Seasonal influenza viruses decay more rapidly at intermediate humidity in droplets containing saliva compared to respiratory mucus

ABSTRACT Expulsions of virus-laden aerosols or droplets from the oral and nasal cavities of an infected host are an important source of onward respiratory virus transmission. However, the presence of infectious influenza virus in the oral cavity during infection has not been widely considered, and thus, little work has explored the environmental persistence of influenza virus in oral cavity expulsions. Using the ferret model, we detected infectious virus in the nasal and oral cavities, suggesting that the virus can be expelled into the environment from both anatomical sites. We also assessed the stability of two influenza A viruses (H1N1 and H3N2) in droplets of human saliva or respiratory mucus over a range of relative humidities. We observed that influenza virus infectivity decays rapidly in saliva droplets at intermediate relative humidity, while viruses in airway surface liquid droplets retain infectivity. Virus inactivation was not associated with bulk protein content, salt content, or droplet drying time. Instead, we found that saliva droplets exhibited distinct inactivation kinetics during the wet and dry phases at intermediate relative humidity, and droplet residue morphology may lead to the elevated first-order inactivation rate observed during the dry phase. Additionally, distinct differences in crystalline structure and nanobead localization were observed between saliva and airway surface liquid droplets. Together, our work demonstrates that different respiratory fluids exhibit unique virus persistence profiles and suggests that influenza viruses expelled from the oral cavity may contribute to virus transmission in low- and high-humidity environments. IMPORTANCE Determining how long viruses persist in the environment is important for mitigating transmission risk. Expelled infectious droplets and aerosols are composed of respiratory fluids, including saliva and complex mucus mixtures, but how well influenza viruses survive in such fluids is largely unknown. Here, we find that infectious influenza virus is present in the oral cavity of infected ferrets, suggesting that saliva-containing expulsions can play a role in onward transmission. Additionally, influenza virus in droplets composed of saliva degrades more rapidly than virus within respiratory mucus. Droplet composition impacts the crystalline structure and virus localization in dried droplets. These results suggest that viruses from distinct sites in the respiratory tract could have variable persistence in the environment, which will impact viral transmission fitness.

transmit efficiently between hosts is therefore critical to better mitigating the circulation of these viruses in future pandemics and ongoing seasonal epidemics.
Influenza virus is spread through the expulsion of virus-containing respiratory fluid aerosols or droplets that remain infectious in the environment and reach the respiratory tract of a new host to initiate a new round of viral replication (2,3).Respiratory activities, such as breathing, coughing, talking, and sneezing, all produce aerosols or droplets spanning a wide range of sizes that may contain infectious virus (4,5).Many studies have indeed confirmed the presence of influenza virus RNA or infectious influenza virus in aerosols expelled from infected individuals (6-10).These virus-containing aerosols or droplets are composed of respiratory fluids originating from distinct parts of the respiratory tract, including the lungs, trachea, oral cavity, and nasal cavity (4).Respiratory mucus, which comes from the epithelial lining of the lungs, consists primarily of water (~95%) (11) and contains a significant number of proteins, lipids, and inorganics, many of which have important functions in immune protection and mucociliary clearance (12,13).Saliva, generated from the salivary glands, contains a higher water content (>99%) (14), and the remaining fraction comprises numerous inorganics and proteins that serve antimicrobial functions as well as roles in digestion and pH buffering (15).Antiviral or antibacterial proteins commonly identified in saliva include mucins and protein-bound sialic acids (16,17).While many of the constituents of respiratory mucus and saliva overlap, like mucins, antibodies, and salts, differences in composition exist, such as lower lipid levels in saliva (15,18) as compared to respiratory mucus (19).Importantly, the relative levels of infectious virus emitted in fluids from different parts of the respiratory tract and their environmental persistence in distinct respiratory fluids are not well understood.
To transmit efficiently, expelled viruses must retain infectivity in the environment.The impact of certain environmental factors, such as humidity and temperature, on virus persistence in droplets or aerosols has been studied extensively (20)(21)(22)(23)(24)(25)(26)(27).Work to date has demonstrated an overall reduction in the persistence of enveloped viruses at intermediate humidities and ambient temperatures in both aerosols and droplets (21,(23)(24)(25)27), although the inverse phenomenon has been observed in some work with nonenveloped viruses (25,26).Unfortunately, these studies have primarily been conducted using laboratory-derived solutions (e.g., cell culture medium and phosphatebuffered saline solution), limiting the relevance of these findings in informing real-world transmission scenarios.
Some studies have recently begun to employ physiologically relevant aerosol or droplet matrices.Influenza virus has been found to maintain infectivity in aerosols and droplets composed of human airway surface liquid collected from human bron chial epithelial (HBE) cell cultures differentiated at an air-liquid interface, representative of respiratory fluid from the lower respiratory tract, at relative humidities (RH) rang ing from ~20% to over 90% (28,29).Similarly, a recent publication using nebulized saliva microdroplets containing human virus surrogates, including bacteriophages MS2, phiX174, and phi6, observed an overall increase in the persistence of the virus in saliva droplets at all RHs compared to water or media after 14 hours (30).However, studies of other mammalian viruses in saliva have found that viruses, such as murine coronavirus and vesicular stomatitis virus, decay rapidly at intermediate RHs (31,32).More research is clearly required to understand how different respiratory viruses retain infectivity in distinct respiratory fluid matrices under a variety of environmental conditions.
The differences in environmental virus inactivation exhibited across different RHs, viruses, and compositions may be due to various factors, including salt or protein content, evaporation kinetics, and/or pH.Prior research has focused on the influence of solutes and proteins on environmental virus decay (22,26), suggesting that protein may provide a protective effect at intermediate RH, while salt may inactivate viruses during the wet phase of droplet drying (22).The role of pH in virus inactivation has been central to recent virus persistence studies, although findings are not consistent.Modeling work demonstrated that rapid acidification of aerosols can render viruses such as SARS-CoV-2 or influenza virus inactive, while another publication showed SARS-CoV-2 inactivation correlated with an increase in pH at high RH (33,34).Experiments in saliva have suggested model viruses are protected by carbohydrates at low RH (31) or susceptible to antiviral proteins at intermediate RH (32).Despite these efforts, the mechanisms driving virus inactivation in distinct respiratory fluids remain largely unknown and may vary depending on virus type.A better understanding of the factors governing viral persistence is important to develop more informed interventions that limit respiratory virus transmission.
In this study, we first established whether influenza virus is present in the oral cavity of experimentally infected ferrets and found substantial levels of infectious virus present in saliva.Given the potential for salivary particles to mediate transmission, we examined the impact of saliva on the environmental stability of influenza virus.To this end, we assessed the persistence of two influenza A viruses, an H3N2 virus [A/ Perth/16/2009 (H3N2); H3N2] and the H1N1 2009 pandemic virus [A/California/07/2009 (H1N1); H1N1pmd09], in 1 µL droplets composed of human saliva or airway surface liquid at low, medium, and high RH (i.e., 20%, 50%, and 80%) and ambient temperature over a 2-hour period.We observed distinct virus decay and droplet morphology patterns in saliva compared to airway surface liquid.Our findings demonstrate the importance of using physiologically relevant matrices when assessing environmental persistence and lend insights into the mechanisms driving virus persistence in different respiratory fluids.Taken together, this information can ultimately be used to help inform strategies for restricting the spread of influenza viruses.

Oral swabs from infected ferrets contain infectious influenza virus
The concentrations of infectious influenza virus found in the saliva of infected individuals over time have not been widely characterized.We therefore assessed the infectious virus levels in the oral cavity of ferrets infected intranasally with two seasonal influenza viruses, H1N1pdm09 and H3N2.Specifically, we swabbed each ferret's tongue, cheeks, hard palate, and soft palate; importantly, we did not swab the back of the throat, which could include virus from the lower respiratory tract.In parallel, we sampled the ferret's nasal cavity by nasal wash to observe the relative amount of virus present in the proximal tip of the ferret nostril.
Influenza virus concentrations from oral swabs followed the same trends observed in nasal wash levels over the course of infection for both H1N1pdm09 and H3N2 viruses, although infectious virus in oral swabs fell below the limit of detection earlier during H3N2 infection (Fig. 1).Infectious influenza virus was consistently detected in oral swabs from intranasally infected ferrets on days 1 through 5 post-infection, with levels as high as ~4-log 10 TCID 50 /swab.Viral titers in the nasal wash of some ferrets were as high as 5.5-log 10 TCID 50 /mL on day 1 or 2 post-infection.Together, these results demonstrate that infectious influenza virus is present at elevated levels within the oral cavity during infection.Expulsion of influenza viruses during breathing, talking, or coughing may therefore act as a source of onward transmission.

Influenza A viruses decay rapidly at intermediate RH in saliva droplets but not in airway surface liquid droplets
The presence of the influenza virus in the host oral cavity on multiple days in experimen tally infected animals highlights the potential role of saliva in transmission.However, the persistence of influenza virus in saliva has not been characterized.We, therefore, investigated the environmental stability of two influenza A virus strains, H1N1pdm09 and H3N2, in 10 × 1 µL saliva droplets after 1 or 2 hours of exposure to 20%, 50%, and 80% RH and ambient temperature (i.e., ~22°C; Fig. 2; Fig. S1).Virus decay in saliva was compared to decay in airway surface liquid, which serves as a biological surrogate for respiratory airway secretions.Our group has previously demonstrated reduced decay of human seasonal influenza viruses in droplets and aerosols consisting of airway surface liquid as compared to cell culture medium (28,29).
Major trends in virus inactivation were similar for both influenza A virus strains evaluated (Fig. 2).As expected, influenza viruses in airway surface liquid droplets were protected from decay, never exceeding 1.1-log 10 decay, on average, over 2 hours.We found that influenza A virus persistence was greatest at 20% RH for both droplet compositions, with maximum average decay for both strains reaching just 1-log 10 after 2 hours (Fig. 2A and D).Interestingly, we observed increased decay of influenza A virus in saliva droplets containing saliva at 50% and 80% RH compared to airway surface liquid droplets (Fig. 2B, C, E and F).At 1 hour of exposure to 50% RH, decay of infectious H1N1pdm09 and H3N2 in saliva droplets was significantly greater than in airway surface liquid (Fig. 2B and E).At 2 hours, virus decay differences for both strains at 50% RH varied significantly by droplet composition, with average influenza virus degradation in saliva droplets exceeding 3-log 10 , while virus decay in airway surface liquid droplets was only ~1-log 10 .Virus decay in saliva droplets at 80% RH was intermediate, with average decay of 2.9-log 10 and 1.7-log 10 for H1N1pdm09 and H3N2 viruses, respectively (Fig. 2C  and F).Our findings establish that influenza A virus persistence in droplets is highly composition dependent.In addition, these two influenza A subtypes decay in a similar manner for a given RH and droplet composition.
To ensure the observed decay was not a side effect of poor virus recovery, we measured RNA recovered from virus-laden droplets at 50% RH in saliva and airway surface liquid at 0 and 1 hour.Results demonstrate no significant loss in viral RNA levels over this period (Fig. 3A; P > 0.05), suggesting that any loss of infectivity observed in our experiments is indeed a result of virus inactivation in droplets (Fig. 3).Because gene copy concentrations were similar over the droplet drying period, but infectious virus levels were reduced, the gene copy to infectious unit ratio increased with time (Fig. 3C).This is an important consideration when using genome-based detection methods to study virus persistence because genome copy levels are likely to overestimate infectious virus levels following longer periods of environmental exposure.

Influenza virus inactivation is not driven by phase, salt content, or protein concentration of the droplet
The inactivation mechanisms driving the differences in virus persistence observed for distinct respiratory fluid droplets are not known.Protein concentration has been suggested to impact virus decay in droplets (22); however, the airway surface liquid and human saliva used in our current study have similar levels of total protein content (Table 1).Therefore, differences in virus decay by droplet composition are not due to overall protein levels, although specific proteins within saliva or airway surface liquid could be responsible.In addition, protein partitioning (e.g., to the air-liquid interface or phase interphases within a droplet) or aggregation could differ between saliva and airway surface liquid, which could also contribute to the observed differences in persistence.Conductivity, a measure of dissolved ions, was also assessed to establish the solute concentrations in these respiratory fluids.Human saliva had lower conductivity, 3.54 mS/cm, than airway surface liquid, with a level of 16.52 mS/cm (Table 1).Elevated salt (i.e., ion) concentrations have been linked to increased virus inactivation (23).Here, we observed the inverse relation, suggesting that other factors were more important in driving virus inactivation in saliva droplets at 50% RH.
Recent work has shown that pH significantly impacts virus persistence in bulk solution, droplets, and aerosols (33,34).All bulk solutions used in our droplet persistence experiments had a pH that was neutral to slightly basic (Table 1), but the pH in small droplets may differ from that of the bulk solution (35).Due to technical limitations, we did not measure pH in 1 µL droplets during drying, so we cannot discern how pH might have affected virus inactivation in these experiments.
Beyond salt and protein content of droplets, previous work has suggested virus inactivation in droplets differs depending on whether the droplet is in the wet phase when it is still evaporating, or the dry phase when the droplet is no longer losing water (27,36).We, therefore, assessed the drying time by measuring the mass of 10 × 1 µL droplets of saliva and airway surface liquid containing H1N1pdm09 over time (Fig. 4, Table 2) (27,36).
As expected, the drying time increased with rising RH.Evaporation kinetics were similar in saliva and airway surface liquid droplets at all tested RHs (Fig. 4; Table 2).Specifically, average drying times differed by at most 4.2 min between airway surface liquid and saliva droplets for a given RH.Average drying time at 20% RH in airway surface liquid and saliva was 15 and 15.6 min, respectively, while drying time at 50% RH was slightly greater, 28.2 min for airway surface liquid and 32.4 min for saliva.In some trials at 20% RH, droplet mass reached a minimum and then gradually increased.We have observed this behavior before (36) and believe it is due to uptake of water vapor by hygroscopic salts that are exposed upon drying.At 80% RH, droplet mass stabilized after ~1.3 hours in both saliva and airway surface liquid.The drying times defined by mass and visual inspection were similar.The only exception was at 80% RH, where droplets did not dry by observation within the 2-hour exposure period.As drying time across droplet matrices was similar (Fig. 4) while decay rates differed (Fig. 2), we can conclude that virus inactivation is not solely a function of evaporation kinetics, and other mechanisms are important.

Droplet composition influences influenza virus inactivation kinetics before and after drying
To better understand influenza virus inactivation kinetics during the wet and dry phases, we examined the short-term inactivation kinetics of H1N1pdm09 in droplets at 50% RH while drying.Decay of influenza virus in saliva droplets after 1 hour at 50% RH was ~2-log 10 , while virus in airway surface liquid droplets exhibited significantly less overall decay, ~0.6-log 10 on average (Fig. 5).Trends in saliva droplet virus decay were distinct across the wet and dry phases, with degradation in each of these phases appearing to follow first-order kinetics, a model commonly applied to describe virus inactivation (37).To accommodate the difference in inactivation rates, we used saliva droplet drying time as the breakpoint to distinguish kinetics in these two phases.Simple linear regression of the data prior to droplet drying indicates that influenza virus in saliva  does not decay significantly during the wet phase (slope = 0.010 ± 0.012 min −1 ; mean ± 95% CI), but after drying, the inactivation rate increases to 0.036 ± 0.020 min −1 .
In contrast to the trends observed in saliva droplets, decay in airway surface liquid continued at a similar rate between the two phases, also following first-order kinetics.We were not able to detect any differences in inactivation rates between the wet and dry phases for airway surface liquid, likely due to the low level of inactivation observed over an exposure period of 1 hour.While the decay of virus in airway surface liquid droplets was low, the virus inactivation rate was significantly greater than zero, 0.010 ± 0.0030 min −1 .Our findings confirm the protective effect of airway surface liquid in comparison to saliva at intermediate RH.Additionally, they highlight considerable differences in the inactivation of influenza virus in saliva during the wet and dry phases and emphasize that the majority of decay at intermediate RH is driven by mechanisms that occur during the dry phase.

Droplet composition and RH influence droplet crystalline structure and nanobead location
It is possible the aggregation or interaction of viruses with other solutes in droplets upon drying could impact virus inactivation.Any observed differences in the crystal structure of airway surface liquid and saliva could therefore inform possible drivers of virus persistence.To assess how crystal structures within droplets vary by RH and droplet type, we captured microscopic images at 10× magnification of 1 µL droplets containing  a The average of two independent replicates is shown in bold, and the range of two independent replicates is shown in parentheses.
H1N1pdm09 composed of human saliva or airway surface liquid following a 2-hour exposure to 20%, 50%, and 80% RH environments (Fig. 6).In line with our evaporation measurements, droplets did not completely dry out at 80% RH, leading to a gelatinous solution without much discernible structure.Droplets at 20% and 50% RH, on the other hand, exhibited extensive crystalline structure.Dried airway surface liquid droplets had feather-like crystals once dry at 20% and 50% RH, while saliva displayed skinnier, line-like crystalline structures.These line-like crystalline structures in the saliva droplets differed at 20% and 50% RH, with more densely packed structures at 20% RH.In addition, a thick ring of structure was observed at the edge of airway surface liquid droplets at 20% and 50% RH, referred to as a "coffee-ring" effect (i.e., movement of solutes within the droplet to the liquid-surface edge) (38,39), while a thin coffee ring was present at the edge of saliva droplets at 20% and 50% RH.The differences in crystal structure formed at variable RH could play a role in virus persistence if virus interaction or incorporation with these crystal structures impacts inactivation.In situ virus particle visualization in droplets presents considerable challenges because of the small size of virions.We therefore used fluorescent nanobeads similar in size to influenza virions as a proxy for virus particles; these beads have a strong fluorescent signal that is easily detectable using standard fluorescence micro scopy techniques.In both respiratory fluids, nanobeads clustered around the outer rim of the droplet upon drying, regardless of RH (Fig. 7; Fig. S2).At 80% RH, the coffee-ring distribution was more dispersed than at and 50% RH.This is likely due to the fact that the droplets had not completely dried by observation, leading to a gelatinous solution lacking in crystalline structures that would allow for nanobead colocalization.At 20% and 50% RH, extensive nanobead accumulation in the airway surface liquid droplets was also observed in the interior of the droplet, where beads colocalized with crystalline structures.To a lesser extent, this phenomenon was also observed in saliva.While the nanobeads in the coffee ring of the airway surface liquid droplets coincided with the thick crystalline complex located around the perimeter of the droplet, the nanobeads in saliva droplets did not appear to collocate with crystalline structures in the coffee ring, but rather collocated with the thin region around the perimeter of the droplet that appeared to be free of any crystalline structures.Additional work to identify exactly which solutes or proteins are found in these regions of the different droplets will be beneficial to uncovering potential interactions of virions and particles that could play an important role in virus inactivation.Together, these data indicate that distinct respiratory fluids exhibit composition-dependent differences that likely influence influenza virus localization, interactions, and stability.

DISCUSSION
Little work to date has looked at infectious influenza virus levels present in the oral vs nasal cavities of infected hosts over time, despite the fact that respiratory fluid particles composed of saliva could contribute to onward spread.Limited clinical research has identified influenza virus nucleic acid or antigen in oral specimens (40)(41)(42).Another study focused on culture-based detection of influenza in humans consistently found nose and throat samples positive for infectious virus (43).Here, we consistently detected infectious virus in oral swabs, as opposed to throat swabs that have been the focus of previous work, sampled from ferrets during influenza A virus infection.Our findings confirm that influenza virus present in the oral cavity of an infected host is likely to be infectious, lending experimental evidence for the possibility of onward transmission from both nasal and oral cavity expulsions.Future work focused on the quantity, size distribution, and composition of emissions from each cavity is needed to ascertain the relative contribution of each anatomical site to virus-laden particles in the environment.this study, we also compared the environmental persistence of influenza virus in droplets of saliva versus airway surface liquid, the latter being representative of mucus-containing respiratory fluid.We demonstrated that two human seasonal influenza A virus strains, H1N1pdm09 and H3N2, are more susceptible to decay at midrange humidity in saliva as compared to in airway surface liquid.Recent work in saliva microdroplets (~30-600 µm in diameter) robust virus stability, with three different bacteriophages decaying by at most 1.5-log 10 after 14 hours over a broad range of humidities (30).Phages exhibited more decay in a buffer solution and water compared to saliva at all RHs, the inverse of what we observed here.On the other hand, work with vesicular stomatitis virus in 2 µL saliva droplets at variable RH resulted in the same U-shaped trend in virus decay that we observed with influenza virus in saliva (32).Murine coronavirus in saliva aerosols also exhibited reduced decay at 20% RH compared to at 50% or 80% RH (31).Although these studies used physiologically relevant fluids, none of them used human respiratory viruses.To our knowledge, our work is the first to evaluate influenza virus decay in saliva.Taken together, these observations suggest there may be substantial differences in virus fate depending on the virus used, respiratory fluid source, RH, and droplet or aerosol size.
By more carefully examining the kinetics of droplet drying and virus decay in saliva and airway surface liquid, we found that while both fluids demonstrated similar drying times, there were distinct differences in influenza virus persistence patterns.At midrange humidity, the virus in saliva droplets exhibited minimal inactivation during the wet phase, but upon drying, decay increased significantly.In contrast to the virus in saliva, virus decay occurred at sustained, low levels in airway surface liquid regardless of droplet phase, although we cannot be certain that there was no shift in inactivation kinetics following drying since overall virus reductions were too low to dependably detect a change.Distinct inactivation trends in different phases of droplet drying have also been observed for influenza and other respiratory viruses in cell culture medium, although the trends are not consistent (27,36), which could be due to different droplet composi tion and/or the timeframe measured.The differences we observed in virus inactivation between the wet and dry phases of saliva droplets underscore the importance of the virus' microenvironment in understanding the mechanisms governing virus persistence.Additionally, our results stress the need to study virus persistence in physiologically relevant solutions, as different matrices clearly exhibit altered trends in virus decay in droplets.
Many factors have been proposed to drive environmental virus decay in droplets or aerosols, but the true causes of inactivation remain unknown.Previous work demon strates an influence of total protein (44), salt content (26), and pH on virus decay (33,34).Our work provides additional insights; despite similar bulk protein levels and slight differences in salt content across respiratory solutions, influenza virus decay was strikingly distinct between airway surface liquid and saliva droplets, indicating that at the concentrations present, these constituents do not appear to be driving the observed trends.Considerable differences in lipid concentrations have been reported for saliva (15,18) and respiratory mucus (19), with levels on the order of ~2 mg/dL and ~750 mg/dL, respectively.It would be interesting to determine whether differences in lipid levels could account for the viral decay observed in the fluids used in this study.Specific chemicals and proteins present in saliva and airway surface liquid could also impact virus persistence.Antiviral salivary components have been identified through hemagglutination inhibition and virus neutralization, including protein-bound sialic acid, salivary scavenger receptor cysteine-rich protein (gp-340), and gel-forming mucin 5B (MUC5B) (16,17,45).MUC5B is also one of the principal mucins found in airway surface liquid (46).While its function in the host includes protection of the epithelial lining, it is unclear what properties this protein exhibits in aerosols and droplets and how expression may alter these properties.How these specific antiviral components of saliva play a role in the distinct patterns of influenza virus decay observed in droplets at various RHs is currently unknown.Their potential presence does not explain the virus stability measured in saliva at 20% RH in our study, unless more rapid evaporation of droplets at low RH resulted in reduced antiviral activity or interactions with virus upon dehydration.Additionally, our controls of virus suspended in bulk saliva solution at ambient tempera ture resulted in little to no reduction in infectious virus over the experimental period (Fig. S3).This suggests that the concentrations of antiviral constituents in bulk saliva not sufficient alone to extensively reduce influenza virus infectivity during this time period, and RH-mediated evaporation changes in small droplet volumes contribute significantly to decay.Comparative analysis of distinct respiratory fluid constituents and further characterization of virus-specific interactions are required to identify the factors that influence virus persistence in droplets under variable environmental conditions.
In examining the droplet microenvironment visually during the dry phase, we noted that the morphology of dried droplets and distribution of 100 nm nanobeads, a proxy for influenza virus particles, differed considerably across RHs and droplet compositions.Previous studies have noted disparate relationships between the coffee-ring effect and virus persistence; less decay of the surrogate virus bacteriophage phi6 in cell culture media droplets was correlated with a thicker coffer ring upon drying (38), while increased vesicular stomatitis virus decay in saliva droplets was associated with increased colocalization of virus with lysozyme, an antiviral protein, in the coffee ring of evaporated saliva droplets at intermediate RH (32).While we observed that the thickest coffee ring was associated with dried airway surface liquid droplets and their viral persistence, virus-containing saliva droplets at 20% RH exhibited a minimal ring, but viruses retained infectivity.Additionally, the spatial distribution of our fluorescent nanobeads generally followed the coffee-ring effect, although nanobeads were observed throughout the airway surface liquid droplets in complexes that could be proteinaceous or solutederived, particularly at 50% RH.Clearly, the significance of the coffee ring of solutes may vary depending on the matrix composition and virus used.It is important to note that the colocalization observed in this study and in other work does not confirm direct interactions of these constituents in dried droplets.Furthermore, while the fluorescent nanobeads in the study are similar in size to influenza viruses, they likely exhibit distinct properties that could differentially impact their distribution within respiratory droplets.Future work characterizing the interactions between actual virions and antiviral or protective constituents in distinct respiratory fluid droplet matrices will be critical to establishing the mechanisms driving inactivation during the droplet's dry phase.
While we have established important differences in influenza virus persistence in droplets composed of distinct respiratory fluids, our work has limitations.We studied influenza virus persistence in 1 µL droplets (i.e., ~1240 µm in diameter), which are at the larger end of the droplet size range measured during respiratory activities (4).We selected this droplet size because it can be manually pipetted while still being more physiologically relevant than larger droplet sizes (e.g., 5-50 µL droplets) that have frequently been used (27,(47)(48)(49)(50). Indeed, a recent study from our laboratory indicated that influenza virus decay in 50 µL droplets does not follow the same trends as smaller droplets (36).Past work with influenza virus has shown similar stability trends in aerosols and droplets (29); however, we cannot be certain that the results we have observed in 1 µL droplets will hold in smaller droplets or aerosols.Future research efforts should focus on representative droplet and aerosol particle sizes using relevant respiratory fluids, like those used in this work, to ensure our findings are generalizable.Also of note, the saliva used in this study was a commercially available pooled saliva sample; the number of samples and relevant traits of the saliva donors were unknown.Factors, such as age, gender, and disease state, could impact the composition of respiratory fluids and may in turn alter influenza virus persistence.Although we observed similar levels of virus decay using airway surface liquid from multiple distinct host cells, more variability might be observed with other samples given these factors.While the use of actual human saliva and airway surface liquid in this study is an important step toward understanding inactivation in the real world, future studies focused on a broad set of respiratory fluid samples will aid in identifying the differences in virus decay that may be expected in the secretions from distinct individuals.In addition, while our work holds for the two seasonal influenza A viruses assessed, persistence in respiratory fluids should be assessed for other strains and additional important respiratory viruses, particularly for those with distinct genome types and structures, including rhinoviruses, adenoviruses, and coronaviruses.The propagation method used generate virus stocks should also be considered, as host origin is known to impact stability (28).
Environmental persistence of influenza viruses is hypothesized as one possible factor contributing to sustained virus transmission in temperate regions during winter months (51,52).Our findings could help explain these seasonal trends.If onward transmission of influenza is driven by aerosols and droplets predominantly composed of saliva, then the decreased persistence of influenza at midrange humidity would be in line with reduced transmission observed in summer months when indoor RH typically ranges from 40% to 60% RH (51).Increased stability of influenza virus in saliva droplets at low RH would be consistent with the sustained transmission in winter months when indoor RH is usually between 10% and 40% (51).While our findings of influenza persistence in saliva align with observed seasonal trends, the ubiquitous survival of influenza virus in airway surface liquid does not.Additional research characterizing the relative propor tion of different respiratory fluids in virus-laden aerosols or droplets contributing to onward transmission is needed to understand whether environmental persistence plays a significant role in influenza virus seasonality.In addition, the characterization of the aerosol size distribution expelled by the mouth versus the nose would help inform the use of correct aerosol sizes and provide insight into the relative composition of aerosols from these cavities.This work has important implications for developing robust interventions to mitigate transmission during peak periods of influenza virus illness and spread.

Influenza virus stocks and quantification
Biological influenza virus A/California/07/2009 (H1N1pdm09) and recombinant influenza virus A/Perth/16/2009 (H3N2) were used in this study as previously described (28).Virus propagation was conducted by growing 1:50,000 CP1 virus stocks on confluent Madin-Darby canine kidney cells, kindly provided by Dr. Kanta Subbarao, in infection medium [Eagles' Minimum Essential Medium (MEM) with L-glutamine (Lonza, Cat.No. BE12-611F) containing 2× antibiotic-antimycotic (Gibco, Cat.No. 15240062), L-glutamine (Lonza, Cat.No. BE17-605E), and TPCK trypsin (Worthington-Biochem, Cat.No. LS003750)] at 37°C.Virus was harvested after significant cytopathic effect was observed.Cellular material was removed through centrifugation at 2,000× g for 10 min at 4°C.The H3N2 virus stock was concentrated through a 30% sucrose cushion to increase the detectable range of virus decay in droplet experiments by ultracentrifugation at 24,000× rpm for 1.5 hours at 4°C after propagated virus was harvested from cells.Following ultracentrifugation, H3N2 virus was resuspended in infection medium and vortexed vigorously.Virus stocks were stored in aliquots at −80°C until use.Infectious virus concentrations were quanti fied using the tissue culture infectious dose 50 (TCID 50 ) Spearman Karber method, as previously described (53).

Animal ethics statement
Ferret experiments were conducted in a biosafety level 2 facility at the University of Pittsburgh in compliance with the guidelines of the Institutional Animal Care and Use Committee (approved protocol 22061230).Animals were sedated with isoflurane following approved methods for all nasal washing and oral swabbing.

Bulk respiratory fluid analyses
The total protein concentration in each respiratory solution was determined by the bicinchoninic acid assay (BCA; Thermo Scientific, Cat.No. 23225), using bovine serum albumin (Thermo Scientific, Cat.No. 23210) as the protein standard.Conductivity, total dissolved salt, salinity, and pH were quantified with a field probe (ThermoFisher, Cat.No. 13-643-124).

Droplet generation
Pooled human saliva (Innovative Research, Cat.No. IR100044P) or airway surface liquid (details of respiratory fluids provided below) was used for virus droplet solutions.In each independent experiment, virus stocks were diluted 1:10 into each droplet solution.10 × 1 µL droplets of each diluted suspension were deposited on polystyrene material (six-well plates, Thermo Scientific, Cat.No. 140675) and exposed to ambient temperature and variable RH (20%, 50%, or 80%) for 1 or 2 hours in a controlled environmental chamber (Electro-Tech Systems, 5532 Series).A Hobo Temperature and Humidity Data Logger (Onset, Cat.No. UX100-011) was used to monitor the temperature and RH for the duration of each experiment (Fig. S1).After the exposure period, droplets were resuspended in 500 µL of MEM with L-glutamine and immediately stored at −80°C until analysis.Droplets were also generated and collected at 0 hours.Three independent replicates were completed with technical triplicates conducted for each RH condition and droplet composition.Control samples of the bulk droplet solution were collected at 0 and 2 hours to ensure no significant influenza virus occurred.

Respiratory fluids
Pooled human saliva was processed as specified by Innovative Research.Briefly, saliva samples were stored at −80°C upon collection and pooled by thawing samples and combining them.The pooled solution was then passed through a cheese cloth before aliquoting and freezing at −80°C until purchase.Details regarding the number of samples and information about saliva donors were not disclosed.HBE cultures were grown at the air-liquid interface as previously described (29).Airway surface liquid was collected periodically from the air-liquid interface by adding 100-150 µL phosphate-buffered saline to wells, incubating cells for 5 min at 37°C, and harvesting the solution following incubation.Airway surface liquid was collected from HBE cultures from multiple donor patients to capture potential heterogeneity in viral persistence.Specifically, airway surface liquid droplet suspensions used in each independent replicate were from a different donor HBE culture.Five different HBE cultures were used in influenza virus persistence work (deidentified culture numbers: 0277, 0284, 0302, 0304, and 0305), and each independent replicate was performed in triplicate.

Examination of droplet drying kinetics
10 × 1 µL droplets composed of virus solutions, generated as described above using H1N1pdm09 virus in either airway surface liquid or saliva, were deposited on a 35 × 10 mm polystyrene petri dish (Falcon, Cat.No. 351008) and placed on a microbalance (Sartorius Cubis I, Model MSE3.6P, readability = 0.0010 mg) within the environmental chamber.The chamber was set at 20%, 50%, or 80% RH.The glass enclosure of the microbalance was removed to ensure droplets were exposed to the desired RH within the chamber.Droplet mass was recorded 10 seconds for 2 hours.The time to droplet drying was determined by assessing the point at which the droplet mass left the linear phase of evaporation.Two independent replicates were conducted for each RH condition and droplet composition.Images of the droplets were taken every 10 seconds to visually observe changes in droplet morphology and drying and are available in Video S1 to S6.While the microbalance was zeroed prior to each experiment, balance sensitivity immediately after droplet addition often resulted in droplet masses dropping below zero.In these cases, the minimum mass achieved over the 2-hour experimental period was assumed to be the zeroed mass for calculations of percentage original mass.This had no impact on the drying time, which was only dependent on when the rate of mass loss diverged from the droplet's linear evaporation phase.The percentage of original mass, shown in Fig. 4, was determined using the following equation: Where the initial droplet mass and droplet mass at time t were m 0 and m t , respec tively.

Short-term inactivation kinetics
10 × 1 µL droplets composed of H1N1pdm09 diluted 1:10 in human saliva or airway surface liquid were deposited on six-well cell culture plates.Technical triplicates were carried out for each of the five independent replicates.Droplets were collected at 0, 10,15,20,25,30,35,40,50, and 60 min following exposure to 50% RH in the environmental chamber.RH and temperature were logged with a Hobo Temperature and Data Logger.One droplet exposure time was conducted at a time, and sample exposure times were randomized.First-order decay was used to fit the data by the following equation:

RNA recovery in droplets after drying
Experiments were conducted before and after droplet drying to ensure droplet recovery was the same regardless of droplet desiccation state.10 × 1 µL droplets containing H1N1pdm09 diluted 1:10 in MEM or respiratory fluid were deposited on six-well tissue culture plates as described above.Droplets were collected following exposure to 50% RH for 0 and 1 hour.Three independent replicates were conducted with each droplet type.Droplets were collected in 500 µL MEM and stored at −80°C until extraction.RNA extractions were carried out with the QIAamp Viral RNA Mini Kit (Qiagen, Cat.No. 52904) following the standard extraction protocol.Extracts were eluted in 60 µL nuclease-free water and stored at −80°C until reverse transciptase quantitative polymerase chain reaction (RT-qPCR) analysis.RT-qPCR targeting the matrix gene [Forward (IAV M25 F): 5′ -AGATGAGTCTTCTAACCGAGGTCG -3′; Reverse (IAV M124mod R): 5′ GCAAAGACACTTTCC AGTCTCTG -3′; Probe (IAV M64 P): 5′ FAM -TCAGGCCCCCTCAAAGCCGA -BHQ -3′; FAM = fluorescein amidite; BHQ = black hole quencher; modified from previously published primers/probe (54)] was performed using the BioRad iTaq Universal Probes One-Step Kit (BioRad, Cat.No. 1725140).In each reaction, primer concentrations were 500 nM, and probe concentrations were 150 nM.Thermocycling was performed on a BioRad CFX Connect Real-Time PCR Detection System (BioRad, Cat.No. 1855201) with the following cycle settings: reverse transcription at 50°C for 10 min, initial denaturation at 95°C for 2 min, and 40 cycles of denaturation and annealing/extension at 95°C for 10 seconds at 60°C for 20 seconds, respectively.In vitro transcribed RNA of the full-length matrix gene was used as the standard.

Bright field microscopy of dried droplet structure
To visualize droplet morphology after drying, 1 µL droplets composed of virus solutions, generated as described above using H1N1pdm09 virus in either airway surface liquid or saliva, were deposited on polystyrene chamber slides (Ibidi, Cat.No. 80826) and exposed to 20%, 50%, or 80% RH for 2 hours at ambient temperature.Droplets were immediately visualized using an inverted microscope (Olympus, Model CKX53) at 10× magnification.

Nanobead visualization in dried droplets
100 nm carboxylate-modified, fluorescent, and polystyrene nanobeads (Invitrogen, Cat.No. F8801) were sonicated for 10 min and subsequently spiked into airway surface liquid or saliva at a final 1:1,000 dilution.1 µL droplets from these solutions were then generated on polystyrene chamber slides.Droplets were immediately placed in the environmental chamber and exposed to 20%, 50%, or 80% RH for 2 hours.The chamber was kept dark during the exposure.Following the 2-hour exposure, fluorescent nanobeads in droplets were visualized using an inverted fluorescence microscope (Olympus, Model IX73P2F) at 10× magnification.Saliva and airway surface liquid droplets without nanobeads were used as negative controls.

Statistical analyses
All statistical analyses were conducted in Prism Version 9.2.0.

FIG 1
FIG 1Infectious influenza A virus levels in the nasal (red) and oral (blue) cavities of ferrets following infection.Ferrets were infected via intranasal inoculation with (A) H1N1pdm09 or (B) H3N2.Sampling of the nasal cavity was conducted by nasal washes, which were collected in 1 mL volumes, while oral cavity samples were collected using swabs.Individual replicates (n = 8) are included for each box and whisker plot.Dashed lines represent the limit of detection for nasal wash samples (NW) and oral swab samples (OS).

FIG 2
FIG 2 Reductions in infectious influenza A virus (i.e., infectious virus decay) in droplets composed of saliva or airway surface liquid.H1N1pdm09 or H3N2 log10 decay at 20%, 50%, or 80% RH after 1 and 2 hours of exposure.Open circles indicate infectious virus decay beyond detection limits.Three technical replicates were conducted for three independent replicates, and the mean decay of three technical replicates is shown for each independent replicate.Unpaired t tests were conducted to establish when influenza virus decay was significantly different between respiratory fluid droplets.* =P < .05;** =P < .005.

FIG 3
FIG 3 Differences in virus decay are not due to poor recovery of viral material.Viral RNA levels recovered in saliva and airway surface liquid droplets exposed to 50% RH for 0 or 1 hour (after drying) (A), infectious virus levels at 0 and 1 hour of drying (B), and the gene copy to infectious virus ratio at 0 and 1 hour of drying (C).Three independent replicates are shown for each droplet composition type, and the mean is indicated with a black line.

FIG 4
FIG 4 Change in droplet mass over time.Saliva or airway surface liquid droplets containing H1N1pdm09 were exposed to 20%, 50%, or 80% RH for 2 hours.Shaded regions designate the drying time.Two independent replicates are shown for each RH condition and droplet composition.

FIG 5
FIG 5 Inactivation kinetics of H1N1pdm09 at 50% RH in 1 µL droplets.Droplets were composed of (A) human saliva or (B) airway surface liquid.Three technical replicates were conducted for each independent replicate, and the mean and SD of five independent replicates are shown.Simple linear regression analysis was used to generate inactivation curves.Breakpoints in inactivation curves were established by average droplet drying time from two independent replicates (Table 2) and are shown as dotted vertical lines.

FIG 6
FIG 6 Bright field images of droplets composed of human saliva or airway surface liquid containing H1N1pdm09 virus after exposure to 20%, 50%, and 80% RH for 2 hours.Images were taken at 10× magnification.Scale bars, 500 µm.One representative image is shown for each experimental condition.

FIG 7
FIG 7 Fluorescence images of droplets composed of human saliva or airway surface liquid with 100 nm fluorescent nanobeads (1:1,000 dilution) after drying at 20%, 50%, or 80% RH for 2 hours.Images were taken at 10× magnification.RFP = red fluorescence protein.Scale bars, 200 µm.One representative image from duplicate droplets is shown for each experimental condition.
where N 0 and N t are the influenza virus concentrations in droplets at time 0 and time t, respectively, and k represents the inactivation rate constant.

TABLE 1
Bulk properties of distinct respiratory fluids, phosphate-buffered saline, and milliQ water a a Mean ± SD of three replicate measurements is shown.Airway surface liquid measurements were conducted using airway surface liquid from three different patients to capture patient-to-patient variability.bBCA protein analysis resulted in a negative value based on the standard curve.Total protein reported as 0 mg/mL.Full-Length Text Applied and Environmental Microbiology February 2024 Volume 90 Issue 2 10.1128/aem.02010-236

TABLE 2
Evaporation rates and drying times of respiratory fluid droplets at variable RH