Parallel Analysis of Cystic Fibrosis Sputum and Saliva Reveals Overlapping Communities and an Opportunity for Sample Decontamination

Abstract

Culture-independent studies of the cystic fibrosis (CF) airway microbiome typically rely on expectorated sputum to assess the microbial makeup of lower airways. These studies have revealed rich bacterial communities. There is often considerable overlap between taxa observed in sputum and those observed in saliva, raising questions about the reliability of expectorated sputum as a sample representing lower airway microbiota. These concerns prompted us to compare pairs of sputum and saliva samples from 10 persons with CF. Using 16S rRNA gene sequencing and droplet digital PCR (ddPCR), we analyzed 37 pairs of sputum and saliva samples, each collected from the same person on the same day. We developed an in silico postsequencing decontamination procedure to remove from sputum the fraction of DNA reads estimated to have been contributed by saliva during expectoration. We demonstrate that while there was often sizeable overlap in community membership between sample types, expectorated sputum typically contains a higher bacterial load and a less diverse community compared to saliva. The differences in diversity between sputum and saliva were more pronounced in advanced disease stage, owing to increased relative abundance of the dominant taxa in sputum. Our effort to model saliva contamination of sputum in silico revealed generally minor effects on community structure after removal of contaminating reads. Despite considerable overlap in taxa observed between expectorated sputum and saliva samples, the impact of saliva contamination on measures of lower airway bacterial community composition in CF using expectorated sputum appears to be minimal.IMPORTANCE Cystic fibrosis is an inherited disease characterized by chronic respiratory tract infection and progressive lung disease. Studies of cystic fibrosis lung microbiology often rely on expectorated sputum to reflect the microbiota present in the lower airways. Passage of sputum through the oropharynx during collection, however, contributes microbes present in saliva to the sample, which could confound interpretation of results. Using culture-independent DNA sequencing-based analyses, we characterized the bacterial communities in pairs of expectorated sputum and saliva samples to generate a model for "decontaminating" sputum in silico Our results demonstrate that salivary contamination of expectorated sputum does not have a large effect on most sputum samples and that observations of high bacterial diversity likely accurately reflect taxa present in cystic fibrosis lower airways.

Keywords: airway infection; microbial community dynamics; microbiome.

FIG 1

Bacterial load (16S rRNA gene copies/ml) in sputum and saliva samples. (A) In the aggregate, bacterial load was greater in sputum samples than in saliva samples, although this difference failed to reach statistical significance (linear mixed model, P = 0.054). (B) For 25 of 37 pairs, the bacterial load of the sputum sample was greater than that of the corresponding saliva sample. Sample pairs collected during antibiotic treatment are marked with a “T.” (C) Bacterial load of all samples (37 saliva and 37 sputum samples) was not significantly different between subjects with early or early/intermediate disease stage (n = 15 sample pairs from 3 subjects) and subjects with intermediate, intermediate/advanced, or advanced disease stage (n = 22 sample pairs from 7 subjects) (linear mixed model, P = 0.655). The top and bottom boundaries of each box indicate 75th and 25th quartile values, respectively; black lines inside each box represent the median values; solid black circles inside each box represent the mean values.

FIG 2

Bacterial community richness (number of observed OTUs) in sputum and saliva samples. (A) In the aggregate, bacterial richness was greater in saliva samples than in sputum samples (paired t test, P = 0.021). (B) For 25 of 37 pairs, the bacterial richness of the saliva sample was greater than that of the corresponding sputum sample. Sample pairs collected during antibiotic treatment are marked with a “T.” The vertical black bars at 0 indicate samples where sputum and saliva samples have equal richness. (C) Richness of all samples (37 saliva and 37 sputum samples) was not significantly different between subjects with early or early/intermediate disease stage (n = 15 sample pairs from 3 subjects) and subjects with intermediate, intermediate/advanced, or advanced disease stage (n = 22 sample pairs from 7 subjects). The top and bottom boundaries of each box indicate 75th and 25th quartile values, respectively; black lines inside each box represent the median values.

FIG 3

Relative abundance of taxa observed in 37 pairs of expectorated sputum and saliva samples collected from 10 subjects, showing individual OTUs with an average relative abundance >1% across all samples. “Other” taxa combine remaining OTUs in each sample. Disease stage is indicated above each subject’s plots.

FIG 4

Community similarity of 37 pairs of expectorated sputum and saliva. (A) Community structures were significantly more similar (theta-YC similarity) in sample pairs from subjects with early or early/intermediate disease stage (n = 15 sample pairs from 3 subjects) than among subjects with intermediate, intermediate/advanced, or advanced disease stage (n = 22 sample pairs from 7 subjects) (t test, P = 0.007). (B) Sample pairs from subjects with later-stage disease tended to have lower Jaccard similarity than subjects with earlier-stage disease, although this difference was not statistically significant (t test, P = 0.105). (C) In sample pairs from subjects with later-stage disease, the relative abundance of the dominant (Max) OTU was significantly greater in sputum than in the respective saliva sample (paired t test, P < 0.001 [later-stage disease group] and P = 0.2386 [earlier-stage disease group]). The top and bottom boundaries of each box indicate 75th and 25th quartile values, respectively; black lines inside each box represent the median values.

FIG 5

Illustration of in silico sputum decontamination procedure using a hypothetical pair of sputum and saliva samples.

FIG 6

(A) Theta-YC based PCoA of bacterial community structures in sputum from five subjects who each provided five pairs of samples. Sputum before (closed circles) and after (open circles) in silico decontamination procedure is shown for each sample (before/after decontamination sputum pairs are connected by lines). Subjects are depicted in different colors. Fifty-nine technical replicates (generous donor controls) are shown as open gray triangles. (B) Using the same ordination data as in panel A, subjects and controls are visualized separately with the axes for each plot set so that the samples fill the ordination space. As a result, the scales between plots are not comparable.