Sample storage conditions significantly influence faecal microbiome profiles

Abstract

Sequencing-based studies of the human faecal microbiota are increasingly common. Appropriate storage of sample material is essential to avoid the introduction of post-collection bias in microbial community composition. Rapid freezing to -80 °C is commonly considered to be best-practice. However, this is not feasible in many studies, particularly those involving sample collection in participants' homes. We determined the extent to which a range of stabilisation and storage strategies maintained the composition of faecal microbial community structure relative to freezing to -80 °C. Refrigeration at 4 °C, storage at ambient temperature, and the use of several common preservative buffers (RNAlater, OMNIgene.GUT, Tris-EDTA) were assessed relative to freezing. Following 72 hours of storage, faecal microbial composition was assessed by 16 S rRNA amplicon sequencing. Refrigeration was associated with no significant alteration in faecal microbiota diversity or composition. However, samples stored using other conditions showed substantial divergence compared to -80 °C control samples. Aside from refrigeration, the use of OMNIgene.GUT resulted in the least alteration, while the greatest change was seen in samples stored in Tris-EDTA buffer. The commercially available OMNIgene.GUT kit may provide an important alternative where refrigeration and cold chain transportation is not available.

Figure 1. Species diversity following incubation under six different storage conditions

. The extent of microbiota structural and composition diversities were measured using (A) Taxa S (species richness), (B) Shannon-Weiner diversity index, (C) Simpson’s evenness index. Each point represents the diversity score for a replicate from collection 1 (●), collection 2 (▲) or collection 3 (■). Error bars represent SEM. Within-group and between-group variations were measured using Kruskal-Wallis one-way ANOVA and Mann-Whitney U-test, respectively. Significant variance is indicated by asterisks; single asterisk (*) indicates p ≤ 0.05, double asterisk (**) represents p ≤ 0.01.

Figure 2. Relative abundance at phylum level for each sample incubated under six different storage conditions.

Storage conditions that differed significantly from the control (−80 °C) are indicated with respective phylum abbreviation and asterisks. The respective phyla were abbreviated as follow: Actinobacteria (A), Bacteroidetes (B), Firmicutes (F), Proteobacteria (P) and Verrucomicrobia (V). Statistical significance was assessed by Mann-Whitney U-test and indicated by asterisks; single asterisk (*) represents p ≤ 0.05, double asterisk (**) represents p ≤ 0.01, and triple asterisk (***) represents p ≤ 0.001.

Figure 3. Clustering of samples due to storage conditions by PCoA, based on Bray-Curtis similarity distance.

The first two principal coordinates are plotted on the x- and y-axes, respectively (representing 57.3% of the total variation). Faecal collections sampled at three different time points are represented by circle (●) for collection 1, triangle (▲) for collection 2, and square (■) for collection 3. Storage conditions are indicated by colour.

Figure 4. The distribution of the major genera in samples incubated under six different storage conditions.

The heatmap shows square root-transformed read counts for the 17 taxa identified as contributing most to variance (86% across all samples) as determined by SIMPER analysis. The dendrogram shows the similarity relationship of genera based on Bray-Curtis distance and Ward’s hierarchical clustering method. Phyla are abbreviated as follows: Actinobacteria (A), Bacteroidetes (B), Firmicutes (F) and Proteobacteria (P). Storage conditions that differ significantly from the control (−80 °C) by Mann-Whitney U-test are indicated with asterisks; single asterisk (*) represents p ≤ 0.05, double asterisk (**) represents p ≤ 0.01, and triple asterisk (***) represents p ≤ 0.001.