Tri-Hydroxyl BAs Exclusively Derived from DCA Oxidations.

As shown in Fig. 1, A–C, we found five unknown tri-hydroxyl BA species (TBA05, TBA09, TBA10, TBA13 and TBA18) in the post-prandial urine and serum of 13 healthy volunteers collected at UNC-CH Hospitals (Zhu et al., 2018). Enzyme digestion techniques demonstrated that they were extensively conjugated, thus, explaining their “invisibility” to those direct analysis techniques without consideration of BA conjugation. The Quad time-of flight MS/MS spectra captured in the digested urine samples showed that they had fragment ions possibly associated with 12-hydroxylation on the skeleton (Lan et al., 2016; Zhu et al., 2018). They were therefore supposed to be structurally associated with DCA and CA, the two major BAs with 12-hydroxylation. Additional analysis of the total BA profiles in spot urine samples from 45 healthy Chinese adults confirmed that they are common BA species in human. Metabolomic analysis of the total urinary BA profiles from 58 healthy adults was employed to examine the biologic associations of the unidentified tri-hydroxyl BAs with known BA species. Despite vast individual differences, the potential biologic relationships were observed using hierarchical cluster analysis of Pearson correlation coefficients between the detected unconjugated BAs. As shown in Fig. 1D, TBA05, TBA09, TBA10, and TBA13 were independently cross-correlated to DCA (r > 0.75), which strengthened the hypothesis that TBA05, TBA09, TBA10 and TBA13 are hydroxylated metabolites of DCA. We subsequently performed in vitro metabolite screening of a series of known unconjugated BAs in human liver microsomes to verify the hypothesis. It was confirmed that these newly detected BAs were not derived from the epimerization of CA (Supplemental Fig. S8) or from hydroxylation of LCA (Supplemental Figs. S9 and S10). They were also not derived from hydroxylation of the other di-hydroxyl BAs including, CDCA, UDCA, isoUDCA, HDCA, isoDCA, MDCA, isoHDCA and 3-deoxyCA (Supplemental Fig. S11–S18). As shown in Supplemental Fig. S7, TBA05, TBA09, TBA10, TBA13, TBA18, in addition to TBA04, a trihydroxy BA that was detected at trace levels in serum and urine, were characterized as hydroxylated metabolites exclusively from DCA species.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Unconjugated tri-hydroxyl BAs (TBA05, TBA09, TBA10, TBA13, and TBA18) are biologically associated with DCA. The extracted ion chromatograms of the mixed BA standards, whose abbreviations and structures are summarized in Supplemental Table S1 (A). TBA05, TBA09, TBA10, TBA13, and TBA18 were detected in the 0- to 2-hour postprandial urine sample from a representative healthy subject (B). TBA09, TBA10, TBA13, and TBA18 were detected in the 1-hour postprandial serum sample from the same subject (C). Four aliquots of the same urine and serum samples were prepared without enzymes (T1), with choloylglycine hydrolase (T2), with sulfatase and β-glucuronidase (T3), and with all three enzymes (T4). The biologic associations of TBA05, TBA09, TBA10, and TBA13 with DCA were highlighted by hierarchical cluster analysis (D) of the Pearson correlation coefficients between the urinary total unconjugated BAs profile in the test population of 58 healthy volunteers (36 men and 22 women).

Characterization of the Stereoselective Oxidation of DCA.

Besides the 3α- and 12α-hydroxyl groups, there is only one unknown oxidation site on the steroid skeleton of TBA04, TBA05, TBA09, TBA10, TBA13 and TBA18. The unusual hydroxylation sites at C-1, C-2, C-4, C-6 and C-19 have been reported in earlier gas chromatography-mass spectrometry–based studies of human BAs and were particularly discussed as being present in early human life (Sjövall et al., 2010). We gratefully acknowledge the contribution from Takashi Iida who shared his negative electrospray ionization/time-of-flight MS data of some early synthesized compounds in the MASSBANK database (Horai et al., 2010), which enabled us to have a relatively reasonable guide for our own synthetic efforts. By matching the high-definition MS/MS spectra (Fig. 2) and retention data (Fig. 1 and 3) of the synthesized standards to those of metabolites captured in vivo or in vitro, TBA04, TBA05, TBA09, TBA10, TBA13 and TBA18 were eventually identified as DCA-19-ol, DCA-6β-ol, DCA-5β-ol, DCA-6α-ol, DCA-1β-ol and DCA-4β-ol, respectively. As far as we know, this is the first report confirming the occurrence of DCA-5β-ol and DCA-19-ol in humans.

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

High-definition MS/MS spectra of the oxidized DCA metabolites captured from the synthesized standards (a) and the metabolites detected in vivo or in vitro (b). All standards were synthesized in the laboratory except 3-dehydroDCA was obtained from TRC (cat. no. O856870). The data of endogenous DCA-6β-ol, DCA-5β-ol, DCA-6α-ol, DCA-1β-ol, and DCA-4β-ol were captured from a representative digested urine sample. The in vitro data of DCA-19-ol and 3-dehydroDCA were captured from the DCA incubation samples in human liver microsomes.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

In vitro regioselective oxidation of DCA in human liver microsomes. TBA04, TBA05, TBA09, TBA10, TBA13, TBA18 (m/z 407 > 407), and 3-dehydroDCA (m/z 389 > 389) were detected after incubation of DCA (m/z 391 > 391) in human liver microsomes (A). The retention data of DCA-19-ol, DCA-6β-ol, DCA-5β-ol, DCA-6α-ol, DCA-1β-ol, DCA-4β-ol, and 3-dehydroDCA corresponded well with the metabolites of DCA (B). The identified oxidation sites cluster on the same plane around the 5β-hydrogen in the three-dimensional structure of DCA (C). The metabolite formation kinetics after incubation of 50 µM DCA (n = 3, data shown as mean ± S.D.) in human liver microsomes and human intestinal microsomes for 240 minutes at a protein level of 0.5 mg/ml (D). The inhibition of metabolite formation after incubation of 50 µM DCA in human liver microsomes for 60 minutes by time-dependent inhibitors (n = 3, data shown as mean ± S.D., student t test compared with control: *P < 0.05; **P < 0.01; ***P < 0.001), including verapamil (25 µM), paroxetine (10 µM), ticlopidine (1 µM), tienilic acid (10 µM), phenelzine (1 µM), thio-TEPA (5 µM), and furafylline (1 µM) (E). The inhibition of metabolite formation after incubation of 50 µM DCA in human liver microsomes for 60 minutes by selective inhibitors, including ketoconazole (0.5 µM), fluconazole (10 µM), DEDC (20 µM), quinidine (1 µM), nootkatone (1 µM), sulfaphenazole (1 µM), quercetin (1 µM), montelukast (0.5 µM), sertraline (10 µM), 2-PCPA (0.2 µM), and α-NF (0.1 µM) (F).

Preliminary fragmentation analysis identified common mass losses corresponding to [M-H-H₂O-CO₂]⁻ (m/z 345.3) and/or [M-H-2H₂O-CO]⁻ (m/z 343.3) in the spectra of DCA-6β-ol, DCA-6α-ol, DCA-1β-ol and DCA-4β-ol. A diagnostic fragment ion, [C₁₉H₂₇O]⁻ (m/z 271.2), was found in the spectra of DCA-5β-ol, which is similar as the ring cleavage fragment of CA, [C₁₉H₂₉O₂]⁻ (m/z 289.2). Diagnostic fragment ions associated with other ring-cleavage mechanisms, [C₂₂H₃₅O₂]⁻ (m/z 331.3) and [C₂₂H₃₃O]⁻ (m/z 313.3), were detected in the spectra of DCA-4β-ol and DCA-19-ol, respectively. These observations were to some extent consistent with the previously identified key role of 12-hydroxylation in the negative fragmentation of unconjugated C24 BAs (Lan et al., 2016). The detailed analysis is to be summarized in our later analytical report.

We have detected both in vivo and in vitro (Figs. 1C and 3A; Supplemental Fig. S19), another major metabolite of DCA, 3-dehydroDCA (Zhu et al., 2018). Consistent with previous reports (Amuro et al., 1986; Deo and Bandiera, 2008, 2009), the 3-dehydrogenation reaction was also observed for LCA, CDCA, UDCA and CA (Supplemental Fig. S20–S23). The 3-dehydrogenation mechanism was proposed to be involved with a 3α,3β-diol intermediate formed by 3β-hydroxylation, based on the evidence that the incubation of isoDCA in human liver microsomes did not produce 3-dehydroDCA (Bodin et al., 2005). Our work has provided additional evidence for the 3β-oxidation mechanism because C3-β was the major oxidation site of 3-deoxyCA (Supplemental Fig. S18). To this end, we have provided strong evidence that DCA is regioselectively hydroxylated at C-1β, -3β, -4β, -5β, -6α, -6β, and -19. As shown in Fig. 3C, the identified oxidation sites cluster on the same plane around the 5β-hydrogen. We have compared the stereoselective oxidation rates of DCA in the pooled microsomes prepared from human liver, intestine and kidney. The human kidney microsomes showed no reactivity to DCA oxidations at any sites (data not shown). Our data shows that the DCA oxidations happen primarily in human liver microsomes and secondarily in human intestinal microsomes (Fig. 3D). No formation of DCA-19-ol and DCA-4β-ol was observed in human intestinal microsomes.

CYP3A Inhibitors Selectively Inhibit DCA Oxidations in Human Liver Microsomes.

A series of in vitro inhibition assays were performed to begin the characterization of the P450 isoenzymes responsible for DCA oxidations. Figure 3, E and F, shows the inhibitory effects of various P450 inhibitors, either time-dependent or not, on the regioselective oxidations of DCA. No inhibition effects were observed for the selective inhibitors for CYP1A2 (furafylline and α-naphthoflavone), CYP2C8 (phenelzine, montelukast, and quercetin), CYP2C9 (sulfaphenazole), and CYP2D6 (quinidine), and the nonselective inhibitors for CYP2A6 (2-PCPA), CYP2C19 (nootkatone), and CYP2E1 (DEDC). Tienilic acid, a selective inhibitor for CYP2C9, showed a mild inhibition to the oxidations at C19 (14%) and C4-β (13%). Thio-TEPA, a nonselective inhibitor for CYP2B6, showed a mild inhibition to the oxidation at C5-β (12%). Ticlopidine, a nonselective inhibitor for CYP2C19, showed mild inhibition to the oxidation at C5-β (7%).

Paroxetine (10 µM), a selective time-dependent CYP2D6 inhibitor, showed mild inhibition to the oxidations at C3-β (17%), C1-β (22%), C5-β (30%), C6-α (7%), and C6-β (26%). Sertraline (10 µM), a nonselective CYP2B6 inhibitor with time-dependent inhibitory effects on CYP3A4 (Masubuchi and Kawaguchi, 2013), showed mild inhibition to the oxidations at C3-β (31%) and C1-β (28%). Fluconazole (10 µM), a strong inhibitor of CYP2C19 and moderate inhibitor of CYP2C9 and CYP3A, showed mild to moderate inhibition of oxidation at C3-β (33%) and C6-β (23%). Verapamil (25 µM), a selective time-dependent CYP3A inhibitor, showed strong inhibition of oxidation at C3-β (60%), C1-β (79%), C5-β (85%), C6-α (65%), and C6-β (100%) and moderately at C19 (27%) and C4-β (37%). Ketoconazole (0.5 µM), an index CYP3A inhibitor, showed strong inhibition of oxidation at C3-β (71%), C1-β (76%), C5-β (80%), C6-α (68%), and C6-β (76%) and, to a lesser extent, at C19 (49%) and C4-β (51%). These results clearly demonstrated that CYP3A is the major isoenzymes responsible for the tertiary stereoselective oxidation of DCA.

CYP3A4 and CYP3A7 Are Responsible for DCA Oxidations Regardless of N-Acylamidation.

We investigated DCA oxidation in a panel of rP450s (0.5 nmol protein/ml) compared with that in human liver microsomes (0.5 mg protein/ml). As shown in Fig. 4A, CYP3A4 and CYP3A7 showed the highest catalytic activity at all oxidation sites, whereas CYP3A5 showed a much lower activity level of 3-dehydrogenation and trace or no oxidation at the other sites. For the other P450s, except for CYP3A, only trace amounts of 3-dehydroDCA and DCA-6β-ol were detected in incubations of CYP2C8. As shown in Supplemental Fig. S24, the specific oxidations of DCA through CYP3A4 and CYP3A7 were consistently observed in two rP450 systems, Supersomes and Bactosomes. The metabolic enzyme phenotyping results were consistent with those of the human liver microsome inhibition assay. We thus concluded that CYP3A4 and CYP3A7 were dominantly responsible for the tertiary oxidation of DCA.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

CYP3A4 and CYP3A7 are responsible for the oxidation of DCA, GDCA, and TDCA. The metabolite formation after incubations of 50 µM DCA (A), GDCA (B), and TDCA (C) for 60 minutes in a panel of 18 rCYPs (CYP1A2, 1B1, 2A6, 2B6, 2C8, 2C9*1, 2C18, 2C19, 2D6*1, 3A4, 3A5, 3A7, 2E1, 2J2, 4A11, 4F2, 4F3B, and 4F12; 50 pmol protein/ml) were compared with the data of human liver microsomes (0.5 mg protein/ml) acquired in parallel. The oxidized metabolites of GDCA and TDCA were detected after being digested by choloylglycine hydrolase. Data were shown as mean ± S.D. (n = 3).

The circulated DCA is conjugated in the liver through N-acyl linkage with either glycine as GDCA or taurine as TDCA. We compared the in vitro metabolite formation of DCA, GDCA, and TDCA in rP450s and human liver microsomes, in which the oxidized metabolites of GDCA and TDCA were detected after being digested by choloylglycine hydrolase (Zhu et al., 2018). Typical ion chromatograms of DCA, GDCA, or TDCA incubated in human liver microsomes, rCYP3A4, rCYP3A5, and rCYP3A7 were shown in Supplemental Fig. S25–S28. Almost the same enzyme specificity with unique contributions of CYP3A4 and CYP3A7 was observed for the oxidation of GDCA and TDCA (Fig. 4). An exception was that the enzyme specificity was lost for 3β-oxidation of GDCA and TDCA, to whom CYP2E1, 2J2, 4A11, 4F2, 4F3B, and 4F12 demonstrated higher reactivity than the other tested P450s. The regioselective activities of CYP3A4, CYP3A5, and CYP3A7 to oxidize DCA, GDCA, and TDCA are comparatively highlighted in Fig. 5. CYP3A4 demonstrated a reverse activity toward 1β- and 6α-oxidation of DCA with and without the N-acylamidation. Except for the oxidation at C-3β, CYP3A4 oxidized DCA predominantly at C-1β, whereas GDCA and TDCA were oxidized by CYP3A4 most frequently at C-6α. Unlike CYP3A4, CYP3A7 exhibited a nearly consistent regioselectivity for DCA, GDCA, and TDCA. N-acylamidation of DCA significantly increased the CYP3A7-mediated 1β, 5β, 6α, 19, and 4β-hydroxylation, with the highest activity observed for TDCA. The activities of CYP3A7 for some reactions were significantly greater than those of CYP3A4. The metabolite formation ratios of CYP3A7/CYP3A4 decreased in the following order: TDCA 19-hydroxylation (81.4), GDCA 19-hydroxylation (70.9), DCA 19-hydroxylation (43.8), GDCA 1β-hydroxylation (11.8), TDCA 1β-hydroxylation (9.1), and DCA 4β-hydroxylation (8.7). Based on enzyme specificity data, it was believed that these reactions are promising as possible novel in vitro probe reactions for CYP3A7 activity. In summary, it was clearly demonstrated that CYP3A4 and CYP3A7 were responsible for the regioselective oxidation of DCA, regardless of its N-acylamidation status.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Regioselectivity of CYP3A4, CYP3A5, and CYP3A7 for the oxidation of DCA, GDCA, and TDCA. Metabolite formation after incubations of 50 µM DCA, GDCA, and TDCA for 60 minutes in rCYP3A4, 3A5, and 3A7 (50 pmol protein/ml). The oxidized metabolites of GDCA and TDCA were detected after being digested by choloylglycine hydrolase. Data were shown as mean ± S.D. (n = 3).

Conjugation Pattern of Tertiary BAs.

Based on the characterized tertiary BA metabolism pathways, we showed the host-gut microbial cometabolism for the downstream metabolism of CA (Fig. 6A) and CDCA (Fig. 6B) in 13 healthy adults during the 2 hour-postprandial period after a high-fat diet. Enzyme digestion techniques demonstrated that all the tertiary BAs were extensively conjugated, resulting in an easily neglected level for their free unconjugated forms, particularly in serum samples. Despite the tertiary BAs having a low serum level, DCA-6α-ol, DCA-1β-ol, DCA-5β-ol, and HDCA were determined as the major tertiary species presenting with total urinary levels comparable with primary and secondary BAs in all 13 subjects. By using the enzyme-digestion technique, the total level of a BA may be categorized in four fractions (Fig. 6C): the free unconjugated form, the amidated form, the glucuronidated/sulfated form, and the “double-conjugates” linked with both glycine/taurine and glucuronide/sulfate. The latter two fractions are both glucuronidated/sulfated species. The average percentage conjugation patterns of serum AUC and urine level for each BA were shown in Fig. 6, A and B. Some data were not available because detections were with a trace level (<10 nM) in some samples. For example, DCA-19-ol was not found in any serum samples and was detected in only 2 of 13 urine samples.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Host-gut microbial cometabolism kinetics of primary, secondary, and tertiary BAs in healthy adults. The time-dependent serum concentrations and their 0- to 2-hour urinary levels (mean ± S.E.M.) were shown for the downstream metabolites of CA (A) and CDCA (B) during the 2-hour postprandial period after a high-fat diet in 13 healthy volunteers. The conjugation pattern (C) was determined by the enzyme digestion techniques, in which the free unconjugated form was detected by T1; the glycine/taurine amidated forms were detected by T2-T1; the glucuronidated/sulfated forms were detected by T3-T1; and the “double-conjugates” linked with both glycine/taurine and glucuronide/sulfate were detected by T4-(T3-T1)-(T2-T1). The apparent renal clearances (D) were calculated based on the total urinary excretion data and the total serum AUC data during the 2-hour postprandial period.

DCA-6α-ol and HDCA, the 6α-hydroxylated metabolites of DCA and LCA, were the most heavily glucuronidated/sulfated tertiary species. The glucuronidated/sulfated fractions of DCA-6α-ol and HDCA were significantly more elevated than DCA and LCA, respectively. HDCA circulated in serum and excreted in urine mainly in glucuronidated/sulfated forms and DCA-6α-ol existed mainly in “double-conjugated” forms. A similar conjugation pattern was observed for DCA-4β-ol and DBA03, an unidentified LCA hydroxylated metabolite that was detected in 6 of 13 subjects. In comparison, DCA-1β-ol and DCA-5β-ol, which did not undergo glucuronidation/sulfation as significantly as DCA-6α-ol did, were excreted in urine, mainly in amidated forms. The glucuronidated/sulfated fractions of DCA-1β-ol and DCA-5β-ol in urine were both significantly lower than that of DCA. DCA-6β-ol and MDCA, the 6β-hydroxylated metabolites of DCA and LCA, were also excreted in urine mainly in amidated forms. Therefore, we deduced that tertiary 6α-oxidation facilitates glucuronidation/sulfation metabolism, whereas 1β- and 5β-oxidation of DCA species do not. Further studies are required to investigate the glucuronidation and sulfation mechanisms of tertiary BA species under the different conjugation patterns of them.

Tertiary Oxidation Facilitates the Renal Excretion of DCA Species.

The free and total 0- to 2-hour urinary levels and time-dependent serum concentrations (mean ± S.E.M.) of each BA detected by T1 and T4 are comparatively showed in Fig. 6 and Supplemental Table S2. The primary BAs, CDCA and CA, are synthesized from cholesterol, N-acylamidated, secreted into the bile, and recycled in the enterohepatic system. The free and total levels of CDCA and CA were significantly higher in serum than urine, manifesting a conservative renal disposition for the primary BAs to facilitate their enrichment in the enterohepatic system. After postprandial release into the intestine, most CDCA and CA (95%) are recovered in the bile by the ileal apical sodium-dependent bile acid transporter, the sinusoidal Na⁺-taurocholate cotransporting polypeptide, and the canalicular bile salt export pump, and the remaining 5% enters the lower gut, where they are further metabolized by gut bacteria into secondary BAs. The major metabolic pathway for secondary BAs is deconjugation of the amidated forms by bacteria with bile salt hydrolase activity and the subsequent 7-dehydroxylation of unconjugated forms by bacteria that carries the bai gene. The resulting LCA and DCA are lipophilic and exogenous to the host. Most of the LCA is excreted in feces because it is much less soluble than DCA, which is consistent with the relative total AUC of DCA/CA (3.68 ± 1.08, mean ± S.E.M., n = 13) being about 40 times higher than that of LCA/CDCA (0.09 ± 0.01); however, the free and total levels of DCA were also significantly higher in serum than urine, indicating that DCA is also inclined to be recycled in the enterohepatic system, as seen for CA and CDCA. Therefore, there might be another mechanism to regulate BA homeostasis by disposing the “toxic” secondary BAs, particularly, DCA which is less hydrophobic and has a much greater exposure than LCA.

In contrast to CA and DCA, the total levels of major tertiary metabolites of DCA (DCA-6α-ol, DCA-1β-ol, DCA-5β-ol, and DCA-4β-ol) were significantly higher in urine than serum, indicating that they are preferentially excreted in urine compared with the primary and secondary BAs. We calculated the apparent renal clearance data based on the total urinary excretion data and the total serum AUC data during the 2 hour-postprandial period. The effects of conjugation metabolism (glucuronidation, sulfation, and N-acylamidation) on renal clearance were included in the apparent clearance data. As shown in Fig. 6D, the tertiary BAs demonstrated significantly higher apparent renal clearances than did the primary and secondary BAs, particularly for the downstream metabolites of DCA. In this regard, we believed that tertiary BA metabolism functions, at least in part, in regulating the BA pool size by facilitating the renal excretion of secondary BAs. Accordingly, the urinary total BA profile is an ideal footprint to understand the functional status of the host-gut microbial cometabolism of BAs in biomedical and transitional researches.