Assessment of a membrane filter coated with hygroscopic glycerol for improved recovery of airborne viable bacteriophage MS2

Abstract

The viability of viruses collected on membrane filters during air sampling is compromised by desiccation. This leads to an underestimation of the concentration of viable viruses in the air. We investigated whether adding a hygroscopic glycerol coating to membrane filters could enhance the recovery of viable bacteriophage MS2 collected through air sampling. Mixed cellulose ester (MCE) filters were coated with glycerol and compared with as-packaged filters (APF) as a control to determine the impact of glycerol on MS2 viability. First, MS2 viability upon direct deposition of MS2 suspension on APF vs. glycerol-coated filters (GCF) was assessed at benchtop. On average 121% more viable MS2 were recovered from GCF than APF (p = 0.031). MS2 was then aerosolized and collected onto APF, GCF, phosphate-buffered saline (PBS)-wetted filters (PWF), and gelatin filters (GF). All filter types collected statistically similar (p > 0.05) amounts of aerosolized virus at 50% RH according to RT-qPCR analysis. Viable virus recovery from GCF was significantly lower than from APF (p = 0.011) at 50% RH but significantly higher (p = 0.0004) at 80% RH. APF and GCF preconditioned for 30 minutes at 80% RH were also assessed for MS2 aerosol sampling. Results showed higher viable virus recovery from GCF than either APF (p = 0.021) or GF (p = 0.006). High RH in aerosols facilitated viability conservation of MS2 on GCF, leading to improved recovery of viable virus from membrane filters.

1. INTRODUCTION

Respiratory viruses cause illnesses such as coronavirus disease-2019 (COVID-19) (Chan et al., 2020; Zhang et al., 2020), common colds, influenza, and measles (Van Doorn & Yu, 2020). Such illnesses are common and significantly impact health, quality of life, and economic well-being. The desire for improved protective health measures and increased awareness of health risks associated with exposure to respiratory viruses have spurred numerous studies on the collection and measurement of airborne viruses (Gollakota et al., 2021; Mainelis, 2020). Nevertheless, knowledge gaps remain, particularly as they pertain to the transmission mechanisms of viral respiratory illnesses and their impact on public health. To effectively address these knowledge gaps, researchers must employ air samplers that can conserve as many viable viruses as possible during sampling (Santarpia et al., 2024). Low concentrations of airborne viruses in ambient air inherently make virus collection and detection challenging (Wang et al. 2021), and the need to conserve viable virus compounds the difficulty.

Physical collection efficiency (PCE) is a common metric for evaluating an air sampler’s performance, which involves the comparison of the quantity of particles entering the sampler to the quantity of particles collected in the sampled air (Hogan et al., 2005). Collecting viable virus necessitates consideration of biological collection efficiency (BCE), which is the proportion of viable microorganisms to non-viable microorganisms collected on a filter to that present in the air (Hogan et al., 2005; Kulkarni et al., 2011). Influenced by PCE and factors like mechanical stress, desiccation, and environmental conditions, which can reduce viability (Thompson et al., 1994; Verreault et al., 2008), BCE is effectively PCE adjusted for viability conservation (Hinds, 2022; Kulkarni et al., 2011). Collecting detectable amounts of virus at low concentrations usually requires long sampling times; unfortunately, desiccation and impaction during sampling can cause inactivation of viruses and directly affect the detectability of viruses in air samples (Verreault et al., 2008). Virus inactivation may also occur if semi-solid collection media dissolves (e.g. gelatin filters) or evaporates (e.g. liquid in impingers) (Reponen et al., 2011). Improvement of sampling methods to conserve a high percentage of viable viruses that are captured during air sampling, have been the focus of recent studies (Pan et al., 2019; Ratnesar-Shumate et al., 2021; Vass et al., 2024).

Filtration is one of the most frequently used sampling methods for the collection of virus-containing airborne particles with an aerodynamic size < 1 μm (Li et al., 2009). Many filter materials have been used to collect airborne viruses, including cellulose, glass fiber, nanofiber, gelatin, polycarbonate, and polytetrafluoroethylene (Verreault et al., 2008). PCE as high as 96% have been reported for both PTFE and gelatin filters during the sampling of MS2 (Burton et al., 2007). PCE of 100% for all particle sizes was reported for glass fiber filters (Li et al., 2018). In comparison, BCE has been reported to have a much lower value. When compared to the Biosampler^® as a standard reference for collecting influenza virus, BCE were reported to be only 32.3%, 22%, and 10% for glass fiber (Li et al., 2009), Teflon, and gelatin filters, respectively (Fabian et al., 2009). The BCE for viruses greatly depends on filter characteristics, virus morphology, hydrophilicity of viruses and filter material, and RH (Tseng & Li, 2005). Accurate measurements of viable viruses are important because they are vital for determining the viable concentration of viruses in the air; such information is essential for estimating exposure and health risks from airborne viruses. A low BCE results in an underestimation of health risks.

Viruses are prone to inactivation due to desiccation during filter sampling and damage during sample processing (Cox, 1933; Verreault et al., 2008). Gelatin filters are frequently used because of their hygroscopic nature, which helps conserve virus viability during sampling. Further, they can be dissolved in extraction media, thereby reducing forces imparted onto viruses to disintegrate their morphology during sample extraction, ultimately improving recovery of viable virus from the filter (Pan et al., 2019). Zhao et al. (2014) reported that the gelatin filter was 100% efficient for total collection of infectious bursal disease viruses. Tseng & Li. (2005) reported that gelatin filters were 10 times better at conserving viability of hydrophilic viruses than Nucleopore polycarbonate filters. However, hydrophobic viruses such as Influenza A Virus (IAV), are less resilient. For example, Fabian et al. (2009) could only recover 23% of viable Influenza A virus (IAV) with gelatin filters. While the ability to dissolve gelatin filters is beneficial to extracting more viable viruses, gelatin filters have other problems. They dry out at low RH and dissolve when the RH is high or when contacting water droplets (Burton et al., 2007). The recommended temperature and RH range for gelatin filters to get high collection efficiency without disintegrating are 22–30 °C and ≤ 60%, respectively (Jaschhof, 1992). Accordingly, most commercially available filters are only suitable for short samplings (< 15 min), as extended sampling duration leads to desiccation and subsequent virus inactivation. (Burton et al., 2005; Li et al., 1999). New sampling techniques are needed to improve the conservation of virus viability during air sampling.

The application of coatings to filter materials has been a growing area of research in recent years (Shah et al., 2022). Coated membranes, in which the pores remain open, may offer improved collection efficiency and conservation of collected microbes, making them suitable for bioaerosol sampling and other similar applications (Regan et al., 2022). Glycerol may be one suitable coating option. It is a colorless, odorless, non-volatile and viscous liquid with properties proper for coating filters (Basiak et al., 2018). Its hygroscopic nature attracts and retains moisture on the coated surface (Koutrakis et al., 1993) and can help conserve virus viability. When glycerol coats the fibers of mixed cellulose ester (MCE) filters, it adheres to the filter surface through physical interactions such as Van der Waals forces, capillary action, and hydrogen bonding, effectively trapping surrounding moisture on the filter surface during filter sampling (Rohaeti et al., 2018). Koutrakis et al. (1993) added glycerol for increasing water molecules over the surface of the filter in their study of ambient ozone measurement with a nitrite-coated filter. Coupland et al. (2000) used glycerol as a plasticizer for moisture content enhancement over whey protein edible films. Other coating materials exist, such as polyethylene glycol (PEG) (Chen et al., 1999), polyvinyl alcohol (PVA) (Wang et al., 2020), and bovine serum albumin (BSA) (Jalvo et al., 2021), but they possess poor water-holding properties compared to glycerol. Further, they may have stronger antimicrobial properties than glycerol. Abd El-Mohdy & Aly (2023) reported significant antimicrobial effectiveness of a phosphorous-polyvinyl alcohol (P-PVA) polymer against bacteria and fungi. Nalawade et al. (2015) reported strong antimicrobial properties of PEG 1000 against S. mutans and E. coli at only 25% strength. Jalvo et al. (2021) reported enhanced antimicrobial properties of a non-woven cellulose fabric coated with BSA. Glycerol, on the other hand, is less toxic than other coating materials and, therefore, suitable for bioaerosol sampling applications (Qi et al., 2015). Furthermore, it has been used with various filter materials for different purposes, such as purification of biopolymers with cellulose filters (Yang et al. (2002), collection of acidic aerosols and gases with Teflon filters (Koutrakis et al., 1988), and measurement of atmospheric particulate matter with quartz, nucleopore and PTFE filters (Finn et al., 2001).

We hypothesized that applying glycerol coating to a membrane filter would enhance the recovery of viable hydrophilic viruses compared to a non-coated filter. Therefore, we conducted experiments using bacteriophage MS2 as a testing agent to assess the effectiveness of glycerol-coated filters under different environmental conditions.

2. MATERIALS AND METHODS

2.1. Preparation of Bacterial and Bacteriophage Stocks

Escherichia coli (E. coli) bacteriophage MS2 (Cat. #15597-B1, ATCC, Manassas, VA), was the biological agent used in this study. MS2 was selected for its well-established role as a safe (Aranha-Creado & Brandwein, 1999) and non-pathogenic proxy for human viruses (Bae & Schwab, 2008; Dawson et al., 2005; Sassi, 2016) in environmental virology studies (Dawson et al., 2005; Shaffer et al., 2023). MS2 has a diameter of around 28 nm and has a single-stranded RNA genome(Golmohammadi et al., 1993). In water, MS2 is electronegative and hydrophilic (Shi & Tarabara, 2018). Bacteriophage MS2 and host E. coli were prepared following the protocol provided by ATCC. The detailed procedure can be found in the supplementary material.

2.2. Test filters and filters preparation

Two types of membrane filters were used in this study: MCE (0.8 μm pore size, 37 mm diameter, 150 μm thickness, Whatman, USA) and gelatin (3 μm pore size, 37 mm diameter, Sartorius, USA). MCE filters are thin, porous membranes made of cellulose acetate and cellulose nitrate blend. They are hydrophilic and suitable for coating with glycerol because glycerol, a polar molecule with multiple hydroxyls groups, exhibits strong affinity for hydrophilic surfaces due to hydrogen bonding and polar interactions (Gentleman & Ruud, 2010; Gu & Jérôme, 2010). The small pore size and thin membrane allow for effective collection of microorganisms on MCE filters vs. other filters (McKinnon & Avis, 1993). Membrane filters are integral in many methods used for evaluating the type and quantity of microbes in diverse matrices. For example, they are used in many approaches utilized for environmental monitoring, indoor air quality assessments, and occupational health studies (Chattopadhyay et al., 2015). Since different types of filters are available, it is important to validate their suitability for a particular application. For example, gelatin filters (GF) were used as a reference for MCE filter performance because of their characteristically high biological collection efficiency (Tseng & Li, 2005). Both filter materials were used in standard 37-mm, 3-piece filter cassette samplers (polypropylene conductive cassette, 3 sections, 37 mm, SKC Inc., USA) at a flow rate of 3.5 L min^-1.

MCE filters were tested in three configurations: as-packaged filter (APF), glycerol-coated filter (GCF), and PBS-wetted filter (PWF). Glycerol was chosen because its strong hygroscopicity can help retain moisture on the filter surface. We aimed to leverage the water-holding property of glycerol to enhance viable virus conservation while avoiding excessive virus inactivation due to its antimicrobial characteristics. Glycerol is a trihydric alcohol (C₃H₈O₃) and a weak antimicrobial agent (Welch et al., 2020), which can affect the growth of bacteria, fungi, and enveloped viruses indirectly when used at a concentration > 65% (Nalawade et al., 2015). Though MS2 is not an enveloped virus, the effect of the coating on MS2 viability was assessed.

To prepare GCF, APF were soaked for 30 minutes in a solution containing 10% glycerol (Difco^™ Glycerol BD 228210, Fisher Scientific Inc., USA) in autoclaved ultrafiltered water prepared with a Nanopure^® water purification system (Barnstead GenPure UF/UV Ultrapure Water System, Thermo Scientific, USA) inside a biosafety cabinet (BSC). The filters were then removed from the solution, placed on an aluminum foil sheet, and baked at 100 °C for 30 minutes to affix the glycerol to the filter (Fu et al., 2022; Kovtun et al., 2024). As the vapor pressure of glycerol at 100 °C (0.025 mmHg) is far lower than water (760 mmHg) (Pure Component Properties, 2007), water evaporated while glycerol largely remained on the filters. Coated filters were then transferred back to the BSC and placed in a Petri dish for further use. Filters were handled under strict and consistent conditions, including within the BSC and inside sterile petri dishes to avoid contamination during transfer between BSC and oven. Additionally, handling was performed using sterile tools to avoid direct contact, and all steps were conducted under aseptic conditions to prevent contamination. Filters were also randomized during experimental setups to avoid positional or procedural biases, and multiple replicates were included to ensure the reliability of results. The instrumentation used for coating and testing was calibrated, and identical parameters were applied to all filters to maintain uniformity in experimental conditions. These rigorous controls ensure that the results accurately reflect the impact of glycerol coating on virus viability, independent of handling effects.

The weights of the APF and GCF, taken by a microbalance (Cahn C-31 Microbalance, model 10931–01E, Serial 71997, 100/120/220/240 Volts) were 42 ± 1 mg and 68 ± 1 mg, respectively. The weight of the coating was calculated as 0.6 ± 0.02 mg glycerol per mg APF. PWF was freshly prepared before the aerosol exposure by pipetting 200 μL of 0.1× PBS onto an APF and leaving the filter in the BSC for 5 minutes to absorb the liquid. PWF was introduced to investigate whether simply wetting an MCE filter can enhance viable virus recovery.

2.3. Testing setup and methodology

Experiments were performed in two phases. In Phase 1, whether glycerol detrimentally impacted MS2 viability was verified. APF and GCF were assessed inside a BSC through benchtop experiments. 200 μL of diluted MS2 suspension was spiked onto GCF and APF, which were then left inside the BSC for 30 minutes. Subsequently, each filter was soaked in 10 mL of extraction media (0.1× PBS) for 10 minutes and vortexed for 1 minute to extract the viruses from the filter. Suspensions produced were used as virus stocks for plating and culturing.

In Phase 2, the relative biological collection efficiencies of all four filter types were tested with aerosolized MS2: GCF, APF, PWF, and GF. Figure 1 shows a diagram of the experimental setup used for aerosol collection at the two RH conditions assessed: 50% and 80%. Solid lines represent setup 1 (50% RH). Dashed lines represent setup 2 (80% RH). A six-jet Collison nebulizer (Model CN25, BGI Inc., Waltham, MA) was used to generate bioaerosols. Aerosols were diluted with dry air (Breathing Grade Air, Airgas Inc., USA) and routed through a dilution dryer. Temperature and RH were measured after the aerosol exited the dilution dryer (Extech CO260 Indoor Air Quality Meter, S/N: 10458346, China). The aerosol was divided into four streams for parallel collection through all filter types. A HEPA filter was used to regulate supply air in excess of what was drawn by the filter pumps. Separately, a water-based condensation particle counter (MAGIC 200/210, Aerosol Dynamics, Berkeley, USA) was connected to the inlet and outlet of each filter cassette to measure upstream and downstream particle number concentrations and pressure drop across filters. Particle counters have either a constant flow or constant pressure designs. The Spider-MAGIC^™ uses a constant flow design to facilitate pressure drop measurements (AMETEK Spectro Scientific, 2019).

Figure 1:

The schematic diagram of Setup 1 is illustrated by solid straight-line connections, used for system validation using four APFs (F1, F2, F3, F4) and for MS2 aerosol sampling at 50% RH using four filter types (APF, GCF, PWF, GF). The dotted line connections show components added for Setup 2 (80% RH) using three filter types (APF, GCF, GF) as PWF was not used in this stage because of heavy pressure drop. Spider-MAGIC^™ was used for the measurement of physical particle size distribution and air pressures across the filters.

Phase 2 was completed in three stages. First, the setup was validated to ensure the similarity of airflow and operation of the sampler pumps connected to the four filters. For setup validation, aerosols containing MS2 were generated at an airflow rate of 3.5 L min^-1. Dry air at 11 L min⁻¹ was added to the dilution dryer to decrease the RH and aerosol concentration. The aerosol was then pulled through each filter by a pump (Aircheck Sampler, Model: 224-PCXR4, Universal PCXR4 Sample Pump, SKC, USA) operated at 3.5 L min^-1. Filter cassettes and sampling pumps were alternated among different sampling ports between trials to reduce any residual bias. Six trials with a 30-minute sampling duration were conducted. The temperature and RH during the trials were 25 ± 1 °C and 50%, respectively.

Stage 2 of Phase 2 involved five trials for MS2 aerosol collection using all filter types at 50% RH (Figure 1). Temperature ranged between 24–26 °C during this stage. Stage 3 was the repetition of Stage 2 with slight modifications in the setup (Figure 1) to increase the RH to 80%. These included operating the Collison nebulizer at 6 L min⁻¹ and adding water vapor by passing 12 L min⁻¹ dry air through a diffuser placed in a beaker of DI water heated by a hot plate (60 °C). The temperature of the resultant aerosol ranged between 23–25 °C. Stage 3 experiments were conducted using only three filter types (APF, GCF, and GF). PWF was not used at this stage because wetted filters choked the airflow in Stage 2 and created operational problems for sampling pumps.

To investigate whether preconditioning affected bacteriophage MS2 recovery with GCF, two additional sets of experiments were conducted with three trial runs each. One set was aerosol sampling at 50% RH, where both APF and GCF were preconditioned at 80% RH for 30 minutes before sampling. The second set involved sampling at 80% RH, where only APF was preconditioned.

2.4. Quantification of sampled virus

Viruses collected on filters during aerosol sampling were extracted from filters following the same procedure as in benchtop experiments. Suspensions existing after virus extractions were cultured to quantify viable virus concentrations using a single-layer bioassay method. The plating for all experiments was done in triplicates (N = 3), and the data from the incubated plates were recorded the next day for infectious count. Total virus counts were determined by RT-qPCR. Only the samples from stage 2 experiments of Phase 2 of the study were also analyzed by RT-qPCR to check whether total genomic equivalence was different for filters in different configurations. Further elaboration on the quantification methods is included in the supplementary material.

2.5. Pressure drops and filter quality assessment

High aerosol collection efficiency in filters can yield high-pressure drops. An important goal in filter development is how to boost efficiency while minimizing pressure drop and penetration (Li et al., 2009). The data for upstream and downstream particle concentrations at 0.3 μm, the standard size for testing the filters (Hinds, 2022), and pressures across the filters were used for determining filter penetration, p, and pressure drop, ΔP, by using Equations 1 and 2, respectively.

$$p=\frac{\#/c{m}^3(Downstream)}{\#/c{m}^3(Upstream)}$$ (1)

$$\mathrm{\Delta }P=P_{Up}-P_{Down}$$ (2)

where # is the number of particles and $P_{up}$ and $P_{down}$ are the upstream and downstream pressures of the filter, respectively. The filter quality $\left(qF\right)$ for all filter types was then calculated using Equation 3 (Hinds, 2022).

$$q_F=\frac{\ln \left(\frac{1}{p}\right)}{\mathrm{\Delta }P}$$ (3)

2.6. Statistical analysis

Data tabulation and unit conversions were conducted in Microsoft Excel. Data were sorted according to “Filter Type”. Sorted data were then saved to a .csv file and loaded into R (Version 4.2) for plotting and statistical analysis using base-10 logarithm. Kruskal-Wallis tests were performed to determine if there were statistically significant differences between the medians of groups. Wilcoxon rank sum tests were used to assess pairwise variances between groups. The two-sided pairwise Wilcoxon signed-rank tests were used to describe differences within groups. Results were considered significant if the p-value < 0.05.

3. RESULTS

Benchtop experiments of Phase 1 (Figure 2) showed 121% more viable MS2 recovery with GCF than APF (p = 0.031 by Kruskal-Wallis Test), illustrating that a coating of 10% glycerol solution was not detrimental to the recovery of viable MS2. This is because the surface of the filter had a saturated (100% RH) condition due to liquid coverage over filter surface. Therefore, the study was continued with coating of 10% glycerol solution for the rest of the experiments.

Figure 2:

Viable MS2 recovered from APF and GCF during benchtop experiments (n = 4). The mean is represented by a triangle, while the median is the middle horizontal line in the boxplots. Top asterisk shows significant difference (p < 0.05) between results of APF and GCF.

Figure S1 shows the results of Stage 1 of Phase 2 experiments which were for the experimental setup validation. Kruskal-Wallis significance test validated that there was no significant difference (p = 0.98) overall in viable virus recovery among the group of four APFs. No significant difference within the groups was found by pairwise comparison using Wilcoxon rank sum exact test with a continuity correction (Table 1). The statistically similar viable MS2 collection validated the experimental setup as unbiased toward any sampling port or pump during the aerosol collection.

TABLE 1.

Significance of the results from different sample types based on experimental conditions.

Significance Test Type	Sample Type	Comparison	P-Values
Two-sided Pairwise Wilcoxon Signed rank Test	Viable MS2 Samples for Experimental Setup Validation	F1-F2	1
		F1-F3	0.5895
		F1-F4	0.3125
		F2-F3	0.4375
		F2-F4	0.5625
		F3-F4	0.5625
Two-sided Pairwise Wilcoxon Signed rank Test	PCR Samples of MS2 aerosol Sampling at 50% RH	GCF-APF	0.25
		GCF-PWF	0.25
		GCF-GF	0.25
		APF-PWF	0.5
		APF-GF	0.25
		GF-PWF	1
Wilcoxon Rank sum Test	Viable MS2 Samples At 50% RH	GCF-GF	0.0666
		GCF-PWF	0.0667
		GCF-APF	0.0116*
Wilcoxon Rank sum Test	Viable MS2 Samples At 80% RH	GCF-APF	0.0039*
		GCF-GF	0.0039*
		GF-APF	0.0117*

^*Indicates significant results according to the Wilcoxon models, * p ≤ 0.05.

Results of MS2 aerosol samples collected at 50% RH analyzed for total (GE/L of Air) and viable (PFU/L of air) viruses are shown in Figure 3a and Figure 3b, respectively. Figure 3a illustrates that all four filter types had a considerable number of total viruses, and no significant difference was observed for total viruses detected among all filter types (p = 0.05723) by Kruskal-Wallis Test. A two-sided pairwise comparison among groups using the Wilcoxon signed rank test, as shown in Table 1, also illustrated no significant difference (p = 0.25) in total virus collection by other filter types compared to GCF (Figure 3a). Analyses of viable virus recovery data revealed worse viability conservation with GCF during air sampling at 50% RH (Figure 3b). Recoveries with GCF were lower than PWF (p = 0.066) and GF (p = 0.067) and significantly lower than APF (p = 0.011), according to the pairwise comparison by the Wilcoxon rank sum test. PWF showed some viable collection in the above experiments; however, a very large pressure drop due to airflow blockages was observed during the sampling. Thus, PWF filters were not used in further trials.

Fig. 3.

Results from MS2 aerosol collection at 50% RH using four filter types (APF, GCF, PWF, GF): (a) total genomic equivalence collection, and (b) viable MS2 recovery from the filters. The mean is denoted by a solid triangle, while the median is represented by a horizontal line in the boxplots. “ns” in (a) indicates no significant differences, and asterisks in (b) indicate significant differences (p < 0.05) between groups.

Figure 4 shows the results for MS2 aerosol samples collected at 80% RH, analyzed for viable concentrations (PFU/L of air). Viable virus recovery from GCF was higher than that from both APF and GF (p = 0.00004), according to Kruskal-Wallis significance analysis. The two-sided pairwise Wilcoxon signed rank test showed significant difference among GCF, GF, and APF (Table 1).

Fig. 4.

Viable MS2 collection by three filter types (APF, GCF, & GF) during MS2 aerosol sampling at 80% RH. No filters were preconditioned with moisture. The mean is denoted by a solid triangle, while the median is represented by a horizontal line in the boxplots. Each red dot represents each measurement. Significance differences (p < 0.05) in viable virus recovery from other filter types compared to GCF have been marked on top of the figure by asterisks.

The results of experiments to evaluate the role of filter pre-conditioning are shown in Figures 5a and 5b. More viable viruses were recovered from GCF at 80% RH compared to APF (p = 0.021) and GF (p = 0.006), even without preconditioning (Figure 5a). When both APF and GCF were preconditioned at 80% RH and then used to collect MS2 at 50% RH, significantly less viable virus was recovered from GCF than from GF (p = 0.0415) and APF (p = 0.0365) (Figure 5b). The results suggest that glycerol coating can improve viable virus recovery from MCE filters in high (80%) RH environments. The results also indicate that high RH in aerosols plays a more critical role in conserving virus viability by GCF during air sampling than preconditioning with moisture.

Fig. 5.

Comparison of viable MS2 collection by three filter types (APF, GCF, GF) during MS2 aerosol sampling to determine the key factor behind high viable virus recovery: (a) results at 80% RH, when only APF was conditioned (b) results at 50% RH when both GCF and APF were conditioned. Conditioning was done at 80% RH for 30 minutes before sampling. The mean is denoted by a solid triangle, while the median is represented by a horizontal line in the boxplots. P-values for a significant difference (p < 0.05) among the groups are marked on top of the figure by asterisks.

Pressure drops and penetrations for all filter types are shown in Table 2. The maximum pressure drop (17.8 kPa) was observed with PWF, and the minimum pressure drop (0.2 kPa) was reported for GF. The pressure drop for GCF was twice that of APF. The filter penetration value for GF (1.05E-03) was the highest and GCF (2.03E-07) was the lowest. The filter quality (q_F) follows the order of GF > GCF > APF > PWF, where q_F for GF was 1.3 times GCF and 1.5 APF.

TABLE 2.

Pressure drops and filter quality assessment for filter types during aerosol sampling. Aerosolized PBS was used for this assessment at 80% RH. Measurements were made by a Spider-MAGIC^® particle mobility spectrometer.

Filter Types	Average Upstream Concentration (#/L of air)	Average Downstream Concentration (#/L of air)	Filter Penetration (Downstream/Upstream)	ΔP (kPa)	qF (kPa−1)
GCF	1.40E+06	2.84E-01	2.03E-07	0.6	25.7
APF	1.48E+06	1.37E+03	9.22E-04	0.3	23.3
GF	1.44E+06	1.49E+03	1.05E-03	0.2	34.4
PWF	1.70E+06	7.81E-01	4.60E-07	17.8	0.82

To assess GCF stability in longer sampling durations, three additional trials were conducted with sampling durations of 3 and 6 hours under the same environmental conditions used in the original experiments but without MS2. These trials revealed no visible structural disintegration or morphological changes in the filters, indicating that the glycerol coating maintained its integrity throughout these extended sampling periods.

4. DISCUSSION

Accurate measurement of infectious airborne viruses improves health risk assessments and the development of effective public health recommendations (Atkinson & Wein, 2008). It enables authorities to gauge the potential of exposures to viruses, evaluate the risk of transmission (Tang et al., 2011), set regulatory standards, develop effective control strategies, and allocate funds for prevention and treatments (Cole & Cook, 1998). Membrane filters can effectively collect particles < 1 μm (Hou et al., 2020) and thereby inform environmental monitoring. Desiccation and inactivation of viruses collected on filters during air sampling, however, hampers the use of these accessible tools for viable virus collection (Verreault et al., 2008), making filters less useful for accurate assessment of the presence of airborne viable virus. This study showed that the application of glycerol coating to MCE filters could potentially improve the conservation of viable viruses collected from aerosols at high relative humidity.

The observed discrepancy between benchtop and aerosol sampling results for MS2 viability on GCF at varying humidity levels highlights the impact of environmental RH on the glycerol coating’s hygroscopic behavior. There was a saturated condition on the filter top due to the deposition of virus suspension during benchtop experiments. Under that saturated condition, virus viability on GCF was conserved significantly higher than APF, as the coating maintained a sufficiently hydrated environment, allowing viruses to retain viability. This high recovery rate under saturated humidity conditions validated the glycerol coating’s efficacy in retaining viable viruses in benchtop applications and justified its further use in aerosol testing.

However, at 50% relative humidity (RH), the hygroscopic nature of the GCF caused it to absorb moisture from the virus particles, leading to desiccation and inactivation of the viruses during sampling. As a result, the GCF performed poorly compared to the APF in retaining viable viruses under low humidity, where the glycerol coating acted as a desiccant. Conversely, at 80% RH, sufficient environmental moisture allowed the GCF to supply moisture to filter surface, thus conserving virus viability more effectively than the APF and other filters. This underscores the coating’s dependence on environmental moisture for optimal performance. Moisture content remaining on GCF after sampling for 30 minutes at 50% RH was 0.05% by a weight-to-weight comparison of post-sampling (wet) coated filters vs. post-baking (dry) filter, while at 80% RH, it was 4% without preconditioning and 6% with preconditioning. The difference in residual moisture between RH levels helps explain the results of this study, i.e., GCF can conserve virus viability at high (80%) RH conditions.

While potential exists for the dissolution of glycerol coating during sampling under high RH conditions due to its high solubility, such a process was unlikely in this study. Strong hydrogen bonds are formed between glycerol hydroxyl groups and MCE filter fibers (Gonçalves et al., 2019). Those bonds likely held the coating intact until the virus extraction process. We did not quantify the amount of residual glycerol on the filters, but we observed from coated and non-coated filter results that any degradation of the coating from absorbed moisture had a negligible effect on the collection and quantification of viable viruses.

Other studies have discussed the low survivability of MS2 at 20% < RH < 80% (Lin et al., 2020; Lin & Marr, 2019). We report here a similar trend, whereby we recovered less viable viruses at 50% RH (Figure 3b, Figure 5b) than we did at 80% RH (Figure 5a). We did see remarkable differences between filter types at 50% RH (Figure 3b), where GCF performed poorly compared to the other filters. These disparities are due to negligible (0.05%) moisture available at the filter surface at 50% RH. Low moisture content likely accelerated moisture absorption from collected aerosols, desiccating viruses, and reducing biological collection efficiency. This may explain the difference between total and viable virus recoveries (Figure 3). All filter types collected similar amounts of PCR-detectable virus (Figure 3a), but significantly less viable virus could be measured using GCF (Figure 3b). It is well-known that desiccation occurring during air sampling with filters (Verreault et al., 2008) can dehydrate viruses (Pan et al., 2019). This work also showed a negative effect at 50% RH caused by coated filters with a hygroscopic material like glycerol.

Our results showed that only high RH within the aerosol environment significantly improved GCF’s performance in conserving virus viability, while any minor ambient moisture absorbed during moisture conditioning and handling had a negligible impact on viability outcomes at lower RH. Studies confirm that glycerol-treated materials enhance moisture retention on surfaces under high humidity, suggesting that handling and preconditioning have minimal effect on glycerol’s hygroscopic properties (Ocampo et al., 2023; Rathinamoorthy & Kiruba, 2022). Our results align with these findings.

The selection of 50% and 80% RH, rather than incremental RH levels (e.g., 60%, 65%, 70%, or 75%), allows for a meaningful understanding of GCF performance in contrasting conditions: one where the glycerol coating inactivates the virus by absorbing moisture and the other where the high RH environment supports virus viability, allowing the coating to function effectively. While indoor air is often maintained at 50% RH, outdoor RH is often above 70% in subtropical region where this study was conducted (Florida), especially in areas like Miami. Our chosen RH levels are, therefore, representative and relevant (Yang et al., 2012). As GCF’s performance at 80% RH had already surpassed that of APF, further testing above 80% was deemed redundant. MS2’s viability with GCF was expected to be zero below 50% RH, as observed in the initial experiment, and testing below this value would not bring values to the comparison of GCF to other filters.

PWF collected some measurable viruses, but PWF proved unsuitable for our experimental design due to large pressure drops (17.8 kPa) that surpassed our pump’s operating limits, resulting in drastically reduced airflow. To determine whether it was the PBS or the liquid film that affected the pressure drop, tests were conducted by wetting the MCE filters with DI water (200 μL, 100 μL, and 50 μL) and checking the airflow through the filters. Airflow was restricted through the filter with the addition of as little as 100 μL. 50 μL was not enough to wet the total surface of the filter. The angle that the liquid of interest forms on a certain solid surface is known as contact angle, which determines the wettability of solid surfaces (Alghunaim et al., 2016). Van Oss (2006) reported that cellulose ester filters have a contact angle of 24°−54° with water. Ding et al. (2006) reported the contact angle between cellulose acetate and water droplet as 62°. Since it is < 90º, water is sucked into pores rather than being repelled by them (Białopiotrowicz & Jańczuk, 2002; Dwyer, 1986). When MCE fibers absorb water, they swell, thus restricting airflow (Kim et al., 2019; Yeganeh et al., 2023). Taken together, we hold that it is imprudent to moisten an MCE filter by applying water or PBS to it before aerosol sampling.

During the experiments at 80% RH, slight condensation formed in our dilution dryer. An improvement over our design may be to insulate the dilution dryer to prevent cooling of the airstream during aerosol distribution. Similarly, it might be good to warm influent dry air to prevent the same condensation effect as aerosol and dry airstreams mix.

GCF had the best filter quality at 80% RH. Viable virus collection with GCF was higher than with GF or APF, both with (Figure 5a) and without (Figure 4) preconditioning. Our results showed that a high RH during sampling was more critical to the conservation of viable MS2 with GCF than conditioning the filter with moisture before aerosol exposure. Interestingly, limited results here may suggest that any antimicrobial or disinfectant effect of glycerol may be counterbalanced by its prevention of desiccation on membrane filters in high RH environments, though this study was not determinative in that regard.

GF are known to disintegrate at high temperatures and dissolve in humid environments (Pan et al., 2019; Verreault et al., 2008). In this study, GF were observed somewhat dissolved inside filter cassettes after 30-min samplings at 80% RH, and they often adhered to the filter support pads. This resulted in GF being torn apart upon removal from the filter support pad. It is typically the case that a significant fraction of viable viruses get inactivated during the extraction process from filters (Li et al., 2018; Tseng & Li, 2005), but this disintegration of GF likely worsened MS2 recoverability. This explains why GF performed worse in the last set of experiments as preconditioning and sampling at 80% RH provided enough moisture for its disintegration. Our work suggests that GCF could be used in instances where GF may fail due to humid ambient conditions.

This study has several limitations due to environmental factors and study design constraints. Glycerol coating is effective for high RH but has constraints at lower humidity, suggesting that alternative coating formulations to enhance viability conservation across diverse conditions may be needed. The study was primarily a proof of concept to evaluate glycerol coating’s efficacy in conserving viral viability at elevated RH, using MS2 as a surrogate virus, which, while informative, may not fully represent human respiratory viruses. Future studies should include pathogenic viruses such as Influenza A, human coronavirus OC43, and SARS-CoV-2 to better understand the glycerol coating’s applicability for human health risk assessment. Additionally, the sampling was conducted for 30-minute durations; future experiments with longer sampling durations are needed to assess the impacts on viability over time, aligning with a typical workday or other longer exposure periods appropriate for exposure assessments related to environmental health hazards. Furthermore, RT-qPCR was not performed on the samples at 80% RH. Therefore, no conclusion can be drawn about total GE under this condition. However, the plaque assays alone provided valuable insight into GCF conservation of viable MS2 relative to other filter types. Despite these limitations, this study has yielded valuable information that can help improve the design of sampling plans and the collection of viruses from aerosols.

5. CONCLUSION

The effect of glycerol coating on MCE membrane filters on the conservation of bacteriophage MS2 viability during air sampling was investigated in this study. The coating significantly improved viable MS2 detection compared to non-coated filters in benchtop assessment. Similar PCR-detectable virus concentrations were recovered from all four filter types during aerosol collection at 50% RH, but coated filters (GCF) conserved less viable MS2. However, at 80% RH, GCF conserved significantly more viable viruses collected from aerosols. The results indicated that glycerol coating can improve viable virus recovery from filters in high (80%) RH environments. Work further showed that it was not the preconditioning step that most improved virus recovery with GCF but the RH of the aerosol. This research offers a method for transforming a commonly used membrane filter to improve viable virus conservation during air sampling in environments where existing filters, like gelatin filters, may degrade or dissolve. The adoption of this method may help yield more accurate characterization of viable virus presence within sampled environments. Such knowledge may, in turn aid regulators and organizational leaders in establishing guidelines that can better protect against environmental exposures to airborne viruses.

Data Availability

Data supporting the findings of this study are available from the corresponding author upon reasonable request.