Human Stool Preservation Impacts Taxonomic Profiles in 16S Metagenomics Studies

Abstract

Microbiotas play critical roles in human health, yet in most cases scientists lack standardized and reproducible methods from collection and preservation of samples, as well as the choice of omic analysis, up to the data processing. To date, stool sample preservation remains a source of technological bias in metagenomic sequencing, despite newly developed storage solutions. Here, we conducted a comparative study of 10 storage methods for human stool over a 14-day period of storage at fluctuating temperatures. We first compared the performance of each stabilizer with observed bacterial composition variation within the same specimen. Then, we identified the nature of the observed variations to determine which bacterial populations were more impacted by the stabilizer. We found that DNA stabilizers display various stabilizing efficacies and affect the recovered bacterial profiles thus highlighting that some solutions are more performant in preserving the true gut microbial community. Furthermore, our results showed that the bias associated with the stabilizers can be linked to the phenotypical traits of the bacterial populations present in the studied samples. Although newly developed storage solutions have improved our capacity to stabilize stool microbial content over time, they are nevertheless not devoid of biases hence requiring the implantation of standard operating procedures. Acknowledging the biases and limitations of the implemented method is key to better interpret and support true associated microbiome patterns that will then lead us towards personalized medicine, in which the microbiota profile could constitute a reliable tool for clinical practice.

Keywords: 16S metagenomics; human gut; microbiota; preservation; stabilizing solution.; standardization; stool.

Figure 1

Illustration of the experimental protocol. Evaluation of ten commercial DNA stabilizing solutions for the storage of fecal samples. (Sc, frozen solid sample; Dc, frozen homogenized sample; S, solid sample without stabilizer; D:, homogenized sample without stabilizer; RT, room temperature). * The whatman FTA card is a cotton-based cellulose matrix containing chemicals that lyse cells, denature proteins and protect DNA; 0.5 ml of homogenized solution was dispatched directly onto the card.

Figure 2

Quantity and quality of DNA extracted from human fecal samples. (A) DNA yield, expressed as ng/µl. (B) A 260/280 nm ratio indicative of the presence or absence of phenol, solvent and protein-type contaminants in the DNA extract. The green range indicates a ratio between 1.7 and 2.0, which here defines an optimal DNA quality.

Figure 3

Mean dispersion using Bray-Curtis distances among technological replicates prior (Sc) and after (Dc) homogenization across 15 fecal samples (SS01 to SS15). The red dotted line indicates equality of dispersion among ‘Sc’ and ‘Dc’ samples.

Figure 4

Summary of community shifts in response to stabilizing solutions over a 14-day storage period. (A) Bray-Curtis distance towards the reference for each patient grouped by stabilizing solution. Median and 5th-to-95th percentile ranges are shown for both interaliquot and interpatient variability. (B) A pairwise paired Wilcoxon test was performed to compare solutions with each other, the color code refers to the log10 fold change of the median performance across patients: blue if the solution on X axis performs better, red if the solution on Y axis performs better. Significance is shown as follows: (*) indicates fdr<0.05, (**) indicates fdr<0.01, (***) indicates fdr<0.001.

Figure 5

Differentially abundant bacterial phyla between samples and their references among the 10 tested DNA stabilizing solutions. The median log2-fold change of average profiles is shown with the significance according the corresponding paired Wilcoxon test. (*) indicates fdr<0.05, (**) indicates fdr<0.01, (***) indicates fdr<0.001.