Here, a novel EBC collection method and device was developed and evaluated in collecting EBC samples from human subjects using culturing and molecular methods. Compared to those currently available devices shown in Table S1
, our device is lightweight with simplicity, reusability, and lower cost. The developed collection device itself costs less than $10, with about $0.5 for consumables (straw and hydrophobic film) per collection. The time needed for 100 µl EBC including sample collection and removal was around 2 min. The physical collection efficiency of the device is shown in . The data points shown in the figure were averages of the EBC samples collected from six volunteers under each of the collection times (1, 2, 3 and 4 min) tested. In general, the amount of EBC sample collected was observed to increase with increasing collection time were observed among subjects. As also observed in , the method has a good reproducibility (small variations). ANOVA analysis indicated that the collection time had a statistically significant effect on the amount of EBC sample collected per unit of time (p-value
0.0026). For the 4 min collection, the volume of collected EBC (168.7 µL) was 1.8 times of that (60.0 µL in average) by 1 min. In our study, when no EBC was collected about 1 µL of liquid was obtained from the hydrophobic surface in an environment with a temperature of 17.9–19.3°C and a relative humidity level of 46–52%. In addition, during the breath sample collection, the collection device had a higher air pressure due to the exhaling, thus it is less likely that environmental air would come into the device. This suggests that environmental water vapor had limited impact on the collection method given the total amount of EBC collected. A recent study indicated that the minimum required volume of EBC was 50 µL for follow-up biological and chemical analysis, such as multiplexed cytokine analysis 
. This on the other hand implies that the EBC device developed in this study can provide adequate amount of EBC sample for rapid analysis. Here, only one type of hydrophobic surface (parafilm) was tested, and in the future different hydrophobic materials should be also explored to improve the overall efficiency.
Figure 3 Physical collection efficiency of the exhaled breath condensate collection device developed in this study under different collection time; control indicates the total volume of sample collected without breathing toward the device; data points represent (more ...)
As listed in Table S1
, currently available EBC collection devices, e.g., the Rtube and the EcoScreen, are comparable to ours with respect to rate of EBC collection. However, our EBC device has advantages in size, weight, and simplicity. In our study, we used a 16 cm long straw for exhaling toward to the super hydrophobic surface without any control of saliva for the possible contamination. However, our collection time was only 1–4 min, and during such short sampling period the sample contamination by saliva is very limited given the length of the straw. Another advantage of our developed device is the one time use of the hydrophobic parafilm (disposable) and exhalation straw with an easy collection of EBC, which thus prevents the possible cross contamination and facilitates the collection of EBC samples from a large number of subjects. This is particularly useful during an influenza outbreak or a man-made bio-terrorism attack in which a rapid screening of exposed persons needs to be conducted immediately.
Here, the EBC samples collected by the developed device from seven human subjects recruited from a respiratory unit of Peking University Third Hospital in Beijing were studied using culturing, DNA stain, SEM and molecular methods. In this study, the particle size distributions trends in a typical exhaled breath were also measured and are shown in . As observed in the figure, the number concentration decreased with increasing particle diameter. For bacterial size ranges (0.65–2.2 µm), a concentration level of 329 to 25819 particles/L was observed, while for larger particles of 2.2–4 µm a concentration level of 60 to 400 particles/L was obtained. In previous studies, similar particle size distribution trend in exhaled breath was also found using the OPC, although the droplet concentrations for respective size ranges were slightly different 
. Nonetheless, due to its rapid evaporation water droplet itself or those adsorbing on bacterial particles in the exhale breath will certainly affect the results obtained here. The results from OPC indicated that particles of larger than 2.5 µm only accounted for 0.4% of the total particles exhaled. According to ICRP (1994), the total lung deposition efficiency for particles larger than 2 µm is more than 80%, while for smaller particles of less than 1 µm, the deposition efficiency is less than 40%, i.e., 60% exhaled out 
. In addition, larger particles could stick to the straw wall. Therefore, in the exhaled breath as well as those collected into DI water droplet smaller particles would dominate.
Figure 4 Particle size distributions in exhaled breath by mouth breathing using an Optical Particle Counter(OPC); x-axis shows the average diameters of 16 channel sizes of the OPC; data points represent averages and standard deviations of 20 min measurements by (more ...)
shows the concentrations of culturable bacterial aerosols in EBC samples collected from seven human subjects. As shown in the figure, bacterial concentration levels ranged from 693 to 6,293 CFU/m3
. ANOVA tests indicated that there were statistically significant differences in culturable bacterial aerosol concentrations for EBC samples collected from different subjects (p-value
.0001). In a recent study, human occupants are also identified as the significant contributors for indoor bacteria, i.e., the emission rate is about 37 million gene copies per person per hour, and a distinct indoor air signature of bacteria was demonstrated to be associated with human skin, hair, and nostrils 
. During human breathing, the bacterial particles from environmental air are continuously inhaled, some of which, i.e., smaller ones, can be exhaled out again by the lung and reside with nostrils. Here, bacterial species Sphingomonas
paucimobilis and Kocuria
rosea were detected using Vitek2 in six EBC samples as shown in Table S2
. Because of limitation of Vitek 2, certain bacterial species were not identified in our study. Among the subjects, subject #6 had substantially higher culturable bacterial concentrations than other subjects. From his medical conditions shown in Table S2
, it was likely that his fever was caused by the bacterial infections. In his EBC sample, we found Kocuria variants which were thought to cause catheter-related bacteremia 
. For other human subjects, the culturable bacterial aerosol concentration levels ranged from 700 to 3000 CFU/m3
paucimobilis, a non-fermenting Gram-negative bacillus, were detected. In a previous study, S.
paucimobilis was found to cause nosocomia bacteremia outbreak 
. For negative control samples, we did not observe the bacterial growth, indicating no contamination during the EBC collection. Ideally, bacterial particles in EBC should be collected using a suitable size-selective sampling tool to investigate the bacterial counts for different size range. However, such device is currently not available yet. Compared to the environmental culturable bioaerosol concentrations, those in EBC samples collected had relatively higher levels, thus representing an important source of bioaerosols particularly in a high human occupancy environment. In addition to viruses, Rhodococcus equi
, a bacterium causing pyogranulomatous bronchopneumonia, were detected in the exhaled air from foals in a recent study 
. When pathogenic bacteria are breathed out, they could pose a serious public health threat.
Figure 5 Culturable bacterial aerosol concentrations detected in exhaled breath condensate samples collected using the device from seven human subjects with symptoms listed in Table S2; F and M indicate Female and Male, respectively, 1–7 indicate the subject (more ...)
shows the qPCR amplification plot from EBC samples collected from seven human subjects in a respiratory clinic. As observed from the figure, bacterial samples were successfully amplified (Ct values were 16–19), while the positive sample (B. subtilis
) had a Ct value of 15 and the negative control had a value of 28. Based on the DNA standards used, the concentrations of bacterial DNA in the EBC samples (Sample 1–7) were in the range of 0.32 µg/µL–3.15 µg/µL. Detection of the bacterial DNA in EBC samples was also confirmed by the melting curve of qPCR amplification as shown in . As observed in the figure, most EBC samples had a peak at 68°C, the same as that of the positive control B. subtilis
. For a few different peaks observed, they might be the possible primer dimer (PD) from the PCR non-specific amplification process. In addition to the qPCR amplification of bacteria in EBC samples collected, DNA stain (AO method) was also performed and the results are shown in Figure S2
. As observed in the figure, both viable (green) and dead (yellow) were found in the EBC samples collected and the positive control B. subtilis
samples, while no cells were detected in the negative control. SEM images with different resolutions and agar plate culturing shown in also indicated that EBC samples (cultured) had various types of bacteria based on their morphologies and colony color. From SEM images, it can be estimated that most bacteria are in the range of 0.5–1.0 µm. According to total particle deposition curve developed by ICRP (1994) 
, more than 60% of bacterial particles of below 1 µm could be exhaled out. These smaller bacterial particles could remain airborne for a prolonged time period, thus playing an important role in airborne transmission of potential diseases. Results shown in , , , and indicate that high levels of bacterial aerosols were detected in the EBC samples collected, and the results on the other hand also implied that the developed device was efficient in collecting bacterial particles in the exhaled breath. These experimental data further confirm that exhaled breath is an important source of bacterial aerosols in the built environments.
Figure 6 Determination of total bacterial aerosols in EBC by qPCR; DNA standards (STD) used were 3.15, 3.15×101, 3.15×102, 3.15×103 ng/µl Bacillus subtilis DNA; Sample 1–7 represent EBC samples collected from seven human (more ...)
Figure 7 Dissociation curve of bacterial aerosols in EBC samples amplified by qPCR; Samples 1–7 were those collected from seven human subjects with their medical conditions listed in Table S2; Bacillus subtilis species was used as the positive control (more ...)
Figure 8 SEM images (different resolutions) of bacteria in EBC samples and images of colony forming units after culturing; the EBC samples were collected from human subjects and cultured using liquid broth overnight; different colored arrows point to likely different (more ...)
In this study, qPCR was also applied to detecting influenza A H3N2 viruses in EBC samples collected by the device. As observed in Figure S3
, H3N2 viruses were detected in the EBC sample collected from subject #3 with a Ct value of 28, while those for subject #1, #2 were shown below the detection limits. In addition, spiking viruses into the samples in general enhanced the overall qPCR signal as observed in Figure S3
. This on the other hand suggests no inhibition or amplification occurred when amplifying H3N2 viruses in EBC samples using qPCR. According to information shown in Table S2
, subject #3 had a fever, but no other information was available at the time of the experiment. In a previous study, it was indicated that use of the RTube for EBC collection had a very low viral detection rate (7%) compared to nasal swabs (46.8%) 
. Recently, a mask-like sampler was also tested and proved to be useful in detecting viruses using PCR in exhaled breath 
. It was indicated that airborne virus detection is difficult due to their low concentration and the presence of a wide range of inhibitors, thus optimized molecular biology should be performed to enhance their detection 
. Although the number of the subjects tested is limited here, the developed method, i.e., EBC collection and qPCR application, was demonstrated successful in detecting viruses from human exhaled breath. This would offer a non-invasive method for diagnosis of respiratory infections by using EBC. In the future, more patients should be tested with the EBC collection device developed here for viral detections.
Exhaled breath holds great promise for monitoring human health and for the diagnosis of various lung and systemic diseases, but analysis challenges remain due to the complex matrix of the breath 
. In this study, different from available devices restricted solely to condensation a simple and low cost EBC collection method using impaction and condensing was developed here for collecting bacteria and virus particles. An important advantage is the reusability of the collection device with a disposable hydrophobic film and an exhalation straw yet with a rapid EBC collection. This would offer the opportunity to collect EBC samples from a large number of subjects, especially during an influenza outbreak or a man-made bioterrorism event, within a shorter time frame. The developed EBC collection method was shown highly successful in detecting bacteria in EBC samples in a clinical setting. The developed EBC collection method was also shown applicable in detecting influenza viruses too. Experimental data here also suggest that exhaled breath, which was shown to contain smaller bacterial particles, could play an important role in airborne transmission of potential diseases. The collection efficiency of other substances including bio-markers (NO,CO, 8-isoprostane, hydrogen peroxide, nitrite, volatile organic compounds) using the developed method here is subject to further investigations. In addition, different exhalation modes should be also investigated with the method in collecting EBC. Besides, the dynamics of the air flow, mixing, and effects of temperatures and humidity, condensation, evaporation, growth of particles during the collection as well as the optimal straw length should be also investigated for improving the developed technique. Overall, our developed method here could be easily made available to a laboratory, and have impacts on current practice of EBC collection. Nonetheless, the reported work is a proof-of-concept demonstration, and its performance in non-invasive disease diagnosis such as bacterimia and virus infections needs to be further validated including effects of those influencing factors described.