Comparative analyses of the bacterial microbiota of the human nostril and oropharynx

Abstract

The nose and throat are important sites of pathogen colonization, yet the microbiota of both is relatively unexplored by culture-independent approaches. We examined the bacterial microbiota of the nostril and posterior wall of the oropharynx from seven healthy adults using two culture-independent methods, a 16S rRNA gene microarray (PhyloChip) and 16S rRNA gene clone libraries. While the bacterial microbiota of the oropharynx was richer than that of the nostril, the oropharyngeal microbiota varied less among participants than did nostril microbiota. A few phyla accounted for the majority of the bacteria detected at each site: Firmicutes and Actinobacteria in the nostril and Firmicutes, Proteobacteria, and Bacteroidetes in the oropharynx. Compared to culture-independent surveys of microbiota from other body sites, the microbiota of the nostril and oropharynx show distinct phylum-level distribution patterns, supporting niche-specific colonization at discrete anatomical sites. In the nostril, the distribution of Actinobacteria and Firmicutes was reminiscent of that of skin, though Proteobacteria were much less prevalent. The distribution of Firmicutes, Proteobacteria, and Bacteroidetes in the oropharynx was most similar to that in saliva, with more Proteobacteria than in the distal esophagus or mouth. While Firmicutes were prevalent at both sites, distinct families within this phylum dominated numerically in each. At both sites there was an inverse correlation between the prevalences of Firmicutes and another phylum: in the oropharynx, Firmicutes and Proteobacteria, and in the nostril, Firmicutes and Actinobacteria. In the nostril, this inverse correlation existed between the Firmicutes family Staphylococcaceae and Actinobacteria families, suggesting potential antagonism between these groups.

FIG 1

Bar graph showing the relative distributions of the major bacterial phyla in the nostril and oropharyngeal samples as detected with a PhyloChip. We used the microarray hybridization intensity to estimate the 16S rRNA gene copy number for each taxon detected on the array and then summed these to estimate the relative prevalence of each phylum in order to compare communities from all participants. Each bar labeled sample 1 to 7 represents 100% of the bacteria detected in a sample by the microarray analysis. Bars labeled AV 1-7 represent the average community composition detected from all 7 seven samples for a site by the microarray. Bars labeled AV 1-4 represent the average community composition detected by the microarray from samples 1 to 4. Bars labeled CL 1-4 represent the average of the relative abundances of phyla in the 16S rRNA gene clone libraries from samples 1 to 4.

FIG 2

Relative abundances of the most common Firmicutes families compared to relative abundances of Actinobacteria families in nostril samples (A) and compared to relative abundances of Proteobacteria families in the oropharynx samples (B) as detected by PhyloChip analysis. For comparison, back-to-back graphs are shown for each site, each with the families from the specified phylum colored as indicated. We used the microarray hybridization intensity to estimate the 16S rRNA gene copy number for each taxon detected on the array and then summed these to estimate the relative abundance of each phylum.

FIG 3

Inverse correlation between the relative prevalences of members of the phylum Firmicutes and another phylum at each site. Dashed lines indicate 95% confidence intervals. (A) Linear regression of the relative prevalences (percentages of the total community) of bacteria from the phylum Firmicutes and the phylum Actinobacteria in the nostril communities. Pearson correlation coefficient = −0.95; P < 0.001. r² = 0.91; P < 0.001. (B) Linear regression of the log₁₀-transformed relative prevalences (percentages of total community) of the Firmicutes family Staphylococcaceae and the Actinobacteria families Corynebacteriaceae and Propionibacteriaceae in the nostril communities. Spearman correlation coefficient (of nontransformed data) = −0.93; P < 0.001. r² = 0.69; P < 0.05. (C) Linear regression of the relative prevalences (percentages of the total community) of bacteria from the phylum Firmicutes and the phylum Proteobacteria in the oropharyngeal communities. Pearson correlation coefficient = −0.994; P < 0.001. r² = 0.99; P < 0.001.

FIG 4

Taxonomic diversity detected by the microarray in samples from the nostril and oropharynx. (A) Total number of taxa detected in each sample and average number (AV) of taxa detected for each site (dark-gray bars). Numbers of taxa detected in each sample and on average per site (AV) that each made up <0.05% of the total community (light-gray bars) are shown. Error bars represent the standard errors of the means. (B) Numbers of taxa that constituted 100% (dark-gray bars), 95% (white bars), and 90% (light-gray bars) of the total bacteria detected by the microarray in each sample and on average per site (AV). Error bars represent standard errors of the means. (C) Simpson’s index of diversity (1 − D) for each sample from the nostril (dark-gray bars) and the oropharynx (light-gray bars) calculated using the estimated 16S rRNA gene copy number derived from microarray hybridization intensity data. Data are graphed as 1 − D, such that the higher the bar, the greater the diversity.

FIG 5

Bacterial communities grouped by site and not by individual. (A and B) Correspondence analysis of the total microarray hybridization profile from each sample performed in MeV v4.4. Black squares, nostrils; gray circles, oropharynges. For clarity the data are shown in two dimensions, with axis 1 graphed against both axis 2 (A) and axis 3 (B). The percentage in parentheses for each axis indicates the percent variation that is explained by that axis. (C and D) Weighted UniFrac analysis of the total microarray hybridization profile from each sample. (C) Each terminal branch represents the total bacterial community detected from one person’s sample from the specified site. All nodes were recovered at 100% using the Jackknife method. (D) UniFrac distances measured within the nostril microbiota samples (w/in N), within the oropharyngeal microbiota samples (w/in OP), between all the nostril and oropharyngeal samples (btwn N and OP), and between the paired nostril and oropharyngeal microbiota for each individual person (btwn paired N and OP). *, statistically different from values for the others as determined by one-way ANOVA with Tukey’s test (set to 0.05) on pair-wise UniFrac values. Error bars represent standard deviations.