Effects of the long-term storage of human fecal microbiota samples collected in RNAlater

Abstract

The adequate storage of fecal samples from clinical trials is crucial if analyses are to be performed later and in long-term studies. However, it is unknown whether the composition of the microbiota is preserved during long-term stool storage (>1 year). We therefore evaluated the influence of long-term storage on the microbiota composition of human stool samples collected in RNAlater and stored for approximately five years at -80 °C. We compared storage effects on stool samples from 24 subjects with the effects of technical variation due to different sequencing runs and biological variation (intra- and inter-subject), in another 101 subjects, based on alpha-diversity, beta-diversity and taxonomic composition. We also evaluated the impact of initial alpha-diversity and fecal microbiota composition on beta-diversity instability upon storage. Overall, long-term stool storage at -80 °C had only limited effects on the microbiota composition of human feces. The magnitude of changes in alpha- and beta- diversity and taxonomic composition after long-term storage was similar to inter-sequencing variation and smaller than biological variation (both intra- and inter-subject). The likelihood of fecal samples being affected by long-term storage correlated with the initial relative abundance of some genera and tend to be affected by initial taxonomic richness.

Figure 1

Study design. Fecal samples were collected from a total of 125 subjects (24 healthy adults and 101 IBS patients, see Supplementary Table S1). Samples were collected by the subjects at home, in RNAlater, and were then processed at the laboratory for dry storage at −80 °C. For each sample, one aliquot was extracted and sequenced to constitute the “reference” (year 0). Extracted DNA from 19 subjects was sequenced a second time, to study technical variability between two sequencing runs (“inter-sequencing runs”). One fecal aliquot from each of 24 subjects was used for DNA extraction and sequencing after long-term (five years) storage at −80 °C. The dataset obtained was compared with the corresponding dataset for reference values (stored DNA resequenced at year 5), to study the effect of long-term storage (“fecal storage”). A second round of stool collection was performed for 93 subjects. This second sample was used for DNA extraction and sequencing in year 0, and the resulting dataset was compared with the reference samples from the same subjects to evaluate biological variability within subjects (“intra-subject”). Inter-subject analysis consisted of an analysis of differences between subjects at year 0. Design Crea Nostra

Figure 2

Boxplot of the relative abundance of phyla and families in fecal samples before and after long-term storage. Phyla or families accounting for more than 10% of the microbiota present in at least one sample are depicted. Microbial taxa were inferred from the SILVA database (version 119).

Figure 3

Scatterplots of the relative abundance of phyla, families, genera and OTUs. Relative abundances of microbial taxa in the conditions tested (inter-sequencing run, fecal storage, intra-subject) were plotted against the corresponding reference dataset. Spearman’s rho correlation coefficients for each set of conditions tested are shown on the graph.

Figure 4

Violin plot of alpha- and beta-diversity metrics for each set of conditions tested, relative to the reference sample. (A) Alpha-diversity variation was evaluated as the positive absolute number of OTUs, Shannon, Chao, and Pielou indices, different from the corresponding reference. (B) Beta-diversity was assessed with Unifrac (unweighted and weighted), Bray-Curtis, and Jensen-Shannon metrics, at the genus and OTU levels.

Figure 5

Stability of individual genera after storage for five years is correlated with mean baseline relative abundance and prevalence. Each point represents a genus (N = 99).

Figure 6

Graphical summary of studies assessing the impact of sample processing and storage on gut microbiota composition. The data presented were extracted from the studies reported in Additional file 1. Sample integrity was determined from the study conclusions, as listed in Additional file 1. For samples treated with a stabilizing buffer, we included in our analysis only those treated with RNAlater.