Development of a high-volume aerosol collection system for the identification of air-borne micro-organisms


Dr Gary L. Andersen, Biology and Biotechnology Research Program, Lawrence Livermore National Laboratory, L-441 PO Box 808, Livermore, CA 94551, USA (e-mail:


Aims: A high-volume aerosol collector was developed to efficiently capture airborne bacteria in order to assess levels of diversity in the air.

Methods and Results: Particulate matter was collected on a device designed to filter 1·4 × 106 litres of air in a 24 h period on a 1-lm pore size polyester membrane. Methods were optimized for extraction of genomic DNA from the air filter concentrate. Preparation times of 90 s with 0·5-0·05 mm diameter zirconia/silica beads yielded the highest concentration genomic DNA that was able to support PCR. A 24-h air sample was taken in Salt Lake City, Utah and the microbial composition was determined by the amplification and sequence analysis of 16S ribosomal DNA fragments.

Conclusions: Sequence analysis revealed a large diversity in the type of microbial species present including clones matching the sequence of Clostridium botulinum. The primary components of the aerosol sample included many different spore-forming bacteria as well as more fragile members of the Proteobacteria division.

Significance and Impact of the Study: The high-volume air collection and genomic DNA recovery system allows for the rapid detection of both cultivable as well as culture-resistant organisms in the environment.


Traditionally, cultivation has been the method of choice for studying bacteria in air (Lighthart and Shaffer 1995; Shaffer and Lighthart 1997; Chang et al. 2001). However, this approach offers just a glimpse of the biological agents present. It has been estimated that less than 1% of bacteria in any environment can be readily cultivated (Amann et al. 1995; Pace 1997; Hugenholtz et al. 1998). The remaining 99% includes bacterial species for which no current method exists for cultivation, and those that have entered a viable but non-cultivable (VBNC) physiological state. DNA-based detection systems allow all bacteria (live, dead, VBNC) in a sample to be examined regardless of the ability to cultivate them.

A high-volume air filtration system for sample collection was designed and DNA extraction protocols for the detection of air-borne micro-organisms were developed. Because of the relatively small amount of biological material in air (Hinds 1982), it is necessary to collect a volume sufficiently large to obtain a representative microbiological sample. The Lawrence Livermore National Laboratory (LLNL) high-volume air filter system routinely samples 1000 litres min−1 of air compared with the widely used all glass impinger sampler (AGI-30, Ace Glass, Vineland, CA, USA) that samples 12 litres min−1 of air (Alvarez et al. 1995). For each environmental sample, genomic DNA is extracted from a capture filter and purified. The small-subunit rRNA gene is amplified by PCR and used for identification of the air-borne micro-organisms. The sequences obtained by this method allow unambiguous identification of both cultivable and culture-resistant bacterial species in a sample. This detection system should be useful for public health departments and for hazard assessments such as monitoring for sick building syndrome.


Air sample collection

A specialized air-sampling device was developed by modification of a conventional high-volume ambient dust particle collector. Air samples were collected on 1 μm pore size 8 × 10 in track-etched Poretics polyester membrane filters (Osmonics, Westborough, MA, USA) in the LLNL high-volume filter unit (Fig. 1). The air sampler contained a three-stage, brushless motor blower with a maximum flow rate of 100 m3 h−1 (58 cfm) against the pressure drop caused by the membrane. The air sampler was typically run for 24 h at 60 m3 h−1, thus processing 1·4 × 106 l air filter−1. All particles > 1 μm were deposited on the surface of the flat membrane filter, which changed colour from white to dark brown from aerosolized microbes and inorganic particles. The air-flow resistance was minimized by selecting a relatively high pore density on the membranes. The flow rate decreased less than 10% in 24 h of air sampling, and was monitored with an exhaust venturi fixed to a dial manometer. The filter was collected in the field under partial suction of the motor blower, folded inward on itself, placed in double plastic bags and transported in a cooler. Half of the filter was stored at 4°C or archived at −20°C for later use. Organisms were washed from the other half filter and their DNA extracted.

Figure 1.

 Schematic diagram of the LLNL high-volume air sampler used for the collection of bioaerosols showing hinged cover, standard 20 × 25 cm filter holder, flow indicator, exhaust Venturi and motor blower. Arrows indicate direction of airflow

Particle extraction from filter

To extract the bacteria and debris, the half filter was cut into 20–30 strips, using sterile scissors and tweezers, which were placed in a 50 ml conical Falcon tube containing 45 ml phosphate buffer Tween solution (PBT: 0·003% Tween-20, 17 mmol l−1 KH2PO4 and 72 mmol l−1 K2HPO4). The Falcon tube containing the filter strips in buffer was vortexed horizontally for 1·5 min at maximum power (Well block combi-shaker; VWR International, West Chester, PA, USA) and sonicated at room temperature for 10 min on power level 9 (Aquasonic Model 75D; VWR International). The tube was vortexed for an additional 5 s and the suspension was poured into a clean Falcon tube. The wash was repeated with an additional 45 ml PBT to remove any residual material from the filter. Both sample washes were centrifuged for 30 min at 4°C, 3500 g (4000 rev min−1) with a Jouan CR422 centrifuge (Jouan, Winchester, VA, USA). The supernatant fluids were immediately drawn off and discarded. The pellets were transferred to 1·5 ml microcentrifuge tubes and centrifuged for 8 min at 16 000 g. The supernatant fluids were discarded and the pellets were recombined with the remaining supernatant fluid to give a total volume of 200 μl.

Genomic DNA extraction

Genomic DNA was extracted directly from the filter concentrate using the MoBio UltraClean Soil DNA kit™ (MoBio Laboratories, Solana Beach, CA, USA) according to the manufacturer's specifications with the following modifications. A silica bead mixture was prepared by mixing equal weights of 0·5 mm, 0·1 mm and 0·05 mm diameter zirconia/silica beads (Fisher Scientific, Pittsburgh, PA, USA), autoclaving for 45 min and then heating in an oven or heat block for a minimum of 1 h at 100°C to dry the beads fully. A 900 mg aliquot of the bead mix was dispensed into individual 2 ml screw-cap tubes. For each sample, 100 μl of the air-filter concentrate and 550 μl of the MoBio Bead Solution (12900-10-BS) were added to a tube containing the bead mix and vortexed for 5 s, together with the recommended volume of the lysis and inhibitor removal solutions supplied with the kit. For increased yield of genomic DNA, a Bio101 Fast Prep 120 machine (Qbiogene, Carlsbad, CA, USA) was used to perform the cell disruption of the microbes in the air-filter concentrate. It was empirically determined that homogenization of the tubes for 90 s at maximum speed was sufficient to recover DNA from a variety of organisms, and support PCR amplification of up to 1·5 kb targets. After centrifugation for 30 s at 10 000 g, 333 μl protein precipitating solution were added to 600 μl of the lysis supernatant fluid. The samples were then chilled for 5 min at 4°C and centrifuged for 1 min at 10 000 g. The supernatant fluid was decanted, mixed with a 2× volume of high salt DNA-binding solution (MoBio S3) and passed through the spin column provided in 700 μl aliquots. The column was washed once with an ethanol/high salt solution (MoBio S4) and then eluted by centrifugation with 50 μl 10 mmol l−1 Tris buffer (MoBio S5).

PCR amplification and sequencing

For amplification of eubacterial 16S rDNA, primers P3 mod (5′ ATTAGATACCCTDGTAGTCC3′) and PC5B (5′ TACCTTGTTACGACTT3′) were used as described previously (Wilson et al. 1990). These sequences amplify an approximately 721 bp product from the 3′ end of the 16S target. The rDNA amplified from the air-filter concentrate was cloned into pGEMT-Easy vector (Promega, Madison, WI, USA) and transformed into 25 μl electrocompetent DH10B cells (Life Technologies, Bethesda, MD, USA). Individual clones containing aerosol organism-specific rDNA fragments were purified using magnetic beads (Skowronski et al. 2000) and sequenced on an ABI3700; the resulting data were analysed using ABI Sequencing Analysis software version 3·2, then assembled and edited using Phred, and Phrap (Ewing and Green 1998; Ewing et al. 1998). Each assembled sequence was compared with sequences in the GenBank and the Ribosomal Database Project II (RDP) 8·1 (Maidak et al. 2001). Individual clones were named for the species of closest similarity based on alignment with RDP sequence.


The amount of microbial material observed and the concentration of genomic DNA recovered varied greatly among environmental samples. Microscopic cell counts were performed on dilutions of the air-filter concentrate using a Petroff-Hausser counting chamber (Hausser Scientific, Horsham, PA, USA). Large insoluble particles, pollen and micro-organisms were visible under the microscope in all samples. To ensure that just the microbes were being counted, only motile cells were tallied. The sludge of organisms and debris from each half filter typically contained between 108 and 1010 motile cells by this method. Dilutions of the filter concentrate were plated on King's medium B (King et al. 1954) agar plates containing 25 μg ml−1 benlate and 40 μg ml−1 natamycin to inhibit fungal growth. Viable colony counts were 1·6 × 106 cfu per half filter from a 24 h air sample taken between 14 October and 15 October, 2000 from Salt Lake City, Utah, while microscopic counts were 2·1 × 109 cells for the same half filter. Therefore, approximately 0·08% of the motile cells were able to be cultured.

Genomic DNA was extracted directly from the filter concentrate using a derivation of the MoBio UltraClean Soil DNA kit™. Various DNA extraction procedures were also evaluated, including other commercially-available DNA extraction kits, a crude lysis method using lysozyme-proteinase K (Wilson et al. 1994), chloroform/SDS extraction (Miller et al. 1999) and an iron bead capture magnetic separation method (Skowronski et al. 2000). The other extraction procedures resulted in inconsistent removal of PCR inhibitors or insufficient DNA recovery from air-filter concentrates (data not shown). To optimize recovery of DNA from the air samples further, tests were performed on endospores of Bacillus thuringiensis, vegetative cells of Escherichia coli and the air-filter concentrate. Recovery of high quality genomic DNA from a diverse set of micro-organisms requires a balance between maximizing the yield with efficient cell disruption and minimizing shear forces that will degrade DNA from more fragile cells. It was found that manipulation of the size and type of beads used, as well as the length of time in a bead mill homogenizer, had the greatest effect on the efficiency of amplification of ribosomal gene products from the different cell types. The smaller-sized zirconia/silica beads (0·5–0·05 mm) resulted in an increased DNA yield over the typically larger garnet beads that have been optimized for soil DNA extraction (data not shown). This may be due to the synergistic effect of naturally-occurring silica particles in the soil and their absence from air-filter concentrates.

Air filters yield as much as 400 ng genomic DNA to less than 50 ng of genomic DNA by gel quantification. However, even the DNA preparations that are undetectable on ethidium bromide-stained agarose gels have been used successfully as PCR templates. Figure 2 depicts the genomic DNA extracted from three sample preparations. The extraction methods worked well to consistently amplify rDNA products up to 1·5 kb from a number of air samples taken over the course of a year. Using air-filter concentrate, it was possible to consistently amplify a product with the bacterial specific primers P3mod and PC5B.

Figure 2.

 Genomic DNA from three air filter preparations loaded onto a 1·5% agarose gel. Lane 1 is a Gibco BRL 500 bp ladder (Life Technologies, Rockville, MD, USA). Lanes 2–4 contain 10 μl of the purified genomic DNA from three representative air filter concentrates. For each sample, around 106 litres of air were concentrated

Using the LLNL high-volume air sampler with the modified DNA extraction protocol, a diverse group of bacterial species, including a number of novel organisms, could be identified in various air samples. As an example, 3 μl out of 50 μl of the purified genomic DNA from the October 2000 air sample taken in Salt Lake City, Utah (see above) were amplified with bacterial primers P3 and PC5B. A total of 319 clones were sequenced and analysed to determine organism identity when possible. The most commonly identified subdivisions included Gram-positive, high G + C bacteria and the Bacillus, Lactobacillus and Streptococcus group (Table 1). Many clones also belonged to the Proteobacteria, including members of the α-Proteobacteria, β-Proteobacteria and γ-Proteobacteria groups (Table 1). Interestingly, three clones possessed a 99–100% sequence identity match to Clostridium botulinum. A full list of clone identifications, RDP phylogenetic codes and percentage match can be found at

Table 1.   Diversity of micro-organisms in a Salt Lake City, Utah air sample Thumbnail image of

There were 245 sequences (77% of clones) with over 93% identity to a known species sequence which were assigned that species name, while 74 sequences (23% of clones) had 93% or less homology to any sequence in the RDP. Analysis of the 74 sequences with Sequence Match version 2·7 from RDP (Maidak et al. 2001) revealed the organism with the most similar sequence from a taxonomic list for 67 of the clones. The remaining seven clones had a similarity score below 0·65 and represented novel, non-chimeric sequences. These unique clones most closely matched the δ-Proteobacteria subdivision, or the Cytophaga, Flavobacteria, Bacteriodes group, and are available at GenBank (AF403186–AF403192). The diversity of the organisms identified from this sample demonstrates the utility of this method in capturing and extracting DNA from a wide variety of microbial species.


The authors thank Anne Marie Erler for assistance in DNA sequencing, Ron Pletcher for air sample collection and Maria Marco for the King's broth agar plates. This work was performed under the auspices of the US Department of Energy by the University of California, Lawrence Livermore National Laboratory under Contract No. W-7405-Eng-48. This work was funded by the Chemical and Biological Non-Proliferation program NN-20 for the Department of Energy.