Influenza virus aerosol exposure and analytical system for ferrets

Abstract

Understanding the transmission ability of newly emerging influenza viruses is central to the development of public health preparedness and prevention strategies. Animals are used to model influenza virus infection and transmission, but the routinely used intranasal inoculation of a liquid virus suspension does not reflect natural infection. We report the development of an inoculation method that delivers an influenza virus aerosol inoculum to ferrets and the characterization of size distribution and viable virus present in aerosols shed from infected ferrets during normal breathing and sneezing. By comparing virus deposition, infectivity, virulence, and transmissibility among animals inoculated intranasally or by aerosols with a human (H3N2) or avian (H5N1) influenza virus, we demonstrate that aerosol inoculations more closely resemble a natural, airborne influenza virus infection and that viable virus is measurable in droplets and droplet nuclei exhaled by infected ferrets. These methods will provide improved risk assessment of emerging influenza viruses that pose a threat to public health.

Influenza viruses cause outbreaks of highly contagious respiratory illness among humans, and most likely have done so since ancient times (1). Despite frequent influenza epidemics and occasional pandemics occurring in more recent times, the relative contribution of the various modes of influenza virus transmission among humans remains poorly understood. Although seasonal and pandemic influenza viruses readily achieve sustained transmission in susceptible humans, avian influenza H5N1 viruses have not, despite having caused hundreds of human infections, primarily resulting from exposure to infected poultry (2). Furthermore, recent reports indicate reduced secondary attack rates in some community settings for the pandemic 2009 H1N1 virus compared with seasonal and previous pandemic strains (3, 4). This notable diversity in the transmissibility of influenza viruses suggests possible differences in the predominant routes of transmission used by different strains under different conditions. There are three modes of transmission: (i) contact transmission, including direct and indirect contact with contaminated surfaces; (ii) droplet transmission, occurring when large (≥5 μm) particles contact a person's conjunctiva or respiratory mucosa; and (iii) droplet nuclei transmission (also referred to as “airborne” or “aerosol” transmission), occurring when small (<5 μm) particles that are capable of prolonged suspension in the air are inhaled. The relative predominance of these different modes of influenza virus transmission in different settings remains controversial and is a major public health question that needs to be addressed. Mammalian models of influenza transmission have been developed to better understand the process of transmission and to rapidly identify the emerging influenza viruses with transmissibility properties that may confer pandemic potential (5–8).

The ferret has become widely recognized as an excellent model of influenza virus transmission and pathogenesis, primarily because this host is naturally susceptible to influenza virus infection, displays clinical signs similar to those of infected humans, and reflects the general transmissibility phenotypes of influenza viruses reported in humans (6, 8–13). Typically, inoculation of animal models is achieved by intranasal (IN) administration of a liquid virus suspension. Although this method is considered standard practice, it does not reflect natural infection by contact or airborne routes. Therefore, to recreate the airborne route of infection, we evaluated the size distribution and viable virus content of aerosols expelled from infected ferrets during normal breathing and sneezing (Fig. S1A) and developed an aerosol inoculation system that can deliver aerosol particles in the same general size range (Fig. S1B). After extensive validation of the aerosol instrumentation (SI Methods), we compared infectivity, virulence, and transmissibility of an H3N2 human influenza virus (A/Panama/2007/1999; PN99) and an H5N1 highly pathogenic avian influenza (HPAI) virus (A/Thailand/16/2004; TH16) in ferrets inoculated IN or by aerosol (AR) inhalation and demonstrated that viable PN99 virus can be detected in droplets and droplet nuclei exhaled by infected ferrets.

Results

Analysis of Aerosols Generated by Infected Ferrets.

To investigate the particle size distribution of aerosols expelled from infected ferrets and the viable influenza virus present in those aerosols, we analyzed exhaled air from anesthetized ferrets for 30 min of closed-mouth, normal breathing followed by 5 min of sneezing stimulation using an aerodynamic particle sizer (APS) and a viable two-stage cascade impactor (Fig. S1A). Four ferrets were inoculated, two IN with 10⁶ pfu and two by AR with 10^4.4 pfu of PN99 virus, and exhaled aerosol samples were analyzed 1, 3, and 5 d postinoculation (dpi) using the impactor and 2, 4, and 6 dpi using the APS. Analysis of particle size distribution revealed that particle counts peaked at <1 μm, 53% for normal breathing and 58% for sneezing (Fig. 1). The analysis of particle size distribution performed on aerosolized PN99 and TH16 virus generated within the aerosol system exposure chamber also peaked at <1 μm (Fig. S2), demonstrating that the particle sizes generated by the aerosol system are comparable to those generated by infected ferrets. Viable virus exhaled from infected ferrets was collected onto two stages of the impactor based on size (0.65–4.7 and >4.7 μm) during normal breathing and sneezing. The recovery rate of this method was determined using known amounts of virus on each plate. For viable virus, the recovery rate was 19 ± 7% for the 30-min collection and 31 ± 11% for the 5-min collection; for viral genetic material, the recovery rate was 54 ± 8% for the 30-min collection and 70 ± 4% for the 5-min collection. All data were normalized to the recovery rate for the applicable procedure. During normal breathing, viable virus was detected in three of four ferrets up to 3 dpi and was five times higher in particles >4.7 μm (Fig. 1A); viral RNA copies were 10–17 times higher than viable virus levels and still were detectable 5 dpi. However, during sneezing, viable virus was detected in all animals 1–5 dpi (Fig. 1B). For both viable virus and viral RNA copies, there was greater variability in peak virus in particles >4.7 μm in size (40–1,180 pfu viable virus; 45–1,678 RNA copies) than in aerosols 0.65–4.7 μm in size, which peaked at 7–10 pfu viable virus and 14–41 RNA copies. Although variability among ferrets is expected, based on collection time and minute volume of respiration, viable virus was shed from infected ferrets at ∼1 pfu every 2 min (1.5 pfu/L) during normal breathing. Nasal washes were collected from each animal after impactor sampling and ranged from 10^4.6–10^5.5 pfu/mL through 5 dpi, demonstrating the additional information gained by analyzing detailed virus shedding profiles of infected animals versus nasal wash sampling alone.

Fig. 1.

Analysis of aerosols generated by infected ferrets. Particle size distribution and viable virus content were assessed for aerosols exhaled from ferrets infected by A/Panama/2007/99 virus during 30 min of normal breathing (A) or 5 min of sneezing stimulation (B). Raw particle counts are plotted against the aerodynamic particle size 2, 4, and 6 dpi. The count median aerodynamic diameter (CMAD) is 0.999 μm [geometric standard deviation (GSD) 1.98] for normal breathing and 0.909 μm (GSD 1.75) for sneezing. Total pfu measured in exhaled aerosols for each size category are shown for individual ferrets. Green data points represent IN-inoculated animals, and orange data points represent ferrets inoculated by AR. Error bars represent mean values ± SE.

Comparison of Aerosol and Intranasal H3N2 Virus Inoculations in Ferrets.

Infectivity, transmission, and virulence were assessed for AR-delivered PN99 virus and were compared with IN-delivered inocula. Serial dilutions of virus were administered either IN or by AR to two to five ferrets each until an end point was reached, and no virus shedding was detected in nasal washes. Once the 50% ferret infectious dose (FID₅₀) values were determined retrospectively for each inoculation, doses were grouped into dose categories as follows: for IN doses, low (1.8–2.3 FID₅₀), intermediate (23 FID₅₀), and high (100–227 FID₅₀); and for AR-presented doses, low (1.5–2.2 FID₅₀), intermediate (15–22 FID₅₀), and high (395–526 FID₅₀). At each dose, transmission was assessed by placing two naïve ferrets in contact with inoculated animals, either in the same cage for contact transmission or in an adjacent cage with a perforated side wall for respiratory droplet transmission. Nasal washes were collected every other day, and clinical signs were monitored daily for at least 14 d. The FID₅₀-IN and FID₅₀-AR were found to be similar at 1.1 and 1.9 pfu, respectively. However, at the low-dose inocula, AR-inoculated ferrets shed significantly higher titers of virus overall (mean peak, log₁₀ 5.6 ± 0.37 pfu/mL) in nasal washes than IN-inoculated animals (mean peak, log₁₀ 4.6 ± 0.47 pfu/mL; P < 0.05; Fig. 2). Naturally infected ferrets (contacts of inoculated animals) shed peak virus titers in nasal washes that ranged from log₁₀ 5.2–5.7 pfu/mL. During contact experiments, transmission between ferrets was observed at all doses regardless of inoculation method. Respiratory droplet transmission was detected in the AR intermediate- and high-dose groups (in two of two ferrets in each group) but was observed only in the high-dose IN-inoculated group, when increased virus shedding was detected in inoculated animals. The observed kinetics of both the IN and AR groups suggest that respiratory droplet transmission depends not only on the level of peak virus shedding (at least 10^4.9 pfu/mL detected in nasal washes) but also on the duration of peak virus shedding, increasing the opportunity for a transmission event to occur. Regardless of dose or method of inoculation, clinical signs exhibited by animals infected by the PN99 virus were similar (Table 1). A transient drop in lymphocytes was observed in both groups (24% at 3 and 7 dpi) with recovery to normal levels by 14 dpi; no substantial changes in blood chemistry were noted. Three ferrets in each group inoculated with a high dose of PN99 virus (IN or AR) were humanely killed 3 dpi, and virus detection in tissues was similar in the two groups (Table 1). For all infected ferrets, nasal turbinate virus titers were highest (10^5.9–10^6.2 pfu/g), followed by tracheal tissues (10^3.2–10^2.5 pfu/g). Virus was not detected in the lungs of any ferrets. Virus was detected just above the level of detection (10^0.7 pfu/g) in brain tissue from all three IN-inoculated ferrets (mean, 10^0.8 pfu/g) and at slightly higher levels in a single AR-inoculated ferret (10^2.2 pfu/g). Together, these findings demonstrate that a low-dose AR inoculum of PN99 virus results in an infection that is indistinguishable from a natural influenza infection in ferrets, based on virus shedding and disease progression, but at higher doses (≥100 FID₅₀), IN- and AR-inoculated animals exhibit more similar virus shedding and tissue dissemination patterns.

Fig. 2.

Virus shedding in A/Panama/2007/99-infected animals. Ferrets were administered a low (1.5–2.3 FID₅₀), intermediate (15–23 FID₅₀), or high (100–526 FID₅₀) dose of virus IN (green) or by AR (orange), and transmission to naïve ferrets was assessed. Virus detection in nasal washes from inoculated animals is represented by transparent curves; virus titers in animals exposed by respiratory droplet contact are shown in light shades; and virus titers in animals exposed by direct contact are shown in dark shades. The level of detection is log₁₀ 0.7 pfu/mL.

Table 1.

Pathogenesis of A/Panama/2007/99 (H3N2) and A/Thailand/16/04 (H5N1) viruses in ferrets inoculated intranasally or by aerosols

			Number of animals
			Clinical signs up to 14 dpi						Virus detection at 3 dpi
Virus	Method of inoculation	Dose range*	Weight loss (%)†	RII	Diarrhea	Dyspnea	Neurological‡	Lethality (MDT)	Lung§	NT	Liver	Brain¶	OB	Spleen	Int**
PN99	IN	High	7.4	1.2	0/5	0/5	0/5	0/5	0/3	3/3	0/3	3/3	2/3	0/3	0/3
PN99	AR	High	3.6	1.0	0/2	0/2	0/2	0/2	0/3	3/3	0/3	1/3	1/3	0/3	0/3
TH16	IN	Low	12.6	2.1	2/2	0/2	0/2	2/2 (6.0)	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.
TH16	IN	Intermediate	12.3	2.0	0/2	0/2	1/2	2/2 (6.0)	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.
TH16	IN	High	15.8	2.2	4/5	3/5	3/5	5/5 (6.2)	2/3	1/3	3/3	1/3	2/3	3/3	3/3
TH16	AR	Low	13.0	2.4	2/2	1/2	2/2	2/2 (6.0)	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.
TH16	AR	Intermediate	18.9	2.1	3/3	3/3	2/3	3/3 (6.3)	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.	N.d.
TH16	AR	High	16.7	2.6	4/6	6/6	6/6	6/6 (6.3)	3/3	3/3	3/3	2/3	2/3	2/3	0/3

AR, inoculation by aerosol; MDT, mean time to death in days; IN, intranasal inoculation; N.d., not determined; NT, nasal turbinates; OB, olfactory bulb region of the brain; RII, relative inactivity index.

*Dose range (FID₅₀ based on pfu): for PN99, high IN (100–227), high AR (74–526); for TH16, low IN (1.2), intermediate IN (12), high IN (62–115), low AR (1.5–1.8), intermediate AR (12–22), high AR (60–87).

^†Percent mean maximum weight loss is shown.

^‡Neurological clinical signs include hind limb paresis, torticollis, seizures.

^§All lobes sampled and pooled.

^¶Including posterior and anterior regions of brain.

**Intestinal duodenum, jejunum, and descending colon sampled and pooled.

Comparison of Aerosol and Intranasal H5N1 Virus Inoculations in Ferrets.

TH16 virus is a HPAI H5N1 virus that has been shown to cause 100% lethality in ferrets inoculated IN with a high dose (10). Infectivity and virulence of this virus was assessed and compared for AR and IN inoculations in ferrets. Serial dilutions of virus were administered to two to six ferrets each IN or by AR to assess infectivity, and disease progression was evaluated at all doses. Inoculations were grouped into dose categories: for IN doses, low (1.2 FID₅₀); intermediate (12 FID₅₀); and high (62–115 FID₅₀); for AR-presented doses, low (1.5–1.8 FID₅₀); intermediate (12–22 FID₅₀), and high (60–87 FID₅₀). Additionally, a 100 FID₅₀ IN dose or 66–87 FID₅₀ AR-presented dose was administered to three ferrets each, and at 3 dpi virus titers in tissues were determined. The infectivity of TH16 virus was similar for the methods of inoculation; the FID₅₀ for IN inoculation was 1.3 pfu, and the FID₅₀ for AR inoculation was 4 pfu. Although this difference is too small to reflect a substantial difference in infectivity of the two methods of inoculations, we wanted to confirm that virus was being deposited in the upper and lower respiratory tracts of ferrets by each method of inoculation. Therefore, we labeled TH16 virus with a fluorescent tag (TH16-FL) to assess deposition of virus. Analysis of the particle size distribution of TH16-FL virus demonstrated that aerosol particle counts peaked at <1 μm, similar to unlabeled viruses (Fig. S2). A ferret was administered an AR inoculum (220 μg/mL) of TH16-FL virus, and a second ferret was inoculated IN with 1 mL of the same inoculum. Both animals were humanely killed immediately after inoculation for ex vivo imaging (Fig. 3). Because the total amount of virus delivered to each ferret was quite different, a quantitative comparison of deposition between the two methods of inoculation was not performed, but virus deposition patterns within individual animals were evaluated. When TH16-FL virus was delivered to a ferret via AR, 70% of virus was deposited in the upper trachea, 1% in the nasal turbinates, 4% in the esophagus, 5% in the bronchi and 14% in peripheral lung tissue. In the IN-inoculated animal, 49% of the inoculum was deposited in peripheral lung tissue, concentrating in the caudal lobes; virus also was detected in the nasal turbinates (2%), upper trachea (26%), esophagus (17%), and bronchi (6%). A greater amount of IN-delivered inoculum was diverted to the digestive tract and collected in the lower lobes of the respiratory tract, whereas a greater amount of AR-delivered inoculum was deposited in the upper trachea because of the impaction of aerosol particles moving through the trachea. Together these findings demonstrate that each method of inoculation delivers virus to the upper and lower respiratory tracts of ferrets, but with AR inoculation, this delivery is accomplished without the use of a liquid inoculum, thus providing a more natural means of inoculation.

Fig. 3.

Virus deposition in ferrets. Fluorescent-labeled A/Thailand/16/04 virus (TH16-FL; 200 μg/mL) was administered to ferrets IN (A) or by AR (B). Each animal was humanely killed immediately and organs were collected for ex vivo imaging. Nasal turbinates are contained within the cap. An increasing fluorescence signal is indicated by brightness from red to yellow.

Similar to the human H3N2 influenza virus, at lower doses AR-inoculated ferrets shed higher titers of virus than did the IN-inoculated animals. Ferrets receiving a low or intermediate dose of TH16 virus inoculum via AR shed nasal wash virus titers >1,000-fold higher (P < 0.05) than shed by ferrets inoculated with the same dose IN. However, when a higher dose of TH16 virus (≥60 FID₅₀) was administered, similar virus titers were achieved by either inoculation method (Fig. 4). Previous studies from this laboratory showed that ferrets inoculated IN with 10⁶ pfu of TH16 virus failed to transmit virus via respiratory droplets to naïve ferrets (14). We confirmed the inefficient transmission phenotype of TH16 virus in AR-inoculated animals (Fig. S3). Ferrets in direct contact with AR-inoculated animals shedding virus titers as high as 10^5.1 pfu/mL did not shed detectable virus or seroconvert. Even at the lowest dose (<2 FID₅₀), TH16 virus infection was fatal in all animals (Table 1). However, there was a marked difference in the clinical spectrum of disease and end-point virus titers in certain tissues, depending on the route of inoculation. Nearly all TH16 virus AR-inoculated ferrets exhibited a gradual progression of severe lethargy, nasal and ocular discharge, dyspnea, diarrhea, and neurological dysfunction regardless of dose. Ferrets inoculated IN also exhibited severe signs of disease, but in many cases onset of severe disease was more sudden, requiring euthanasia and thus limiting the opportunity to observe the full spectrum of disease exhibited by the AR-inoculated group (Table 1). By 3 dpi, lymphocytes dropped by 61% in IN-inoculated ferrets and by 36% in AR-inoculated animals; at later time points, both groups exhibited 75% lymphocyte reduction. Comprehensive analysis of serum analytes at these time points indicated progressive hepatic and renal dysfunction in both groups of ferrets. Virus was detected throughout the respiratory tract of all high-dose AR-inoculated animals by 3 dpi, and no virus was detected in the intestinal tract, whereas all IN-inoculated ferrets had detectable virus in intestinal tissue (Table 1 and Fig. S4B); this difference may have been caused by differences in the amount of virus swallowed during inoculation. End-point (6 or 7 dpi) virus titers in nasal turbinates, brain tissues, and liver were significantly reduced in ferrets inoculated IN with a low or intermediate dose of TH16 virus compared with ferrets inoculated with a similar dose by AR (P < 0.05; Fig. S4A). Mean virus titers were similar in end-point lung tissue samples in the IN and AR groups (10^5.0 and 10^5.2 pfu/g, respectively), as were mean death times. The noted difference in disease progression may be caused by the differences in virus dissemination in IN- and AR-inoculated animals (Table 1 and Fig. S4A); nevertheless, when ferrets were administered a high dose of inoculum IN or by AR, end-point virus titers were similar in all tissues tested (Fig. S4B).

Fig. 4.

A/Thailand/16/04 virus titers in ferret nasal washes. Ferrets were inoculated IN or by AR with a low or intermediate dose (≤22 FID₅₀) and virus titers in nasal washes were determined on the days shown. The limit of detection is log₁₀ 0.7 pfu/mL. ^†Statistically significant (P < 0.05).

Discussion

Influenza viruses spread readily among humans through the transmission of respiratory secretions that are expelled from infected individuals. This process has been modeled in ferrets, but until now, the virus shedding and particle size profiles associated with virus exhaled from infected ferrets have not been reported, and IN administration of a liquid virus suspension has been the standard method of inoculation used in animal models. We report here the characterization of size and viable virus content of aerosols shed from infected ferrets and the development of an aerosol method of inoculation, and we further demonstrate that this method of inoculation more closely resembles a natural influenza virus infection in ferrets. These advances in laboratory methods will give us a better understanding of how these viruses are transmitted and cause disease in mammals.

When measured using a variety of technologies, the majority of aerosol particles expelled from humans during breathing and sneezing are <1 μm (15–17). A recent study of aerosols collected in a hospital during influenza season demonstrated that 53% of virus content was detected by PCR in aerosols ≤4 μm (18). Using our ferret model, we have shown that viable virus is present in aerosols ≤4.7 μm exhaled during breathing and sneezing, highlighting one reason why this virus is transmitted so easily in the ferrets and perhaps in humans. However, the level of virus shedding and the particle sizes containing viable virus, although important, are not the only factors affecting the transmissibility of influenza viruses. Additional factors such as environmental influences, deposition patterns, and differences in receptor-binding preference no doubt contribute to the process of transmission. In humans, aerosol particles ≤5 μm are capable of reaching deep into the respiratory tract and being deposited in alveolar tissues (19). Aerosols of this size, as well as larger particles, can be deposited via inertial impaction in the upper airway, where airflow velocity is greatest, at points in the respiratory tract of airflow directional change. The deposition patterns of influenza virus following inoculation have not been reported previously in ferrets, but we have shown that fluorescent-labeled influenza virus, when administered using our aerosol system, is deposited in the upper and lower regions of the ferret respiratory tract.

The aerosol system we have developed provides a method of influenza virus inoculation for ferrets that more closely resembles a natural, airborne infection and provides the ability to assess the full range of disease progression. The FID₅₀ of AR-delivered virus was 1.9 pfu for PN99 virus and 4 pfu for TH16 virus, similar to the reported 50% human infectious dose of aerosols 1–3 μm in size, which ranges from 0.6 to 3 50% tissue culture infectious dose (TCID₅₀) (20). A recently developed nose-only influenza aerosol system generates particles in the same size range as our system, but characterization of infectivity, transmission, and deposition in ferrets was not reported (21, 22). Aerosol inoculation of mice and guinea pigs has been reported, but aerosol system validation and dosimetry were not described (23, 24). Previous ferret studies in which high-dose IN inoculations (≥10,000 FID₅₀) of PN99 and TH16 viruses were administered for pathotype and transmissibility assessment report peak nasal wash titers similar to those achieved by any of the AR inoculations performed here (8, 10, 14). However, a difference in kinetics was observed; high-dose IN inoculations resulted in earlier peak titers (1 dpi) compared with lower dose IN or AR inoculations. The overall kinetics of lower dose PN99 virus inoculations (IN or AR) were more similar to those observed in naturally infected animals (contacts of inoculated ferrets), but peak virus titers in low-dose IN-inoculated ferrets were significantly reduced compared with both AR-inoculated and naturally infected ferrets. Consequently, a higher IN dose was needed for respiratory droplet transmission to occur. Reduced virus shedding was also observed in ferrets inoculated IN with a low or intermediate dose of TH16 virus compared with similar AR inoculations. A possible explanation is that the 1-mL volume of inoculum alters the surface tension of the mucous layer lining the respiratory tract, and such alteration in surface tension has been shown to result in reduced virus shedding (25), although this effect seems to be overcome by increasing the viral load in the IN inoculum. Ferrets inoculated via AR with TH16 virus exhibited a broader spectrum of disease than IN-inoculated animals, but lethality was similar. The presence of a liquid inoculum did not affect virus replication in lung tissue, but virus dissemination outside the respiratory tract was reduced, perhaps accounting for the increased clinical signs observed in AR-inoculated animals.

Although the relative importance of different routes of transmission used by influenza A viruses remains controversial, it is likely that the modes of transmission are not mutually exclusive, and, whether transmission occurs at close range or long range, aerosols are likely to be involved. We have developed an aerosol inoculation method that more closely recreates a natural influenza virus infection in the laboratory and have demonstrated the presence of viable virus in aerosol particles generated by ferrets infected with the H3N2 influenza virus. This aerosol delivery and analytical system will increase our understanding of the processes involved in influenza virus transmission and pathogenesis and will provide an improved platform with which to perform pandemic risk-assessment studies to guide public health preparedness efforts.

Methods

Viruses and Plaque Assays.

Human H3N2 influenza virus, PN99, and HPAI H5N1 virus, TH16, were used in this study. Virus stocks were propagated in the allantoic cavity of 10-d-old embryonated hens’ eggs as previously described (8). Virus titers of stocks were determined by standard plaque assay in Madin–Darby canine kidney cells (MDCK) as described (26) for determination of pfu. Fluorescent-tagged virus (TH16-FL) was generated using formalin-inactivated TH16 virus (27) and a SAIVI Antibody Alexa Fluor 680 Labeling kit (Invitrogen) per the manufacturer's instructions for virus deposition experiments. HPAI viruses were handled in biosafety level 3 containment, including enhancements required by the US Department of Agriculture and the National Select Agent Program (28–30).

Analysis of Aerosols Expelled by Ferrets.

Four ferrets were inoculated, two IN with 10⁶ pfu and two by AR with 10^4.4 pfu of PN99 virus. Ferrets were anesthetized as described (10), and exhaled aerosol samples were analyzed for viable virus 1, 3, and 5 dpi using a viable two-stage cascade impactor (Tisch Environmental) followed by collection of a nasal wash sample as described (10) to confirm virus shedding at each time point. Exhaled aerosol samples also were collected 2, 4, and 6 dpi for size distribution analysis using an APS (TSI Inc.) for 30 min of closed-mouth, normal breathing followed by 5 min of sneezing stimulation as described (10). Aerosol samples were collected on alternating days to avoid duplicate sample collection during a single day, and ferrets were anesthetized for all procedures for consistency of sample collection. Analyses of APS particle size distribution are reported as the total particle counts collected at each aerodynamic diameter particle size from 0.5–20 μm measured during the aerosol collection period. When the cascade impactor is operated at a flow rate of 28.3 L/min, aerosols are separated onto two stages based on size; the top stage collects particles >4.7 μm, and the bottom stage collects aerosols 0.65–4.7 μm in size. The impactor flow rate was calibrated immediately before each aerosol collection session using an airflow calibrator (BGI, Inc). Each impactor collection plate contained a bottom layer of agar, a Nuclepore PC-TEM filter (Whatman) with a pore size of 0.05 μm, and a top layer of gelatin. The volumes used in each stage were adjusted to 27 mL total to maintain the proper distance between the critical orifice of each stage and the collection medium to ensure correct aerosol segregation. After sample collection, the top layer of gelatin was collected, melted at 37 °C, and subjected to plaque assay, RNA extraction, and real-time RT-PCR. Viral RNA copy numbers were extrapolated using a standard curve based on samples of known virus titers (pfu/mL). Virus recovery rates were assessed by spiking impactor plates with known concentrations of virus (10–100 pfu) and passing sterile air through the impactor for 30 or 5 min. Total viable virus and RNA copy numbers were normalized to the recovery rate of the respective collection procedure. Environmental conditions inside the laboratory were monitored daily and were 21 ± 1 °C and 30 ± 10% relative humidity.

Ferret Experiments.

Ferrets, 6- to 11-mo-old (Triple F Farms) and serologically negative against currently circulating influenza viruses and H5N1 viruses, were used in this study and were housed in cages within a Duo-Flo Bioclean mobile clean room (Lab Products). Ferrets were sedated as described (10) before all inoculations. For AR inoculations, each ferret was placed in a disposable Tyvec sleeve, with only the snout exposed, inside a stainless steel mesh restraint cage to secure the animal while preventing contamination of fur. The restraint cage was placed inside the plethysmograph chamber for minute volume measurement and then was transferred immediately to the aerosol exposure chamber. Each exposure session was conducted at 21 °C and 30% relative humidity and lasted 15 min followed by a 5-min wash to allow aerosolized virus to evacuate the chamber before the animal was removed from the restraint cage and returned to the holding cage. IN inoculations were performed by administration of virus in 1 mL of PBS distilled into the ferrets’ nares. (For this study, “IN inoculation” refers to the method of inoculation and is not representative of inoculation solely of the nasal mucosa.)

FID₅₀ values (the pfu equivalent of 1 FID₅₀) were determined for PN99 and TH16 virus after IN or AR delivery of virus inocula based on virus detection in nasal washes (31). For both PN99 and TH16 virus inoculations, ferrets served simultaneously for transmission, virulence, and FID₅₀ assessment. Two to six ferrets in each group were administered a low, intermediate, or high dose of inoculum IN or AR (Table 1). IN doses of PN99 virus were as follows: low (1.8–2.3 FID₅₀), intermediate (23 FID₅₀), high (100–227 FID₅₀); AR-presented doses were low (1.5–2.2 FID₅₀), intermediate (15–22 FID₅₀), and high (395–526 FID₅₀). IN doses of TH16 virus were as follows: low (1.2 FID₅₀), intermediate (12 FID₅₀), and high (62-115 FID₅₀); AR-presented doses were low (1.5–1.8 FID₅₀), intermediate (12-22 FID₅₀), and high (60-87 FID₅₀). To determine the dose at which the ferrets inoculated with PN99 virus become contagious, two naïve ferrets for each transmission experiment were placed either in the same cage as inoculated ferrets or in an adjacent cage with a perforated adjoining wall, to assess contact or respiratory droplet transmission, respectively, as described (8). During a contact transmission experiment, transmission may occur by direct or indirect contact, droplets, or droplet nuclei. During a respiratory droplet transmission experiment, transmission may occur by droplets or droplet nuclei. Nasal washes were collected every other day for at least 11 dpi or dpc for estimation of virus-shedding kinetics. Clinical signs including body temperature, weight, and activity level were monitored daily, as described (10), in all animals for at least 14 d. Blood samples collected on dpi 0, 3, 7, and 14 were subjected to complete blood cell counts (CBCs) and chemistry analyses per manufacturer's instructions. CBCs with white blood cell differential and blood chemistry analysis were performed as described (32). Any ferret losing more than 25% of body weight or exhibiting neurological dysfunction was humanely killed and submitted to postmortem examination. To assess virus dissemination, three additional ferrets in each group were inoculated with 74–121 FID₅₀–AR or 100 FID₅₀-IN of PN99 virus or 66–87 FID₅₀-AR or 100 FID₅₀-IN of TH16 virus. These ferrets were humanely killed 3 dpi for postmortem examination and collection of tissues for virus titration as described (10). Tissue specimens were collected aseptically in the following order: intestines (duodenum, jejunum, and descending colon), spleen, kidneys, liver, trachea, lungs (each lobe was sampled and pooled), brain posterior section, brain anterior section, brain olfactory bulb, and nasal turbinates. The pfu in nasal washes and tissues was determined by plaque assay in MDCK cells as previously described (26). Deposition experiments were performed by inoculating ferrets with 220 μg/mL or 75 μg/mL TH16-FL virus either IN or by AR. Immediately after inoculation, both animals were humanely killed, and organs were excised for ex vivo imaging using a Spectrum in vivo imaging system and Living Image 4.0 Software (Caliper Life Sciences). Deposition data were normalized against background fluorescence. All animal procedures were conducted with the approval of the Centers for Disease Control Institutional Animal Care and Use Committee and the Office of Safety, Health and Environment.

Statistics.

The statistical significance of differences observed in virus titers measured in nasal washes was determined from the area under the curve using the Mann–Whitney test. Virus titers in tissues from infected animals were analyzed using the Student's t test.