Associations between dog keeping and indoor dust microbiota

Abstract

Living with dogs appears to protect against allergic diseases and airway infections, an effect possibly linked with immunomodulation by microbial exposures associated with dogs. The aim of this study was to characterize the influence of dog ownership on house dust microbiota composition. The bacterial and fungal microbiota was characterized with Illumina MiSeq sequencing from floor dust samples collected from homes in a Finnish rural-suburban (LUKAS2, N = 182) birth cohort, and the results were replicated in a German urban (LISA, N = 284) birth cohort. Human associated bacteria variable was created by summing up the relative abundances of five bacterial taxa. Bacterial richness, Shannon index and the relative abundances of seven bacterial genera, mostly within the phyla Proteobacteria and Firmicutes, were significantly higher in the dog than in the non-dog homes, whereas the relative abundance of human associated bacteria was lower. The results were largely replicated in LISA. Fungal microbiota richness and abundance of Leucosporidiella genus were higher in dog homes in LUKAS2 and the latter association replicated in LISA. Our study confirms that dog ownership is reproducibly associated with increased bacterial richness and diversity in house dust and identifies specific dog ownership-associated genera. Dogs appeared to have more limited influence on the fungal than bacterial indoor microbiota.

Figure 1

Box-plots of bacterial richness and Shannon entropy in dog and non-dog homes in two cohorts. Box-plots of bacterial richness (Chao1) in (A) LUKAS2 and (B) LISA, and Shannon entropy in (C) LUKAS2 and (D) LISA in house dust samples collected from dog homes (grey) and non-dog homes (white). In LUKAS2, additional comparison is made by samples taken during snow cover (snow) and without snow cover (no snow). p-values are from Mann–Whitney U-test in LUKAS2 and Weighted Mann–Whitney U-test in LISA. The boxplots present 5^th percentile, first quartile, median, third quartile, and 95^th percentile of the values.

Figure 2

Box-plots of fungal richness in dust samples from dog and non-dog homes in two cohorts. Box-plots of fungal richness (Chao1) in house dust samples collected from dog homes (grey) and non-dog homes (white) in (A) LUKAS2 and (B) LISA cohorts. In LUKAS2, additional comparison is made by samples taken during snow cover (snow) and without snow cover (no snow). p-values are from Mann–Whitney U-test in LUKA S2 and Weighted Mann–Whitney U-test in LISA. The boxplots present 5^th percentile, first quartile, median, third quartile, and 95^th percentile of the values.

Figure 3

Plots of two bacterial axes scores by dog houses stratified by snow cover in LUKAS2. Plots of the first (PCoA1) and the second (PCoA2) axes scores from bacterial weighted and unweighted UniFrac based Principal Coordinate Analyses in LUKAS2 cohort by houses with (black dots) or without (grey dots) dog(s). Percentages of the variance explained by the axis scores are in the parentheses. Top two plots were made using dust samples from the whole LUKAS2 cohort (excl. farmers), two middle plots with samples collected without snow cover, and two plots in the bottom with samples when snow was on the ground.

Figure 4

Bacterial taxa associated with dog and non-dog homes in LUKAS2 cohort, determined by ANCOM. Cladogram is additionally stratified by the sampling time: snow cover and no-snow cover. From the highest rank to the lowest, the major taxonomic ranks are domain (in the middle), phylum, class, order, family and genus (in the farthest layer). Genera that differ between the dog homes and non-dog homes are numbered in the figure. U. denotes “unassigned”, O. “other” genus within a family, and brackets indicate candidate taxonomy. Three datasets were created [(A) the whole cohort, (B) only samples, when snow was on the ground, and (C) only samples without snow on the ground] for this phylogenetic tree. If the top of triangle is pointing up, the relative abundance of the taxa is higher in dog homes. If the top of triangle is pointing down, the relative abundance of the taxa is lower in dog homes than non-dog homes. Yellow = the difference is found in all three datasets; black = the difference is seen only when the whole cohort is used (dataset A), but the difference disappears when data stratified by snow cover; light blue = the difference is only found when snow is on the ground (dataset B); dark blue = the difference is found in the whole cohort and when snow was on the ground (datasets A and B); light red = the difference is seen only without snow cover (dataset C); dark red = the difference is seen with the whole cohort and without snow cover (datasets A and C), but not when snow cover.