The aim of the present study was to extend our previously established method for the quantification of human serum BAs in routine laboratory diagnostics [18, 30], to rodent BA profiles applicable to different sample materials. Furthermore, we aimed to keep the method run time short in order to achieve sufficient sample throughput and to increase the sensitivity by using a state-of-the-art LC–MS/MS system.

As a first step, commercially available BAs were used to find and optimize mass spectrometric settings in negative ion mode including declustering potential (DP), collision energy (CE), and collision cell exit potential (CXP). As previously described [18], the main fragment ions of m/z 74 and 80 were observed for glycine and taurine conjugated species, respectively. Most unconjugated BAs did not show a prominent product ion. Therefore, we used a mass transition without fragmentation, and only for UDCA, CA and DCA additional fragment ions were included to increase confidence in their identification and quantification (Table S1).

Since isomeric BAs cannot be differentiated by mass spectrometry, they need to be separated by chromatography. Previously, we successfully used a water–methanol gradient and an RP18 column at basic pH to separate human BA [18]. Despite adaptation and optimization of the LC-gradient, we were not able to efficiently separate MCA species (data not shown). Therefore, we tested a biphenyl column for MCA species separation. While this stationary phase showed superior resolution of MCA species, co-elution of conjugated and unconjugated species of the same BA required a decrease in ammonia concentration from 0.1 to 0.01% to achieve co-elution of conjugated and unconjugated species of the same BA (data not shown). Co-elution of free BAs and its glyco- and tauro-conjugates not only facilitates easy identification in the absence of the corresponding BA standard, but may also be advantageous to account for the absence of a corresponding SIL-IS (see discussion below). Using these parameters, we were able to separate 24 of the 30 targeted BA species in less than 6.5 min (Fig. 1). It was not possible to separate isomeric β- and ω-MCA, as well as their conjugates, differing only in the orientation of the hydroxy group at C6 position.

Chromatogram of a representative BA calibrator sample. Isomeric BAs measured with the same mass transition are shown in the same color

Compared to most existing methods for the separation of murine BA species [19,20,21,22,23,24,25,26,27,28], this method allows a more than twofold increase in sample throughput. Sangaraju et al. [29] report a run time of 10 min, also using basic LC mobile phases. However, they did not report on ω-MCA species, which may co-elute with β-MCA similar to the present method. Furthermore, they chose to report α-MCA and β-MCA (including their tauro-conjugates) together, as these isomers were not separated by the baseline.

Quantification was based on 22 stable isotopically labeled (SIL) BA species used as IS and 6-point calibration lines for most of the target analytes. As previously demonstrated, suitable ISs (best matching stable isotope labelled) are essential to compensate for matrix effects and to achieve accurate and reproducible BA quantification [30]. Therefore, for those species for which no matching SIL-IS was available, the IS with the closest retention time was used for quantification (Table S1). Calibration lines were linear in the validated range with Pearson coefficients > 0.99 (Table 1). Of note, the calibrator was made from a BA standard mixture prepared for routine diagnostics of human serum supplemented with MCA species. No commercially available standard was available for GαMCA and GγMCA. Therefore, calibration lines for TαMCA and TγMCA were used. As the analytical response of tauro- and glycol-conjugates may not be similar, these concentrations should be considered as an approximation. Compared to existing methods [19,20,21,22,23,24,25,26,27,28,29], however, the comprehensive set of SIL-BAs included here represents an advance, as matching SIL-ISs were missing for only 6 analytes.

The current method is intended for research purposes, not for patient diagnosis. Therefore, we decided to focus method validation on key analytical metrics, rather than following the comprehensive guidelines typically used for biomedical assays (see for example [31]). The performance of our method was first evaluated by assessing intra- and inter-day precision and accuracy for two serum quality control (QC) samples prepared from pooled human serum. Precisions for almost all BAs and both QCs were better than 10% CV (Table 1), demonstrating good reproducibility of the method. Higher variation was observed for QC1 for β- and ω-MCA due to concentrations close to the limit of quantification (LoQ; see below) and for inter-day CVs of GHDCA and THDCA for both QCs. No matching SIL-ISs were available for GHDCA and THDCA, which most likely caused the high analytical variation. Due to the lack of appropriate SIL-ISs, target concentrations were not available for HDCA and its conjugates (D5-HDCA has recently become commercially available). Since target values were also not available for MCAs added to the QCs, we did not calculate accuracies for these analytes. The accuracies for the remaining BA species were within ± 20% of the target values established for the human BA species. Only GUDCA showed a systematic shift in accuracy, most likely due to concentration deviation in the calibrator, as the reproducibility was excellent and both QCs showed similar variation.

The next step was to determine the limit of detection (LoD). Unfortunately, for the majority of the analytes, the IS blanks interfered with the SIL-IS due to insufficient isotopic purity (data not shown). For these analytes, the application of S/N to estimate the LoD is impossible. Therefore, we decided to roughly estimate the limit of quantitation (LoQ) from a serial dilution of QC1 (1:10, 1:100, 1:1000, and 1:10000; n = 4, respectively). The concentration levels for which both the CV was < 20% and the accuracy of the dilution was within ± 20% were defined as LoQ. For most BA species, the LoQs were less than 100 nM and ranged from 3 to 152 nM. Due to the tenfold dilution steps, the LoQs shown in Table 1 should be considered as estimates. It should also be noted that analyte interference resulting from SIL-IS can increase the LoQ. For example, TCA showed the highest LoQ and D4-TCA showed a fraction of 0.36% unlabeled analyte. Therefore, the use of isotopically pure SIL-ISs could significantly improve the sensitivity of the analysis in the low nM range.

For free BAs, the MS2 transitions for CA and DCA showed lower LoQs. Furthermore, quantification using MS2 transitions should be considered more specific than without fragmentation, and therefore, MS2 transitions should be preferred for quantification. However, for UDCA, fragmentation leads to a significant decrease in sensitivity, and therefore, quantification may only be possible without fragmentation.

Next, sample carryover was evaluated by repeated injection of the highest calibrator, followed by solvent blanks. Although carryover was less than < 0.2% for most BA species, a blank is recommended after samples with very high concentrations to avoid misquantification.

In addition to spiked human serum QCs, the performance of the method was tested in mouse plasma, bile, and liver samples. For plasma, sample volume requirements were tested by using 1, 5, 10, 25, and 50 µL as sample volume and filling each to a total volume of 50 µL with water (n = 4, respectively). The CV and precision of the dilution (based on 50 µL sample volume) were calculated. While the reproducibility was still sufficient for several BA species at 1 µL plasma volume, the dilution integrity was not met, also due to insufficient isotopic purity of SIL-IS (see above). As expected, CVs increase with lower volumes, as shown in Fig. 2 for TβωMCA and TCDCA. Using the same criteria as for the determination of the LoQ (see above), the following analytes could be quantified at 1 µL (βωMCA), 5 µL (TUDCA, TCA, CA, and TβωMCA), 10 µL (TDCA, DCA, TαMCA), 25 µL (UDCA, TCDCA), and 50 µL (HDCA, THDCA, TγMCA). For the quantification of major BA, 5 µL can be considered as sufficient sample amount, which represents a very economical use of sample material. However, for the quantification of minor BA, as shown in Fig. 2, larger volumes of up to 50 µL plasma are required. To cover the major BA species in bile, 50 µL of a 1000-fold diluted sample (equivalent to 0.05 µL of native bile) and 1 mg of liver tissue are required. BA retention times did not shift in these samples, permitting identification of BA species (see Figure S1 for representative chromatograms for mouse plasma, bile, and the liver).

Accuracy and repeatability as a function of the plasma volume used. Precisions (dashed) and concentrations ± SD (solid) of TCDCA (green, upper panel) and TβωMCA (blue, lower panel) for n = 4 replicates of a plasma sample as a function of the volume used for protein precipitation. The CV threshold for accurate quantification is highlighted by a dashed red horizontal line

Finally, we determined BA concentrations in wild-type C57/BL6/N mice fed a chow diet. As expected, the plasma profiles differed from those in the liver and bile (Fig. 3). While secondary and unconjugated primary BA such as DCA, TDCA, CA, and βωMCA are detectable at relatively high levels in plasma, this is not the case for bile and the liver. As expected, bile and liver BA profiles are dominated by conjugated primary BA such as TCA and TβωMCA, and only traces of other BAs are detectable due to hepatic metabolism of these gut microbiota-derived metabolites [13, 32]. The observed concentrations were in good agreement with previously published concentrations and profiles of BAs in plasma, bile, and the liver [13, 33,34,35].

BA concentrations in murine plasma, bile, and liver. Bile acid concentrations in plasma, bile, and liver samples from WT mice (n = 3–4) are shown. Plasma sample volume was 50 µL. Bile and plasma data are based on pooled samples. Concentration is shown in µM, mM, and pmol/mg wet tissue ± SD for plasma, bile, and the liver, respectively. The mean concentration of the respective BA species is given above the bar