High-throughput sequencing-based identification of airborne Aspergillus species and their triazole susceptibilities

Abstract

Aspergillus is a common fungus presenting everywhere in our environments. The genus Aspergillus currently includes 339 diverse species in 19 sections of Aspergillus, of which about 50 species are known as allergenic and/or pathogenic for human. The pathogenicity of this fungus mostly depends on species-specific characteristics. For instance, two most prevalent pathogenic Aspergillus species, A. fumigatus and A. terreus, produce the smallest conidia that can reach the terminal alveoli to cause pulmonary infection whereas the other pathogenic species producing larger conidia choose the larger routes of infection in our bodies, such as mouth and nose. Plus, the concentrations of airborne Aspergillus species have varied by season and by location. Therefore, from an environmental health point of view, to appropriately understand the properties and the compositions of airborne Aspergillus species is important in our environment with various conditions. Though, the difficulty in to differentiating the closely related species of Aspergillus from environmental specimens presents when high-throughput sequencing-based method is adapted to sensitively detect unculturable and low abundant fungal species together due to an insufficient variability of targeted marker and a potential inaccuracy of reference sequence databases used. In addition, to diagnose the antifungal susceptibility levels of airborne pathogenic Aspergillus species has been difficult owing to a current inconvenience of isolating tremendous amounts of environmental fungal colonies collected by culture plate-based air samplings. To this end, the goal of this study is to facilitate to identify and to measure the concentrations of airborne Aspergillus species and their levels of triazole susceptibilities. Thus the specific aims include to monitor the most powerful antifungal agent, triazole, susceptibilities in airborne Aspergillus species with triazole-containing selective sampling medium (Chapter II), to characterize the accuracy of high-throughput sequencing-based method to identify the species of Aspergillus when adopting alternative markers analysis and curated databases (Chapter III), and to apply this method to identify and to estimate the abundances of airborne Aspergillus species in our environment (Chapter IV). In Chapter II, I estimated the concentration of triazole resistant fungi and Aspergillus in outdoor air of Seoul Capital Area. Although emerging triazole resistant fungi are a concern with the increased use of environmental triazoles, little is known about the levels of triazole susceptibility in outdoor airborne fungi and Aspergillus making it difficult to assess the risks of inhalation exposure to airborne, antifungal-resistant pathogenic Aspergillus. To this end, the impactor air sampling with triazole-containing nutrient agar plates was performed to selectively screen for airborne fungal isolates based on their triazole susceptibilities. The study estimated that 0.17% of all the culturable fungi belong to the pathogenic thermotolerant taxa, among which each isolate of Aspergillus niger and Aspergillus tubingensis showed a minimum inhibitory concentration (MIC) of 2 μg/mL or greater for a clinical triazole, itraconazole, with another pathogenic fungal species, Paecilomyces variotii. The results confirmed the presence of airborne pathogenic species of Aspergillus with high MICs for itraconazole in ambient air. Though vigilance was still required to estimate the relative abundance of these and other pathogenic Aspergillus species in the air due to the limitation of detecting non-culturable and low abundant fungal communities with this culture-based method. In Chapter III, I characterized the accuracy of high-throughput amplicon sequencing to identify the species within the genus Aspergillus. Although high-throughput sequencing has been introduced for fungal ecology studies to sensitively detect target organisms from samples with low abundance such as clinical specimens or air filter samples without requiring culture process, the short amplicons typically used for high-throughput sequencing may result in inaccurate taxonomic assignments due to the limited information available for DNA markers. To maximize the advantage of using high-throughput sequencing to identify pathogenic Aspergillus species in environmental specimens with more accuracy, three DNA markers, the internal transcribed spacer 1 (ITS1), β-tubulin (BenA), and calmodulin (CaM), were sequenced in this study from eight reference Aspergillus strains with known identities using 300-bp sequencing on the Illumina MiSeq platform. The identifications with sequences longer than 250 bp were accurate at the section rank, with some ambiguities observed at the species rank mostly due to cross detection of sibling species. Additionally, in silico analysis was performed to predict the identification accuracy for all species in the genus Aspergillus, where 107, 210, and 187 species were predicted to be identifiable down to the species rank based on ITS1, BenA, and CaM, respectively. The genus-rank identification of Aspergillus was successful to classify more than 99% of Aspergillus species as the genus Aspergillus in silico. Except A. terreus having significant intra-specific variability, the identifiable Aspergillus species defined in silico had more than 96.1, 97.9 and 99.9% genus-rank accuracy

95.4, 98.3 and 99.5% section-rank accuracy

95.4, 97.7 and 99.4% species-rank accuracy in vitro with ITS1, BenA and CaM sequinning, individually. The results were reproducible across biological duplicates both at the species- and section-rank, but not strongly correlated between ITS1 and BenA, in air filter samples, suggesting the Aspergillus detection can be taxonomically biased depending on the selection of the DNA markers and/or primers. In Chapter IV, I used the high-throughput sequencing-based method to study seasonal abundances and compositions of airborne Aspergillus species in the Seoul Capital Area. The detections of Aspergillus species from air filter samples were found to be biased dependent on the selection of the DNA markers. For instance, some sections, such as Raperi, Versicolores, Restricti, Aspergillus, Circumdati, and Ochraceorosei were preferentially detected by ITS1 (ITS1/BenA >1)

other sections, such as Fumigati, Falvi, Nidulantes, Candidi, Cremei, Sparsi, Cervini, Terrei, Clavati and Aenei were preferentially detected by BenA (ITS1/BenA <1) in 18 air filter samples. By combining ITS1 and BenA, however, a total of 16 pathogenic Aspergillus species were detected while 9 and 8 species were detected only by ITS1 and BenA, respectively, highlighting the complementarity of the two markers in detecting pathogenic Aspergillus species from ambient air. The study found A. fumigatus as the most abundant pathogenic species with 89.6% and 98.3% of relative abundances based on ITS1 and BenA, respectively. Additionally, the taxon-specific concentrations were determined by multiplying relative abundances of Aspergillus species based on the high-throughput sequencing and total fungal concentrations measured by quantitative PCR. The concentration of total pathogenic Aspergillus species was highest in June at the campus but in October at the farm. The concentrations were about 4 times higher at the farm (85 GCN m-3) than at the campus (21 GCN m-3) on average. The concentrations estimated in gene copy numbers (GCN m-3) are not comparable to the doses calculated in culture-based method (CFU m-3). However, as the measurement of the method combining the high-throughput sequencing-based method established in this study and qPCR was reproducible with the defined markers and primers with higher identification accuracy for the species of Aspergillus, the method further can be utilized to conveniently and accurately understand the communities of Aspergillus in a large environmental sample set collected in different locations and in different seasons. In conclusion, this study characterized seasonal compositions and concentrations of airborne pathogenic Aspergillus species and their triazole susceptibilities. The study detected the airborne pathogenic Aspergillus strains with high MICs of a medical triazole of itraconazole from outdoor air of the Seoul Capital Area in South Korea, implying risks of inhalation exposures to triazole-resistant fungal pathogens in general environments. Owing to the abundances of airborne Aspergillus species with high MICs of itraconazole being detected from ambient air, the study further explored the possibility of using the high-throughput sequencing to identify pathogenic species within the genus Aspergillus. In vitro and in silico analyses showed that the high-throughput sequencing was useful to identify Aspergillus species with the ITS1, BenA, and CaM sequences longer than 250 bp. By combining this high-throughput sequencing-based method established in this study that is reproducible with the defined markers/ primers with higher accuracy for the species identification of Aspergillus, and the conventional PCR-based quantification method, more pathogenic Aspergillus species were detectable and quantified with higher accuracy in air filter samples. As the airborne Aspergillus species concentrations estimated by location and season have had different patterns, the method established in this study is thought to be useful to conveniently and accurately understand the communities of Aspergillus in a large environmental sample set collected in different locations and in different seasons.