Characterization and quantification of the fungal microbiome in serial samples from individuals with cystic fibrosis

Abstract

Background: Human-associated microbial communities include fungi, but we understand little about which fungal species are present, their relative and absolute abundances, and how antimicrobial therapy impacts fungal communities. The disease cystic fibrosis (CF) often involves chronic airway colonization by bacteria and fungi, and these infections cause irreversible lung damage. Fungi are detected more frequently in CF sputum samples upon initiation of antimicrobial therapy, and several studies have implicated the detection of fungi in sputum with worse outcomes. Thus, a more complete understanding of fungi in CF is required.

Results: We characterized the fungi and bacteria in expectorated sputa from six CF subjects. Samples were collected upon admission for systemic antibacterial therapy and upon the completion of treatment and analyzed using a pyrosequencing-based analysis of fungal internal transcribed spacer 1 (ITS1) and bacterial 16S rDNA sequences. A mixture of Candida species and Malassezia dominated the mycobiome in all samples (74%-99% of fungal reads). There was not a striking trend correlating fungal and bacterial richness, and richness showed a decline after antibiotic therapy particularly for the bacteria. The fungal communities within a sputum sample resembled other samples from that subject despite the aggressive antibacterial therapy. Quantitative PCR analysis of fungal 18S rDNA sequences to assess fungal burden showed variation in fungal density in sputum before and after antibacterial therapy but no consistent directional trend. Analysis of Candida ITS1 sequences amplified from sputum or pure culture-derived genomic DNA from individual Candida species found little (<0.5%) or no variation in ITS1 sequences within or between strains, thereby validating this locus for the purpose of Candida species identification. We also report the enhancement of the publically available Visualization and Analysis of Microbial Population Structures (VAMPS) tool for the analysis of fungal communities in clinical samples.

Conclusions: Fungi are present in CF respiratory sputum. In CF, the use of intravenous antibiotic therapy often does not profoundly impact bacterial community structure, and we observed a similar stability in fungal species composition. Further studies are required to predict the effects of antibacterials on fungal burden in CF and fungal community stability in non-CF populations.

Figure 1

Characterization of the fungal communities of sputum samples from CF subjects. (A) Fraction of pyrosequencing reads assigned to each of the top four fungal taxa detected in CF subjects immediately after an exacerbation and before starting an antibacterial therapy (pre) and approximately 2 weeks afterwards (post) or while hospitalized (inter). The legend indicates the color assigned to each indicated fungal species or genus (Malassezia sp. unknown). (B) Relative abundance of all fungal genera, extended to the species level for the clinically important Candida species (colored dots), is plotted against the prevalence of each genus and Candida species in the six CF subject samples (pre- and post-treatment) and show that fungi that are highly abundant in a single patient are also highly prevalent across patients. The colored dots indicate the Candida species and the genus Malassezia that are both highly abundant and highly prevalent and the empty dots represent all remaining fungal genera identified in our study.

Figure 2

Relatedness of microbial communities. Box-and-whisker plots of pairwise Bray-Curtis distances of the mycobiome and microbiome of cystic fibrosis (CF) subjects. (A) Fungal samples from unrelated subjects on different treatments (UNR) are no more different than samples from unrelated subjects on the same treatments (STX). Mycobiome samples from the same subject (SPT) were significantly more similar to each other (P <0. 001) by a Tukey’s honest significant test than mycobiome samples from the same treatment group, suggesting that patients’ fungal communities are specific to patients and remain relatively stable during treatment. (B) Bacterial genera from the same subject (SPT) were marginally more similar than unrelated subjects on different treatments (UNR) or unrelated subjects on the same treatments (STX) but these differences were not significant by a Tukey’s honest significant test.

Figure 3

Characterization of the bacterial communities of sputum samples from CF subjects. Fraction of reads assigned to each of the top 13 bacterial genera detected in CF subjects before treatment (pre), during treatment (inter), or after completion of treatment (post).

Figure 4

Comparison of fungal and bacterial richness within samples. Scatter plot of the subsampled mean normalized number of bacterial genera vs. the number of fungal genera found in the sputum of CF subjects. Neither treatment nor high numbers of bacterial or fungal genera have an obvious impact on each other. The pre-treatment samples are depicted as circles, the intermediate (inter) samples collected during treatment shown as triangles, and the post-treatment samples are depicted squares. The colors represent the different subjects.

Figure 5

Comparison of fungal communities in the same subject. Distribution of fungal taxa detected in CF subjects either immediately after an exacerbation and before starting an antibacterial therapy (pre), approximately 2 weeks afterwards (post) or while hospitalized (inter) presented in Euler diagrams for pre-post samples (subjects #1–6) and Venn diagrams for pre-inter-post samples (subjects #8 and 9) diagrams. The numbers in parenthesis describe the total number of taxa detected in a sample; the numbers in the circles represent either the unique number of taxa in a sample or the number of shared taxa in the overlap regions. The numbers in brackets represent the total number of reads for each sample.

Figure 6

Quantification of fungal and bacterial burden in CF subjects. (A) Quantification of the fungal burden of sputum samples from CF subjects. Total fungal burden was determined by amplifying the 18S rDNA locus by qPCR. (B) Quantification of the bacterial burden of sputum samples from cystic fibrosis subjects. Total bacterial burden was determined by amplifying the 16 s rRNA locus by real-time qPCR.