Long‐term stability of microbiome diversity and composition in fecal samples stored in eNAT medium

Abstract Fecal samples collected for microbiome analyses are typically frozen to avoid postcollection changes in microbial composition. eNAT is a guanidine thiocyanate‐based medium that stabilizes microbial DNA and allows safe specimen handling and shipping by inactivating microorganisms. We collected fecal samples (n = 50) from children undergoing hematopoietic stem cell transplantation. We divided samples into three aliquots: (a) stored in RNAlater and immediately transferred to −80°C; (b) stored in eNAT medium and immediately transferred to −80°C; and (c) stored in eNAT medium at ambient temperature (~20°C) for 30 days prior to transfer to −80°C. Mean (standard deviation) Shannon diversity and Chao1 indices in sample aliquots were 2.05 (0.62) and 23.8 (16.6), respectively. Comparing samples frozen immediately in RNAlater to samples frozen immediately in eNAT, there were no differences in Shannon diversity (p = .51), Chao1 richness (p = .66), and overall microbiome composition (p = .99). Comparing eNAT samples frozen immediately to samples stored at ambient temperature, we identified no differences in Shannon diversity (p = .65), Chao1 richness (p = .87), and overall microbiome composition (p = .99). Storage of fecal samples in eNAT at ambient temperature for 30 days did not alter microbiome richness, diversity, or composition. eNAT may be a useful medium for fecal microbiome studies, particularly when cold chain storage is unavailable.

bacteria and stabilizes DNA (Choo, Leong, & Rogers, 2015). Use of these stabilization media may be particularly useful for microbiome studies that involve the collection of clinical samples outside of healthcare settings. For instance, these media enable fecal samples collected in the home environment to be stored at ambient temperature, obviating the need for storage in a refrigerator or freezer alongside food items. Additionally, stabilization media may be essential to microbiome studies conducted in regions of the world where there is no cold chain storage and where quickly transporting samples to a freezer is not possible or would be cost-prohibitive (Choo et al., 2015). Finally, use of stabilization media and storage at ambient temperature could minimize DNA degradation associated with freeze-thaw cycles during sample transport or processing (Song et al., 2016).
Tris-EDTA buffer and 70%-99% ethanol have historically been the most common stabilization media used in microbiome research, but there are several challenges to the use of these media (Choo et al., 2015;Vandeputte, Tito, Vanleeuwen, Falony, & Raes, 2017).
First, buffers containing EDTA may not optimally preserve the microbial composition of fecal samples (Reidmiller et al., 2006). In one study, samples stored in Tris-EDTA at ambient temperature had lower abundances of Bifidobacterium and Anaerostipes and higher abundances of Bacteroides and Proteobacteria than samples stored in Tris-EDTA buffer and frozen immediately to −80°C (Choo et al., 2015). Similarly, shifts in microbiome diversity and composition have been reported with the storage of fecal samples in 70% ethanol at ambient temperature (Song et al., 2016). In contrast, 95% ethanol effectively preserves gut microbial diversity and composition at ambient temperature for at least 8 weeks (Song et al., 2016), but this solution is highly flammable and expensive to transport (Bentley, 2007;Nagy, 2010).
In this study, we sought to evaluate the potential utility of eNAT ® (Copan Italia,) as a stabilization medium for microbiome research.
eNAT is a commercially available medium developed for the preservation and stabilization of DNA and RNA in clinical specimens used for nucleic acid tests. This medium contains a protein denaturant (guanidine thiocyanate) that completely inactivates microorganisms, does not require special shipping precautions, and is marketed as being able to stabilize DNA at room temperature for up to 4 weeks (Chomczynski & Sacchi, 2006). We compared microbiome diversity and composition of fecal samples stored in a standard microbiome medium (RNAlater solution and immediately frozen) to matched fecal samples stored in eNAT and immediately frozen. Additionally, we evaluated the stability of the gut microbiome in fecal samples stored in eNAT at ambient temperature for 30 ± 2 days.

| Sample collection
This study made use of fecal samples collected between January 2018 and June 2018 through a prospective cohort study of children and adolescents (<18 years of age) undergoing hematopoietic stem cell transplantation at Duke University. The primary objective of this study is to evaluate the utility of serial sampling of the gut microbiome for the prediction of infections after stem cell transplantation.
Fecal samples were collected into 50-mL sterile collection tubes containing a small amount (~2.5˗5.0 ml) of RNAlater solution and placed immediately into a 4°C refrigerator. RNAlater solution inhibits bacterial growth but leaves bacterial cells intact and viable (Life Technologies, 2011).

| Sample processing
Fecal samples were collected in the hospital and transported to the laboratory daily (Monday-Friday) by research team members for processing and storage. Samples were vortexed in the collection tube, and aliquots of the resulting stool slurry were distributed as follows: (a) 0.5 ml was placed into a cryovial containing 1.0 ml of RNAlater solution and immediately transferred to a −80°C freezer; (b) 0.5 ml was placed into a cryovial containing 1.0 ml of eNAT medium and immediately transferred to a −80°C freezer; and (c) 0.5 ml was placed into a cryovial containing 1.0 ml of eNAT medium, allowed to sit in the laboratory at ambient temperature (~20°C) for 30 ± 2 days, and then transferred to a −80°C freezer.

| DNA extraction and bioinformatics
Total genomic DNA was extracted using an established protocol involving mechanical and enzymatic lysis (LaTuga et al., 2011). PCR was used to amplify the variable V4 region of the bacterial 16S ribosomal RNA gene using an Illumina MiSeq instrument (Illumina Inc.).
Sequencing reads were split, quality-trimmed, and demultiplexed with the use of QIIME tools (Caporaso et al., 2010). A DADA2 pipeline was used to remove chimeric variants and to identify amplicon sequence variants (Callahan et al., 2016). Taxonomic assignments were made based on alignment with the Greengenes database (Larsen et al., 2006). Samples with less than 1,000 sequencing reads were pruned. In addition, amplicon sequence variants with a mean relative abundance of less than 5 × 10 -6 were filtered.

| Statistical analyses
The Shannon diversity and Chao1 index were calculated for each sample (McMurdie & Holmes, 2013). We performed a mixed-model ANOVA model where the independent variable of interest was the sample preparation method, and the outcome was either the Shannon diversity index or the Chao1 index (Moser, 2004). We applied a natural log transformation of the Chao1 index before ANOVA because this statistic was not normally distributed in study samples. To account for children having repeated samples and the fecal samples being split into aliquots, we included a random effect of a participant, where the sample was nested within the participant. A Bland-Altman plot was used to describe the agreement between the Shannon diversity index, and separately the Chao1 index, with pairwise sample preparation methods (Altman & Bland, July, 1983). The Bray-Curtis nonmetric multidimensional scaling matrix was calculated. We used permutational multivariate analysis of variance using distance matrices (PERMANOVA) to determine whether there were global differences in fecal microbiota composition according to the sample preparation method (Oksanen et al., 2019). We implemented zero-inflated Gaussian mixture models using the R package metage-nomeSeq to identify genera that were differentially abundant by preparation method (Paulson, Pop, & Bravo, 2013) All analyses were conducted by a statistician blinded to the sample preparation method. The analyst was blinded to sample preparation method during analysis.

| RE SULTS
The 50 fecal samples included in this study were collected from 13 subjects. After pruning of low-read samples (<1,000 reads), there were matched aliquots of RNAlater solution followed by immediate freezing to −80°C and eNAT followed by immediate freezing to −80°C for 39 fecal samples. Similarly, there were matched aliquots of eNAT followed by immediate freezing and eNAT with storage at ambient temperature for 39 fecal samples. A total of 2,145,517 high-quality 16S ribosomal RNA sequences (mean of 17,876 sequences per sample aliquot) were obtained from the 120 sample aliquots included in this study. Rarefaction curves were constructed to ensure complete coverage of the bacterial diversity at the given sequencing depth ( Figure S1). The mean (SD) Shannon diversity and Chao1 indices in sample aliquots were 2.05 (0.62) and 23.8 (16.6), respectively. Sequences were assigned to 526 amplicon sequence variants (after filtering) representing 117 genera from 10 phyla. Table 1 shows the relative abundances of the 20 most abundant bacterial genera identified in sample aliquots included in these analyses. The most highly abundant bacterial genera were Bacteroides (34.6%), Enterococcus (17.0%), Parabacteroides (7.3%), and Clostridium (6.2%). Figure 1 shows nonmetric multidimensional scaling matrices of fecal microbiota communities by sample preparation method. We observed no difference in the overall microbiome composition of samples prepared in RNAlater with immediate freezing, samples prepared in eNAT with immediate freezing, and samples prepared in eNAT and stored at 20°C for 30 ± 2 days (PERMANOVA, p > .99).   (Table A1).  and stored at ambient temperature. Global gut microbiota composition did not differ by sample preparation method (PERMANOVA; p = .99). Moreover, we found that the relative abundances of the 20 most abundant bacterial genera were not significantly different by the sample preparation method (Figure 3d). The median difference in relative abundance in paired sample aliquots was 0.06% for all 20 bacterial genera tested (Figure 3d). There were no significant differences in the relative abundances of these bacterial genera by sample preparation method (Table A1).

| D ISCUSS I ON
This study demonstrates that the microbial diversity and composition of human fecal samples are not significantly different between storage in eNAT medium compared with RNAlater solution, a standard microbiome medium. Moreover, we found that the diversity and composition of the gut microbiome were stable after prolonged storage in eNAT at ambient temperature. These results suggest that eNAT is a suitable medium for fecal microbiome studies and may be particularly well suited for studies that include the home collection of fecal samples or are conducted in areas where cold chain storage is unavailable.
A variety of media has been used in the collection of fecal samples in studies of the gut microbiome. In this study, we compared samples processed in eNAT to samples processed in RNAlater solution, which is among the most frequently used preservation methods in microbiome studies (Choo et al., 2015). RNAlater is a storage reagent that stabilizes bacterial RNA and is particularly useful for bacterial gene expression profiling. However, the storage of fecal samples in RNAlater solution at ambient temperature results in substantial shifts in gut microbiome composition (Flores et al., 2015; Hale, Tan differences of relative abundances of the 20 most abundant bacterial genera are between 1.6% higher in eNAT frozen immediately to −80°C and 2.1% higher in eNAT stored at ambient temperature achieved with Whatman FTA ® matrix cards, which contain protein denaturants and buffers that lyze microbial cells and stabilize DNA (Rajendram et al., 2006 In conclusion, we provide the first controlled experiment to assess the potential utility of eNAT as a stabilization media for fecal samples to be used in microbiome analysis. Our results suggest that eNAT may be a useful medium for studies of the gut microbiome.
The ability of eNAT and other similar media to stabilize fecal samples at ambient temperature has the potential to improve the feasibil-  Acquisition.

CO N FLI C T O F I NTE R E S T
eNAT media used in this experiment were provided by Copan Italia.

E TH I C S S TATEM ENT
Written informed consent was obtained from a legal guardian by study staff after a detailed explanation of the study procedures and the rights and protections of research participants. The study protocol was approved by the Duke University Institutional Review Board.

DATA AVA I L A B I L I T Y S TAT E M E N T
The dataset supporting the conclusions of this study is available in the Sequence Read Archive (https://www.ncbi.nlm.nih.gov/biopr oject /PRJNA 578756). The statistical code used for data analyses has also been made publicly available (https://github.com/rebec catre e/eNAT).