Comparison of Five Bacteriophages as Models for Viral Aerosol Studies

Abstract

Bacteriophages are perceived to be good models for the study of airborne viruses because they are safe to use, some of them display structural features similar to those of human and animal viruses, and they are relatively easy to produce in large quantities. Yet, only a few studies have investigated them as models. It has previously been demonstrated that aerosolization, environmental conditions, and sampling conditions affect viral infectivity, but viral infectivity is virus dependent. Thus, several virus models are likely needed to study their general behavior in aerosols. The aim of this study was to compare the effects of aerosolization and sampling on the infectivity of five tail-less bacteriophages and two pathogenic viruses: MS2 (a single-stranded RNA [ssRNA] phage of the Leviviridae family), Φ6 (a segmented double-stranded RNA [dsRNA] phage of the Cystoviridae family), ΦX174 (a single-stranded DNA [ssDNA] phage of the Microviridae family), PM2 (a double-stranded DNA [dsDNA] phage of the Corticoviridae family), PR772 (a dsDNA phage of the Tectiviridae family), human influenza A virus H1N1 (an ssRNA virus of the Orthomyxoviridae family), and the poultry virus Newcastle disease virus (NDV; an ssRNA virus of the Paramyxoviridae family). Three nebulizers and two nebulization salt buffers (with or without organic fluid) were tested, as were two aerosol sampling devices, a liquid cyclone (SKC BioSampler) and a dry cyclone (National Institute for Occupational Safety and Health two-stage cyclone bioaerosol sampler). The presence of viruses in collected air samples was detected by culture and quantitative PCR (qPCR). Our results showed that these selected five phages behave differently when aerosolized and sampled. RNA phage MS2 and ssDNA phage ΦX174 were the most resistant to aerosolization and sampling. The presence of organic fluid in the nebulization buffer protected phages PR772 and Φ6 throughout the aerosolization and sampling with dry cyclones. In this experimental setup, the behavior of the influenza virus resembled that of phages PR772 and Φ6, while the behavior of NDV was closer to that of phages MS2 and ΦX174. These results provide critical information for the selection of appropriate phage models to mimic the behavior of specific human and animal viruses in aerosols.

INTRODUCTION

The airborne route can transmit several viral diseases. Most of them can also be transmitted through other means, such as direct contact with infected persons. However, the importance of the airborne route compared to that of the other routes is still less understood. While aerosols are now recognized to be important mechanisms of transmission for viruses like influenza virus, the severe acute respiratory syndrome coronavirus, respiratory syncytial virus, porcine corona virus, norovirus, and foot-and-mouth disease virus (1–6), it has been pointed out that more standardized studies are needed to better understand the behavior of these viruses under airborne conditions (7). Indeed, little is known about virus aerosolization and persistence in the aerosol state (7, 8). Moreover, the efficacy of recovery of viruses by air sampling devices still remains to be thoroughly analyzed (9). The field of aerovirology is aimed at, among other things, studying the topics mentioned above (9).

Bacteriophages are believed to represent good surrogates for studies of airborne viruses. Bacterial viruses are safe for laboratory workers, and their study does not require specialized biocontainment precautions. They are relatively easy to produce in large quantities, and several purification procedures are available (10). Phages are highly diversified from a genetic and morphological standpoint, thereby providing a large pool of viruses to choose from (11). Interestingly, some phages also display structural features similar to those of eukaryotic viruses (12). Over the years, specific phages have been studied and then used as surrogates for eukaryotic viruses. For example, the U.S. Food and Drug Administration (FDA) has recognized the utilization of phage PR772 as a virus model to test filtration systems in the biopharmaceutical industry (13).

Tailed phages with a double-stranded DNA (dsDNA) genome (Caudovirales order) are by far the most studied among bacterial viruses and are used in a wide range of fields, including in aerosol studies (9). However, since eukaryotic viruses are tail-less, members of the Caudovirales order, such as coliphages T4 and T7, might not be the most suitable models. Tail-less phages like MS2, Φ6, and ΦX174 were also explored as viral aerosol models (9). However, only a few studies have compared the effects of aerosolization and sampling on these tail-less phages (14, 15). Moreover, culture-based methods were often used in these studies. Some authors pointed out that PCR quantification of total viral particles collected by a sampler rather than the culture recovery of infectious viruses is essential to assess the physical stress caused to the virus by the aerosolization and air sampling (16). Considering their structural composition and their genetic makeup, it is expected that very distinct viruses react differently to the inevitable mechanical stresses caused by aerosolization and sampling.

Virus infectivity in aerosols is also influenced by environmental conditions, such as temperature, relative humidity, and UV light (17–23). Therefore, the behavior of model viruses under various environmental conditions should be known. Finally, it is worth noting that transmissible viruses are often aerosolized by natural processes like sneezing and coughing and thus are found in complex mixtures containing body fluids, salts, microbial cells, etc. Few studies have investigated the effect of medium composition on the aerosolization process and virus infectivity (16, 24, 25). These studies revealed that organic matter can have a protective effect on phages T3 and Φ6 in the aerosol state as well on influenza virus on dry surfaces. In the latter studies, influenza virus was not tested in the aerosol state and phage aerosols were analyzed only by culture.

In this study, we have analyzed five tail-less phages as possible model viruses for aerovirology studies: MS2 (Leviviridae family), Φ6 (Cystoviridae family), ΦX174 (Microviridae family), PM2 (Corticoviridae family), and PR772 (Tectiviridae family). These viruses were chosen on the basis of their similarities (morphology, envelope, capsid size, and genome material) with known pathogenic viruses (see Table 1). Their behavior during aerosolization and under different sampling conditions was monitored using plaque assays and quantitative PCR (qPCR) data. We also determined the ability of organic matter in aerosols to preserve viral infectivity. Finally, we compared the resistance of the five phage models to aerosolization and sampling with that of two pathogenic viruses, human influenza A virus H1N1 (Orthomyxoviridae family) and the poultry virus Newcastle disease virus (NDV; Paramyxoviridae family).

TABLE 1.

Bacteria and viruses used in this study

Bacterial or viral strain	Growth conditions and characteristicsa	Reference(s)
Bacteria
HER-1036	Escherichia coli, TSB, 37°C, 200 rpm	30
HER-1102	Pseudomonas syringae var. phaseolicola, TSB, 22°C, 100 rpm	14
HER-1221	E. coli, TSB, 37°C, 200 rpm	13
HER-1254	Pseudoalteromonas espejiana, NB-AMS, 30°C, 200 rpm	26
HER-1462	E. coli, TSB, 37°C, 200 rpm	14
Viruses
HER-036	Phage ΦX174, 25 nm, unenveloped, linear ssDNA, 5,386 bases, bacterial host HER-1036	30
HER-102	Phage Φ6, 85 nm, enveloped, segmented dsRNA, 13,385 bp, bacterial host HER-1102	14
HER-221	Phage PR772, 80 nm, unenveloped, linear dsDNA, 14,492 bp, bacterial host HER-1221	13
HER-254	Phage PM2, 60 nm, unenveloped, circular dsDNA, 10,079 bp, bacterial host HER-1254	26
HER-462	Phage MS2, 25 nm, unenveloped, linear ssRNA, 3,569 bases, bacterial host HER-1462	14
ATCC VR-95	Human influenza A virus H1N1 A/PR/8/34, 80–120 nm, enveloped, segmented ssRNA, 13,588 bases, grown on SPF chicken eggs	53, 54
NDV B1	Newcastle disease virus Hitchner B1, 100–200 nm, enveloped, linear ssRNA, 15,186 bases, grown on SPF chicken eggs	28

^assDNA, single-stranded DNA; dsRNA, double-stranded RNA; dsDNA, double-stranded DNA; ssRNA, single-stranded RNA.

MATERIALS AND METHODS

Bacterial and viral strains.

The bacteria and phages used in this study are listed in Table 1 and were provided by the Félix d'Hérelle Reference Centre for Bacterial Viruses (www.phage.ulaval.ca). Culture media were purchased from Difco Laboratories (Detroit, MI). Phages MS2 and ΦX174 were amplified in Trypticase soy broth (TSB) as reported previously (14). Phages Φ6 and PR772 were cultivated on Trypticase soy agar (TSA) with TSB soft agar (0.75%) and TSB soft agarose (0.75%), respectively, as described previously (13, 14). Phage PM2 was grown on Pseudoalteromonas espejiana host cells in 30 ml nutrient broth containing artificial marine salt (NB-AMS) as instructed elsewhere (26). All phage lysates were titrated on their respective bacterial host using standard plaque assay (27) with TSA and TSB soft agar for Φ6, PR772, MS2 and ΦX174, as well as NB-AMS agar and NB-AMS soft agar for PM2. To assess the host specificity of the five phages, all phages were plated on all bacterial hosts used in this study. No plaque was observed with the five phages when they were plated on other bacterial hosts (data not shown).

The human influenza A virus H1N1 A/PR/8/34 strain was obtained from the American Type Culture Collection (ATCC VR-95). The Newcastle disease virus Hitchner B1 strain was purchased as a live vaccine (Wyeth Animal Health). Both strains were used after one passage in chicken embryos. The influenza virus and NDV strains were grown on specific-pathogen-free (SPF) chicken eggs as described previously (28). The infectious titer, expressed as 50% egg infectious dose (EID₅₀) units, was calculated according to the method of Reed and Muench (29) using five eggs per dilution, as described elsewhere (28).

Aerosolization in a controlled chamber.

The aerosolization setup is described elsewhere (30). Aerosols were generated with an atomizer (model 9302; TSI Inc., Shoreview, MN) at a dispersion rate of 3 liters/min. The nebulizer was filled with 1 ml of each phage lysate (10¹⁰ PFU), and the volume was completed to 70 ml using phage buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM MgSO₄). For the nebulization with organic fluid, phages lysates were mixed with 1 ml of allantoic fluid 16 h before the experiment. The allantoic fluid was harvested from noninfected eggs, pooled, mixed, filtered on a 0.22-μm-pore-size membrane, and stored at 80°C until use. Immediately before aerosolization, phage buffer was added to the phage-allantoic fluid mixtures to obtain a final volume of 70 ml. Aerosols were also generated with a 6-jet Collison nebulizer (BGI, Waltham, MA) filled with 1 ml of phage PR772 lysate (10¹⁰ PFU) diluted with 49 ml of phage buffer at a dispersion rate of 7 liters/min. The third nebulizer tested, the Aeroneb Lab nebulizer (Aerogen Inc., Galway, Ireland), was filled with 7 ml from a phage dilution (1 ml of phage PR772 lysate at 10¹⁰ PFU/ml in 49 ml of phage buffer). The aerosol was driven into the chamber at a flow rate of 4 liters/min.

Once produced, the aerosols were passed through a desiccator (model 306200; TSI Inc.) to form aerosol droplet nuclei, before entering a GenaMini chamber (SCL Medtech Inc., Montreal, QC, Canada) from which the aerosols were sampled (see below). Aerosols inside the chamber were mixed with the dilution air at a rate of 23 liters/min with the TSI 9302 atomizer, 22 liters/min with the Aeroneb Lab nebulizer, and 19 liters/min with the Collison 6-jet nebulizer and were maintained in a continuous flow throughout the sampling. Medical-grade compressed air was used for aerosol generation and dilution. The experiments were conducted at temperatures between 28°C and 31°C. The relative humidity in the aerosol chamber was below 20% with the Aeroneb Lab nebulizer and the TSI 9302 atomizer and was 30% with the 6-jet Collison nebulizer. The aerosol size distribution and concentration were monitored every 30 min during each experiment with an aerodynamic particle sizer (APS; model 3321; TSI Inc.) to ensure proper nebulizer function.

Aerosols were sampled with an SKC BioSampler (SKC Inc., Eighty Four, PA) for 20 min at about 11.3 to 13.3 liters/min (the maximum of the critical orifice capacity of the SKC BioSampler) driven by a Gilair Aircon II pump (Levitt Sécurité, Montréal, QC, Canada) and loaded with 20 ml of phage buffer. Following sampling, the residual liquid in the SKC BioSampler was measured and collected. Air samples were also collected using Centers for Disease Control and Prevention (CDC), National Institute for Occupational Safety and Health (NIOSH), two-stage cyclone bioaerosol samplers (CDC, NIOSH, Morgantown, WV) mounted with a 15-ml tube (first stage), a 1.5-ml tube (second stage), and a 0.8-μm-pore-size 37-mm backup polycarbonate filter in an open-face cassette (SureSeal; SKC Inc.) and connected to a Gilair Aircon II pump at a flow rate of 3.5 liters/min and 10 liters/min for 60 min. The samples were recovered by vortexing for 15 min at a 70% motor speed and a 100% frequency in 5 ml (first stage), 1.5 ml (second stage), and 5 ml (filter) of phage buffer using a multipulse vortex apparatus (Glas-Col, Terre Haute, IN). Controls were performed to confirm that the vortex conditions did not affect viral infectivity.

For each aerosolization assay, aerosol samples and the nebulizer content were collected and analyzed at the same time. When several phages were nebulized together, samples were diluted and plated on the bacterial hosts to determine the infectious titer of each specific phage. The viral genome concentration, corresponding to the total concentration of infectious and noninfectious viral particles, was evaluated using qPCR. Every aerosolization condition was repeated at least three times in distinct experiments (n ≥ 3). The number of experiments is displayed in the legends to the figures.

Quantification of total viral particles using qPCR.

Phage ΦX174, PM2, and PR772 genomes were quantified from two dilutions of each sample without DNA extraction. For quantification of Φ6, MS2, NDV, and influenza virus, genomic RNA was extracted from two dilutions of all samples using a QIAamp viral RNA minikit (Qiagen, Chatsworth, CA) as described by Gendron et al. (14). Briefly, the RNA carrier was omitted from the Qiagen AVL buffer, and the RNA was eluted from the column with two volumes of 40 μl TE buffer, pH 8.0 (10 mM Tris, 0.1 mM EDTA). All RNA samples were stored at −86°C. Influenza virus, MS2, and Φ6 cDNA synthesis was performed using an iScript cDNA synthesis kit (Bio-Rad Life Sciences, Mississauga, ON, Canada) (14).

The primers and probes used in this study are listed in Table 2. The probes were labeled with 6-carboxyfluorescein (FAM) at the 5′ end and Iowa black FQ (IABlkFQ) or black hole quencher 1 (BHQ) at the 3′ end. Influenza virus and PM2 probes were also labeled with a ZEN quencher. Primers and probes for the detection of phages PM2 and PR772 were designed with Beacon Designer (version 4.02) software (Premier Biosoft International, Palo Alto, CA). The qPCRs used to detect all phages were done separately. Primer and probe specificity was analyzed using the BLAST tools at NCBI. Moreover, the five phage primer and probe sets were tested with all phage models used in this study. All primer and probe sets give qPCR fluorescence signals only with their designated phages (data not shown).

TABLE 2.

Primers and probes used in this study

Primer	Sequence	Target	Reference
ΦX174for	5′-ACAAAGTTTGGATTGCTACTGACC-3′	ΦX174 genome positions 508–531	30
ΦX174rev	5′-CGGCAGCAATAAACTCAACAGG-3′	ΦX174 genome positions 630–609	30
ΦX174probe	5′-FAM-CTCTCGTGCTCGTCGCTGCGTTGA-BHQ-3′	ΦX174 genome positions 533–556	30
Φ6Tfor	5′-TGGCGGCGGTCAAGAGC-3′	Φ6 S segment positions 430–446	14
Φ6Trev	5′-GGATGATTCTCCAGAAGCTGCTG-3′	Φ6 S segment positions 530–506	14
Φ6Tprobe	5′-FAM-CGGTCGTCGCAGGTCTGACACTCGC-BHQ-3′	Φ6 S segment positions 450–474	14
PR772for	5′-CCTGAATCCGCCTATTATGTTGC-3′	PR772 genome positions 4538–4560	This study
PR772rev	5′-TTTTAACGCATCGCCAATTTCAC-3′	PR772 genome positions 4663–4641	This study
PR772probe	5′-FAM-CGCATACCAGCCAGCACCATTACGCA-IABlkFQ-3′	PR772 genome positions 4639–4614	This study
PM2for	5′-CAAGTGGTCAGGCGTTTATCAG-3′	PM2 genome positions 4058–4079	This study
PM2rev	5′-TGCTCGGCTTTGGCATCTTC-3′	PM2 genome positions 4157–4138	This study
PM2probe	5′FAM-AATTGCCGC-ZEN-ATCTTCACTCTCAACACCGTT-IABlkFQ-3′	PM2 genome positions 4128–4099	This study
MS2 1 for	5′-GTCCATACCTTAGATGCGTTAGC-3′	MS2 genome positions 1261–1284	14
MS2 1 rev	5′-CCGTTAGCGAAGTTGCTTGG-3′	MS2 genome positions 1420–1401	14
MS2 1 probe	5′-FAM-ACGTCGCCAGTTCCGCCATTGTCG-BHQ-3′	MS2 genome positions 1391–1367	14
M+4213	5′-TCCTCAGGTGGCCAAGATAC-3′	NDV genome positions 4213–4232	31
M-4350	5′-TGCCCCTTCTCCAGCTTAGT-3′	NDV genome positions 4331–4350	31
M-4268	5′-FAM-TTTTAACGCTCCGCAGGCAC-IABlkFQ-3′	NDV genome positions 4249–4268	31
InfAfor	5′-GACCRATCCTGTCACCTCTGAC-3′	Influenza virus M segment positions 160–181	55
InfArev	5′-AGGGCATTYTGGACAAAKCGTCTA-3′	Influenza virus M segment positions 242–265	55
InfAprobe	5′-FAM-TGCAGTCCT-ZEN-CGCTCACTGGGCACG-IABlkFQ-3′	Influenza virus M segment positions 215–238	55

The assay components per 25 μl were 5 μl of phage sample or 2 μl of cDNA, 12.5 pmol of primers for all phages and 20 pmol of primers for influenza virus, and 12.5 μl of 2× master mix of the iQ Supermix (Bio-Rad). Five picomoles of dually labeled probe was added for phage PR772, ΦX174, and PM2 detection, 7.5 pmol was added for Φ6 detection, 3.75 pmol was added for MS2 detection, and 4 pmol was added for influenza virus detection. The PCR program for all the phages was as follows: 5 min at 94°C and then 40 amplification cycles including denaturation at 94°C for 15 s and annealing and elongation at 60°C for 60 s, followed by fluorescence measurement. For influenza virus detection, the PCR program was 5 min at 94°C and then 40 cycles of 15 s at 94°C and 30 s at 55°C, followed by fluorescence measurement.

One-step detection of the NDV genome was performed according to the method of Wise and colleagues (31). Briefly, the assay components per 25 μl were 0.5 μl of the enzyme supplied with the kit, 5 μl of template RNA, 12.5 pmol of primers, 5 pmol of dually labeled probe, and 12.5 μl of 2× master mix of the iScript one-step reverse transcription-PCR (RT-PCR) kit for probes (Bio-Rad). The RT-PCR program was as follows: 15 min at 50°C, 5 min at 95°C, and then 40 cycles including denaturation at 94°C for 15 s and annealing and elongation at 60°C for 60 s, followed by fluorescence measurement.

All experiments were performed using an Opticon 2 system (MJ Research, Waltham, MA). Data were analyzed with the Opticon software supplied with the apparatus. For each PCR run, a standard curve was generated in duplicate with plasmid DNA (or RNA for NDV). Serial 10-fold dilutions ranging from 10⁰ to 10⁷ molecules per reaction tube were used to establish standard curves. For each sample of unknown concentration, two 10-fold dilutions made in duplicate were analyzed, and the concentration was determined using the standard curves. The background was subtracted using the average over cycle range function of the software. Threshold cycle (C_T) values were determined automatically with the software. The plotting of C_T as a function of the logarithm of the amount of DNA template gave a straight line. The slope of this graph line gave the PCR efficiency (E) according to the equation E = (10^−1/slope − 1) × 100. Results were considered accurate when E was over 85% and the error between standard points and the regression curve was lower than 0.1.

The DNA and RNA for standard curves for phages MS2, Φ6, and ΦX174 and for NDV were obtained in previous studies (14, 28, 30). The DNAs for standard curves for influenza virus and phages PR772 and PM2 were prepared as follows. The region of interest of each virus was amplified from purified genomes (cDNA for influenza virus) by PCR using the primer sets InfAfor/InfArev, PM2for/PM2rev, and PR772for/PR772rev. The amplification products were cloned into the pDrive cloning vector using a Qiagen PCR cloning kit (Qiagen). The DNA constructions were transformed into Escherichia coli MC1061 competent cells using the rubidium chloride method (32). Plasmid DNAs were purified from E. coli using a Qiagen plasmid minikit and were quantified with a GeneQuant pro UV/visible spectrophotometer (Biochrom Ltd., Cambridge, United Kingdom).

Material preparation and blank controls.

Air samplers were sterilized by autoclave before each experiment. After each experiment, the nebulizers, tubing, and air samplers were decontaminated by soaking for 20 min in 1/20 Virox 5 disinfectant. The material was then rinsed with water and dried. Once cleaned, the tubing and nebulizers were reassembled onto the system and run with clean medical-grade compressed air to purge the diffusion dryer and the GenaMini chamber. The system was purged until zero particles were counted from the chamber with the APS.

Blank aerosol experiments were performed as described above for the aerosol experiments, except that the nebulizer was filled with sterile phage buffer. The samples collected in the blank aerosol experiments were processed and analyzed by culture and qPCR with the other air samples. No virus was detected by culture and qPCR from the samples collected during the blank aerosol experiments. Controls in the absence of template were performed at every qPCR run. Moreover, controls without reverse transcriptase were used for qPCR detection of RNA viruses (MS2, Φ6, influenza virus, and NDV).

Calculations.

The number of infectious viruses collected per liter of air (number of PFU/liter) for each sample was calculated as described in equation 1.

$$I_{\mathrm{P}\mathrm{F}\mathrm{U}}=\frac{C_{\mathrm{P}\mathrm{F}\mathrm{U}}\times v}{f\times t}$$ (1)

where I_PFU is the concentration of infectious viruses collected by the air sampler (number of PFU/liter), C_PFU is the air sample concentration (number of PFU/ml), v is the air sample volume (ml), f is the air sampler flow rate (liters/min), and t is the sampling time (min).

The number of viral genomes per liter of air (I_qPCR) was also calculated for all air samples using the same method by replacing C_PFU by the air sample concentration in number of genomes/ml (C_qPCR) in equation 1. The number of total viral particles in the GenaMini chamber as well as inside the nebulizer varied from one experiment to the other. The relative recovery in numbers of PFU (PFU relative recovery) was calculated as described in equation 2 to compensate for the interexperimental variation, and these values were used for the analyses.

$$\mathrm{P}\mathrm{F}\mathrm{U}\hspace{0.17em}\mathrm{r}\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\hspace{0.17em}\mathrm{r}\mathrm{e}\mathrm{cov}\mathrm{e}\mathrm{r}\mathrm{y}=\frac{I_{\mathrm{P}\mathrm{F}\mathrm{U}}}{n_{\mathrm{P}\mathrm{F}\mathrm{U}}}$$ (2)

where I_PFU is the concentration of infectious viruses collected by the air sampler (number of PFU/liter) and n_PFU is the concentration of infectious viruses inside the nebulizer (number of PFU/ml).

The relative recoveries by qPCR were obtained by equation 2 by replacing I_PFU and n_PFU by I_qPCR and n_qPCR, respectively, where I_qPCR is as defined above and n_qPCR is the concentration of infectious viruses inside the nebulizer in number of genomes/ml. Relative ratios of the number of PFU obtained by culture to the number of genomes obtained by PCR were also calculated to allow comparison of the results for viruses cultured on embryonated chicken eggs and those of the phage plaque assays. This ratio was calculated by dividing the percentage of infectious virus in the samples by the percentage of infectious virus in the nebulizer. Data were analyzed using 2-way analysis of variance or an unpaired t test (when mentioned) of the data. To fulfill the normalization and variance assumptions, variables were log transformed, and P values were reported from these transformations. The significance level that was used to assign significant differences was a P value of <0.05.

RESULTS

Effect of nebulizer.

We first compared three nebulizers (the TSI 9302 atomizer and the Aeroneb Lab and Collison 6-jet nebulizers) for the relative recovery of the dsDNA Tectiviridae phage PR772. We also tested two nebulization buffers (with and without organic fluid) using the TSI 9302 atomizer. In all experiments, phage particles were collected using two air samplers (SKC BioSampler and a NIOSH two-stage cyclone bioaerosol sampler at 10 liters/min). We analyzed phage recovery by culture (which gave the number of PFU) and qPCR (which gave the genome copy number). These experiments were conducted only with the FDA-approved phage PR772. The organic fluid could not be used with the Collison 6-jet and the Aeroneb Lab nebulizers, as the organic supplement clogged the Aeroneb Lab nebulizer as well as produced too much foam in the Collison 6-jet nebulizer.

The characteristics of the aerosols delivered into the GenaMini chamber using each nebulizer are presented in Table 3. The aerosols produced with the Collison 6-jet nebulizer and the TSI 9320 atomizer were similar in size and concentration. However, the aerosols delivered by the Aeroneb Lab nebulizer were larger and less concentrated. The addition of organic fluid into the nebulization buffer slightly increased the number of particles and decreased the particle size produced by the TSI 9320 atomizer.

TABLE 3.

Characteristics of aerosols delivered into the GenaMini chamber by each nebulizer measured with an APS^a

Nebulizer	MMADb (μm)	Total count (no. of particles/cm3)
Collison 6-jet nebulizer	0.779–0.855	1.43e4–2.30e4
Aeroneb Lab nebulizer	1.000–1.320	3.30e3–6.60e3
TSI 9302 atomizer	0.890–1.100	1.68e4–4.97e4
TSI 9302 atomizer with organic fluid	0.869–0.877	4.18e4–5.51e4

^aThe values displayed are the minimum and maximum measured.

^bMMAD, median mass aerodynamic diameter.

In our setup, the TSI 9302 atomizer and Collison 6-jet nebulizer gave similar performances when phage recovery was measured by both culture and qPCR (Fig. 1). Still, the Aeroneb Lab nebulizer led to lower relative recovery of phage by culture as well as qPCR (Fig. 1, P < 0.05). This result can be explained by the fact that fewer particles were delivered into the aerosol chamber when this nebulizer was used. Overall, our results showed that qPCR detected 1,000-fold more genome copies than infectious PR772 particles, suggesting that the aerosolization and/or the sampling strongly affected its structural integrity. When the results are taken altogether, the TSI 9302 atomizer performed better in our setup.

FIG 1.

Effect of nebulizer and nebulization buffer on relative recoveries of phage PR772 by qPCR and culture. Samples were taken with the SKC BioSampler (black symbols) or the NIOSH two-stage cyclone bioaerosol sampler at 10 liters/min (gray symbols). The black and gray bars indicate the median relative recoveries obtained with each sampler (n = 3). §, P = 0.03 for relative recovery by qPCR and P < 0.001 for relative recovery by culture (PFU) for comparison of the SKC BioSampler and the NIOSH two-stage cyclone bioaerosol sampler; *, P < 0.05 for comparison of the relative recoveries obtained with both air samplers when samples were aerosolized with different nebulizers.

The relative recovery of PR772 was 100-fold higher with the SKC BioSampler than the NIOSH two-stage sampler when the air samples were analyzed by plaque assays (P < 0.0001). However, the differences between the two air samplers in the relative recovery of phage PR772 by qPCR were less significant (P = 0.038). These data suggest that the NIOSH sampler is more damaging to the integrity of phage PR772 and/or is less efficient at collecting particles of the size range generated by the nebulizers. Finally, the PFU relative recovery of PR772 by culture was higher (P = 0.03) when it was aerosolized in buffer containing an organic fluid (Fig. 1).

Five phage models.

The five selected phages were then aerosolized all together using the TSI 9302 nebulizer. Phage particles were collected using three air samplers (the SKC BioSampler and the NIOSH two-stage cyclone bioaerosol sampler set at 3.5 liters/min and 10 liters/min). Again, we analyzed phage recovery by qPCR and culture. For the plaque assays, the collected air samples containing the five phages were plated on each individual and phage-specific bacterial host to calculate the titer of each phage contained in the samples. The qPCRs for the detection of all the phages were done separately. The five primer and probe sets were tested with the five phage models. Each primer and probe set gave a qPCR fluorescence signal only with its designated phage (data not shown).

The relative recovery of all phages by qPCR was similar with all samplers (Fig. 2). However, the results by plaque assays were different between phages and samplers. The most infectious particles recovered were those of phage MS2. Interestingly, the relative recoveries by qPCR and culture were of the same order of magnitude with this phage. With phages Φ6, PR772, and ΦX174, the relative recovery by culture was between 100- and 1,000-fold less than that by qPCR (P < 0.0001). Strikingly, infectious phage PM2 particles were poorly recovered (Fig. 2). PM2 plaques were found for only 3 out of 18 samples analyzed. In fact, the relative recovery of phage PM2 by culture was 6.3 × 10⁷ times less than the relative recovery by qPCR.

FIG 2.

Comparison of relative recovery by qPCR and relative recovery by culture of the five phage models nebulized all together and sampled with the SKC BioSampler (black symbols) and the NIOSH two-stage cyclone bioaerosol sampler at 3.5 liters/min (dark gray symbols) and 10 liters/min (light gray symbols). The bars indicate the median relative recoveries obtained with each sampler (n = 6). P values were determined by an unpaired t test of normally distributed log-transformed data. *, comparison of the SKC BioSampler and the NIOSH two-stage cyclone bioaerosol sampler at 3.5 liters/min and 10 liters/min for relative recovery by culture. P was <0.0001 for phage ΦX174. The use of a in the left panel indicates that the relative recoveries by qPCR for these phages were not significantly different. The use of a also indicates that the qPCR and culture relative recoveries were not significantly different for phage MS2 with all air samplers. For a comparison of the qPCR and culture relative recoveries with all air samplers, P was <0.0001 for phages PM2, Φ6, PR772, and ΦX174. a, b, c, and d, comparison of the relative recoveries of phages by culture (PFU). P was <0.05 with the NIOSH two-stage cyclone bioaerosol sampler at 10 liters/min when comparing phages in cluster b with phages in clusters a, c, and d. P was <0.05 between clusters c and a with all air samplers. P was <0.05 between clusters c and d with the SKC BioSampler. P was <0.05 between clusters a and d with all air samplers.

The relative recoveries by culture were similar with the NIOSH samplers at 3.5 liters/min and 10 liters/min for all phages except PM2. The SKC BioSampler led to a better recovery (100-fold higher) of infectious ΦX174 than the NIOSH sampler (Fig. 2, P < 0.0001). No significant difference between the samples collected with the two samplers was observed for Φ6, PR772, and MS2 (Fig. 2).

Comparison of the relative recoveries of the phages by culture divided the five phages into four clusters (Fig. 2). The relative recoveries of MS2 and PM2 by culture were different from those of the other phages with both samplers (Fig. 2, clusters a and b). There was no statistically significant difference between the relative recovery of phages Φ6 and PR772 by culture with both air samplers (Fig. 2, cluster c). There was no statistically significant difference between the relative recovery of PR772, Φ6, and ΦX174 by culture with the NIOSH samplers at 3.5 liters/min and 10 liters/min, but there was a difference with the SKC BioSampler (Fig. 2, cluster d).

Phages and organic fluid.

We investigated if the use of organic fluid in the buffer used to aerosolize the phages could improve the recovery of infectious viral particles using the NIOSH sampler (Fig. 3). Our data show that the utilization of organic fluid in the nebulization buffer did increase the relative recovery of Φ6 by culture by 100-fold (unpaired t test, P = 0.0018) but that it had no significant impact on the recovery of infectious particles for MS2, PM2, and ΦX174 (data not shown).

FIG 3.

Comparison of the relative recovery by culture of phage Φ6 aerosolized from phage buffer or phage buffer supplemented with organic fluid and sampled using the NIOSH two-stage cyclone bioaerosol sampler at 3.5 liters/min (dark gray symbols) and 10 liters/min (light gray). The bars indicate the median relative recoveries obtained with each sampler (n = 3). *, P = 0.0018 by comparison of the two nebulization buffers with air sampling using the NIOSH two-stage cyclone bioaerosol sampler. An unpaired t test was performed on normally distributed log-transformed data.

Correlation with pathogenic viruses.

Finally, we studied the aerosolization and the sampling of two pathogenic viruses, the poultry virus NDV as well as human influenza A virus H1N1. For culture and biosafety reasons, these viruses were aerosolized one at a time under the same conditions used for the phage models. We compared the percentage of infectious viruses collected with the SKC BioSampler with the percentage of infectious viruses in the nebulizer at the beginning of the experiment to obtain the relative infectious ratio (Fig. 4). We used these ratios for comparison with the results obtained with the phage models. As shown in Fig. 4, NDV was the most resistant virus tested in this study. The strain of influenza A virus tested displayed intermediate resistance to aerosolization and sampling.

FIG 4.

Comparison of the relative culture/genome ratios for the five phage models with those for influenza A virus H1N1 (InfA) and NDV. The viruses were sampled with the SKC BioSampler. The bars indicate the median relative culture/genome ratio obtained for each virus. The relative ratios of PFU obtained by culture to genomes obtained by PCR could not be calculated for phage PM2 because no plaques could be detected for this phage under the conditions used in the six experiments. Data are for six experiments for all phages and three experiments for influenza A virus and NDV. a and b, comparison of the relative culture/genome ratios of the viruses. An unpaired t test was performed on normally distributed log-transformed data. P was >0.05 between viruses in cluster a, and P was >0.05 between the viruses in cluster b. P was <0.05 between the viruses in clusters a and b.

DISCUSSION

Here, we have studied the behavior of five very distinct tail-less phages and two pathogenic viruses under various aerosolization and sampling conditions using culture assays and qPCR data. One clear result from this study was that the relative recovery of virus was almost always higher when using qPCR than when using culture, with exceptions being phage MS2 and the poultry virus NDV. It has already been documented that virus relative recovery is higher when qPCR is used than when culture is used (14, 30, 33–35). For example, it was previously shown for RNA phages MS2 and Φ6 that the number of viral genome copies estimated by qPCR was 10-fold (MS2) to 1,000-fold (Φ6) higher than the number of infectious particles (14). Similar data were obtained here with both phages, although minor differences were noted, likely due to the distinct sampling devices used in the two studies.

qPCR is often used in aerosol chamber studies to quantify the total number of viral particles collected and compare it with the amount of infectious virus to assess the damage caused to the viruses by aerosolization and sampling (14, 28, 30, 33, 34). To use qPCR, one must ascertain the genome stability in the air sampler. Such genome degradation has been reported for influenza virus after long-term air sampling using the NIOSH two-stage bioaerosol sampler (33). However, no degradation was observed in the SKC BioSampler for influenza virus and porcine reproductive and respiratory syndrome virus (33, 35). In our experiments, the relative recovery by qPCR was similar for the five phages when sampled with the SKC BioSampler for 20 min and the NIOSH two-stage bioaerosol sampler at 10 liters/min and 3.5 liters/min for 1 h, demonstrating that the genome stability of these phages was similar under all our sampling conditions.

The relative recoveries of the five phage models by culture were remarkably different (Fig. 2). Phage MS2 was the most robust among the five phages tested, as it could be detected by qPCR and plaque assays at similar levels. This phage demonstrated a good resistance to aerosolization and sampling, which is in line with the results obtained by others (14, 23). Not surprisingly, phage MS2 has been used as a surrogate for pathogenic viruses in several sterilization, air filtration, and aerosolization studies (14, 17, 22, 23, 36–45). Its facility to be detected by culture is certainly one of the reasons for its widespread use. However, this phage may not be the best suitable surrogate for all pathogenic viruses. For example, phage MS2 is 7 to 10 times more resistant to aerosolization, sampling, and UV light than a coronavirus (23). MS2 is also very resistant to air sampling under dry conditions, as demonstrated here and by others (14), while the opposite was observed for influenza virus (34).

Discrepancies between quantification by plaque assay and genome quantification from virus preparations were previously reported for phages Φ6, MS2, and ΦX174 as well as for influenza virus (14, 33, 34, 46). To shed further light on these discrepancies between our qPCR and plaque assay data, we compared the numbers of PFU and genome copies for the five phages in crude lysates and in phage preparations purified through a CsCl gradient. The free genome can be eliminated from phage preparations using a CsCl buoyant density centrifugation, as the density of RNA and DNA is higher than the density of phage particles (47). The CsCl gradient did not decrease the ratio of the number of genomes to the number of PFU in pure phage preparations (data not shown). Thus, the discrepancy between the qPCR and culture results is unlikely to come from free genomes. Moreover, we observed with PR772 that the long-term storage of the lysate of this phage led to a decrease in the number of PFU and in the number of genome copies; however, the overall ratio of the number of genomes/number of PFU increased (data not shown). These data suggest that phage PR772, and possibly other phages as well, lost its infectivity before genome degradation.

In order to limit the difference between the number of genomes and the number of infectious phage particles in our initial material, we always worked with fresh phage lysates obtained from the same stock. The phage titer was measured before each experiment, and the same relative amount of each phage was used each time. Also, the concentration of genomes and the numbers of PFU in the nebulizer were used to calculate the relative recoveries by qPCR and culture to make sure that if differences were observed, they originated from damage caused during aerosolization and sampling and not from the starting material. Such precautionary measures should be implemented when testing viruses in aerosol studies.

The NIOSH two-stage bioaerosol sampler (48) has previously been shown to efficiently sample phages (4, 46, 49–51) and influenza viruses (33). In one comparative study, the NIOSH sampler outperformed the SKC BioSampler in sampling tailed phages in industrial settings (51). In our experiments, this sampler set at 3.5 liters/min as well as 10 liters/min was as efficient as the SKC BioSampler for the collection of phages, as demonstrated by the qPCR analysis of our air samples. Moreover, the operation of the NIOSH sampler at 10 liters/min did not cause more damage to our phage models than operation at 3.5 liters/min. There was no statistically significant difference between the two samplers in the relative recovery of phages Φ6, PR772, and MS2 by culture. On the other hand, the SKC BioSampler was more efficient in recovering infectious phage ΦX174. Because the NIOSH two-stage bioaerosol sampler can be used for an extended period of time and collect a larger air volume and because the samples can be resuspended in a small volume, this sampler may be more suitable than the SKC BioSampler for the collection of rare events. This is supported by our data with phage PM2, which poorly resisted aerosolization and sampling, but we were still able to recover infectious viruses only with the NIOSH sampler set at 10 liters/min after 1 h of air sampling.

Viruses aerosolized by natural process such as sneezing and coughing are in a complex and variable medium containing mucus, saliva, etc. Reports have also suggested that the particles emitted by healthy or sick patients are of different sizes (52). Our results suggest that the aerosolization medium can influence the resistance to aerosolization and sampling of some viruses. Further studies are warranted to identify the most suitable agent to be used to improve, if needed, viral recovery.

This study illustrates the diversity of viral behaviors in aerosols. A summary of our results, as well as general recommendations for sampling of these five phages in aerosols and their utilization as surrogates, is presented in Table 4. In our setup, phages MS2 and ΦX174 were the best surrogates for NDV and phages PR772 and Φ6 were the best surrogates for influenza virus (Fig. 4 and Table 4). As more information becomes available, it may be possible to correlate phage behavior to the behavior of human and animal viruses under various conditions.

TABLE 4.

Summary of results obtained in this study with the recommended sampler, detection method, and nebulization conditions for five phages

Phagea	Recommended sampler	Recommended detection method	Organic fluid in aerosolization buffer	Good model for pathogenic virus used in this study
MS2 (ssRNA)	SKC BioSampler or NIOSH two-stage cyclone bioaerosol sampler at 3.5 and 10 liters/min	qPCR or culture	No effect	NDV
Φ6 (dsRNA)	SKC BioSampler or NIOSH two-stage cyclone bioaerosol sampler at 3.5 and 10 liters/min	qPCR	Protective effect	Influenza virus
ΦX174 (ssDNA)	SKC BioSampler	qPCR	No effect	NDV
PR772 (dsDNA)	SKC BioSampler or NIOSH two-stage cyclone bioaerosol sampler at 3.5 and 10 liters/min	qPCR	Protective effect	Influenza virus
PM2 (dsDNA)	NIOSH two-stage cyclone bioaerosol sampler at 10 liters/min	qPCR	No effect

^assRNA, single-stranded RNA; dsRNA, double-stranded RNA; ssDNA, single-stranded DNA; dsDNA, double-stranded DNA.