BAL fluid samples were collected from patients undergoing bronchoscopy procedures for lung mass evaluation at Severance Hospital, Seoul, Korea from May to November 2015. The bronchoscope was inserted through the mouth, and BAL was collected according to a standardized protocol at the opposite side of the lung mass; 10 mL BAL fluid was acquired from the patients with approximately 30 mL sterile 0.9% saline. Sample was collected in accordance with the Institutional Review Board of Severance Hospital (IRB No. 4-2014-1014); informed consent was obtained from the patients.

BAL fluid samples were cultured on the day within 6 hours of collection. Undiluted BAL fluid (100 µL) was inoculated onto an agar plate using a spreader. Different agar plates with aerobic and anaerobic conditions were used: blood agar (aerobic, 35℃, 5% CO₂), chocolate agar (aerobic, 35℃, 5% CO₂), MacConkey agar (aerobic, 35℃, 5% CO₂), Brucella agar (anaerobic, 35℃), phenylethyl alcohol agar (anaerobic, 35℃), and Sabouraud dextrose agar (aerobic, 30℃, ambient air). Bacterial colonies were isolated after 48 hours of culture. Sabouraud dextrose agar (SDA) plates used for fungal culture were observed for up to four weeks of incubation.

An average of 20 colonies with different morphologies and sizes were selected from each agar plate and applied to a Microflex LT matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) system with Biotyper software 3.1 (Bruker Daltonics, Bremen, Germany) for simultaneous identification and subculture. When an accurate identification was not available (score <1.7) using only α-cyano-4-hydroxycinnamic acid, formic acid was added. Strains that could not be reliably identified by MALDI-TOF MS were subjected to further analysis; the subcultured colonies were analyzed using 16S rRNA sequencing for bacteria and ITS sequencing for fungi performed with primers 27F-1492R and ITS4, respectively (Table 1). To detect Staphylococcus spp., Klebsiella spp., and Enterobacter spp. in samples from which bacterial species were isolated solely on culture plates, we performed PCR amplification using organism-specific primers as previously described [121314] (Table 1).

DNA was extracted from BAL fluid within 6 hours of sample collection by using the DNeasy Blood & Tissue kit (Qiagen, Venlo, Limburg, Netherlands), according to the manufacturer's protocol. DNA extraction was performed simultaneously with bacterial culture. The DNA samples were stored at –70℃ until further analysis. The 16S rRNA gene from each DNA sample was amplified using barcoded fusion primers targeting the V1 to V3 regions (27F-518R), because this region has been reported to be reliable for accurate taxonomic classification of bacterial sequences [15] (Table 1). Amplification was carried out under the following conditions: initial denaturation at 95℃ for 5 minutes; followed by 30 cycles of denaturation at 95℃ for 30 seconds, primer annealing at 55℃ for 30 seconds, and extension at 72℃ for 30 seconds; and a final elongation step at 72℃ for 7 minutes. PCR products were confirmed using 2% agarose gel electrophoresis and visualized with a Gel Doc system (BioRad, Hercules, CA, USA). The amplified products were purified using the QIA-quick PCR purification kit (Qiagen). Equal concentrations of purified products were pooled together, and short fragments (non-target products) were removed using the AMPure beads kit (Agencourt Bioscience, Beverly, MA, USA). Quality and product size were assessed on a Bioanalyzer 2100 (Agilent, Palo Alto, CA, USA) using a DNA 7500 chip. Mixed amplicons were obtained by emulsion PCR and then deposited on picotiter plates. Sequencing was carried out at Chunlab, Inc. (Seoul, Korea) using the GS Junior Sequencing system (Roche, Branford, CT, USA) according to the manufacturer's instructions. Reads obtained from different samples were sorted using the unique barcodes on each PCR product. The barcode, linker, and primer sequences were removed from the original sequencing reads. Any reads containing two or more ambiguous nucleotides, low-quality score (average score <25), or reads shorter than 300 bp were discarded. Potential chimera sequences were detected using the UCHIME algorithm [16]. Following removal of the chimera sequences, the taxonomic classification of each read was assigned against the EzTaxon-e database (http://eztaxon-e.ezbiocloud.net) [17].

The operational taxonomic units (OTUs) identified at the genus level were listed with the number of reads, reflecting the relative frequency of detection. Next, the number of cultured colonies was compared with the number of OTUs determined by NGS in the same sample to analyze the difference in identification by NGS and culture.

A total of 2,272 microbial colonies were isolated from 27 BAL fluid samples and identified using culture and MALDI-TOF MS. The number of genera identified per sample ranged from 4 to 14, with an average of 8. The total number of genera identified was 29; these included 85 species. Streptococcus spp. were the predominant isolates with a total number of 1,213 colonies (Table 2). Culture on blood and MacConkey agar resulted in the highest and the lowest number of microorganisms, respectively. Forty (1.76%, 40/2,272) colonies were not identified by MALDI-TOF MS. These colonies were subcultured and re-identified using MALDI-TOF MS with the aid of formic acid, yielding mainly Streptococcus spp. (47.5%, 19/40). Four colonies (0.18%, 4/2,272) could not be identified with repeated MALDI-TOF MS; these were identified as Actinomyces odontolyticus, Actinomyces gravenitzii, Fusobacterium periodonticum, and Aspergillus fumigatus by sequencing using 16S rRNA and ITS gene amplification.

Sequencing using the GS Junior Sequencing system generated 10,332 valid reads per sample on average (range 6,761–19,764). At the genus level, a total of 288 OTUs were identified with a mean of 55.8 (range 27–78) per sample. Considering those with a relative abundance greater than 1%, the mean number of OTUs identified at the genus level was 10.0 (range 6–17) per sample (Fig. 1), which is comparable to the number of bacterial genera identified using conventional culture and MALDI-TOF MS. Prevotella spp. was the most frequent and abundant OTU identified at the genus level; it was identified in all 27 samples (relative abundance 6.5–54.5% with a mean of 29.2%). The next most prevalent OTUs (mean relative abundance) were Streptococcus (11.5%), Neisseria (10.5%), Haemophilus (10.1%), and Veillonella (9.1%).

NGS revealed more OTUs compared with the number of genera identified by culture (Fig. 1). The relative abundance of bacterial reads identified simultaneously by culture was 47.1% on average, ranging from 3.2% to 88.0% in the 27 samples. The bacteria present in BAL fluids according to the identification method (NGS, culture, or both) are shown in Fig. 2. Of the 156 bacterial genera identified solely by NGS with >1% relative abundance, Prevotella spp. was the most common (14.1%) and frequent (22 of 27 samples, 81.5%) genus. When considered as relative abundance, the genus Prevotella spp. comprised over 50% of the bacteria that were sequenced but not cultured. Interestingly, some bacteria were identified solely by culture; of these, Staphylococcus spp., Clostridium spp., and Enterobacteriaceae including Klebsiella spp. were common, comprising 66.7% (18/27) of the 27 samples and 56.4% (22/39) of the total number of colonies that were cultured but not sequenced (Table 3). We per-formed organism-specific PCR using the primers detailed in Table 1 to detect Staphylococcus spp., Klebsiella spp., and Enterobacter spp. All PCRs failed to amplify the target organisms in the samples, suggesting DNA instability in clinical samples. Candida and Aspergillus spp. were only cultured because they cannot be identified by 16S rRNA PCR.