Impaired oxidoreduction by 11β-hydroxysteroid dehydrogenase 1 results in the accumulation of 7-oxolithocholic acid

11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) mediates glucocorticoid activation and is currently considered as therapeutic target to treat metabolic diseases; however, biomarkers to assess its activity in vivo are still lacking. Recent in vitro experiments suggested that human 11β-HSD1 metabolizes the secondary bile acid 7-oxolithocholic acid (7-oxoLCA) to chenodeoxycholic acid (CDCA) and minor amounts of ursodeoxycholic acid (UDCA). Here, we provide evidence from in vitro and in vivo studies for a major role of 11β-HSD1 in the oxidoreduction of 7-oxoLCA and compare its level and metabolism in several species. Hepatic microsomes from liver-specific 11β-HSD1-deficient mice were devoid of 7-oxoLCA oxidoreductase activity. Importantly, circulating and intrahepatic levels of 7-oxoLCA and its taurine conjugate were significantly elevated in mouse models of 11β-HSD1 deficiency. Moreover, comparative enzymology of 11β-HSD1-dependent oxidoreduction of 7-oxoLCA revealed that the guinea-pig enzyme is devoid of 7-oxoLCA oxidoreductase activity. Unlike in other species, 7-oxoLCA and its glycine conjugate are major bile acids in guinea-pigs. In conclusion, the oxidoreduction of 7-oxoLCA and its conjugated metabolites are catalyzed by 11β-HSD1, and the lack of this activity leads to the accumulation of these bile acids in guinea-pigs and 11β-HSD1-deficient mice. Thus, 7-oxoLCA and its conjugates may serve as biomarkers of impaired 11β-HSD1 activity.

To determine apparent K m and V max values, lysates were incubated for 10 min at 37°C in a total volume of 20 l containing 500 M NADPH and 7-oxoLCA at concentrations between 62.5 nM and 4 M. To determine the relative content of product formed, lysates at a high total protein concentration (to achieve almost complete substrate conversion) were incubated with 1 µM of 7-oxoLCA for 10 min. Reactions were terminated by adding 80 µl of acetonitrile containing 100 nM of CDCA-d4. Concomitantly, lysates were used in parallel experiments to assess the conversion of cortisone to cortisol. Reaction products were directly measured by mass spectrometry as described previously ( 26,31 ). The rate of all reactions was kept below 25% of substrate conversion, with the exception of the comparison of product formation from 7-oxoLCA shown in Fig. 1A .
The expression of recombinant C-terminally FLAG-tagged 11 ␤ -HSD1 in different transfection experiments was determined semiquantitatively by immunoblotting. Protein concentrations were determined by the bicinchoninic acid (BCA) method. An amount of 25 µg of total protein was resolved on 12% Bis-Tris gels (NuPage®, Invitrogen) using 1X MES as buffer (Invitrogen, NuPAGE® MES SDS Running Buffer) and then transferred to nitrocellulose membranes (iBlot®, Invitrogen). Thereafter, membranes were blocked with Odyssey® blocking buffer (LI-COR, Biosciences, Lincoln, NE) overnight at 4°C. Immunoreactions were carried out with primary anti-FLAG antibody M2 (Invitrogen) to detect the recombinant enzyme and with secondary goat antimouse Alexa Fluor® 790 antibody, respectively (Invitrogen). After immunoreaction of FLAG-tagged enzymes, the membranes were stripped and incubated with an antibody against ␤ -actin (Sigma). All detection and quantifi cation reactions were performed using a LI-COR Odyssey infrared imaging system (LI-COR, Biosciences, Lincoln, NE).

Impact of bile acids on the interconversion of glucocorticoids by 11 ␤ -HSD1 of six species
The inhibitory impact of bile acids on the interconversion of glucocorticoids by 11 ␤ -HSD1 of six species was performed as described previously ( 26 ). Briefl y, lysates were incubated for 10 min at 37°C in a total volume of 22 l containing 200 nM and 10 nCi of [1,[2][3] H]cortisone or [1,2,6, H]cortisol and 500 M cofactor NADPH or NADP + , respectively, and vehicle or various concentrations of bile acids. Following the interconversion of radiolabeled glucocorticoids and termination of reactions by adding methanol containing 2 mM of unlabeled cortisone and cortisol, 15 l were spotted on Polygram SIL G-25 UV254 silica plates (Macherey-Nagel, Oensingen, Switzerland). Plates were dried, and cortisone and cortisol were resolved using a solvent system of 9:1 (v/v) chloroform/methanol. The separated steroids were analyzed by scintillation counting.
Circulating concentrations of glucocorticoids and their metabolites are not suitable biomarkers because their levels in plasma and urine are a result of combined activities of 11 ␤ -HSD1 and 11 ␤ -HSD2. Furthermore, an ideal biomarker should be independent of mechanisms of negative feedback regulation, such as the control of circulating glucocorticoid levels by the hypothalamuspituitary-adrenal (HPA) axis and stress-induced fl uctuations ( 18,19 ).
In the present study, we aimed to i ) obtain further in vitro and in vivo evidence for a role of 11 ␤ -HSD1 in the oxidoreduction of 7-oxoLCA, ii ) assess species-specifi c differences in the 11 ␤ -HSD1-dependent oxidoreduction of 7-oxoLCA, and iii ) test our hypothesis that 7-oxoLCA might accumulate in transgenic 11 ␤ -HSD1-defi cient mice and provide initial evidence that this bile acid might serve as a biomarker for 11 ␤ -HSD1 defi ciency.
Committee of the University of Birmingham (Birmingham, United Kingdom) in accordance with the UK Animals (Scientifi c Procedures) Act, 1986. Mice were kept in a climate-controlled facility, housed under standard conditions on a 12 h light/dark cycle, and fed ad libitum with standard chow and free access to drinking water.
To assess the impact of 11 ␤ -HSD1 on the circulating and intrahepatic levels of 7-oxoLCA and its conjugated metabolites, 15week-old wild-type mice (n = 16), 11 ␤ -HSD1 global KO (n = 8), and 11 ␤ -HSD1 liver-specifi c KO (n = 16), previously described in Ref. 33 , were fasted overnight, and blood samples were collected by intracardiac puncture. Plasma was prepared immediately by centrifugation, and samples were stored at Ϫ 80°C until further processing.

Quantifi cation of 7-oxoLCA and its taurine and glycine conjugates by LC-MS/MS
Extraction and quantifi cation of bile acids in plasma samples was performed essentially as described previously ( 31 ). The extraction of bile acids from liver tissue was performed by homogenizing 100 mg of tissue in 200 µl 50% methanol. Samples were spiked with 300 µl of deuterium labeled bile acids as internal standards (CDCA-d4) at a fi nal concentration of 1 µM in acetonitrile, followed by protein precipitation by adding 1.5 ml of alkaline (5% NH 4 OH) ice-cold acetonitrile. Thereafter, samples were mixed continuously for 1 h and centrifuged at 11,000 g for 10 min. The supernatant was transferred to a new tube, the solvent was evaporated, and the residue was reconstituted in 100 µl of 50% methanol, followed by an additional centrifugation step to remove insoluble particles. The method was qualifi ed on the basis of extraction effi ciency, intraday accuracy, and precision for 7-oxoLCA, 7-oxoLC-Tau, and 7-oxoLC-Gly (data not shown). The method presented acceptable extraction effi ciency, accuracy, and precision for the bile acids studied.

Calculation of enzyme kinetic parameters
Enzyme kinetics was analyzed by nonlinear regression using four-parameter logistic curve fi tting. For statistical comparisons, the ratio t -test in GraphPad Prism 5 software was used. Results (mean ± SD) were obtained from at least three independent experiments. For calculation of V max , the expression level of the FLAG-tagged enzyme was normalized to the expression signal of the internal control ␤ -actin.

Microsomal preparations and activity assays with hepatic microsomes
Microsomes from livers of wild-type and liver-specifi c 11 ␤ -HSD1-knockout mice were prepared as described earlier ( 32 ). The quality of microsomal preparations was validated by using a kit to measure cytochrome c reductase activity (Sigma, Saint Louis, MO). 11 ␤ -HSD1 reductase activity was determined by incubation of microsomes (0.05 mg/ml) for 60 min at 37°C in a total volume of 25 l containing TS2 buffer, 500 M NADPH or 1 mM glucose-6-phosphate (G6P), 1 M 7-oxoLCA or 1 µM cortisone, and vehicle or 5 M of the 11 ␤ -HSD1 inhibitor T0504 as indicated. Substrates and inhibitors were diluted from 10 mM stock solutions in DMSO or methanol. The fi nal solvent concentration in all reactions was kept below 0.2%. Reactions were started by adding microsomes into freshly prepared reaction mixture and stopped by adding 500 µl of acetonitrile containing CDCA-d4 and cortisol-d4 at a concentration of 100 nM as internal standards for LC-MS/MS analysis. Thereafter, the organic phase was evaporated to dryness, and samples were reconstituted in 50% methanol/water solution, followed by injection into the LC-MS/ MS instrument.

Animal experimentation
Animal studies were conducted under Home Offi ce license and following approval of the Joint Ethics and Research Governance for 10 min to convert most of the substrate and to assess the relative composition of products formed. Human 11 ␤ -HSD1 preferentially formed CDCA with minor amounts of UDCA as reported earlier ( 26 ). Canine and hamster 11 ␤ -HSD1 almost completely converted the substrate and presented similar stereoselectivity to human 11 ␤ -HSD1, thus mainly producing the 7 ␣ -hydroxylated bile acid CDCA, with minor amounts of the 7 ␤ -hydroxylated UDCA. Rat and mouse 11 ␤ -HSD1 also almost completely converted the substrate and reduced 7-oxoLCA to signifi cant amounts of both CDCA and UDCA ( Fig. 1A ). Surprisingly, we could detect only background activity of 7-oxoLCA oxidoreduction when using HEK-293 cell lysates expressing recombinant guinea-pig 11 ␤ -HSD1 or when using guinea-pig liver microsomes ( Fig. 1A, B ). Under the same conditions, cortisone was effi ciently reduced by recombinant guinea-pig 11 ␤ -HSD1 ( Fig. 1C , Table 1 ), with an activity at 4 µM cortisone that was about two times higher than that of the human enzyme, in line with an earlier report on kinetic parameters for glucocorticoids as substrates of 11 ␤ -HSD1 ( 29 ). On the other hand, 11 ␤ -HSD1 from human, mouse, rat, hamster, and dog efficiently converted 7-oxoLCA ( Table 1 ) with K m values ranging 1.2-5.0 µM and V max values ranging 24-107 nmol × mg Ϫ 1 × h Ϫ 1 . None of the 11 ␤ -HSD1 enzymes was able to catalyze the conversion of either CDCA or UDCA to 7-oxoLCA at pH 7.4 and in the presence of NADP + , suggesting that exclusively the reduction is catalyzed under physiological conditions.

Inhibition of the 11 ␤ -HSD1-dependent interconversion of glucocorticoids by bile acids
Next, we investigated whether 7-oxoLCA, CDCA, or UDCA can inhibit 11 ␤ -HSD1 reductase and dehydrogenase activity, respectively, and assessed species-specific differences. A preferential inhibition of the reductase reaction was observed when using 7-oxoLCA, except for the guinea-pig enzyme, which showed weak inhibition with all three bile acids analyzed, independent of whether reduction or oxidation was measured ( Table 2 ). The IC 50 values obtained for both reductase and dehydrogenase activity did not refl ect the ability of a given bile acid to serve as a substrate/product. Human and hamster 11 ␤ -HSD1 showed lowest IC 50 values for 7-oxoLCA on cortisone reduction, while rat, mouse, and canine 11 ␤ -HSD1 reductase activity was equally well inhibited by 7-oxoLCA and CDCA and less efficiently by UDCA. Interestingly, although CDCA potently inhibited cortisone reduction by 11 ␤ -HSD1,

Statistical analysis
Data are presented as mean ± SD. Statistical signifi cance was assessed by Student t -test. P р 0.05 was considered signifi cant.

Docking studies
The ligands were drawn using ChemBioDraw Ultra 12.0 and the 2D structures converted into 3D structures using ChemBio3D Ultra 12.0 (1986-2010 CambridgeSoft). The docking studies were performed using GOLD ( 34,35 ), which uses a genetic algorithm to predict binding modes for small molecules in a protein binding site. The crystal structures for human and guinea pig 11 ␤ -HSD1 were downloaded from the Protein Data Bank (PDB, www. pdb.org) ( 36 ). PDB-entry 2BEL, chain A ( 37 ), was chosen for human protein, and 3LZ6, chain A ( 38 ), for guinea-pig protein. For human 11 ␤ -HSD1, the binding site was defi ned as an 8 Å sphere, centered with hydroxyl-oxygen of Ser170 (x: 3.84, y: 22.49, z: 13.34). For the guinea-pig protein, the binding site was defi ned as an 8 Å sphere, centered by hydroxyl-oxygen of Tyr158 (x: 13.21, y: 22.34, z: 45.45). For both enzymes, GoldScore was used as a scoring function, and the program was set to defi ne the atom types for proteins and ligands automatically. The proteins were kept rigid and ligands fl exible during the docking run. To give the steroidal ligands more fl exibility, the program was set to fl ip ring corners while searching possible binding orientations for the ligands. For each ligand, a maximum of ten binding orientations were generated, but in case the three best-ranked solutions were within RMSD of 1 Å of each other, the program was allowed to terminate the run earlier. Using these settings, the program successfully reproduced the binding orientations of the co-crystallized ligands carbenoxolone (2bel) and N-adamantan-2-yl-1-ethyl-D-prolinamide (3lz6), thus validating the docking settings.

Comparison of the oxidoreduction of 7-oxoLCA by 11 ␤ -HSD1 from six species
We previously showed that human 11 ␤ -HSD1 preferentially converts 7-oxoLCA to CDCA ( 26 ). Evidence from studies using rat liver microsomes indicated the formation of both CDCA and UDCA from 7-oxoLCA. In the present study, we investigated whether these observations are due to species-specifi c differences in the stereoselective product formation by 11 ␤ -HSD1 or whether additional enzymes might be involved in the carbonyl reduction of 7-oxoLCA in rats and mice. Recombinant 11 ␤ -HSD1 from six species were expressed in HEK-293 cells, and the carbonyl reduction of 7-oxoLCA was determined by incubation of lysates at high total protein concentration with 1 µM of 7-oxoLCA The results are mean ± SD (n = 3).
cytochrome c reductase activity in microsomes from livers of wild-type and liver-specifi c 11 ␤ -HSD1-defi cient mice and found no differences between the two preparations, indicating that the preparations were qualitatively comparable (data not shown).

Comparison of circulating concentrations of 7-oxoLCA and its conjugates in serum from various species
Because 7-oxoLCA oxidoreductase activity was absent in guinea-pig liver microsomes and in lysates of HEK-293 cells expressing recombinant guinea-pig 11 ␤ -HSD1, but it was present in all other species tested, we hypothesized that 7-oxoLCA and its conjugated metabolites accumulate in the serum of guinea-pigs but not in the other species. Therefore, we determined the concentrations of 7-oxoLCA, 7-oxoLC-Tau, and 7-oxoLC-Gly in the serum of various species, including guinea-pig (Dunkin-Hartley), there was almost no end-product inhibition of the reduction of 7-oxoLCA, and this bile acid was almost completely reduced by 11 ␤ -HSD1 ( Fig. 1A ). 11 ␤ -HSD1 dehydrogenase activity of all species, except guinea-pig, was most potently inhibited by CDCA, with weaker effects by 7-oxoLCA and UDCA.

Relative contribution of 11 ␤ -HSD1 to the oxidoreduction of 7-oxoLCA
To clarify whether the oxidoreduction of 7-oxoLCA is mainly catalyzed by 11 ␤ -HSD1 or whether other enzymes are involved, we performed experiments with microsomes prepared from livers of wild-type and liver-specifi c 11 ␤ -HSD1-knockout mice ( 33 ). To distinguish between NADPHdependent enzymes that are oriented to the cytosol, including cytochrome P450 enzymes, aldoketoreductases, and shortchain dehydrogenase/reductase enzymes, and enzymes that are facing the endoplasmic reticulum (ER) lumen, such as 11 ␤ -HSD1, we compared activities in the presence of NADPH (for cytoplasmic enzymes) or G6P (for luminal enzymes). In the ER lumen, NADPH is regenerated by hexose-6-phosphate dehydrogenase (H6PDH), which is primarily dependent on G6P under physiological conditions ( 39 ). Thus, intact liver microsomes, where the luminal compartment is protected by the ER membrane, contain an endogenous NADPH-regenerating system ( 40 ). After 60 min of incubation of liver microsomes from wildtype mice with G6P, approximately 60% of initially supplied 7-oxoLCA was converted to about two times more CDCA than UDCA ( Fig. 2 ). The reaction was almost completely inhibited by the selective 11 ␤ -HSD1 inhibitor T0504 (also known as Merck-544) ( 41 ). When NADPH was added to the reaction mixture, we detected only traces of CDCA and UDCA, which were abolished by T0504, suggesting that this activity was due to a small fraction of microsomes with inverted orientation. Importantly, in experiments using liver microsomes from liver-specifi c 11 ␤ -HSD1-knockout mice, no conversion of 7-oxoLCA could be detected, regardless of whether G6P or NADPH was supplied to the reaction mixture. As control, we measured The results are mean ± SD (n = 3).  these fi ndings by comparing 11 ␤ -HSD1 from six species and observed remarkable species-specifi c differences. The catalytic activities of hamster and canine 11 ␤ -HSD1 resembled those of the human enzyme. These two species preferentially reduced 7-oxoLCA to CDCA, suggesting that they are suitable animal models to study potential physiological and toxicological consequences of 7-oxoLCA accumulation due to 11 ␤ -HSD1 defi ciency. In contrast, rat and mouse 11 ␤ -HSD1 were not stereoselective and produced substantial amounts of both 7 ␣ -and 7 ␤hydroxylated forms. Interestingly, for the substrate 7-oxocholesterol, the human, rat, and mouse enzymes generated 7 ␤ -hydroxycholesterol exclusively, whereas the hamster enzyme was not stereoselective and formed both 7 ␣ -and 7 ␤hydroxycholesterol ( 23 ). Importantly, recombinant guinea-pig 11 ␤ -HSD1 and guinea-pig liver microsomes were unable to reduce 7-oxoLCA, providing an explanation for the very high levels of 7-oxoLCA and 7-oxoLC-Gly in the circulation of guinea-pigs compared with the other species investigated ( Table 3 ). Our fi ndings are in line with an earlier study suggesting that 7-oxoLCA is a primary bile acid in this species, accounting for 30% of the total bile acids contained in bile ( 42 ). Moreover, the high IC 50 values of 7-oxoLCA, CDCA, and UDCA for 11 ␤ -HSD1dependent cortisone reduction ( Table 2 ) suggest that the binding of these bile acids is not optimally stabilized in the substrate pocket of the guinea-pig enzyme.
The comparison of the superimposed 3D structures of human and guinea-pig 11 ␤ -HSD1 alone did not explain the species differences in 7-oxoLCA metabolism (see supplementary Fig. I for alignment data). Therefore, 7-ox-oLCA and cortisone were docked to human and guinea-pig 11 ␤ -HSD1. In the reduction reaction, the hydrogen atoms are transferred to the ligand from the catalytic amino acid Tyr and the cofactor NADPH ( 43, 44 ). Therefore, the ligand needs to be oriented such that the carbonyl group is located in the vicinity of the catalytic Tyr and the cofactor. This is the case for cortisone, which binds similarly in human and guinea-pig 11 ␤ -HSD1 ( Fig. 3A , B ). However, 7-oxoLCA has different binding orientations in human and guinea-pig 11 ␤ -HSD1. In the human enzyme, as reported earlier ( 26 ), 7-oxoLCA binds in an inverted orientation with its A-ring aligning with the D-ring of cortisone and with the carbonyl group in position 7 in close proximity to the catalytic center, like the carbonyl group in position 11 of cortisone ( Fig. 3C ). In contrast, a fl ipped binding mode of 7-oxoLCA is predicted for the guinea-pig enzyme, where the carbonyl group points away from the catalytic amino acids ( Fig. 3D ). This fl ipped binding orientation may be explained by steric reasons: 7-oxoLCA fi ts to the binding pocket, but because of the bulky side-chain in position human, mouse (balb/c and C57bL/6), rat (Han Wistar and Sprague-Dawley), canine ( Canis familiaris , beagle bred), and hamster (golden Syrian). We observed very high levels of 7-oxoLCA and its glycine conjugate in guinea-pig serum ( Table 3 ). In contrast, in all other species that were investigated and that express 7-oxoLCA oxidoreductase activity, the circulating levels of 7-oxoLCA and its conjugates were very low. Taurine conjugation plays a minor role in guinea-pigs, and 7-oxoLC-Tau levels were negligible in this species.

Impact of 11 ␤ -HSD1 disruption on circulating levels of 7-oxoLCA and its conjugates in mice
We hypothesized that 11 ␤ -HSD1 defi ciency and, therefore, the inability to reduce 7-oxoLCA, leads to elevated circulating levels of 7-oxoLCA and its conjugated metabolites. To test this hypothesis, we measured the concentrations of 7-oxoLCA and its taurine and glycine conjugates in serum and in liver tissue of liver-specifi c and global 11 ␤ -HSD1-defi cient mice ( Table 4 ). The glycine conjugation plays a minor role in mice, in contrast to guinea-pigs, and concentrations of 7-oxoLC-Gly were very low in serum and liver tissue of liver-specifi c 11 ␤ -HSD1-knockout and wildtype mice. In liver tissue of liver-specifi c knockout and global knockout mice, 7-oxoLCA levels were approximately two times higher, although the differences did not reach statistical signifi cance. In contrast, intrahepatic 7-oxoLC-Tau was 7-fold and 4-fold higher in liver-specifi c 11 ␤ -HSD1defi cient mice and in global 11 ␤ -HSD1-defi cient mice compared with wild-type controls. As expected, 7-oxoLC-Tau levels were generally higher than 7-oxoLCA levels in liver tissue. Circulating levels of 7-oxoLC-Tau were somewhat higher than free 7-oxoLCA levels. Importantly, 7-oxoLCA concentrations signifi cantly increased (6-fold) in serum of liver-specifi c 11 ␤ -HSD1-defi cient mice and further increased (24-fold) in global-knockout mice. Circulating 7-oxoLC-Tau levels were 40-fold higher in liver-specifi c 11 ␤ -HSD1knockout mice compared with wild-type mice, and they were even somewhat lower in serum of global-knockout mice (20-fold higher than wild-type mice).

DISCUSSION
Recently, we reported that 7-oxoLCA and its taurine and glycine conjugates are substrates of human 11 ␤ -HSD1 and that they are effi ciently converted to the 7 ␣ -hydroxylated CDCA and, to a lesser extent, to the 7 ␤ -hydroxylated UDCA ( 26 ). At physiological pH, 11 ␤ -HSD1 exclusively catalyzed the reduction of 7-oxoLCA, and neither CDCA nor UDCA served as substrate. In the present study, we The results are expressed in nM as mean ± SD (n = 8). Underlined values represent below lower limit of quantifi cation. ND, not detected.
Analysis of the 7-oxoLCA levels in liver tissue indicated a trend to increase (approximately 2-fold) in liver-specifi c and global 11 ␤ -HSD1-knockout mice. Serum levels, however, were signifi cantly increased 6-fold and 24-fold. Thus, serum 7-oxoLCA is a better biomarker for 11 ␤ -HSD1 deficiency than is intrahepatic 7-oxoLCA. Importantly, a signifi cant increase in both hepatic and circulating levels of 7-oxoLC-Tau was obtained with a more pronounced increase in serum (40-fold and 20-fold) compared with liver tissue (7-fold and 4-fold). The reason why 7-oxoLC-Tau is higher in liver-specifi c compared with global knockout mice remains unclear but may be due to compensatory mechanisms or differences in gut microbiota. In liverspecifi c and global 11 ␤ -HSD1-knockout mice, unlike in guinea-pigs, taurine conjugation of 7-oxoLCA is the main route of its elimination in mice and glycine conjugation can be neglected. Based on these findings, we propose that 7-oxoLCA and 7-oxoLC-Tau, as well as the sum of these two bile acids, may represent suitable biomarkers of 11 ␤ -HSD1 defi ciency.
17 of the steroid backbone, the binding mode is unfavorable for the reduction reaction. However, when binding to the enzyme, 7-oxoLCA prevents other ligands from binding, explaining its action as a weak competitive inhibitor. These docking results support our biological fi ndings that cortisone is metabolized by both human and guinea-pig 11 ␤ -HSD1 but that 7-oxoLCA is not reduced by the guinea-pig enzyme.
Our data clearly show that guinea-pig 11 ␤ -HSD1 is able to reduce cortisone. Thus, glucocorticoid activation is similarly catalyzed in guinea-pigs and the other species investigated. 11 ␤ -HSD1 is a highly effi cient enzyme, and upon pharmacological application, cortisone and prednisone are almost completely converted to cortisol and prednisolone in the liver. Thus, the lack of the reduction of 7-oxoLCA (produced by intestinal microorganisms) by 11 ␤ -HSD1 in the liver may be responsible, at least in part, for the high circulating levels in this species. Because 11 ␤ -HSD1 seems to be the only enzyme responsible for the reduction of 7-oxoLCA, we hypothesized that this bile acid and its conjugates would accumulate in 11 ␤ -HSD1-defi cient mice. The results are expressed in nM as mean ± SD. Wild-type (WT) (n = 24), liver-specifi c 11 ␤ -HSD1-defi cient mice (LKO) (n = 16) and global 11 ␤ -HSD1-defi cient mice (KO) (n = 8). Data represent mean ± SD. * P р 0.05, ** P р 0.01, *** P р 0.001. other investigators ( 53 ). Similar to humans and guineapigs, glycine conjugation seems to prevail in hamsters ( 54 ), whereas in rodents and canines, taurine conjugation prevails and only minor amounts of glycine conjugated bile acids were detected, in line with observations by other investigators ( 52,55 ). Nevertheless, the applicability of 7-oxoLCA and its conjugates as potential biomarkers of 11 ␤ -HSD1 defi ciency and inhibition requires further investigation in vivo using specifi c 11 ␤ -HSD1 inhibitors or in clinical situations, such as patients suffering from cortisone reductase defi ciency.

CONCLUSIONS
Our results provide further evidence for an exclusive role of 11 ␤ -HSD1 in the reduction of 7-oxoLCA, which is generated by intestinal microorganisms, reaches the liver after entering the enterohepatic cycle, and is rapidly metabolized to CDCA and UDCA. Unlike glucocorticoids, the substrate 7-oxoLCA and particularly its taurine conjugate were markedly elevated in the serum of liver-specifi c 11 ␤ -HSD1-defi cient mice. Because 7-oxo and 7-oxoLC-Tau are not under the control of the HPA axis and they do not act through classical bile acid receptors, their concentrations do not seem to be regulated by feedback mechanisms, and they may be used as biomarkers to detect impaired 11 ␤ -HSD1 activity in clinically relevant situations. Future studies are needed to validate the use of these bile acids as biomarkers for reduced 11 ␤ -HSD1 activity, for example, in animal models and patients treated with 11 ␤ -HSD1 inhibitors and in patients with cortisone reductase defi ciency.
Biomarkers for impaired 11 ␤ -HSD1 activity are of great interest for clinical researchers as well as for the pharmaceutical industry to assess the in vivo effi cacy of therapeutic inhibitors. Despite the fact that their effi ciency can also be assessed through traditional and indirect clinical markers, such as fasting plasma glucose and lipid profi les, biomarkers for direct assessment of 11 ␤ -HSD1 inhibition in preclinical and clinical studies are still unavailable. It has been reported that 11 ␤ -HSD1-defi cient mice have slightly elevated plasma levels of corticosterone and adrenocorticotropic hormone (ACTH); however, these changes seem to be strain-dependent ( 45,46 ). Interestingly, circulating corticosterone levels were not or were only mildly elevated in global 11 ␤ -HSD1-knockout, liver-specific 11 ␤ -HSD1knockout, H6PDH-knockout, and 11 ␤ -HSD1/H6PDH double-knockout mice when compared with wild-type mice ( 33,(47)(48)(49). Therefore, it is reasonable to assume that glucocorticoid levels are not substantially altered after a therapeutic intervention with 11 ␤ -HSD1 inhibitors, thus excluding the applicability of glucocorticoid levels as a biomarker of 11 ␤ -HSD1 defi ciency or enzyme inhibition. Likewise, urinary levels of 11-dehydrocorticosterone (11-DHC), corticosterone, and their tetrahydro metabolites are also of limited value as a measure of hepatic inhibition because they were not signifi cantly changed in liver-specifi c 11 ␤ -HSD1 knockouts. Similarly, deletion of one allele in heterozygous 11 ␤ -HSD1-knockout mice, a scenario similar to 50% enzyme inhibition, is insuffi cient to elicit a signifi cant change in urinary corticosteroid metabolites compared with wild-type controls ( 33,47 ). H6PDH-knockout, 11 ␤ -HSD1-knockout, and 11 ␤ -HSD1/H6PDH double-knockout mice do have altered urinary biomarkers, but this refl ects total body loss or 11 ␤ -HSD1 reductase activity, and the value of these biomarkers to determine tissue-specifi c effects remains uncertain. It will be interesting to investigate in a follow-up study whether 7-oxoLCA and 7-oxoLC-Tau levels are elevated in heterozygous 11 ␤ -HSD1-defi cient mice and in mice treated with selective 11 ␤ -HSD1 inhibitors.
Importantly, although 7-oxoLCA and 7-oxoLC-Tau are substrates of 11 ␤ -HSD1, they are independent of the HPAaxis feedback regulation, and they are not involved in adaptive metabolic changes. Further, they are devoid of overt toxicity, yet they have been shown to reduce the biliary lithogenic index ( 50 ). Moreover, 7-oxoLCA and 7-oxoLC-Tau are not ligands for the vitamin D receptor (VDR) ( 51 ), and they do not activate farnesoid X receptor (FXR) to the same extent as do CDCA and UDCA (data not shown). Thus, it is unlikely that these bile acids infl uence the activity of these nuclear receptors. Due to the fact that, in humans, circulating glycine-conjugated bile acids are more abundant than taurine-conjugated metabolites ( 52 ), we propose that 7-oxoLCA, 7-oxoLC-Gly, and the sum of these two bile acids may be used as biomarkers for impaired 11 ␤ -HSD1 function. In guinea-pigs, as in humans, circulating glycine-conjugated bile acids are more abundant than taurine-conjugated metabolites. Only traces of 7-oxoLC-Tau were detected in guinea-pig serum, whereas the concentration of 7-oxoLC-Gly was approximately 5.5 µM ( Table 2 ). These fi ndings are in line with observations by