Fitness for purpose of stabilized stool samples for bile acid metabolite analyses

Biobanks and cohort studies are increasingly utilizing chemical stabilizers to collect and store stool samples for downstream DNA-based microbiome analyses. While stabilizers permit ambient-temperature collection and storage of samples for gut microbiome studies, the use of the same sample type for downstream metabolomics assays has not been explored. Microbiome-metabolomics analysis of fecal samples is increasingly getting attention to further elucidate the mechanisms by which the gut microbiota influences the host. In this study, we evaluated fitness-for-purpose of OMNIgene-GUT-collected stool samples for downstream metabolomics assays in the scope of fecal bile acids (BA) quantification. Biocrates Bile Acids Kit was used for the quantification of BA from eight healthy donors’ samples collected in (1) OMNIgene-GUT kit and (2) snap frozen in −80 °C in duplicates. A highly selective reversed phase LC–MS/MS analysis method in negative ion multiple reaction monitoring (MRM) detection mode was applied to determine the BA concentrations in each sample.Total fecal BA levels were detectable in OMNIgene-GUT-collected samples (range: 29.9–903.7 pmol/mg). Paired t-test confirmed that there was a significant difference in the total BAs between the OMNIgene-GUT and snap frozen samples (p < 0.05). Extractions from snap frozen samples resulted in higher concentrations of total BAs (range: 243.7–1136.2 pmol/mg). Qualitative differences between individual donors’ BA profiles were detectable using the two sample collection methods. No significant difference was found in the relative concentrations of primary (CA, CDCA) or secondary (DCA, LCA, UDCA) unconjugated BAs to the total BA concentrations in OMNIgene-GUT-collected samples as compared with the snap frozen samples (Wilcoxon-Mann–Whitney test, p > 0.05). Passing-Bablok method comparison and correlation analyis showed a high degree of correlation in the relative concentrations of CA, CDCA, DCA and LCA between OMNIgene-GUT and snap frozen samples. For these four bile acids, the two methods are comparable at an acceptability bias of 30%. We conclude that the OMNIgene-GUT-collected stool samples are fit-for-purpose for downstream fecal bile acids analysis.

Previous studies have established a bidirectional relationship between the host bile acid homeostasis and the gut microbiota, and bacterial overgrowth has been associated with intestinal inflammation and reduced bile acid concentrations in the gut [1][2][3] . Bile acids (BAs) are natural products of cholesterol synthesis that aid in the emulsification and absorption of dietary fats in the small intestine and are considered to be as important as hormones 4 . The primary BAs, cholic acid (CA) and chenodeoxycholic acid (CDCA), are synthesized in the liver and largely conjugated with glycine and taurine. After performing their critically important function of promoting intestinal fat absorption, approximately 95% are absorbed by ileal BA transporters for recycling back to the liver. The remaining 5% pass to the colon where most undergo other microbiota-mediated biotransformations (e.g. 7-alpha-dehydroxylation from primary to secondary bile acids). Only a small amount of secondary bile acids are reabsorbed by passive diffusion, most is excreted in faeces 5 .
The gut microbiota can change the amount and composition of the BA pool through their effects on BA metabolism, specifically in synthesis, deconjugation and conversion of primary to secondary BA 1 . The quantification of fecal bile acids and the profiling of the gut microbial community are increasingly getting attention to further elucidate the mechanisms by which the gut microbiota influences the host 3,5 . One study which evaluated adults with non-alcoholic fatty liver disease (NAFLD) concluded that gut microbiota dysbiosis is associated with altered bile acid (BA) homeostasis, which renders the patients at increased risk of hepatic injury 6  www.nature.com/scientificreports/ To be able to make meaningful statements about the correlation between the BA homeostasis and gut microbiota composition, it is essential that the same stool sample be used for both analyses. The collection method, which is considered as the gold standard method for downstream analysis of metabolites and the gut microbiome composition, is the immediate freezing of stool samples 7,8 . Previous studies showed that freezing samples in − 20 °C or lower immediately after collection preserves the microbial composition similar to analysis of a fresh sample and also avoids potential influence of added preservative. While preserving the microbial composition, it also preserves the detectability of metabolites 7,8 . However, this approach is not always feasible in large-scale, population-based studies, home collection and collection in remote areas because of the high costs associated with cold-chain shipping requirements.
On the other hand, biobanks and large-cohort studies usually collect stabilized stool samples with collection kits, such as the OMNIgene-GUT kit, which has extensively been validated for downstream DNA-based microbiome analyses 9 . As per the manufacturer, the mechanism of stabilization is through inhibition of microbial growth and DNA degradation 10 .
While such a kit permits ambient-temperature collection and storage of samples for gut microbiome studies, the use of the same sample type for downstream targeted bile acid quantitative assays has not been explored. Integrated microbiome-metabolomics analyses of fecal samples are increasingly getting attention to further elucidate the mechanisms by which the gut microbiota influences the host. Therefore in this study, we evaluated the fitness-for-purpose of the OMNIgene-GUT-collected stool samples for downstream metabolomics assays in the scope of the fecal bile acids (BA) quantification.

Results
The concentration of total fecal BA levels could be measured in OMNIgene-GUT-collected samples (mean: 323.8 pmol/mg; range: 29.9-903.7 pmol/mg) (Figs. 1, 2 and Table 1). Paired t-test confirmed that there was a significant difference in the total BAs between the OMNIgene-GUT and snap frozen samples (p < 0.05). Extractions from snap frozen samples resulted in higher concentrations of total BAs (mean: 578.1 pmol/mg; range: 243.7-1136.2 pmol/mg). This is caused by the fact that the OMNIgene-GUT-collected samples are diluted in stabilizer (approximately 500 mg of sample in 1.5 ml stabilizing liquid), resulting in lower total fecal BA concentrations.
Variation between each donors' BA profile is detectable using the two sample collection methods (e.g. the BA profiles of Donors 1 and 2, which are samples collected from children under the age of 2, are readily differentiated among the rest of the donors) (Fig. 3).
No significant difference was found in the relative concetrations of primary unconjugated BAs (CA, CDCA) to the total BA concentrations in OMNIgene-GUT-collected samples as compared with the snap frozen samples (Wilcoxon-Mann-Whitney test, p > 0.05) ( Table 1 and Fig. 3). The same result was observed with the relative concentrations of secondary unconjugated BAs (DCA, LCA, UDCA) to the total BA concentrations in both sample collection methods.
To evaluate the degree of correlation in relative concentrations of bile acids between OMNIgene-GUT vs snap frozen samples, we performed Passing-Bablok method comparison and correlation analysis for bile acids with average relative concentration higher than 0.01. Passing-Bablok regression showed a high degree of correlation in the relative concentrations of CA, CDCA, DCA and LCA between OMNIgene-GUT and snap frozen samples (Fig. 4). For these four bile acids, the two methods are comparable at an acceptability bias of 30%.

Discussion
Excellent reviews of preanalytical aspects of fecal metabolomics have been published 11,12 , highlighting the variability in the protocols used for fecal sample collection, processing (metabolite extraction) and metabolite analysis, the impact of such variations on the recovery of fecal metabolites, and the need for both preanalytical and analytical guidelines. It has specifically been shown that the type of solvent used, as well as the ratio fecal:solvent influence the accuracy of analyses by nuclear magnetic resonance (NMR), while freeze thaw and sonication have less impact 13,14 . In this study, we used a fully validated analytical method [Pham et. al, 2016] that has been optimized by Biocrates for use with plasma samples 15 . This validated protocol was adapted for stool samples using a modified protocol published by the same manufacturer 16 .A study by De Spiegeleer et al. using lyophilized fecal samples, has shown that freeze thawing affects the polar more than the lipid metabolome, with 10% and 7% of metabolites showing statistically significant differences after one freeze thaw cycle 17 . Previous studies have also shown an impact of freeze thawing, probably due to reactivation of microbial biochemical reactions upon thawing, however such changes were small relative to inter-individual differences 18 . In our study, all samples underwent two freeze thaw cycles before analysis, and it can be assumed that the potential bias was the same in all samples.
RNAlater is a very commonly used stabiliser for nucleic acids in different types of biospecimens, including feces. However, it has been repeatedly shown that RNAlater is not fit for purpose for any metabolomics analyses, probably due to high sodium sulphate content 7,8,19 . There has been a previous preanalytical study, including OMNIgene-GUT tubes, and comparing them to snap frozen fecal samples, and samples collected in ethanol, RNAlater or on FTA cards, in the scope of 16S rRNA gene sequencing, and untargeted and targeted metabolomics 7 . This study showed that OMNIgene-GUT stabiliser is suboptimal for untargeted metabolomics analyses, due to lower sensitivity, but fit for purpose for targeted metabolomics on short chain fatty acids (SCFA), with high concordance with the snap frozen samples. A more recent study by Lim et. al. suggested that it is possible to extract both microbiome and metabolite data from a single stool sample collected using OMNIgene-GUT tubes 20 . Our study has also found fitness for purpose of OMNIgene-GUT tubes for targeted metabolomics for another class of metabolites, the bile acids.
We conclude that the OMNIgene-GUT-collected stool samples are fit-for-purpose for downstream metabolomics assays in the scope of quantitative fecal bile acids analysis. However, the reference ranges of the total and individual fecal BA concentrations are different in OMNIgene-GUT-collected samples and in snap frozen samples. Therefore, the values obtained from the two collection methods cannot be compared directly. However, there is a high degree of correlation in the relative concentrations of CA, CDCA, DCA and LCA between OMNIgene-GUT and snap frozen samples.
Finally, at the time of writing of this article, a new stool collection device has been commercialized by DNA Genotek, the OMNImet-GUT stool collection device, which is declared by the manufacturer to be specifically fit for purpose for metabolomics analyses.   www.nature.com/scientificreports/ transport conditions. After storage at RT, the sample in the OMNIgene-GUT tube was aliquoted into two 2-ml cryovial containing 1 ml of stool suspension each and frozen at −80 °C for two weeks.

Methods
For the snap frozen sample, approximately 1 g of the fresh donor sample was placed in a plain 2 ml cryovial. This cryovial was frozen immediately in −80 °C for two weeks.
Sample preparation. Eight OMNIgene-GUT-stabilized samples and eight snap frozen samples were thawed and each sample was further aliquoted into another cryovial. The weight of each sample was recorded prior to re-freezing. A total of 32 frozen stool samples were sent to Biocrates Life Sciences AG(Innsbruck, Austria) for the quantification of endogenous bile acids. See Table 2 for sample details.
To extract metabolites from feces, threefold volume of ethanol/phosphate buffer (85:15 v/v) was added to each stool sample and was vortexed for 3 min. The homogenized samples were then placed in a plate shaker for 30 min at 200 rpm at 0 °C. Following this step, the samples were sonicated at 70 W for 5 min at 0 °C and centrifuged at 800 g for 10 min at 0 °C. The supernatant from each tube was then transferred into a new reaction tube and centrifuged at 19,000 g for 10 min at 4 °C. The new supernatants from this step were transferred into new reaction tubes and were used for analysis. 10µL of supernatant/fecal extracts were used for the Biocrates' Bile Acids Kit according to the manufacturers' instructions.
Quantification of fecal BA. The endogenous bile acids were quantified in collaboration with the Metabolic Phenotyping Services Center of Biocrates Life Sciences AG, (Innsbruck, Austria) using a mass spectrometric-based metabolomic approach with the commercially-available Biocrates' Bile Acids Kit 21 . A highly selective reversed phase LC-MS/MS analysis method in negative ion multiple reaction monitoring (MRM) detection mode was applied to determine the concentrations of bile acids. The samples were extracted via dried filter spot technique in 96-well plate format. Sample extracts were measured by LC-ESI-MS/MS with a tandem mass spectrometry instrument (TSQ Vantage, Thermo Fisher Scientific TSQ). For highly accurate quantification, 7-point external calibration curves and 10 stable isotope-labeled internal standards were applied. Data of bile acids were quantified using the appropriate MS software (Thermo Fisher Scientific Xcalibur) and the results were finally imported into Biocrates MetIDQ software for further analysis.
Accuracy of the measurements (determined with the accuracy of the calibrators) was in the normal range of the method (deviations from target ≤ 20%) for all analytes. Quality control samples were within the pre-defined tolerances of the method.