Bacterial lipopolysaccharide reduces the stability of avian and human influenza viruses

Commensal bacteria can promote or reduce the severity of viral infection and disease progression in their hosts depending on the specific viral pathogen1. Influenza A virus (IAV) has a broad host-range, comprises many subtypes, and utilizes different routes of transmission including the fecal-oral route in wild birds2. It has been previously demonstrated that commensal bacteria can interact with the host’s immune system to protect against IAV pathogenesis3,4. However, it is unclear whether bacteria and their products may be interacting directly with IAV to impact virion stability. Herein we show that gastrointestinal (GI) tract bacterial isolates in an in vitro system significantly reduce the thermal stability of IAV. Moreover, bacterial lipopolysaccharide (LPS), found on the exterior surfaces of bacteria, was sufficient to significantly decrease the stability of both human and avian viral strains at the physiological temperatures of their respective hosts, as well as in the aquatic environment. Subtype and host-origin of the viruses were shown to affect the extent to which IAV was susceptible to LPS. Furthermore, using a receptor-binding assay and transmission electron microscopy, we observed that LPS binds to and affects the morphology of influenza virions.


3
IAV infection 3, 4 . Moreover, mice pretreated with bacterial lipopolysaccharide (LPS), a product present on the exterior surface and shed by all Gram-negative bacteria, triggered a TLR4 mediated antiviral response to protect the hosts from lethal infection with IAV 12,13 . In contrast, LPS binds directly to the capsid protein of poliovirus, increasing cell attachment and the ability of the virions to remain infectious at increased temperatures 14 , and LPS binding to mouse mammary tumor virus (MMTV) enables increased immune evasion and successful transmission of the virus 15,16 . In the case of influenza, it is unclear whether commensal bacteria and LPS are interacting directly with IAV in addition to their indirect effects on the immune system.
We first tested whether GI tract-derived bacteria can affect the stability of IAV. A panel of thirteen heat-killed bacteria derived from the GI microbiome (Table S1), standardized by protein content, and a water control were incubated individually with a human WSN H1N1-GFP virus for 1h at 48 °C. Stability was assessed by measuring the percentage of infected cells (those expressing GFP) by flow cytometry after an overnight infection. Eleven of the thirteen bacterial strains significantly decreased the stability of the virus after incubation compared to the water control (Fig. 1B). Our results indicate the reduction in infectivity was not due to the diluted bacterial products being cytotoxic (Fig.   S1A), nor was the decrease in infectivity due to the presence of bacteria or their products limiting the susceptibility of host-cells to infection (Fig. S1B). This suggests that the observed reduction in thermal stability may be due to the virus interacting directly with the bacterial cells and/or their products.
Because there was a wide range in the extent to which bacterial strains reduced viral stability, the specific composition of a host's microbiome could influence its susceptibility to infection by IAV. A recent study found that the cloacal microbiome of wild mallards infected with IAV differed from those who were IAV-negative, specifically in the Firmicutes, Proteobacteria and Bacteroidetes phyla 17 .
Although it is challenging to determine whether these differences in the microbiome preceded or were the result of IAV infection, bacteria representing these phyla used in our study all reduced IAV stability.
Given that bacterial LPS is present in milligrams/milliliter concentrations in the animal GI tract 18 , and that it interacts directly with poliovirus 14,19 and mouse mammary tumor virus (MMTV) 16,20 , we hypothesized that LPS may be interacting with IAV and affecting its stability. Incubating a human WSN H1N1-GFP and an avian H3N8 subtype influenza with 1 mg/mL of E. coli O111:B4 LPS for 1 h significantly reduced the stability of both viruses in a temperature-dependent manner (Fig. 1C). Intriguingly, the stability of the human and avian viruses were respectively reduced by 82-fold and 16-fold in LPS compared to water (p <0.001 and 0.01) at 37 °C and 42 °C, the physiological temperatures of their respective hosts. Furthermore, with temperature held constant (37 °C), LPS reduced the stability of IAV in a concentration dependent manner (Fig. S2). We used our in vitro system to determine whether the observed reductions in infectivity in the thermal stability assays were due to LPS interacting with the virus directly or through the indirect response of infected cells. Cells were either pretreated with LPS prior to infection or not and then were treated concurrently with a mixture of virus and LPS. Across a range of virus titers tested, the addition of LPS did not reduce the expected titer in our system for both scenarios (Fig. 1D), suggesting that the reduction in infectivity observed was due to LPS interacting directly with the virus, rather than the tissue-culture cells mounting an antiviral response. This was further supported by the observation that detoxified LPS, a modified version of LPS lacking the immunogenic lipid A portion of the molecule, also reduced viral stability (Fig. S3). We also observed that biotinylated-LPS binds to H1N1 PR8 virions in a solid-phase receptor binding assay (Fig. 1E). With 2.0x10 7 virions, the dissociation constant (Kd) was 24 µg/mL of LPS (R 2 =0.99). Since the HA glycoprotein of IAV binds to sialylated receptors on host cells, we postulated that LPS binding is mediated through HA.
However, pretreating the virus with antibodies against HA or its sialic acid receptors failed to block binding to LPS (Fig. S4). LPS could be interacting with virions in an HA-independent manner, either through the viral neuraminidase or potentially host-proteins derived from the host cell membrane as with MMTV 16 .
Aquatic habitats serve as long-term reservoirs for avian IAVs and are important to sustain transmission of the viruses in wild birds 21,22 . Abiotic factors, including temperature, pH, and salinity, that affect the persistence of IAV in water are well established 23 , but it is unclear how biotic factors, including bacteria and their products, affect virus stability. Avian H3N8 and human H1N1 PR8 viruses persisted significantly longer in water compared to water containing 100 µg/mL of LPS at 25 °C (respective p= 0.002 and <0.002) (Fig. 1F). Using the slopes of the regression analyses, for each treatment we calculated the log10 reduction time (Rt), defined as the length of time for the viral titer to decrease by 90% 24 . LPS reduced the Rt of the avian virus compared to water by 58%, and to a greater extent for the human H1N1 by 66%. Viruses circulating in the aquatic environment must also be able to withstand the natural freeze-thaw cycles that rapidly inactivate the virions 25 . We found that LPS significantly reduces the freeze-thaw stability of the human H1N1 PR8 virus by 34-fold compared to water (p< 0.001), whereas the avian H3N8 virus was not significantly affected (Fig. 1G). Bacterial products such as LPS may be reducing persistence and stability of IAV in the environment, which has as significant implications for modeling the emergence of novel strains from animal reservoir hosts.
Our long-term aquatic persistence results suggested that the avian isolate may be more resistant to LPS than the human virus tested. In addition, it has been reported that the microbiome content of IAV infected mallards was significantly affected by hemagglutinin (HA) subtype of the virus 17 .
To examine the extent to which the reduction in stability may be conserved across viruses we compared the thermal stabilities in E. coli LPS and water at 48 °C for 1 h of a panel of viruses that represented i) phylogenetically-distinct HA subtypes H1, H3, H4, H5, and H12 ( Fig. 2A) and ii) different host-origins (avian, human, seal). All of the strains tested had significantly reduced stability in LPS compared to water ( Fig. 2B), suggesting that LPS appears to have a universal effect of reducing the thermal stability of influenza viruses. Measurements of relative stability of each virus in LPS compared to water (LPS:water) did not follow a normal distribution (Shapiro-Wilks test, p<0.05) and were tested with a non-parametric Kruskall-Wallace H test. A significant difference was detected between subtypes (p<0.001), as well as host origin (p<0.001). The subtypes H4, followed by H5 and H12 (all avian origin) showed the highest stability in LPS, compared to H1 and H3 (Fig. 2C). We further analyzed H1 and H3 subtypes because for these two subtypes measurements of LPS:water stability were made for two or more hosts. For H1, virus originating from avian hosts was significantly more stable in LPS, compared to human viruses (DF=1, p=0.006). For H3, avian and seal viruses were more stable in LPS compared to human viruses, though not at a statistically significant level (DF=2, p=0.055).
Our data suggest that subtype and host-origin of IAV influence the ability of the virus to withstand inactivation (Fig. 2D). Most notably, avian-derived strains of IAV were more stable in LPS compared to the human-derived stains. Given avian IAV transmits primarily via the fecal-oral route 9 , compared to the respiratory transmission of human IAV 26 , and that the bacterial load in the GI tract is higher than airways 27,28 , avian strains may be adapted to the high concentrations of LPS that they encounter inside the host and in feces.
Because LPS affects IAV in a temperature dependent manner via direct binding, reducing the stability of the virus at increased temperatures, we next hypothesized that LPS may be compromising the structural integrity of the virions' envelopes. H1N1 PR8 virions incubated in water at 37 °C for 1 h retained their spherical morphology by transmission electron microscopy (TEM), whereas virions incubated with LPS had envelopes that were more noticeably deformed (Fig. 3A). To quantify whether there was a change in virion morphology, we measured the length:width ratio of the virions (Fig. S5). LPS treated virions had significantly different length:width ratio distribution compared to control virions using a Kolmogorov-Smirnov (KS) test (p <0.001) (Fig. 3B), but did not result in increased virion aggregation (Fig. 3C). Therefore LPS causes an increase in the proportion of virions with envelope deformations, which may account for the decrease in stability observed.

7
In conclusion, our study provides new evidence that LPS directly affects the stability of IAV by binding directly to and altering the morphology of influenza virions, suggesting that bacteria within hosts and in the external environment can limit the transmission of influenza virus. Our results are presented mainly in the context of IAV transmission within its natural avian reservoir, however we found that human-derived viruses could be more susceptible to bacterial products. Understanding the extent to which bacteria interact with and affect the infectivity of IAV may suggest novel pathways to the development of therapeutics to prevent or treat respiratory infections. Moreover, the overuse of antibiotics in poultry and swine, in addition to generating and disseminating antibiotic resistant strains of bacteria 29 , may be increasing their risk of IAV infection. However, our results stand in contrast to those found with poliovirus 14,19 leading us to propose that interaction with LPS is highly dependent on the virion structure. Additional studies monitoring the effects of antibiotics and changes in the bacterial microbiome on the susceptibility of host animals, as well as humans, to pathogenic viruses will help clarify the extent to which bacterial-viral interactions may be influencing the severity and epidemiology of viral infection.

Bacterial strains, viruses and cell-lines:
Cultures of GI tract and fecal microbiome bacterial isolates from humans and mice (see Table S1) were heat-killed at 95 °C for 10 min and standardized by their protein content using a bicinchoninic acid (BCA) assay (ThermoFisher). Viruses used in this study are listed in Table S2. We also used a reverse-genetics engineered H1N1 WSN PB1flank-eGFP (obtained from the laboratory of Jesse Bloom), referred to in this study referred to as H1N1 WSN-GFP, in which the PB1 gene's coding sequence was replaced by a gene encoding enhanced green fluorescent protein (eGFP) 30 . Virus titers and infectivity. H1N1 WSN PB1flank-eGFP was initially titered as described previously 30 .
Briefly, serial dilutions of the stock virus were allowed to infect 1.0x10 5 MDCK-SIAT1-CMV-PB1 cells for 16 h at 37 °C with 5% CO2. The % infected cells was determined by flow cytometry (Accuri C6, Accuri Cytometers), using uninfected control cells as the baseline and setting 0.5% of the control cells as GFPpositive (FlowJo). We then used the Poisson equation to calculate the number of infectious units (IU)/mL of the virus in initial stock. For the stability assays, viruses were added to cells as described above along with a standard curve of known quantities of virus in parallel to ensure they were in the linear range of the assay. To compare samples across independent replicates, we normalized the % infected cells to that of the water control. All other viruses were titered in MDCK cells by tissue-culture infectious dose 50 (TCID50) assays. Briefly, 3.0x10 4 MDCK cells were seeded into 96-well plates (VWR) overnight and then viruses were serially diluted across the plate, incubated for 2 h to allow attachment, washed, and then returned to incubate for 72 h at 37 °C with 5% CO2. Presence or absence of cytopathic effect (CPE) was observed and the TCID50/mL was calculated using the Reed-Muench method 32 .

Thermal stability assays:
To test whether bacterial products may be affecting the thermal stability of influenza virus, 1.0x10 6 infectious units (IU)/mL of H1N1-GFP was mixed into water or the heat-killed bacterial isolates, standardized to 1 mg/mL protein in water, and incubated at 48 °C for 1 h. Similarly, to test the effects of LPS on influenza stability, 1.0x10 6 IU/mL of H1N1 WSN-GFP or 5.0x10 6 TCID50/mL of the other viruses (Table S2)  Aquatic environmental stability: For the long-term persistence experiment, 1.0x10 4 TCID50/mL of avian H3N8 or human H1N1 PR8 were added to water or water with 100 µg/mL of LPS and incubated at 25 °C.
Viruses were periodically removed and titered immediately over 7 days to determine the quantity of infectious virions remaining. Data points were log10 transformed and then fit with a linear regression using GraphPad to calculate the log10 reduction time (Rt) which is the time required for infectivity to decrease by 90% (or 1 log10 TCID50/mL) 24,34 . An F-test was performed to compare the slopes of the linear regressions.
Freeze-thaw stability was tested by adding 1.0x10 6 TCID50/mL of human H1N1 PR8 or avian H3N8 virus to either water or 1 mg/mL of E. coli LPS. Samples were frozen at -20 °C for 3-5 days and then thawed at 25 ° C. Stability was assessed similar to the thermal stability experiments by measuring the titer prior to freezing and after thawing the viruses. Table S2) was selected based on diversity of hemagglutinin (HA) subtypes and host-origin of the viruses. The nucleotide sequences of the viruses' Segment 4 gene, which encodes HA, were used to construct a phylogenetic tree. We used MEGA7 35 to compute the evolutionary distances via maximum composite likelihood method 36 and assessed the evolutionary history using the Neighbor-Joining method 37     and LPS (N=579) binned by b, the length:width ratio of the virions to assess differences in viral envelope morphology, or c, the number of virions per viral cluster to assess differences in virion aggregation.

Phylogenetic analysis: A panel of viruses (see
Fractions of virions in were calculated based on the total number of virions in each micrograph. Data are represented as the mean ± SEM of 9-10 micrographs. Statistical significance was assessed by a student's t-test. ***p-value <0.001, ** <0.01, * <0.05.