The present results demonstrate that considerably more diverse microbial flora is observed from the same material by PCR and clone library sequencing compared to traditional cultivation. The fungal diversity in indoor dust by the former method was high, consisting of nearly 400 OTU types, 45% of which were identified at the species level. The theoretical diversity calculated by using nonparametric estimators was about 100 to 400 OTU per building per season, a richness comparable to that found in soil and plant organs by the same methodology (47
). In addition to the overall high diversity of fungi in both buildings, a constant increase in fungal diversity from winter through spring and summer to fall was observed. This is explained by the amplification and diversification of natural outdoor fungal source communities in accordance with their annual succession. With certain exceptions (39
), this kind of seasonal variation has not been observed in viable counts or diversity from house dust samples (28
). The present findings reflected the various sources of microbes in indoor environments. The largest diversity in cold seasons originated from the ustilaginous basidiomycete subclass Exobasidiomycetidae
, consisting mainly of Malassezia
yeasts that belong to human skin flora (51
). In general, winter samples were dominated by yeasts, and all other seasons were dominated by filamentous fungal species. Sequences belonging to or closely related to various phylloplane and saprotrophic fungi, including Aureobasidium pullulans
sp., Cladosporium herbarum
and other Cladosporium
sp., and Rhodotorula
sp. were observed in both buildings throughout the year, and these were among the most abundant fungi in the studied material. Also, several DNA sequences belonging to or related to phytopathogenic, mycorrhizal, and lichenous fungi in various subclasses were detected, emphasizing the strong influence of outdoor fungal flora on indoor microbiota. In the summer and fall samples the outdoor load of OTU belonging to decomposing filamentous basidiomycete species increased and finally crowded out some of the ascomycetous diversity in our clone libraries. This predictable phenomenon also explained most of the observed seasonal increase in the total diversity.
Except for Cladosporium
spp., which were repeatedly encountered in all clone libraries, few typical indoor air molds were observed in our clone libraries, although these were shown to be present in the material by cultivation techniques. According to the literature, the genera most commonly found in house dust by cultivation are Cladosporium
, and yeasts (11
). Moreover, Aureobasidium
, basidiomycetes, and nonsporulating or unidentifiable isolates have been frequently isolated from dust (11
). The Sphaeropsidales
, and Aspergillus
were the major components of culturable mycoflora in our material. Clone sequences originating from Alternaria
, and Fusarium
were found, but none of these was abundant or prevalent in the material. Sequences from Aureobasidium pullulans
were present in all clone libraries, and the genus was also present in substantial amounts in about half of the samples as determined by cultivation. Isolates placed as being Sphaeropsidales
group by culture were prevalent and were annotated to represent Phoma herbarum
sp. based on their ITS sequences. Corresponding sequences were also found in clone libraries in most samples.
In cultivation studies dustborne yeasts have not usually been characterized further, and yet their prevalence in most studies is high (11
). In our study, the cultivation results showed that yeasts were abundant especially in the reference building, but this difference between buildings was not discernible in clone libraries. Yeasts were predominant in winter samples in clone libraries but were present in considerable amounts during other seasons as well. In a cultivation study conducted in The Netherlands the predominant yeast species were Rhodotorula glutinis
, R. minuta
, R. mugilaginosa
, Cryptococcus albidus
, and C. laurentii
). Here, the yeast floras consisted of several species of the genera Candida
, and Trichosporon
, as well as the species Udeniomyces pannonicus
, Saccharomyces cerevisiae
, Debaryomyces hansenii
, and Phaeococcomyces nigricans
. In our study most of the yeast species were detected by clone library analysis only. Basidiomycetes were not quantified by cultivation in our study but accounted for half of the diversity and also about half of the total clone number in our libraries.
In addition to the above-mentioned and several other well-characterized species and groups, we observed a large variety of previously unknown sequence types originating from fungi that have not been cultivated or whose ITS sequences have not been deposited in public databases. These unknown sequence types occurred, on average, in lower counts in clone libraries and also in a smaller number of samples than the known species. The OTU similarities with known sequences were, on average, highest in the winter samples and gradually decreased toward fall. Similarity coefficients were, on average, higher between buildings than between seasons. This suggests that the main factor behind the observed high fungal diversity was the seasonally fluctuating outdoor fungal load.
In order not to miss part of the diversity, which was expected to be high on the basis of previous studies (29
), we used large-scale sequencing instead of screening clones or PCR products by methods such as DGGE or amplified ribosomal DNA restriction analysis. For the same reason, a strict 99% OTU threshold was used. The observed high diversity of fungi at various genetic levels supported these decisions. It was observed, however, that even the 99% similarity threshold may not be high enough to differentiate between some closely related species.
The amounts of ergosterol measured in the house dust samples in the present study are comparable to those reported by Saraf et al. (61
), who measured ergosterol concentrations of between 2 and 16.5 ng mg−1
in 17 floor dust samples. The highest ergosterol concentrations in both buildings were observed during the spring and summer seasons, which presumably reflects the influence of fungi originating from the outdoor air. The low concentrations observed during the fall were unexpected, since the effect of outdoor air should also have been seen then (39
), and abundant fungal diversity was readily visible in clone library results. The observed seasonal trend was supported by the summed qPCR spore counts of the assayed fungi. The amounts of culturable fungi also roughly followed the concentrations of ergosterol except for the summer sample of the reference building, which had low viable counts compared to the measured ergosterol level. The highest concentrations of culturable fungi in our study were observed during the spring. Evidently, the importance of the season on the culturable fungal counts of house dust is not straightforward, since the results of Koch et al. (39
) showed the highest concentrations of viable fungi during the summer months, especially August.
Differences between studied buildings.
In subarctic climates indoor sampling is generally recommended to be done during the cold season since the outdoor airborne spore load is then very low due to frozen soil, and the effect of indoor sources becomes most visible (57
). During winter, we observed an elevated ergosterol concentration and somewhat higher culturable diversity in dust sampled from the index building than in dust from the reference building. Using clone-library analyses an increased diversity and larger proportion of unknown fungi was observed for the index building compared to the reference building in winter, but—based on the knowledge of moisture-indicating species (59
)—the details of the clone data for fungal flora of winter samples did not clearly refer to a difference in the moisture conditions of the buildings. However, it is difficult to interpret the differences between buildings for such a small sample number.
In combined seasonal libraries the qualitative difference between the buildings was noticeable at the OTU level but became indeterminate when higher taxonomic groups or morphological groups were examined. Most OTU sporadically occurred in one or two seasons, and none of the observed prevalent species was unique to just one building. The observed difference in the fungal diversity between the two buildings nevertheless most likely relates to real differences in microbial source populations rather than to low sample coverages, since the highest number of common OTU in the buildings was observed in the fall when the diversity was high, library coverages were low, and the highest proportion of unknown species was present. The opposite was found in winter when the fungal flora was least similar between the buildings, the coverage values were highest, and the highest number of identifiable species was present.
Considering fungal species that might contribute to building related illnesses, some potentially toxic or allergenic fungi were found by ITS clone library sequencing, yet their occurrence was sporadic in most cases. Stachybotrys echinata
sp., and Mucor hiemalis
were found only in the index building. Stachybotrys echinata
, which is often found growing in wet cellulose materials, may produce cytotoxic trichothecene mycotoxins, as well as several griseofulvins (32
sp. has been associated with hypersensitivity pneumonitis (67
). Furthermore, Mucor hiemalis
is a common food spoilage mold that produces toxic ergoline alkaloids (41
). The potentially allergenic Alternaria tenuissima
) was detected from the reference building only.
Comparison of methods.
The observed fungal diversity was substantially higher in all seasons when analyzed by the direct cloning approach compared to the traditional culturing method. Using cloning, the number of OTU was on average seven times higher than the number of identified culturable fungal species or groups in the sample (see Table S1 in the supplemental material). The sequences of almost half of the cultivated fungal strains were found in clone libraries. In the majority of the cases, the morphological identification at genus level corresponded to the sequence analysis. The community composition observed by the two methods diverged profoundly both qualitatively and quantitatively. Species affiliated with subclasses Ustilaginomycetidae, Exobasidiomycetidae, Agaricomycetidae, Taphrinomycetidae, Saccharomycetidae, Pezizomycetidae, Lecanoromycetidae, and Erysiphomycetidae were only detected and identified by cloning, whereas subclasses Dothideomycetidae, Eurotiomycetidae, and Leotiomycetidae were accentuated in the cultured diversity, and sequences affiliated with Eurotiomycetidae were practically missing from clone libraries.
Based on cultivation results that were obtained in the beginning of the study and on previous experience with indoor dust, seven mold species or groups were picked for quantitative analysis by qPCR. The occurrence of these did not correlate well with either cultivation or clone library analysis except for Cladosporium cladosporioides
, which was found to be the most prevalent species with all of the methods used. The spore counts from the qPCR assay Cclad1 for C. cladosporioides
were 2 to 3 orders of magnitude higher than the culturable Cladosporium
counts. Comparable ratios have been encountered before in studies comparing viable counts with qPCR results in dust (45
) and viable counts in air samples using direct microscopic counts (70
). The Cclad1 assay probe matched with the major Cladosporium
isolate and two OTU sequences in clone libraries. According to the database comparisons, this assay lacks the ability to differentiate between Cladosporium
species C. cladosporioides
, C. tenuissimum
, C. cucumerinum
, and C. oxysporum
due to the high sequence similarity of these species' ITS regions.
High CFU counts of Penicillium
were observed using cultivation, but this group was not detected by qPCR. Here, the absence of assayed Penicillia
, as well as Aspergillus sydowii
, in qPCR was explained by single mismatches in the probe sequences compared to the actual sequences of cultivated strains of the named species. Single clones of Penicillium commune
and P. crustosum
were found in the clone libraries. The unexpected absence or rareness of Penicillium
, as well as some other fast-growing abundantly sporulating fungi has been encountered before in several studies utilizing different PCR primer sets with clone library sequencing (29
). It seems that, for some unknown reason, cloning methods detect these fungi in much lower numbers than do traditional methods. On the other hand, it is probable that cultivation methods overestimate the proportions of Penicillium
and other ubiquitous fast-growing fungi since on plate culture conditions these may compete by overgrowth with the more fastidious organisms (8
Concerning other species studied using qPCR, two of the quantified fungi, T. viride and E. amstelodami, showed high spore counts but were absent from all clone libraries, as well as plate cultures of most of the samples. When these species were encountered in culture, the viable counts were about 500 times lower than the qPCR result, which suggests low spore viability under the studied conditions. In our study setup, a long collection period, exposure to daylight, and desiccation during the collection probably affected the cultivation results of the more sensitive species. Comparison of the methods is also inherently affected by the fact that CFU often consist of chains of spores or clumps of conidia, whereas qPCR measures single conidia. In addition, dust can be inhomogeneous, and since the sample was divided into aliquots for the different analyses, the separate aliquots may have harbored slightly different flora, although the dust was homogenized before it was divided into aliquots. qPCR inhibition was observed in one sample (the index winter sample). This effect was not removed by further purification or dilution of the samples.
Although offering major advantages to microbial community analysis, DNA-based methods inevitably also bias the results. The profound differences in the relative contributions of diverse taxa such as Penicillium
and basidiomycetes to the dust mycoflora revealed by DNA- versus culture-based methods underscore the practical importance of the biases inherent in each of these methods. Technological factors such as the extractability of genomic DNA, the rDNA copy number, primer matching and performance, PCR amplification efficiency of the target region, and the cloning efficiency of the amplified PCR fragments may vary between microbial species and affect the OTU frequencies in clone libraries (75
). Besides differential amplification of target species, another considerable source of bias is artificial diversity mainly originating from PCR errors and chimeric PCR products (35
). In the present study PCR conditions were optimized to produce as small a bias as possible in this regard. The proportion of chimeric PCR products was significantly lower among sequences originating from simple PCR than from seminested PCR (0.02% versus 8.7%, respectively), a finding that is consistent with previous studies reporting a positive correlation between frequency of chimeric products and PCR cycle number (68
The Fun18Sf-ITS4 primer pair has been designed to be fungal specific in silico and in microbial pure cultures. It provides robust PCR amplification result from fungal pure cultures and environmental samples and has produced ITS libraries free of plant sequences when used in fungal community analysis of waste composting process (Hultman et al., unpublished data). In the present study, however, this primer pair amplified plant and, to a lesser extent, algal DNA. This behavior was unexpected and calls for reevaluation of the primer performance and required stringency conditions before the primer pair is further used in samples that harbor plant material.
The comparison of clone sequencing with qPCR and cultivation in the present study supported the idea that clone libraries do not reliably reflect the original frequencies of species in a sample. Nevertheless, the inherent differences of the methods we used complicated the comparison of the results. For instance, the comparison of the cultivation and library cloning methods with qPCR appeared partly irrelevant since the widely used U.S. Environmental Protection Agency qPCR probe set was not suitable for some of the local fungal strains. It is also possible that the ITS sequences in clone libraries are not always correctly identified by similarity searches against databases due to incomplete coverage of available reference sequences or due to falsely annotated reference sequences. For this, corresponding sequence analysis of cultivated strains and sequence alignment with clone sequences is crucial for comparing the results.
Despite the discussed methodological limitations, the present study supported the idea that the cloning and sequencing method offers a very sensitive and accurate tool, especially for the detection of rare or uncultivable fungi. This method may provide a useful account of total fungal diversity in the screening of poorly known environments. However, it is not suitable as a sole method for quantitative analyses of fungal communities.
To our knowledge, this is the first time that the fungal flora of indoor dust has been characterized using sequencing of rDNA clone libraries. Based on our data, the answers to the issues posed in the first part of the present study can be stated as follows. (i) The diversity of the fungal flora in indoor dust is wide, comprising many basidiomycetous species in addition to the ascomycetes found by cultivation in earlier studies. (ii) The fungal flora observed by the cultivation and clone sequencing methods differs markedly. Diversity was found to be wider and more evenly distributed between fungal groups using the clone sequencing. Uncultivable fungi and species whose detection would be challenging by cultivation were detected by clone analysis. On the other hand, both methods failed to detect some taxa readily detected by the other. (iii) qPCR did not support the findings in clone libraries or culture. This calls into question the reliability of characterizations of microbial populations based on a single method. (iv) Seasonal variation of fungal communities can be observed by clone analysis. Trends in the occurrence of fungal subclass level groups and morphological types were detected. The differences between the water-damaged index building and undamaged reference building occurred at the species level, but no fungi characteristic to either building were found. Drawing further conclusions about the relevance of the results in this respect would require more prominent differences between the buildings or a much larger data set.