingImage 4.5 software package (XenonCorp). For mouse imaging, regions of interest (ROIs) were drawn by creating a circle region around the chest area. For cell culture plates, ROIs were drawn using the standard grids with matrices consistent with cell culture plates. Bioluminescence signal from ROIs were defined and expressed as average photon flux (photons/s/cm²/sr). Data are presented as mean ± SEM.

Column and time-course data are presented as mean ± SEM. Differences between treatment groups were determined on log-normalised data using the t-Test, ANOVA with post hoc test for linear trend, or ANOVA with Dunnett’s multiple comparison test as appropriate. P-values of less than 0.05 were deemed statistically significant. Analyses were performed using GraphPad Prism 10.