New Class of Broad-Spectrum Antivirals Improves In uenza Virus Vaccine Development


 Enveloped viruses can cause devastating zoonotic diseases and are the most likely to cause global pandemics. We identified a new class of small-molecule sulfur-containing antiviral compounds (XM series) that broadly inhibit enveloped viruses. The antivirals’ mechanism of action was explored via various multidisciplinary approaches, concluding that the XM antivirals alter membrane lipid chemical compositions, increase membrane order deep within the hydrophobic region of the bilayer, and increase membrane phase transition temperatures. Such effects cause inhibition of membrane fusion and viral entry, while leaving the viral glycoproteins and genomes largely unaffected. Consequently, we tested whether these features would lead to effective whole inactivated enveloped virus vaccines. As a proof-of-principle, we generated an inactivated influenza virus (IIV) vaccine using compound XM-01 (XM-01-IIV). We compared this new vaccine to traditional paraformaldehyde-inactivated, to control compound-inactivated, and to live virus vaccines, using a mouse model of disease. Excitingly, compared to a traditional IIV vaccine, XM-01-IIV vaccination improved neutralizing antibody responses against both hemagglutinin and neuraminidase, and decreased mouse morbidity and mortality following influenza virus challenge. Therefore, this study uncovers a novel class of broadly acting antivirals that enhances influenza virus vaccine development and offers great potential for broadly generating future highly-potent inactivated enveloped virus vaccines.


Introduction
The 2020 World Health Organization's (WHO) list of emerging pathogens most likely to cause major pandemics and require urgent research and development is comprised entirely of enveloped viruses, including Ebola, SARS-CoV-2, and Nipah (NiV) viruses [1]. Annually, enveloped viruses are responsible for millions of human deaths and trillions of dollars of economic cost [2][3][4]. While established vaccines reduce the human and economic cost of viral outbreaks, new vaccine candidates against emerging pathogens do not always generate adequate protection from illness [5]. Therefore, the ability to rapidly generate effective vaccines for emerging or established viruses is imperative.

Whole inactivated virus (WIV) vaccines have advantages over live attentuated virus (LAV) vaccines,
including that WIV are replication-incompetent and thus typically safer, while still offering a broad repertoire of antigens presented in the context of viral particles as compared to sub-unit or genetic vaccines [6,7]. However, WIV vaccines often induce insu cient immunity or even vaccine-enhanced disease often due to antigen conformational differences caused by viral inactivation [7] via chemical (typically formalin or β-propiolactone [BPL]) or physical (heat, UV light) methods. Attempts to generate new viral inactivation methods for WIV vaccines face many pitfalls, including the degradation of protein conformations and genetic components [8][9][10][11][12][13][14][15][16]. Thus, developing new methods to fully inactivate viruses while leaving their glycoproteins in their native and immunogenic conformations is highly desirable.
Current WIV vaccines for in uenza viruses (IV), responsible yearly for 350,000-650,000 human deaths and up to a billion human infections worldwide, have rather low e cacies between 10% to 60%. [17,18]. This is partially due to a lack of heterologous immunity and abrogated immunogenicity due to the WIV preparation process [13,16,19], despite a vaccine including 3-4 IV strains. WIV vaccination shortcomings often point to the chemical methods of virus inactivation. For example, the properties that make formalininactivated vaccines ideal to inactivate a wide range of viruses, namely non-speci c cross-linkers that react with amino acids, RNA, and/or DNA, can also cause destruction, alteration, or sub-optimal presentation of antigens to the immune system [11,15,16]. Indeed, in early clinical trials, formalin-based WIV vaccination of respiratory syncytial virus (RSV) resulted in enhancement of disease due to impaired glycoprotein structure [20,21]. Therefore, maintaining the structure of viral glycoproteins is key to WIV vaccine development.
In the last decade, several potential broad-spectrum antivirals were identi ed which act speci cally on the viral membrane, often via light activation, including LJ001, JL122, and the hydrogen sul de (H 2 S)producing compound GYY4137. These compounds display inhibitory properties against multiple enveloped viruses including RSV, human metapneumovirus (hMPV), NiV, and others [22][23][24]. Targeting the viral membrane is an attractive option in preventing viral entry/infection; however, the abundance of host cellular membranes may limit the use of such compounds as therapeutics.
In this study, we identi ed a new class of antiviral compounds (XM series) that target the viral membrane while maintaining the integrity of viral surface glycoproteins, characteristics ideal for vaccine development. A multi-disciplinary characterization of the biological activity of XM compounds indicated that these light-independent persul de-radical producing agents embed into viral membranes, alter their biophysical and chemical properties, and inhibit viral-cell membrane fusion, viral entry, and productive infection. To test whether these characteristics of XM compounds made them suitable for WIV vaccine development, we optimized the inactivation of a mouse-adapted in uenza virus strain, and corroborated that XM treatment completely inactivated the virions without affecting viral glycoprotein functions. Further, mice vaccinated with XM-IIV developed more highly neutralizing antibody responses to both hemagglutinin (HA) and neuraminidase (NA) glycoproteins, and upon viral challenge yielded reduced morbidity and mortality compared to other IIV approaches. Finally, we demonstrated that in ammation and immune responses in XM-WIV vaccinated mice at 5 days post challenge were comparable or improved relative to other vaccine groups. Thus, the new lipid-membrane targeting XM class of compounds constitute a breakthrough for the development of enveloped virus whole inactivated vaccines.
We then evaluated the cytotoxicity levels of these ve inhibitory XM compounds in Vero cells using a CCK-8 cytotoxicity kit at 1, 3 and 10 µM for 30 min to 24 h (Fig. 1B). Based on these results, we proceeded with XM-01 for further characterization. XM-01 cytotoxicity was then tested from 1 µM -1 mM in Vero cells (Fig. 1C) and 1-100 µM in MDCK cells (Fig. S3)  XM-01 perturbs viral membranes without affecting viral glycoproteins or RNA. To determine if viral infection is inhibited by XM-01 at viral entry or post-viral entry steps, we incubated Vero cells with XM-01 at different time points post-viral infection. Cells were infected with pNiV for 2, 4, 6, 8, 18, or 24 h, after which unbound pNiV was washed away from the cells, and 100 uM XM-01 in media was added for the remainder of 24 h at 37°C. pNiV infection was then measured via luciferase activity ( Fig. 2A). No inhibition by XM-01 was observed compared to the DMSO vehicle control at any time point, indicating that once the virus has entered cells, XM-01 does not exert inhibitory activity, consistent with XM-01 inhibiting viral entry. We next incubated either Vero cells or pNiV for 30 min with XM-01 at various concentrations prior to viral infection (Fig. 2B). Viral infection assessed 24 hours post infection (h.p.i.) was decreased only when pNiV virions, but not when cells, was pre-incubated with XM-01, suggesting a direct effect of XM-01 on virions rather than on host cells.
Further, to narrow down whether viral membranes or viral glycoproteins were affected by XM-01, we tested whether XM-01 affected the conformations of NiV-F or -G. pNiV virions were incubated with 10 uM XM-01 for 30 min, followed by measurement of binding of established conformational antibodies to NiV F or G via ow virometry, to determine potential structural changes in F or G [29,[31][32][33]. As a control we used LJ001, a compound known to affect viral membranes but not viral glycoprotein conformations [23,27]. In both cases, no statistically signi cant changes were observed in conformational antibody binding to either glycoprotein (Fig. 2C), indicating that native glycoprotein conformations remained upon XM-01 treatment. To further test for potential effects in protein function, we used in uenza virus (IV) Ca09. When IV Ca09 was treated with up to 1,000 mM XM-01 for 4 h, neither neuraminidase (NA) nor hemagglutinin (HA) activity were signi cantly impacted as compared to the untreated control ( Fig. 2D-E). These were the same conditions con rmed to yield fully-inactivated in uenza virus (Fig. 2D), and used for vaccine development as further described below for Figs. 4, 5, and S6.
Further, to visually look for gross viral structural changes, we treated pNiV with 10 uM XM-01 or LJ001 control compound for 30 min, then imaged particles via electron microscopy. Most viral particles treated with XM-01 or the LJ001 control displayed compromised membranes, altered morphologies, and frequent RNA spillage compared to the DMSO control, consistent with XM-01 and LJ001 affecting viral membranes (Fig. 2F) [34,35]. Then to test the effect of XM-01 on viral RNA function, we in vitro transcribed CHIK reporter virus, which contains luciferase in the hypervariable domain of non-structural protein 3 (nsP3). RNA was treated with different compounds including XM-01 and decapping enzyme, and then electroporated in BHK cells. After 2 h, enough time for the CHIK reporter to initiate translation of non-structural protein and the luciferase to be produced, cells were lysed and luciferase activity measured. No effects of XM-01 were observed for XM-01 on viral RNA (Fig. 2G). Combined, these data indicate that XM-01 modulates viral entry by affecting viral membranes and not the viral glycoproteins or RNA.
XM-01 inactivates viral membrane physical properties. We then investigated whether XM-01 affected the membrane's physical properties and whether XM-01's mechanism of action may involve radical formation, similarly to LJ001/JL122 [34,35]. We rst performed electron spin resonance on large multilamellar vesicles treated with XM-01 using lipids with spin labels at various depths within the membrane. We detected spectral changes upon XM-01 binding to the spin-labeled lipid 1-palmitoyl-2-(16doxyl stearoyl) phosphatidylcholine (16-PC) (Fig. 3A). At 25 °C, a shift of the high eld peak upon XM-01 binding towards a lower frequency indicated that 16-PC was in a more hydrophobic environment ( To further understand the effect of XM-01 on membrane structure, we measured lipid order (ΔS 0 ). The ΔS 0 of liposomes with and without XM-01 binding was calculated for ve XM-01:lipid Mol:Mol ratios, with a greater S 0 indicating a higher ordered lipid alignment. The lipid order in the headgroup region of dipalmitoyl phosphatidyl tempo (2,2,6,6-tetramethyl-1-oxy) choline (DPPTC) was unchanged upon XM-01 binding (Fig. 3B), but increased in the 16-PC region. We also observed that XM-01 increased the phase transition temperature of both pure POPC (~5.3 °C) and POPC/POPG (4:1) (~5.8 °C) membranes, consistent with XM-01 turning the liquid-ordered phase into a relatively more gel-ordered phase (Fig. 3C). Altogether, these data indicate that XM-01 intercalates deep in the hydrophobic region of the lipid bilayer, induces membrane ordering, and increases the phase transition temperature of the membrane, all consistent with a decrease in the capacity for membrane fusion.
Since the activity of control compound LJ001 involves light-dependent radical formation, we then investigated whether the activity of XM-01 would also be light-dependent, and possibly induce radical formation. Whereas compound LJ001 was not signi cantly active in the dark, XM-01 was (Fig. S2). Further, we next examined the chemical properties of XM-01. In the case of other H 2 S donors, acyl groups on sulfur are transferred to nucleophiles and ultimately generate persul des, which can be oxidized into perthiyl radicals [44]. We thus analyzed XM-01 and its reaction with butylamine in CH 2 CL 2 , and recovered cysteine polysul des, the decomposition products of persul des ( Fig. 3D and 3E). The formation of these products indicates the presence of a persul de intermediate (RSSH) derived from XM-01. In cellular environments, the generation of persul des is known to produce radicals [45], which suggests that XM-01 and its perthiyl radical products may induce chemical changes to the viral membrane by a radicalforming mechanism.
Since XM-01 intercalates within the lipid bilayers and is able to form persul de radicals, we hypothesized that XM-01 may cause viral membrane lipid chemical changes. To test this, we performed lipidomic analyses on fully inactivated IV Ca09, using 1 mM XM-01, 0.02% paraformaldehyde (PFA), or 10% DMSO as a negative control, all treated for 4 h at room temperature (RT). The extracted membrane lipids were analyzed via liquid chromatography and tandem mass spectrometry (LC-MS/MS), and data analysis showed several classes of lipids to be markedly affected by XM-01 treatment ( Fig. 3F-H), including changes in ratios of key lipids identi ed as crucial for initial steps in membrane fusion. These data suggest that XM-01 acts directly upon membrane lipids to change their relative distribution/presence, both at the lipid class and individual lipid structure levels (e.g., acyl chain length and degree of unsaturation). Strikingly, the ratio of phosphatidylcholine (PC) to phosphatidylethanolamine (PE) lipids, required to be ~0.9-2:1 PC:PE for e cient membrane fusion (stalk formation), changed drastically in XM-01 treated membranes ( Fig. 3G-H) [46]. The PC:PE ratio in XM-01 treated virus increased to 10.97:1 PC:PE, as the concentration of PC slightly increased and concentrations of PE fell sharply. This may be due to a signi cant portion of PE species in mammalian membranes being plasmalogens, which alleviate oxidative stress via free radical scavenging [47]. Overall, these changes are consistent with a decreasing amount of intrinsic negative curvature upon XM-01 treatment, making fusion stalk formation and thus membrane fusion energetically unfavorable.
Vaccination with XM-01-inactivated virus improves neutralizing antibody responses and protection against live IV challenge. As a proof of principle, we sought to exploit the unique properties of XM-01 described in this study, namely being a membrane fusion antiviral that leaves viral glycoproteins largely unaffected, to produce an improved inactivated in uenza virus (IIV) vaccine. We optimized complete inactivation of IV Ca09 with XM-01, JL-122 [23], and 0.02% PFA treatment (a traditional viral inactivation process). Treating 1x10 5 PFU/mL Ca09 with 1 mM XM-01, 0.02% PFA, or 10 µM JL122 for 4h at RT completely inactivated Ca09, while DMSO vehicle treatment did not signi cantly inhibit virus viability. This was con rmed via plaque assays and 3 supernatant passages of treated virus in cell culture and in chicken eggs (Fig. 2D, S5). Importantly, such XM-01 inactivation conditions preserved HA and NA functions (Figs. 2D-E).
We then vaccinated groups of 10 mice (5 males + 5 females) intramuscularly (IM) with XM-01-IIV, JL-122-IIV, PFA-IIV, live IV (Ca09), or phosphate buffer saline (PBS) as a mock control group, all mixed 1:1 with alum. Although live IV IM vaccination is not approved for human use, while intranasal (IN) IV challenge is lethal in mice, IM vaccination using live virus in mice does not lead to pathology/infection, and is a benchmark for ideal IV vaccines to achieve in mouse models, providing a robust and protective immune response [48]. Mice received 2 vaccinations, 2-weeks apart, and serum was collected 7-10 days postvaccination to assess neutralizing antibodies (NAbs) using hemagglutination inhibition (HI) and neuraminidase inhibition (NI) assays ( Fig. 4A-C) [49,50]. Excitingly, the XM-01-IIV vaccinated mice generated statistically higher NAb titers against HA and NA compared to the PFA-IIV or JL-122-IIV vaccinated mice, and notably performed nearly as well (for NA NAbs, Fig. 4B) or as well (for HA NAbs, Fig.  4C) as the live-virus IM vaccinated mice ( Fig. 4B-C). These results are particularly exciting, as most conventional IIV vaccines do not induce strong anti-NA NAb responses, [51] and NAbs for the XM-01 treatment were statistically signi cantly improved over the PFA group at the higher Ab concentrations (Fig. 4B). Notably, in female mice after two vaccinations, XM-01-IIV NAbs against both HA and NA were even superior to those generated by the live-virus vaccine (Fig S6A), and after a third dose the NAbs for the live IM vaccine reached the levels of the XM-01 vaccination NAbs (Fig. S6).
After these encouraging results, all mice were then challenged intranasally at 5 LD 50 (1,000 PFU/mouse) with IV Ca09. 100% of mice vaccinated with the XM-01-IIV or live-virus vaccines survived the challenge, whereas only 80% of PFA-IIV vaccinated mice survived, and all mock vaccinated mice succumbed to the infection within 6 days post-challenge (dpc) (Fig. 4D). Weight loss is a key predictor of IV morbidity in mice [52]. Importantly, mice vaccinated with XM-01-IIV and live-virus control suffered signi cantly lower morbidity, as shown by a lower amount of weight loss compared to all other IIV-vaccinated or control groups (Fig. 4E).
In a complementary experiment with fewer mice (female only) animals were euthanized at 5 dpc for histological analysis of various tissues, and ow cytometric analyses of immune cells (Fig. 5). Histological examination revealed peribronchiolar and perivascular in ammation with lymphocytes and granulocytes in all mice (Fig. 5A-J). Mice vaccinated with mock or JL-122-IIV vaccines developed more severe peribronchiolar and perivascular in ammation with degenerate neutrophils and cellular debris in airways (Fig. 5F, H). Mice vaccinated with PFA-IIV, XM-01-IIV, and live-virus had less in ammation and no signs of intraluminal degenerate neutrophils or debris in airways (Fig 5G, I, J). Flow cytometric analysis of immune cells corroborated the observation of reduced lung in ammation in the PFA-IIV, XM-01-IIV, and live-vaccinated groups when compared to JL-122-IIV and saline-vaccinated groups, with fewer lung neutrophils and CD8 + T cells (Fig. 5K-L). Additionally, the proportion of germinal center B cells was increased in the mediastinal lymph nodes of PFA-IIV, XM-01-IIV, and live-vaccinated groups compared to mock-vaccinated control (Fig. 5M, N), indicating a positive and speci c response to vaccination.
Overall, viral inactivation using XM-01 generated a more effective IIV vaccine compared to traditional PFA treatment, by eliciting a relatively more protective neutralizing antibody immune response, correlating with reduced morbidity and mortality upon viral challenge.

Discussion
We report a new class of broad-spectrum antivirals with broad inhibitory properties against enveloped viruses. Compounds XM-01, 02, 03, 06, and 12 possess antiviral activity and display low levels of cytotoxicity. A series of mechanistic analyses demonstrated that XM-01 affects the virus membrane lipid bilayer chemically and physically, inhibiting viral entry, without signi cantly affecting the viral glycoproteins. Given these results, we explored the potential of XM-01 for improved WIV vaccine development, as one of the major reasons for the poor performance of many WIV vaccines is the damage of the viral glycoprotein antigens during the inactivation process [8, 9,12,15,19,[53][54][55][56][57][58][59]. Excitingly, using a mouse IAV model, we demonstrated that XM-01 treatment improved the in vivo generation of NA and HA neutralizing antibodies, in turn lessening both morbidity and mortality upon viral infection challenge.
We observed that XM-01 physically and chemically changes the lipid bilayer of viral membranes by intercalating into the lipid bilayer, increasing membrane order and rigidity deep within the membrane, and increasing the phase transition temperature, all consistent with making the negative curvature events required for membrane fusion energetically unfavorable. One exciting feature of XM-01 treatment is that the PC:PE ratios are drastically altered from the optimal range of 0.9-2.0 PC:PE [60] to the observed PC:PE ratio in XM-01 treated IAV of 10.97, mainly due to reduction in PE lipid species. This may be due to a signi cant portion of PE species in mammalian membranes being plasmalogens, which alleviate oxidative stress via the free radical scavenging capability of a vinyl-ether bond and sn-2 polyunsaturated fatty acid acyl chain [47]. Other global lipid changes included a global increase in lipid unsaturation and an increase in lysolipids, both of which can contribute to making the negative membrane curvature required during the membrane fusion process energetically unfavorable [61]. These chemical changes within the membrane may explain the observed changes in its physical properties. As with LJ001 and related compounds, XM compounds can exhibit effects against both cells and viruses, although at different concentrations, indicating no speci city in the type of membrane targeted. However, cells have an impressive capacity to repair their membranes while viruses lack lipid repair mechanisms, which may explain why the viral particles were signi cantly more sensitive than cells [62]. Importantly, we do not anticipate that XM or related compounds would be used as therapeutic agents unless they were speci cally made to target viral membranes. This is because the abundant host cell membranes would likely sequester the XM compounds. However, all these XM characteristics, including the relative lack of changes in glycoprotein conformations or functions make the XM compound inactivation strategy ideal for WIV vaccine development.
As a proof-of-principle, we utilized the unique properties of XM-01 to generate a WIV in uenza vaccine. WIV vaccines are advantageous over subunit vaccines since WIV vaccines can present multiple viral proteins and genetic material to the immune system in the stoichiometric context of the naturally immunogenic viral particle [63]. However, many viral inactivation strategies for the generation of WIVs, such as heat, UV light, or the traditional PFA and BPL methods commonly used for IIV generation, are known to often damage the viral glycoprotein antigens during the inactivation process, which in turn can lead to ineffective, or even deleterious (e.g. for RSV), immune responses to vaccination [8-11, 13, 15, 19, 64, 65]. Importantly, XM-01 fully inactivated IAV while leaving NA and HA enzymatic activity unaffected. Excitingly, XM-01-IIV vaccination induced high levels of NAbs against both HA and NA, comparable to live-virus IM vaccination, performing better than the traditional PFA IIV vaccination. Although both anti-HA and anti-NA Nab responses can be protective, an increase in NAbs against NA is exciting, as NA NAbs are thought to provide relatively more broadly-protective responses against multiple in uenza strains [51]. Indeed, previous studies showed that passive transfer of serum containing antibodies speci c only to NA completely prevented death in lethally-challenged mice [51]. Further, traditional IIV human vaccination does not seem to yield high NA NAb responses, which may contribute to its poor performance [51]. Notably, female mice were found to have more effective anti-HA and anti-NA NAb responses following fewer vaccinations of XM-01-IIV than other vaccinations types, including that of live-virus vaccination (Fig. S4). These data suggest that XM-01 derived vaccines may allow faster seroconversion rates after a single boost. It has been previously reported that sex-associated immune responses for both in uenza infection and vaccination are more robust in female mice, which may explain why this difference was more prominently noticed in this gender [66].
Histological analyses of various tissues and ow cytometric analyses of immune cells from vaccinated mice at 5 dpc supported the ndings of enhanced induction of immunity by the XM-01 derived vaccine. Notably, histological examination of lung sections did not reveal quantitative differences between livevirus-vaccinated, XM-01-IIV, and PFA-IIV. Similarly, ow cytometric analysis of lung immune cells indicated reduced in ltration of neutrophils and CD8 + T cells, indicative of reduced lung in ammation upon challenge following vaccination with live-virus, XM-01-IIV, or PFA-IIV. However, the signi cant enhancement of NAbs to both HA and NA of the XM-01-IIV compared to all other IIV-vaccine groups correlated with reduced morbidity and mortality. We speculate that these encouraging results may be due to both the quantity and quality of neutralizing antibodies, in turn due to increased glycoprotein preservation in the XM-01-IIV group, a major indicator of WIV vaccine potential [15]. Animal studies have shown that passive transfer of even non-NAbs against IV challenge can be fully protective against mortality [67]. This may explain why PFA-IIV had lower NAb titers yet maintained similar histological and speci c immune cell responses as compared to the live-vaccinated and XM-01-IIV groups. Ultimately, however, the PFA-IIV and JL-122-IIV groups did not perform as well as the live-vaccinated or the XM-01-IIV groups in overall morbidity (weight loss) and mortality, indicating that not only the overall strength of generated immune response impacts the outcome, but the speci city to key neutralizing antibody epitopes is likely critical for successful vaccination results. In summary, our results demonstrate a rst proof-of-principle study whereby membrane inactivators are shown useful for vaccine development. A new class of antivirals, represented by XM-01, effectively inactivates enveloped viruses by physically and chemically changing their membranes, while minimizing perturbation of their glycoproteins. Excitingly, our data indicates that these characteristics of XM compounds make them useful to improve WIV vaccine development. Further mechanistic chemical studies, such as those that optimize the rate of persul de radical formation, may lead to future derivatives that possess improved antiviral activity, perhaps further improving vaccine development. The exciting results of this proof-of-principle study may thus help us better prepare against current endemic as well as emerging enveloped viral pathogens and their potential pandemics, which pose a major threat to human and animal global health.
Methods Small-molecule compounds tested. XM-01 and other sulfur-containing compounds tested were used in prior studies as H 2 S donors. These compounds were synthesized using known protocols [68]. LJ001 and JL122 compounds were synthesized at the University of California, Los Angeles (UCLA) by Dr. Michael Jung's group [23,27]. All stock solutions for these compounds were prepared in 100% DMSO, stored at -20 °C, and used within 6 months of reconstitution. Cytotoxicity assay. Vero cells were incubated with each compound for 30 min to 24 h, as indicated, at the speci ed concentrations. This was followed by incubation with a cell counting kit reagent (CCK-8) (Dojindo Molecular Technologies, Japan) for 1-2 h, absorbance was measured at 450 nm using an in nite M100 microplate reader (Tecan Ltd). The quantity of the formazan dye produced when WST-8 (Dojindo) is reduced by dehydrogenases is directly proportional to the number of living cells (i.e. cell viability).
Transmission electron microscopy (TEM) imaging. 5 µL of concentrated VLPs or VSV-NiV were pipetted onto a 200-mesh Formvar-coated nickel grid and allowed to settle for 20 min at RT. Excess liquid was removed by wicking with lter paper before coating the deposited sample with 5 µL of 1% uranyl acetate (UA) (Polysciences, Inc). After 2.5 min, excess UA was wicked off using lter paper and dried overnight in a desiccator. TEM micrographs of the samples were recorded under high vacuum with an electron beam strength set at 200 kV using the FEI Technai G2 20 Twin TEM (FEI Corp., Hillsboro, OR).
RNA Stability Assay. Brie y, we in-vitro transcribed CHIK reporter virus containing a luciferase gene in the hypervariable domain of nsP3. 20 µg of the nsP3:luciferase construct was treated with decapping enzyme at 1, 3, 10, 30, and/or 100 µM XM-01 concentrations for 1 h on ice, and then electroporated in BHK cells. After 2 h, enough time to initiate translation of nsP3 and luciferase, cells were lysed and luciferase activity measured. Tris-HCl at pH 7.5, and 1 mM EDTA) at 110,000 x RCF for 2 h. The treated virus was resuspended in NTE buffer, then stained as previously described [69]. We used conformational anti-NiV F and/or anti-NiV G speci c rabbit primary antibodies (Anti NiV F Ab 66, or anti-NiV G Ab 213) [29,32,33] at 1:100 dilution for 1 h, followed by a FACS buffer (1% FBS in PBS) wash and incubation with secondary Alexa 647 goat antirabbit antibodies (Life Technologies, NY) for 30 min followed by one more FACS buffer wash. We then measured the relative levels of antibody binding through ow virometry, using a Guava easyCyte8HT ow cytometer (EMD Millipore, MA) [69]. Background mean uorescence intensity (MFI) was obtained by Electron Spin Resonance. Prepared MLVs (POPC, POPG and 0.5% (mol: mol) spin-labeled lipids), were resuspended and hydrated in a buffer consisting of 5 mM HEPES, 10 mM MES, 150 mM NaCl (pH 7) for 2 h at RT. Varying amounts of XM-01 (1 mg/mL in DMSO) were added to the MLV dispersion for 1 h at RT, along with the vehicle control. ESR spectra of ultracentrifuged MLVs were collected on an ELEXSYS ESR spectrometer (Bruker Instruments, Billerica, MA) at X-band (9.5 GHz) using an N2 Temperature Controller (Bruker Instruments, Billerica, MA). The ESR spectra were analyzed by the NLLS tting program based on the stochastic Liouville equation [41,43] using the MOMD or Microscopic Order Macroscopic Disorder model, as in previous studies [37-39, 73, 74]. Each experiment (and subsequent t) was repeated 2 or 3 times to check reproducibility and estimate experimental uncertainty.
XM-01 Decomposition. Butyl amine (47.4 mg, 0.65 mmol) was added to a solution of XM-01 (50 mg, 0.13 mmol) in CH 2 Cl 2 (10 mL). The reaction mixture was stirred for 2 h at RT, then concentrated and subjected to column chromatography (30% Ethyl acetate/Hexane) to separate the products (33 mg) as a mixture of disul de and polysul des.  Vaccine generation and vaccination. A/California/04/2009 (~2x10 5 PFU/mL) was thawed on ice before being incubated for 4 h at RT with 1 mM of XM-01 in 10% DMSO, 10 µM JL-122 in 10% DMSO, 0.02% PFA, and a mock control with vehicle only (10% DMSO). The solutions were then mixed 1:1 with alum for 30 min under light rocking before 100 µL was injected intramuscularly to mice.
In uenza virus (IV) propagation and IIV passage in egg culture. All infections were done in certi ed pathogen free eggs following standard protocols [49]. Brie y, eggs (10-12 days old) were surface decontaminated in a biosafety cabinet, then 100 µl of inactivated virus or a low dose (~1x10 3 PFU/mL) of A/California/04/2009 was injected into the amniotic uid. The injection site was sealed with glue, and placed in an egg incubator for 72 h. Eggs were moved to 4 °C for 24 h before collecting amniotic uid. Amniotic uid was cleared by centrifugation (1000 x g for 5 min) and supernatant was aliquoted and snap-frozen in liquid nitrogen before being stored at -80 °C. Treated IV was passaged as described above, and amniotic uid was assayed for infectious virus or passaged again in eggs up to three times to propagate any infectious particles.
Hemagglutination Inhibition (HI) Assay. Serum samples in HI buffer were added to a V-bottom 96-well plate, and diluted by 2-fold serial dilutions, then 8 HA units of Ca/04/2009 H1N1 in HI buffer was added.
The plate was incubated for 30-45 min at RT, followed by the addition of 0.8% rooster blood in HI buffer.
Results were read 30-45 min later and the HI titer value was read as the inverse of the lowest dilution of serum that completely inhibited hemagglutination [49]. Three or more independent experiments were performed, error bars represent ±SD. Unpaired t-tests were performed using Graphpad PRISM software, with Bonferroni corrections.
Neuraminidase Inhibition a(NI) ssay. NI assays were performed using a protocol adapted from Leang and Hurt [50]. Brie y, sera taken from mice was assayed using 2-fold serial dilutions, then incubated with A/California/04/2009 virus in at-bottom plates in assay buffer for 30-45 min at RT before the addition of the MUNANA substrate. The sealed plate was incubated at 37 °C for 1 h before the addition of the stop solution. Results were read using a Tecan Spark plate reader set to 355 nm excitation measuring absorbance at 460 nm.
Histology. For histologic examination, lung tissues were collected directly after euthanasia and placed in 10% formalin for >72 h. Following para n embedding, 4 m tissues sections were prepared and stained with hematoxylin and eosin (H&E).
Flow cytometry. Lymphocytes were extracted from lung, mediastinal lymph node, and spleen, as previously described [77]. Cells were stained in 30 L of 1:200 antibodies in PBS for 20 min at RT, washed, and then resuspended for ow cytometry. Fluorescent antibodies used for the study are listed as target (clone; catalog #). Antibodies against mouse antigens included CD16/32 (93; 14-0161-85) Fc block, CD3 Eight-week-old C57BL/6 mice were used for this study. Mice were intranasally inoculated with 1x10 3 PFU/animal. Following challenge, mice were monitored and weighed daily and euthanized at predetermined humane criteria following approved protocols, generally when weight loss reached 20% from day of challenge or mice became moribund with a clinical score >3 on a 5-point scale (58  ANOVA and Dunnett's multiple comparisons test (N = 3; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).