An Amplicon-Based Sequencing Approach for the Study of Aeromycology ^†

Introduction

Fungal spores are ubiquitous in indoor and outdoor air and their diversity and biomass quantity vary depending on the geographical location, environmental conditions, and the presence of sources such as hay, agricultural crops or growth on building material [1,2]. Fungal exposure can be explained by the presence of decaying materials like, peat, wood dust, manure, biosolids, and organic wastes like compost [3]. Waste treatment plants and dairy farms are locations with high fungal exposure [4,5,6]. The health effects linked to this exposure are likely underestimated due to the presence of undocumented fungi [7]. Culture methods are still widely used to describe fungal aerosols. However, there are inherent biases associated with these methods when applied to fungal diversity analyses because most fungal species are difficult to isolate from culture. Amplicon-based sequencing is a good alternative that offers a more detailed analysis of the microbial composition of aerosol samples due to the millions of sequences produced. Amplicon-based sequencing approaches depend on the critical choice of which DNA region is used as the barcode. Universal markers are based on different criteria such as their presence across taxa and sufficient sequence variation between taxa. For fungi, the internal transcribed spacer (ITS) region of rDNA is considered the best barcode for most fungal groups [8,9,10]. The ITS region is composed of three parts: ITS1-5.8S-ITS2. The limitations imposed by the sequencers in term of amplicon size forces the use of ITS1 or ITS2 to study fungal diversity of bioaerosols.

The objective of this work is to use air samples from waste treatment facilities and dairy farms to make a systematic comparison of the performance of ITS1 and ITS2 in determining the fungal diversity of bioaerosols.

Materials and Methods

Various air samples were collected: composting sites, biomethanization facilities and dairy farms. In each environment, the sampling sites were chosen according to where workers were most exposed to bioaerosols. Each composting plant treats different raw materials: household green waste (domestic), manure and hay (vegetal), and pig carcasses and placenta (animal). All of the composting plants were located in the province of Quebec, Canada. Samples were also collected from two biomethanization facilities. One facility processes primary and secondary sludge from wastewater treatment plants, as well as industrial waste. The second one handles municipal waste from domestic sources. Air samples were collected from five dairy farms in Eastern Canada during summer 2016. The buildings at each farm exhibited differences in building type and building characteristics (age, volume, ventilation), the number of animals presents (cows), the methods of milking (automatic or manual) and types of animal feed animals were given. For the three environments, and during each visit, three samples were taken: at the beginning, the middle and the end of work shifts.

A liquid cyclonic impactor Coriolis µ® (Bertin Technologies, Montigny-le-Bretonneux, France) was used for collecting air samples. The sampler was set at 300 L/min for 10 minutes (3m³ of air per sample) and placed within 1-2 meters of the source. The air flow in the sampler creates a vortex through which air particles enter the Coriolis cone and are impacted in the liquid. Fifteen milliliters of a phosphate buffer saline (PBS) solution with a concentration of 50 mM and a pH of 7.4 were used to fill the sampling cone.

A MoBio PowerLyser® Powersoil® Isolation DNA kit (Carlsbad, CA, U.S.A) was used to extract the total genomic DNA from the samples following the manufacturer’s instructions. Then, DNA was eluted in a 100µl MoBio buffer and the isolated DNA samples were stored at -20°C until subsequent analyses.

Amplification of the fungal ITS1 and ITS2 regions, equimolar pooling and sequencing were performed at the Genomic analysis platform (IBIS, Université Laval, Quebec City, Canada). Detailed information about the protocols used for MiSeq Illumina Sequencing is presented in the original study report (Mbareche et al., 2018. ITS1 or ITS2 for metabarcoding analyses of fungal aerosol populations? [unpublished materials]).

The bioinformatics workflow used to treat the sequences and analyze the diversity was based on the one previously described study [4].

Results and Discussion

In compost samples, 3 871 313 raw reads clustered into 1 208 OTUs (Operational Taxonomic Units) after quality filtering for ITS1, and 3 680 926 raw reads clustered into 772 OTUs for ITS2. In biomethanization samples, 675,642 raw reads clustered into 1142 OTUs and 730,688 sequences into 330 OTUs for ITS1 and ITS2, respectively. In dairy farm samples, 354,262 sequences led to 1015 OTUs and 310,362 sequences led to 218 OTUs for ITS1 and ITS2, respectively.

The ITS1 barcode produced significantly higher richness and diversity measures (Observed OTUs, Chao1, Shannon, Simpson) per sample compared to the ITS2 barcode in the three environments (data not shown). These results suggest ITS1 is a better diversity indicator than ITS2. The taxonomic profiles were analyzed more carefully to compare classes of fungi across samples according to the barcode used in each one of the three environments studied. The results presented in Figure 1 include all three environments combined. Of the 20 classes of fungi represented, Dothideomycetes, Eurotiomycetes, Saccharomycetes, Sordariomycetes and Agaricomycetes are the most dominant accounting for 90% of the total relative abundance. The most striking difference between ITS1 and ITS2 was in the class Saccharomycetes. It was 2.5 times more abundant in results from ITS2 (20%) compared to ITS1 (8%). Of the less abundant taxa, Wallemiomycetes, Exobasidiomycetes and Taphrinomycetes were detected only when the ITS1 region was used and Glomeromycetes, Tritirachiomycetes, Mucoromycotina, Rozellomycota and Lecanoromycetes were detected only by ITS2. Saccharomycetes were detected in all three environments when the ITS2 region was used while no Saccharomycetes were detected with the ITS1 region. This discrepancy may be linked to a 3’ terminal mismatch associated with ITS1 primers [11]. For example, the ITS1F forward primer specific to fungi is known to have mismatches with some classes of fungi including Chytridiomycota, Saccharomycetes and some genera of Dothideomycetes. The occurrence of introns between primer sites in many taxonomic groups of Ascomycota could also explain the difference in results obtained using the ITS1 and ITS2 regions [12,13,14].

Figure 2 presents a summary of the species that were consistently (in the three environments) more abundant when either ITS1 or ITS2 was used (based on Man-Whitney U test). Abundance comparisons for all of the species in Figure 2 are significant at P<0.005. The fungal species that were found exclusively when used with either the ITS1 or ITS2 barcode should be examined more closely when designing a study of fungal diversity in bioaerosols. The differential abundance results presented herein are not exhaustive. All of the information is presented in the supporting material associated with the original study report (Mbareche et al., 2018. ITS1 or ITS2 for metabarcoding analyses of fungal aerosol populations? [unpublished materials]). Primer mismatch could explain the potential biases linked to the taxa identified only by ITS1 or ITS2 but not both [15,16].

Other factors may impact the performance of ITS1 and ITS2 as fungal barcodes in amplicon-based sequencing studies. For example, the GC content is known to have an effect on PCR and sequencing efficiencie [17]. While the GC content was not addressed in this work, a recent study used sequences from an ITS database to examine the GC content of ITS1 and ITS2 sequences. ITS1 had significantly lower GC content than ITS2 in the taxa studied which may be advantageous for amplification and sequencing compared to ITS2 [18]. Bioinformatics analyses can also impact diversity analyses depending on the clustering algorithms, the percent identity threshold and taxonomy assignment tools (BLASTn vs Naïve Bayesian Classifier). In this work, the performances of ITS1 and ITS2 were evaluated using the same bioinformatics tools in order to avoid adding any further biases to the diversity analyses.

Conclusions

The results presented in this conference paper are a short summary of a more exhaustive analysis comparing ITS1 and ITS2 barcodes in air samples from different environments as well as an analysis of the use of the culture approach to describe fungi in bioaerosols. The goal of this work is to offer aerosol scientists a guide for designing studies to address the fungal populations in aerosols using molecular approaches. Although ITS1 outperformed ITS2 in richness measures, the results obtained suggest that neither one of the barcodes evaluated is perfect in terms of distinguishing all species. Using both barcodes offers a wider view of the fungal aerosol population present. However, we strongly recommend the use of the ITS1 region as a universal fungal barcode for quicker general analyses of diversity and when limited financial resources are available.