A high-throughput LC-MS/MS method for the measurement of the bile acid/salt content in microbiome-derived sample sets

Due to the physicochemical properties of bile acids/salts (i.e., hydrophobic and ionizable), the application of reverse-phase liquid chromatography-tandem mass spectrometry (LC-MS/MS)-based methods are ideally suited for the measurement of these compounds in a host of microbiologically-relevant matrices. Here, we provide a detailed bioanalytical protocol that contains several modifications of a method previously described by Wegner et al. [1]. Briefly, this modified method exhibits the following advantages for the measurement of cholic acid (CA), taurocholic acid (TCA), and deoxycholic acid (DCA) in microbiome-relevant sample matrices: i) fecal sample processing has been streamlined by the elimination of lyophilization and manual homogenization steps; ii) the Sciex 6500 QTRAP hybrid triple-quadrupole/linear ion trap mass spectrometer has sufficient sensitivity to perform the measurement of bile acids/salts in negative ion mode – ammonium adducts of bile acids/salts are not required for detection; and, iii) assay throughput has been boosted by more than 5-fold by shortening the chromatographic duty cycle of a single sample injection from 45 min to 8.4 min. Recently, the method was used to perform 508 sequential injections (72 calibration standards, 52 blank-internal standard sample, and 368 MiniBioReactor Array (MBRA)-derived samples) from four separate batches over a 4-day time period.


Specifications
Polyvinyl difluoride (PVDF) membrane filter plates with 0.2 μm pores were from Thermo Fisher Scientific. High-Performance Liquid Chromatography (HPLC) separations were performed using a Raptor C18 (2.7 μm superficially porous silica particle (SPP), 100 mm (L) x 2.1 mm (ID), 90 Å pore size; Cat #9304A12) analytical column and an Ultra C18 (5.0 μm silica, 10 mm (L) x 2.1 mm (ID), 100 Å pore size; Cat #917450212) guard column from Restek (Bellefonte, PA, USA). MiniBioReactor Array (MBRA) devices were constructed and used as described previously [11] . Biological Materials The custom bioreactor medium used for this study has been described previously [11] . The protocols for the harvesting and processing of bacterial cells from human fecal communities used for this study were described previously [11] . The protocols for the collection and processing of mouse fecal pellets used for this study were described previously [12] .

Biochemical background
In the human liver, primary bile acids chenodeoxycholic acid (CDCA) and cholic acid (CA) are formed by catabolism of cholesterol [ 2 , 3 ], and are conjugated to glycine and taurine to form the hydrophilic bile salts glycocholic acid (GCA) and glycochenodeoxycholic acid (GCDCA), and taurocholic acid (TCA) and taurochenodeoxycholic acid (TCDCA), respectively ( Figure 1 ) [ 4 , 5 ]. After formation, bile salts enter the enterohepatic cycle where they are: i) exported with other hydrophobic components to form bile in the gall bladder [6] ; ii) secreted into the duodenum where they aid in the absorption and digestion of hydrophobic dietary components; and, iii) are reabsorbed (~95% of the pool) from the chyme in the distal ileum and are transported back to the liver via the portal vein thus completing a single revolution of the enterohepatic cycle [6] . In the human gut, a small proportion (~5%) of the bile salt pool undergoes microbiota-mediated deconjugation of glycine and taurine to reform the hydrophobic, cytotoxic, and potentially antimicrobial primary bile acids CDCA and CA [ 7 -9 ], and these compounds may be metabolized further to the secondary bile acids, deoxycholic acid (DCA) and lithocholic acid (LCA) by microbiota in the large bowel [10] .

Solution preparations
Internal standard (IS) solution preparations   Table 1 ). These Calibrators are refrigerated at + 4 °C while not in use, and are discarded after 1 week of storage.

Bioreactor sample collection, filter-sterilization, and extraction procedures
Bacterial cells from human fecal communities are grown in the bioreactor medium contained in the MBRA devices [11] . Bioreactor medium contains bovine bile as a source of bile salts. Samples are removed from communities and cells are pelleted by centrifugation at 30 0 0 g for 5 min.

Mouse stool collection, filter-sterilization, and extraction procedures
Fecal samples are collected from human microbiota associated mice and stored at −80 °C [12] . 3. If samples are provided in 96-well plates, then the entire sample volume contained in each well is transferred to individual 0.6 or 1.5 mL Eppendorf tubes, and processed further according to the following protocols. All samples are vortex-mixed for 2 min using a multi-tube vortexer. Stool sample extracts are centrifuged at 10,0 0 0 g for 10 min to settle the debris prior to sample aliquoting and dilution. After sample processing has been completed for the entire batch, the residual sample volumes contained in the Eppendorf tubes are stored frozen at −80 °C pending re-analysis.

10-fold sample dilutions for stool (CA, DCA, and TCA) and MBRA-derived bioreactor (TCA only) samples
1. For this study, it was determined that a 10-fold dilution was suitable for the levels of TCA contained in the MBRA-derived bioreactor samples, and for the CA, DCA, and TCA contained in the mouse fecal samples. Dilution factor suitability should be assessed for each new study prior to sample analysis. 2. A 10-fold dilution is performed directly in a polypropylene autosampler injection vial by mixing a 10 μL volume of the undiluted bioreactor/stool extract sample in 90 μL of the WIS-A Solution.
The diluted sample is then vortex-mixed for 15 s, and a 5 μL volume of sample is injected onto the LC-MS/MS system.

Column efficiency characterization
The column efficiency characterization for the Raptor C18 analytical and Ultra C18 guard column combination was performed by computing the pertinent experimental parameters from a mid-level calibration standard (Calibrator D -prepared at 62.5 ng/mL for CA, DCA, and TCA, and 225 ng/mL for the IS compounds, D 4 -CA and D 4 -DCA) using previously described methods [13] . Briefly, the height equivalent to a theoretical plate (HETP; Eq (1) ) and the number of theoretical plates per meter (N/m; Eq (2) ) were calculated using the following experimental parameters for each analyte and IS compound: i) full-width at ½ max for the chromatographic peaks ( W 1/2 ) in minutes; ii) the chromatographic retention time ( τ R ) in minutes; and, iii) the column length (L) specified in the units of cm for the computation of HETP ( Eq. (1) ), and in the units of m for N/m ( Eq. (2) ). All experimental parameters used to perform the column efficiency characterization have been tabulated in Table 2 below.

Method performance characteristics
A total of four analytical batches were prepared across two separate days -each batch consisted of a calibration curve, 13 interspersed blanks, and 96 MBRA-derived media samples that had been diluted according to the procedures described above. Calibration curves were constructed for each analyte by plotting the instrument response (IR = A analyte / A IS ) factor of each Calibrator against their respective nominal concentration. From this plot, a least-squares, linear regression with weighting (1/x) was used to calculate the line of best fit for each analyte, and yielded the following representative calibration curves for each analyte: CA: IR CA = 0.00526 * [CA] + 0.00651, R 2 = 0.991; TCA: IR TCA = 2310 * [TCA] + 3045, R 2 = 0.9856; DCA: IR DCA = 0.00337 * [DCA] + 0.00641, R 2 = 0.999. Limit of Detection (LOD) and Limit of Quantitation (LOQ) estimates were calculated using the standard deviation of the y-intercepts and the mean slope of the four calibration curves calculated for each analyte as described previously [14] . The LOD and LOQ estimates for each analyte are tabulated with the mean linear regression parameters for the four calibration curves ( ± Standard Error of the Mean (SEM)) in Table 4 .

Discussion
Herein, we report on a high-throughput LC-MS/MS-based bioanalytical method that is suitable for the quantitation of CA, TCA, and DCA in microbiome relevant matrices such as stool and filtersterilized MBRA-derived bioreactor broth media. Using this method, TCMC-MSL staff are capable of processing up to 192 stool or bioreactor samples in a single 8 h day, and the LC-MS/MS acquisition of all calibrators, blanks, and unknown samples will require ~36 h to complete for the two analytical batches. This method is highly adaptable, and with the purchase of authentic reference standards, can be modified to include other bile acids/salts that may be of interest TCMC-MSL collaborators. Table 4 Mean weighted (1/x) linear regression parameters for cholic acid (CA), taurocholic acid (TCA), and deoxycholic acid (DCA) calibration curves ( n = 4) over a dynamic range of 0.977 to 10 0 0 ng/mL.

Analyte
Internal

Declaration of Competing Interest
The authors declare no conflicts of interest