Abstract
Ertugliflozin is primarily cleared through UDP-glucurosyltransferase (UGT)–mediated metabolism (86%) with minor oxidative clearance (12%). In vitro phenotyping involved enzyme kinetic characterization of UGTs or cytochrome P450 enzymes catalyzing formation of the major 3-O-β-glucuronide (M5c) and minor 2-O-β-glucuronide (M5a), monohydroxylated ertugliflozin (M1 and M3), and des-ethyl ertugliflozin (M2) metabolites in human liver microsomes (HLMs). Fractional clearance (fCL) from HLM intrinsic clearance (CLint) indicated a major role for glucuronidation (fCL 0.96; CLint 37 µl/min per milligram) versus oxidative metabolism (fCL 0.04; CLint 1.64 µl/min per milligram). Substrate concentration at half-maximal velocity (Km), maximal rate of metabolism (Vmax), and CLint for M5c and M5a formation were 10.8 µM, 375 pmol/min per milligram, and 34.7 µl/min per milligram and 41.7 µM, 94.9 pmol/min per milligram, and 2.28 µl/min per milligram, respectively. Inhibition of HLM CLint with 10 µM digoxin or tranilast (UGT1A9) and 3 µM 16β-phenyllongifolol (UGT2B7/UGT2B4) resulted in fraction metabolism (fm) estimates of 0.81 and 0.19 for UGT1A9 and UGT2B7/UGT2B4, respectively. Relative activity factor scaling of recombinant enzyme kinetics provided comparable fm for UGT1A9 (0.86) and UGT2B7 (0.14). Km and Vmax for M1, M2, and M3 formation ranged 73.0–93.0 µM and 24.3–116 pmol/min per milligram, respectively, and was inhibited by ketoconazole (M1, M2, and M3) and montelukast (M2). In summary, ertugliflozin metabolism in HLMs was primarily mediated by UGT1A9 (78%) with minor contributions from UGT2B7/UGT2B4 (18%), CYP3A4 (3.4%), CYP3A5 (0.4%), and CYP2C8 (0.16%). Considering higher ertugliflozin oxidative metabolism (fCL 0.12) obtained from human mass balance, human systemic clearance is expected to be mediated by UGT1A9 (70%), UGT2B7/UGT2B4 (16%), CYP3A4 (10%), CYP3A5 (1.2%), CYP2C8 (0.5%), and renal elimination (2%).
SIGNIFICANCE STATEMENT This manuscript describes the use of orthogonal approaches (i.e., enzyme kinetics, chemical inhibitors, and recombinant enzymes) to characterize the fraction of ertugliflozin metabolism through various UDP-glucuronosyltransferase (UGT) and cytochrome P450 (CYP) enzyme-mediated pathways. Phenotyping approaches routinely used to characterize CYP hepatic fractional metabolism (fm) to estimate specific enzymes contributing to overall systemic clearance were similarly applied for UGT-mediated metabolism. Defining the in vitro metabolic disposition and fm for ertugliflozin allows risk assessment when considering potential victim-based drug-drug interactions perpetrated by coadministered drugs.
Introduction
Ertugliflozin (Fig. 1) is an orally administered, potent, and selective sodium-dependent glucose cotransporter (SGLT) 2 inhibitor (Mascitti et al., 2011) approved for the treatment of adults with type 2 diabetes mellitus (T2DM) at single daily doses of 5 or 15 mg as either monotherapy, fixed-dose combinations with metformin or sitagliptin, or in combination with antidiabetic medications (https://www.accessdata.fda.gov/scripts/cder/daf/index.cfm?event=overview.process&applno=209803; http://www.ema.europa.eu/ema/index.jsp?curl=pages/medicines/human/medicines/004315/human_med_002241.jsp&mid=WC0b01ac058001d124; Markham, 2018). Diabetes mellitus is a highly prevalent disease; there are approximately 463 million adults living with diabetes mellitus in 2019, a number expected to rise to 700 million by 2045 (https://www.idf.org/aboutdiabetes/what-is-diabetes/facts-figures.html). According to the National Diabetes Statistics Report from 2020, 34.2 million people in the United States have T2DM, of whom 7.3 million are undiagnosed (https://www.cdc.gov/diabetes/data/statistics/statistics-report.html). SGLT2 inhibitors such as ertugliflozin are indicated to treat T2DM as an adjunct to diet and exercise to improve glycemic control through inhibition of glucose reabsorption in the kidneys (Terra et al., 2017). Glucose homeostasis in the kidney is maintained by SGLT1, SGLT2, and SGLT3 transporters located in the brush border of the epithelial cells of proximal renal tubules (Wright, 2001). SGLT2 is the primary transporter responsible for glucose reabsorption (80%–90%) with a lesser role of SGLT1 (10%–20%), substantiated by compelling clinical data (Plosker, 2014; Deeks and Scheen, 2017; Frampton, 2018), preclinical investigation (Freitas et al., 2008; Han et al., 2008), and in vitro human kidney experiments (Hummel et al., 2011; Ghezzi and Wright, 2012; Vrhovac et al., 2015). Ertugliflozin is >2000-fold more selective as a SGLT2 inhibitor over SGLT1 (Mascitti et al., 2011). After a 5 or 15 mg once daily dose of ertugliflozin for 26 weeks to subjects with T2DM demonstrating inadequate glycemic control despite diet and exercise, fasting plasma glucose levels decreased and placebo-adjusted glycated hemoglobin levels decreased 0.99% and 1.16%, respectively (Terra et al., 2017). SGLT2 inhibitors demonstrated cardioprotective and renoprotective effects with treatment resulting in significant reductions in fasting plasma glucose, glycated hemoglobin, body weight, and blood pressure, as well as improved lipid profiles (Hollander et al., 2019; Tentolouris et al., 2019).
Ertugliflozin human relevant metabolites identified from a mass-balance study.
Absorption of ertugliflozin is rapid and complete with oral exposure increasing in a dose-proportional manner from 0.5 to 300 mg and an absolute oral bioavailability of ∼100% (Raje et al., 2018). The absorption, distribution, metabolism, and excretion (ADME) of ertugliflozin was determined after a single oral dose of 25-mg [14C]ertugliflozin (100 µCi) to healthy male subjects, and structures of key metabolites identified are presented in Fig. 1 (Miao et al., 2013). The primary clearance mechanism is metabolism with metabolite profiling in excreta indicating that the major route is glucuronidation with a minor contribution from oxidative metabolism. The major metabolite eliminated in human urine was ertugliflozin-3-O-β-glucuronide (M5c, previously M4c) (Miao et al., 2013), and a minor glucuronide was identified as ertugliflozin-2-O-β-glucuronide (M5a) (Fig. 1). Oxidative metabolism was minor with monohydroxylated ertugliflozin (M1 and M3) and des-ethyl ertugliflozin (M2) present in urine and feces. Ertugliflozin is completely absorbed (fraction absorbed of 111%) (Raje et al., 2018) with unchanged parent drug excreted in feces (∼40%) presumably the result of glucuronide metabolite(s) that are excreted in the bile, hydrolyzed back to the parent drug in the intestine, and eliminated via the feces. Renal excretion of unchanged ertugliflozin was low (<2% of dose). Overall, the contribution of glucuronidation (86%), oxidative metabolism (12%), and renal excretion (2%) to ertugliflozin systemic clearance was based on the scaling of the human ADME study findings (Supplemental Table 1).
Preliminary reaction phenotyping or recombinant enzyme mapping indicated that UDP-glucuronosyltransferases (UGTs) 1A9 and 2B7 were able to catalyze ertugliflozin glucuronidation, and oxidative metabolism involved CYP3A4 (Kalgutkar et al., 2011). Since oxidative metabolism is a minor clearance (CL) pathway, clinically significant drug-drug interactions are not expected after coadministration with cytochrome P450 (CYP) inhibitors. Detailed quantitative UGT reaction phenotyping is required to predict the clinical drug-drug interaction risk when ertugliflozin is coadministered with UGT inhibitors. The objective of this study was to characterize the enzymes involved in the metabolism of ertugliflozin using in vitro human matrices. Reaction phenotyping for UGT and CYP enzymes was conducted using enzyme kinetics, recombinant enzyme systems, and isoform-selective chemical inhibitors.
Materials and Methods
Materials
Ertugliflozin or (1S,2S,3S,4R,5S)-5-[4-chloro-3-(4-ethoxybenzyl)phenyl]-1-hydroxymethyl-6,8-dioxabicyclo[3.2.1]octane-2,3,4-triol, M1 and M3, M2, M5c, M5a, and 16β-phenyllongifolol were obtained from Pfizer Global Research and Development Chemistry. Alamethicin, UDP glucuronic acid (UDPGA), MgCl2, NADPH, diclofenac, terfenadine, furafylline, 2-phenyl-2-(1-piperdinyl)propane (PPP), montelukast, sulfaphenazole, benzylnirvanol, quinidine, ketoconazole, CYP3cide, tranilast, digoxin, 1 M potassium phosphate dibasic solution, 1 M potassium phosphate monobasic solution, Tris, bovine serum albumin (BSA; crude BSA product no. A7906), and DMSO were purchased from Sigma Aldrich (St. Louis, MO). Pooled human liver microsomes (HLMs) prepared from 50 mixed-gender donors and recombinantly expressed human UGT enzymes (rhUGTs) (1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15, 2B17, and nontransfected control) were obtained from Corning Life Sciences (Woburn, MA). Pooled human kidney microsomes (HKMs) prepared from eight mixed-gender donors were obtained from Xenotech (Kansas City, KS). HLM 3A5*1*1, HLM 3A5*3/*3 microsomes pools were prepared as previously described (Tseng et al., 2014). Recombinantly expressed human CYPs (1A2, 2B6, 2C8, 2C9, 2C19, 2D6, 3A4, and 3A5) were obtained from either Invitrogen (Life Technologies, Waltham, MA) or BD BioSciences (Franklin Lakes, NJ).
General UGT Incubation Conditions
UGT incubations were performed in a buffer mixture containing 100 mM Tris HCl buffer (pH 7.5 at 37°C) with or without 2% BSA, 5 mM MgCl2, and microsomal or recombinant protein. The buffer mixture was preincubated with alamethicin (10 µg/ml) on ice for 15 minutes to allow for alamethicin pore formation. The incubation mixture was warmed to 37°C prior to the addition of ertugliflozin. Reactions were initiated by the addition of 5 mM UDPGA and terminated by protein precipitation with acetonitrile containing internal standard (IS).
UGT Substrate Saturation in HLMs and rhUGTs
Using the general UGT incubation conditions described above, reactions with 0.25 mg/ml protein (i.e., HLM, rhUGT1A9, and rhUGT2B7) and ertugliflozin (1–1000 µM for HLMs, 0.33–1000 µM for rhUGTs) were initiated by the addition of UDPGA and terminated at 30 or 40 minutes by protein precipitation with acetonitrile containing IS. All samples were analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS), where M5c and M5a formation were quantified against standard curves. The experimental conditions selected for enzyme kinetics were chosen after conduct of preliminary reaction rate experiments to ensure linear reaction velocity with respect to protein concentration and incubation time. HLM and rhUGT incubations for M5a formation were performed in triplicate. Incubations for M5c formation were conducted on several days (n = 6) in both HLMs and rhUGTs.
UGT Phenotyping
Ertugliflozin was incubated in the presence of 13 rhUGTs (1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B10, 2B15, and 2B17) and nontransfected control to assess which isoforms contribute to glucuronide formation. The general incubation mixture without 2% BSA contained 0.5 mg/ml protein, either rhUGT or vector control, and ertugliflozin (7, 77, and 770 µM) representing approximately 0.1×, 1×, and 10× the HLM apparent substrate concentration at half-maximal velocity (Km) for the primary glucuronidation (M5c) pathway (Table 1). The incubation mixtures were prewarmed to 37°C prior to the addition of UDPGA and terminated over a time course by protein precipitation with acetonitrile containing IS. Samples were analyzed by LC-MS/MS, where ertugliflozin depletion and both M5c and M5a formation were measured by analyte/IS area ratios. All incubations were performed in duplicate.
Enzyme kinetic parameters for ertugliflozin glucuronide metabolite (M5a and M5c) formation in pooled HLMs
Data are expressed as the means ± S.D. from triplicate experiments, except M5c formation with 2% BSA (n = 6).
Ertugliflozin (5 µM) was incubated with HLMs in the presence of 2% BSA with and without UGT chemical inhibitors for a quantitative assessment of UGT isoform contribution. General incubation conditions and sample preparation were similar to those described for UGT substrate saturation experiments. Reversible inhibitors were incubated at the following concentrations: 10 µM tranilast (UGT1A9), 10 µM digoxin (UGT1A9), and 3 µM 16β-phenyllongifolol (UGTs 2B4/2B7). Control incubations were performed in the absence of chemical inhibitors (HLM control). The incubation mixture was warmed to 37°C for 5 minutes prior to the addition of ertugliflozin. Reactions were initiated by the addition of UDPGA and terminated over a time course by protein precipitation with acetonitrile containing IS. Samples were analyzed using LC-MS/MS, where M5a and M5c formation were quantified against standard curves. Selective chemical inhibition experiments were conducted in duplicate.
CYP Substrate Saturation in HLMs and Recombinantly Expressed Human CYP Enzymes
HLM incubation mixtures contained 100 mM potassium phosphate buffer (pH 7.4), 3 mM MgCl2, 0.1 mg/ml HLMs, and ertugliflozin (0.01–1000 µM). Incubations were warmed to 37°C prior to the addition NADPH (1.2 mM). Reactions were terminated after 10 minutes by transferring aliquots of the incubation mixture to acetonitrile containing IS. Incubations were conducted in duplicate. For substrate saturation experiments in recombinantly expressed human CYP enzymes (rhP450s) (2C8, 3A4, and 3A5), incubation conditions were similar to those described previously in HLMs. The rhP450 concentration was 10 pmol/ml (0.11 mg/ml) protein, the incubation time was 10 minutes, and ertugliflozin concentrations ranged from 0.01 to 500 µM. All samples were analyzed by LC-MS/MS and quantified against standard curves of M1, M2, and M3. The experimental conditions selected for enzyme kinetics were chosen after conduct of preliminary reaction rate experiments in HLMs to ensure linear reaction velocity with respect to protein concentration and incubation time.
CYP Phenotyping
Ertugliflozin was incubated in the presence of eight rhP450 (1A2, 2B6, 2C8, 2C9, 2C19, 2D6, 3A4, and 3A5) at a concentration of 10 pmol/ml. Potassium phosphate buffer (100 mM), rhP450s or nontransfected control, and ertugliflozin (10 µM for M1 and M2 formation, or 50 µM for M3 formation) were prewarmed to 37°C. Incubations were initiated with 1.2 mM NADPH followed by addition of substrate. Reactions were terminated at 10 minutes by protein precipitation with acetonitrile containing IS. The recombinant CYP panel was incubated in triplicate. All samples were analyzed by LC-MS/MS and quantified against standard curves of M1, M2, and M3.
Ertugliflozin (10 or 50 µM) was incubated with HLMs in the presence of CYP selective chemical inhibitors for the determination of CYP isoform contribution. General incubation conditions and sample preparation were similar to those described for CYP substrate saturation experiments. Reversible inhibitors were incubated at the following concentrations: 0.64 µM montelukast (CYP2C8), 10 µM sulfaphenazole (CYP2C9), 3 µM benzylnirvanol (CYP2C19), 1 µM quinidine (CYP2D6), and 1 µM ketoconazole (CYP3A4/5). Inactivators were preincubated at the following concentrations: 10 µM furafylline (CYP1A2), and 3 µM PPP (CYP2B6). For incubations containing reversible inhibitors, microsomes were preincubated with inhibitors for 5 minutes prior to initiating the reactions by the addition of NADPH or buffer minus NADPH control followed immediately by ertugliflozin. Incubations containing the inactivator furafylline or PPP were preincubated with NADPH and microsomes for 20 minutes to achieve complete inactivation of the respective CYP isoform prior to the addition of ertugliflozin. Control incubations were performed in the absence of chemical inhibitors (HLM control). Selective chemical inhibition experiments were conducted in triplicate. All samples were analyzed by LC-MS/MS and quantified against standard curves of M1, M2, and M3.
To discern the relative contributions of CYPs 3A4 and 3A5 to the metabolism of ertugliflozin, additional experiments were conducted in a pooled lot of HLMs prepared from individual HLM lots selected for either the *1/*1 genotype (HLM 3A5 *1/*1) associated with CYP3A5 extensive metabolizers or *3/*3 genotype (HLM 3A5 *3/*3) associated with CYP3A5 poor metabolizers (Tseng et al., 2014). Microsomal incubations with ertugliflozin (10 and 50 µM) were conducted as described above for HLMs (0.1 mg/ml protein, 10-minute incubation) and included 1 µM ketoconazole (CYP3A4/5 inhibitor) or 1 µM CYP3cide (CYP3A4 inhibitor). Pretreatment of the incubates varied based on the presence of the competitive inhibitor ketoconazole (no preincubation with NADPH) or the inactivator CYP3cide (10-minute preincubation with NADPH). The control incubation for these experiments was conducted with no preincubation. These experiments were conducted in duplicate. All samples were analyzed by LC-MS/MS and quantified against standard curves of M1, M2, and M3.
Ertugliflozin Clearance in HLMs and HKMs by Substrate Depletion
The clearance of ertugliflozin was investigated in HLMs and HKMs in the presence of the individual and combined cofactors (e.g., UDPGA and/or NADPH) to estimate the in vitro fraction of metabolic clearance (fCL) by UGT or CYP enzyme families. Incubations were performed using the general UGT incubation conditions described above with and without 2% BSA containing 2 mg/ml HLMs or HKMs. The incubation mixture was warmed to 37°C prior to the addition of UDPGA (5 mM) or NADPH (1.6 mM), either as a mixture or individually. Reactions were initiated by the addition of ertugliflozin (1 µM final) and terminated over a time course from 0 to 180 minutes by protein precipitation with acetonitrile containing IS. All incubations were performed in duplicate. Samples were analyzed by LC-MS/MS and quantified against standard curves of ertugliflozin, M5c, and M5a.
Determination of Unbound Fraction of Ertugliflozin in HLMs
Using the equilibrium dialysis method previously described (Di et al., 2012), the fraction unbound (fu) in HLMs was determined using 1 µM ertugliflozin. For UGT HLM incubation conditions, binding was determined with and without 2% BSA at a protein concentration of 0.25 mg/ml HLMs. Additional components of the donor chamber mimicked the incubation mixture described in Materials and Methods on UGT Substrate Saturation in HLMs and rhUGTs, except that UDPGA was omitted. The receiver chamber contained Tris buffer containing 5 mM MgCl2. For binding under CYP incubation conditions, fu was determined in the incubation components described in Materials and Methods on CYP Substrate Saturation in HLMs and Recombinantly Expressed Human CYP Enzymes, except that NADPH was omitted and the protein concentration was 0.76 mg/ml HLM. Samples were placed in a CO2 incubator (5% CO2, 75% relative humidity) for 6 hours at 37°C. Ertugliflozin recovery, stability, and binding samples were matrix matched and protein precipitated with acetonitrile containing IS. Samples were analyzed for ertugliflozin and the IS by LC-MS/MS. Peak area ratio of analyte/internal standard was used to calculate fu, recovery, and stability.
Data Analysis
Clearance by Substrate Depletion.
The unbound in vitro intrinsic clearance (CLint,u) and scaled in vivo whole organ intrinsic clearance (CLint,sc,u) was determined at a single ertugliflozin concentration and determined by depletion using the following equations:(1)
(2)
(3)where the elimination rate constant (kel) is negative slope of natural log of the average peak area ratio versus time, and fu is the fraction unbound in the incubation (fu,inc). For HLMs, X and Y were 45 and 21, respectively. For HKMs, X and Y were 12.8 and 4.5, respectively. The CLint,u was also determined by metabolite formation at a single ertugliflozin concentration using the equation
(4)where formation rate is the slope of metabolite formation. Using eq. 3, the CLint,sc,u was calculated and used to estimate systemic clearance for the relevant metabolic organ. The fCL for renal metabolism was calculated using eq. 5:
(5)
Clearance in HLMs Using Enzyme Kinetic Parameters.
In general, substrate concentration [S] and velocity (v) data were fit to the appropriate enzyme kinetic model by nonlinear least-squares regression analysis to derive the apparent enzyme kinetic parameters, maximal rate of metabolism (Vmax), and Km or S50. The Michaelis-Menten model (eq. 6), the substrate activation model (eq. 7) which incorporates the Hill coefficient (n), and the substrate inhibition model (eq. 8) were evaluated for best fit of the data. Best fit was determined using R2 goodness of fit, Eadie-Hofstee fit, and Akaike information criterion.(6)
(7)
(8)The exception was formation of M1 in rhCYP2C8. An unweighted linear fit was used the calculate the formation slope [i.e., apparent intrinsic clearance (CLint)] since M1 formation was not saturable (Supplemental Fig. 3) with the highest concentration of ertugliflozin in incubation (500 µM). The CLint,u was calculated using either eq. 9 for typical kinetics or for M1 formation in rhCYP2C8, CLint was equivalent to the linear slope. Equation 10 was used to calculate CLint,u atypical kinetics incorporating the Hill coefficient.
(9)
(10)The contribution of each metabolic pathway (as represented by each metabolite) to the total clearance of ertugliflozin (fCL) was calculated using eq. 11:
(11)where CLint,u,x is the unbound apparent intrinsic clearance for metabolite x, and CLint,tot,u is the sum of unbound scaled intrinsic apparent clearance values for all metabolites measured.
Fraction of Metabolism Using Selective Chemical Inhibitors in HLMs.
The effect of selective chemical inhibitors on the formation of each metabolite was calculated using eq. 12:(12)To determine the isoform specific contribution to total clearance, metabolite specific contributions were adjusted for fractional clearance:
(13)where Inhibitionx is the contribution of a specific isoform to the formation of metabolite x, and fm,CLtot,u,x is the fraction of total clearance represented by metabolite x. Since two inhibitors were used for UGT1A9, the mean % inhibition was used for the final fraction of metabolism (fm) estimate.
Using Relative Activity Factor or Intersystem Extrapolation Factor to Scale Recombinantly Expressed Enzyme Clearance to HLM Clearance.
The rhUGT 1A9 and 2B7 CLint,u was calculated using eq. 9, which was then scaled to HLM CLint,u using eq. 14:(14)where RAF is the relative activity factor for an individual enzyme based on selective probe substrate CLint values (eq. 15), rhCLint is the unbound intrinsic apparent clearance in rhUGTs, and CLint is the unbound intrinsic apparent clearance scaled to liver. The RAF value for UGT1A9 (0.8) was generated using mycophenolic acid as the substrate. The UGT2B7 RAF value of 2.4 was generated using zidovudine.
(15)For rhCYP 3A4, 3A5, and 2C8 incubations, the rhCLint,u was calculated (eq. 9). An intersystem extrapolation factor (ISEF) was applied in conjunction with the abundance of the CYP in the liver (Supplemental Table 2) to scale to a hepatic CLint,u.
(16)
Delineation of the Fraction Metabolized via CYP3A4 versus CYP3A5.
The linear portion of the mean regression line (n = 2) was used to determine a single slope of each incubation condition (kel). The slope of the line was used to determine the half-life (t1/2; minutes) at each incubation condition (eq. 1), which was then extrapolated to the CLint, like eq. 2. The contribution by CYP3A5 was calculated using the following equations (Tseng et al., 2014):(17)
(18)where CLint,cnt is the clearance in HLMs without inhibitor.
Results
UGT and CYP Substrate Saturation in HLMs.
The metabolite formation kinetics for ertugliflozin glucuronides M5c (major) and M5a (minor) were determined after incubation of ertugliflozin in HLMs in the absence and presence of 2% BSA (Fig. 2; Table 1). In the absence of BSA, M5a exhibited substrate inhibition kinetics, and the estimated apparent Km value (3130 µM) exceeded the highest ertugliflozin concentration in incubation (1000 µM). All other profiles displayed typical Michaelis-Menten kinetics. In general, the Km,u values were lower in the presence of 2% BSA for M5c (4-fold) and M5a (41-fold) resulting in Km,u values of 10.8 and 41.7 µM, respectively. Inclusion of BSA had minimal effect on M5c Vmax, whereas M5a Vmax decreased in the presence of BSA from 734 ± 363 to 94.9 ± 1.2 pmol/min per milligram. Overall, the formation CLint,u for M5c and M5a increased >3.8-fold in the presence of BSA primarily driven by a lower Km,u. The highest total CLint,u for glucuronidation (37 µl/min per milligram) was observed in the presence of BSA with M5c (34.7 µl/min per milligram) the primary and M5a (2.28 µl/min per milligram) the minor metabolic route.
Substrate saturation profiles of ertugliflozin glucuronide metabolites, M5c and M5a, derived from pooled HLMs, rhUGT1A9, and rhUGT2B7. HLM data were generated in the absence (open circles) and presence of 2% BSA (blue circles), whereas rhUGT kinetics were conducted in the presence of 2% BSA as described under Materials and Methods. Data are expressed as the means ± S.D. from triplicate experiments, except M5c formation with 2% BSA (n = 6).
CYP enzyme kinetic and intrinsic clearance values were determined for the formation of M1, M2, and M3 in HLMs (Fig. 3; Table 2). Visual inspection of the Eadie-Hofstee plots for the M1, M2, and M3 reaction rates demonstrated nonlinear profiles consistent with activation kinetics. For all three metabolites, a Michaelis-Menten equation including a Hill coefficient best described the formation kinetics and was selected for determination of enzyme kinetic parameters. S50,u, Vmax, and Hill coefficient values were 73.0 µM, 166 pmol/min per milligram, and 1.54 for M1, 51.6 µM, 31.4 pmol/min per milligram, and 1.59 for M2, and 93.0 µM, 24.3 pmol/min per milligram, and 1.56 for M3, respectively. Based on these kinetic parameters, the respective M1, M2, and M3 CLint,sc,u values were 1.12, 0.297, and 0.129 ml/min per kilogram.
Substate saturation profile of ertugliflozin oxidative metabolites M1, M2, and M3 derived from pooled HLMs. The kinetic profiles exhibited a sigmoidal behavior, as illustrated by the Eadie-Hofstee graphs. Data are expressed as the means from duplicate experiments.
Enzyme kinetic parameters for ertugliflozin oxidative metabolite (M1, M2, and M3) formation in pooled HLMs
Data are expressed as the means from duplicate experiments after HLM incubations as described under Materials and Methods.
The overall UGT-mediated CLint,sc,u in the presence of BSA was 35.0 ml/min per kilogram, whereas the overall CYP mediated CLint,sc,u for ertugliflozin was 1.55 ml/min per kilogram. With a combined UGT- and CYP mediated CLint,sc,u of 36.4 ml/min per kilogram, the in vitro fm for UGT- and CYP mediated metabolism was 0.96 and 0.04, respectively (Fig. 6). Using the UGT-mediated CLint,sc,u without 2% BSA (9.07 ml/min per kilogram), the combined UGT- and CYP mediated CLint,sc,u decreased to 10.6 ml/min per kilogram.
UGT Phenotyping of Ertugliflozin.
UGT phenotyping or enzyme mapping with rhUGTs (UGT1A1, 1A3, 1A4, 1A6, 1A7, 1A8, 1A9, 1A10, 2B4, 2B7, 2B10, 2B15, and 2B17) was conducted after incubation of ertugliflozin with rhUGTs in the absence of BSA to investigate formation of M5c and M5a at ertugliflozin concentrations approximately 0.1×, 1× and 10× the primary glucuronide (M5c) HLM Km (7, 77, and 770 µM ertugliflozin) (Fig. 4; Supplemental Table 3). At low substrate concentrations (7 µM ertugliflozin), four UGTs (1A9, 1A3, 1A7, and 1A8) were able to catalyze formation of the major glucuronide (M5c) with UGT1A9 as the apparent dominant isoform. At higher ertugliflozin concentrations, M5c was quantifiable in other UGT1A enzyme incubations (i.e., 1A1, 1A3, 1A4, 1A6, 1A7, 1A8) and UGTs 2B7, 2B15 and 2B17, albeit trace or minor relative to UGT1A9. The minor glucuronide (M5a) formation as was quantifiable with UGTs 1A9, 2B7, and 2B4 at low substrate concentration and relative product formation increased at higher substrate concentrations with metabolic turnover observed for UGTs 1A4, 1A1, 1A3, 1A7, and 1A8.
Formation of ertugliflozin glucuronide metabolites M5c and M5a incubated with 13 rhUGT enzymes in the absence of 2% BSA. Ertugliflozin was incubated at concentrations of 7, 77, and 770 µM representing 0.01×, 1× and 10× the M5c (Table 1). Data are expressed as the means from duplicate experiments.
To provide a quantitative assessment of the contribution of the apparently major UGTs 1A9, 2B4, and 2B7 observed in UGT enzyme mapping (Fig. 4) toward M5c and M5a formation, the effects of chemical inhibitors on the metabolism of ertugliflozin in HLMs were investigated (Supplemental Fig. 1; Table 3). Formation of M5c was inhibited 88.0% by tranilast (UGT1A9), 81.8% by digoxin (UGT1A9), and 13.9% by 16β-phenyllongifolol (UGT2B4/2B7). The inhibition results were scaled to 100% to achieve a mean fm UGT1A9 and UGT2B4/2B7 of 0.86 and 0.14, respectively, for formation of the major in vitro glucuronide. For M5a, mean UGT1A9 inhibition and UGT2B4/2B7 inhibition resulted in fm values of 0.20 and 0.80, respectively. The combined inhibition of overall in vitro glucuronidation in HLMs resulted in a fm for UGT1A9 and UGT2B4/2B7 of 0.81 and 0.19, respectively.
Inhibition of ertugliflozin (5 µM) glucuronide formation in pooled HLMs using UGT-selective chemical inhibitors
Data are expressed as the means from duplicate experiments after HLM incubations as described under Materials and Methods. The relative contribution values (fm) were calculated from data scaled to 100% inhibition.
Enzyme kinetic characterization describing formation of M5c and M5a in rhUGT 1A9 and 2B7 were performed in the presence of 2% BSA (Fig. 2; Table 4). All profiles exhibited typical Michaelis-Menten kinetics. For M5c, the Km,u and Vmax in rhUGTs 1A9 and 2B7 was 14.7 µM and 375 ± 5 pmol/min per milligram (rhCLint,u = 25.4 µl/min per milligram), and 22.4 µM and 2.63 ± 0.18 pmol/min per milligram (rhCLint,u = 0.177 µl/min per milligram), respectively. For M5a, the Km,u and Vmax in rhUGTs 1A9 and 2B7 were 14.9 µM and 5.09 ± 0.15 pmol/min per milligram (rhCLint,u = 0.343 µl/min per milligram), and 25.6 µM and 6.55 ± 0.28 pmol/min per milligram (rhCLint,u = 0.256 µl/min per milligram), respectively. The rhCLint,u was scaled to hepatic CLint,u using the appropriate RAF (Table 4). For UGT1A9, a mycophenolic acid generated RAF of 0.8 (Lin et al., 2015) was used to scale M5c and M5a formation to a hepatic microsomal CLint,u of 20.4 and 0.274 µl/min per milligram, respectively. Zidovudine was used to generate a UGT2B7 RAF of 2.4 (Lin et al., 2014). UGT2B7 mediated hepatic CLint,u for M5c and M5a formation was 2.81 and 0.613 µl/min per milligram, respectively. Overall, addition (M5c and M5a) of the RAF-adjusted scaled HLM CLint for UGT1A9 (20.7 µl/min per milligram) and UGT2B7 (3.42 µl/min per milligram) results in the combined UGT-mediated CLint,u of 24.1 µl/min per milligram (CLint,sc,u = 22.7 ml/min per kilogram) or RAF-adjusted fm UGT1A9 and UGT2B7 of 0.86 and 0.14, respectively. (The rhUGT CLint prior to RAF scaling yields fm UGT1A9 and UGT2B7 of 0.986 and 0.014, respectively).
Enzyme kinetic parameters for ertugliflozin metabolite formation in recombinantly expressed CYP and UGT enzymes
Data are expressed as the means ± S.D. from triplicate incubations or means from duplicate experiments. Recombinant enzyme kinetics were scaled to HLM CLint using the HLM Scaling Factors (RAF for UGTs and ISEF for CYPs) as described under Materials and Methods.
CYP Phenotyping of Ertugliflozin.
CYP reaction phenotyping or enzyme mapping was conducted examining ertugliflozin metabolite formation in eight rhP450s (CYP1A2, 2B6, 2C8, 2C9, 2C19, 2D6, 3A4, and 3A5), which revealed M1 formation was quantifiable in rhCYPs 1A2, 3A4, and 3A5, M2 was quantifiable in rhCYPs 1A2, 2C8, 2C19, 2D6, 3A4, and 3A5, and M3 was quantifiable in rhCYPs 2D6, 3A4, and 3A5 (Supplemental Fig. 2).
The effect of CYP selective chemical inhibitors on the metabolism of ertugliflozin in HLMs was also investigated. The metabolic formation rates of M1, M2 and M3 in HLMs were most significantly inhibited in the presence of ketoconazole resulting in percent inhibition values of 91.9%, 86.8%, and >81.8%, respectively (Table 5). After scaling inhibition responses to 100%, fm for CYP3A4/5 was predicted to be 0.72 for M1, 0.15 for M2, and >0.080 for M3. M2 was also significantly inhibited in the presence of montelukast resulting in a percent inhibition value of 24.7% and a predicted fm CYP2C8 of 0.042. Minor contributions by the remaining CYP isoform inhibitors were observed for M1 (2C8, 2C9, 2D6), M2 (1A2, 2B6, 2C9, 2D6), and M3 (2B6, 2C8, 2C9), although inhibition did not achieve statistical significance (P > 0.01) and was considered negligible.
The effect of CYP selective chemical inhibitors on ertugliflozin metabolism in pooled HLMs
Data are expressed as the means from duplicate experiments. The percent inhibition data were scaled to 100% as described under Materials and Methods. The relative contribution values (fm) were calculated from data scaled to 100% inhibition.
To further discern the contribution of CYP isoforms 3A4 and 3A5 to ertugliflozin metabolism, incubations with the CYP3A4 selective inactivator CYP3cide and dual CYP3A4/5 inhibitor ketoconazole were conducted in CYP3A5 high (HLM 3A5 *1/*1) and low expressing (HLM 3A5 *3/*3) microsomes (Fig. 5; Supplemental Table 4). There were statistically significant differences (P < 0.01) in the percent inhibition of M1 and M2 resulting from incubation with CYP3cide (75% and 68%) when compared with ketoconazole (84% and 86%) indicating a contribution by CYP3A5 in metabolism of ertugliflozin, albeit small. The calculated in vitro contribution of CYP3A5 to overall CYP3A metabolism of M1 and M2 was 9.7% and 18%, respectively (Supplemental Table 4). In HLM 3A5 *1/*1, the percent inhibition of M3 by CYP3cide (81%) was not statistically different when compared with ketoconazole (88%). When incubated in microsomes devoid of CYP3A5 expression (HLM 3A5 *3/*3), there was no statistically meaningful difference between the effects of CYP3cide and ketoconazole on any of the metabolites.
Percent inhibition of CYP metabolites M1, M2, and M3 by ketoconazole (CYP3A4/5) and CYP3cide (CYP3A4) in pooled HLM 3A4 *1*1 microsomes. CYP3A5 contribution was calculated as the difference between ketoconazole and CYP3cide inhibition. The percent inhibition was used to calculate CYP3A4 and CYP3A5 contributions to M1, M2, and M3 formation (Fig. 6). Mean data represent duplicate incubations.
Ertugliflozin kinetic parameters were generated for all three metabolites in rhCYP 3A4, 3A5, and 2C8 (Supplemental Fig. 3; Table 4). rhCYP3A4 kinetics was best described by the substrate inhibition model with Km,u, Vmax, and inhibition constant for substrates exhibiting substrate inhibition kinetics (Ksi) values of 120 µM, 37.0 pmol/min per picomole and 1090 µM for M1, 135 µM, 11.0 pmol/min per picomole and 587 µM for M2, and 213 µM, 15.6 pmol/min per picomole and 397 µM for M3, respectively. Based on these kinetic parameters, the respective M1, M2, and M3 CLint,sc,u values were 4.78, 1.27, and 1.13 ml/min per kilogram (after ISEF and abundance scaling). rhCYP2C8 M2 metabolite formation was best fit to a substrate inhibition model for determination of enzyme kinetic parameters. Km, Vmax, and Ki values for M2 formation were 32.0 µM, 0.0361 pmol/min per picomole, and 568 µM. The resulting hepatic CLint,sc,u value for M2 was 0.0654 ml/min per kilogram. Kinetic parameters for M1 with rhCYP2C8 were not calculated due to insufficient solubility to achieve saturation Vmax. A CLint,sc,u value for M1 of 0.0174 ml/min per kilogram was calculated from the slope of substrate concentration versus formation rate data. M3 formation with rhCYP2C8 was detectable but did not have sufficient measurable concentration data to generate a kinetic curve. The total scaled, ISEF-adjusted, CLint,cs,u for CYP2C8 contribution to ertugliflozin metabolism was 0.083 ml/min per kilogram. In rhCYP3A5, M1 and M2 reaction rates were fit to a substrate inhibition model, and M3 was fit to a Michaelis-Menten model for determination of enzyme kinetic parameters. Km,u, Vmax, and Ki values for M1 and M2 formation in rhCYP3A5 were 175 µM, 5.47 pmol/min per picomole, and 1660 µM and 163 µM, 4.93 pmol/min per picomole, and 771 µM, respectively. The Km,u and Vmax values for M3 formation are 495 µM and 1.66 pmol/min per picomole. Based on these kinetic parameters, the respective M1, M2, and M3 rhP450 CLint,u values were 0.0313, 0.0302, and 0.0044 µl/min per picomole, respectively. HLM CLint,u values were not calculated due to lack of CYP3A5 ISEF. The total CLint,sc,u for CYPs was 7.26 ml/min per kilogram, with CYPs 3A4 (7.18 ml/min per kilogram) and 2C8 (0.0828 ml/min per kilogram) representing a fm of 0.99 and 0.01, respectively (Table 4).
Clearance of Ertugliflozin in HLMs and HKMs at 1 µM Ertugliflozin.
Ertugliflozin clearance was investigated in HLMs and HKMs in the presence of combined and individual UGT and CYP cofactors. As measured by substrate depletion, the scaled unbound human intrinsic clearance (CLint,sc,u) obtained from incubations in HLMs and HKMs was 27.1 and 2.93 ml/min per kilogram, respectively (Table 6), resulting in a combined CLint,sc,u of 30.0 ml/min per kilogram. The resulting fCL was 0.90 and 0.10 for liver and kidney, respectively.
Ertugliflozin (1 µM) clearance estimates and fraction of clearance in HLMs and HKMs by both substrate depletion and metabolite formation
Data are presented as the means of two replicate experiments.
Metabolite formation kinetic analyses after ertugliflozin incubation in HLMs were also conducted to measure CLint,u with individual UGT and CYP cofactors. The HLM UGT-mediated CLint,sc,u was 33.9 ml/min per kilogram, and the CYP mediated CLint,sc,u was 1.55 ml/min per kilogram. The combined HLM CLint,sc,u was 35.5 ml/min per kilogram. The UGT-mediated CLint,sc,u for HKMs was 3.66 ml/min per kilogram. The combined HLM and HKM CLint,sc,u was 39.2 ml/min per kilogram corresponding to a fCL of 0.91 and 0.09 for liver and kidney, respectively. Using the well stirred model, the systemic clearance (CLb) was 1.55 ml/min per kilogram, where the fCL for liver and kidney was 0.90 and 0.10, respectively.
Nonspecific Binding of Ertugliflozin in HLMs.
Under UGT experimental conditions, the fu measured in an equilibrium dialysis experiment was 0.54 and 0.11 in HLMs (0.25 mg/ml protein) in the absence and presence of 2% BSA, respectively. The fu,HLM at 0.76 mg/ml was 0.397. The fraction unbound for CYP HLM (0.1 mg/ml protein), rhCYP3A4 (0.11 mg/ml protein), rhCYP3A5 (0.16 mg/ml), and rhCYP2C8 (0.06 mg/ml protein), were calculated as 0.83, 0.81, 0.76, and 0.90, respectively, using the differential protein binding equation (Austin et al., 2002).
Discussion
Ertugliflozin ADME characterized in a human [14C] mass-balance study demonstrated the primary clearance mechanism was UGT-mediated metabolism (86%) with minor contributions from CYP-mediated metabolism (12%) and renal clearance (2%) (Miao et al., 2013). Reaction phenotyping studies reported herein defined UGT1A9 as the primary enzyme responsible for ertugliflozin metabolic CL with minor contributions from UGT2B7, UGT2B4, CYP3A4, CYP3A5, and CYP2C8. Quantitative UGT reaction phenotyping has advanced significantly (Miners et al., 2010; Zientek and Youdim, 2015; Di, 2017), allowing consistent approaches to conduct core in vitro UGT and CYP phenotyping experiments (Zhang et al., 2007; Miners et al., 2010; Zientek and Youdim, 2015; Di, 2017). Core studies involved identification of human relevant metabolites from elimination routes, defining fCL from enzyme kinetics, qualitative rhUGT mapping, and quantitative detailing with isoform-selective chemical inhibitors to define in vitro fm. Orthogonal phenotyping approaches informed core studies, which involved scaling recombinant enzyme kinetics (quantitative assessment) using RAF or ISEF, and genotyped tissue fraction studies (i.e., CYP3A5). Based on consistent and robust results defining glucuronidation and UGT1A9 as the primary metabolic enzyme, UGT correlation analyses (Achour et al., 2017) were not pursued.
Human relevant metabolites (Fig. 1) initially identified in human hepatocyte incubations (Kalgutkar et al., 2011) and quantified from human mass-balance studies included three primary regioisomer glucuronides (M5a, M5b, M5c) with M5c representing >80% of UGT-mediated CL (Kalgutkar et al., 2011; Miao et al., 2013). The minor glucuronide (M5b) accounted for <5% of human glucuronidation and was not formed in appreciable amounts in rhUGTs or HLMs to conduct phenotyping. UGT reaction phenotyping of the two primary glucuronides (M5c and M5a), representing >95% of human glucuronidation, was conducted. Oxidative metabolism resulted in elimination of three primary metabolites (M1, M2, M3) and secondary glucuronides (M6a, M6b), assumed secondary to oxidative metabolism, which collectively accounts for a minor (12%) human CL route.
Enzyme kinetic assays defining fCL were conducted in HLMs since UGT-selective chemical inhibitors are not qualified for use in human hepatocytes. HLM formation kinetics for M5c (major) and M5a (minor) were performed with 2% BSA to sequester fatty acids released during microsomal incubations since fatty acids could inhibit activity for some UGT isoforms (e.g., UGT1A9 and UGT2B7), lowering the unbound Km and hence increasing CLint,u (Rowland et al., 2007; Manevski et al., 2011; Walsky et al., 2012; Miners et al., 2017). Ertugliflozin UGT-mediated CLint,u (9.59 µl/min per milligram) increased 3.9-fold (37.0 µl/min per milligram) in the presence of 2% BSA, primarily driven by a lower unbound Km with BSA (Table 1), similar to UGT1A9 substrates propofol and mycophenolic acid (Rowland et al., 2008; Gill et al., 2012) or UGT2B7 substrates zidovudine (Manevski et al., 2011; Walsky et al., 2012) and naloxone (Kilford et al., 2009; Gill et al., 2012). In vitro formation of M5c (34.7 µl/min per milligram) was approximately 15-fold greater than M5a (2.28 µl/min per milligram), consistent with human disposition studies indicating M5c as the major human glucuronide (Miao et al., 2013). In vitro formation of oxidative metabolites (1.64 µl/min per milligram) was significantly lower than glucuronidation, resulting in fCL estimates of hepatic glucuronidation versus oxidative metabolism of 96% and 4%, respectively. These fCL estimates predict major human conjugative CL mechanisms well (88%) when not accounting for renal CL, and somewhat underestimate observed human oxidative CL (12%). A possible explanation for the underestimation of observed human oxidative CL is the potential reabsorption of ertugliflozin, which is generated from hydrolysis of biliary secreted ertugliflozin glucuronides in the gastrointestinal tract. The potential reabsorption and subsequent metabolism by CYP3A4 in the intestine may account for the increased contribution of oxidative metabolism to ertugliflozin metabolism in vivo, resulting in an underestimation of CYP-mediated metabolism using in vitro studies. However, a physiologically based pharmacokinetic model for ertugliflozin, which does not contain this reabsorption component, adequately characterizes ertugliflozin pharmacokinetics across a dose range of 0.5–300 mg, indicating that any potential reabsorption does not significantly contribute to ertugliflozin disposition and the role of oxidative metabolism in ertugliflozin metabolism remains small (Callegari et al., 2020).
Qualitative recombinant UGT enzyme mapping identified UGT1A9 as the apparent major enzyme responsible for M5c glucuronide formation, whereas extrahepatic UGTs 1A7 and 1A8 also demonstrated catalytic turnover (Fig. 4). At higher concentrations, formation of M5c was also catalyzed by UGT1A3 and UGT2B7, albeit at apparently low turnover rates. In addition to UGT1A9, subsequent kinetic experiment with rhUGTs (Table 4) and chemical inhibition studies (Table 3) indicated a minor but more substantial role for UGT2B7 in M5c glucuronidation (Fig. 2). This increase in CLint is likely due to the inclusion of 2% BSA in rhUGT kinetic assays and HLM chemical inhibitor studies, which have demonstrated significant increases in rhUGT CLint (9-fold) for UGT2B7 substrates such as zidovudine (Walsky et al., 2012). In turn, M5a glucuronidation was apparently primarily catalyzed by UGTs 1A9, 2B7, and 2B4 (Fig. 4) with low turnover rates from other UGT1A enzymes. Subsequent phenotyping studies demonstrated a more significant contribution of UGT2B7 relative to UGT1A9 than apparent from rhUGT mapping.
Quantitative chemical inhibition studies conducted in the presence of digoxin and tranilast (Lapham et al., 2012) demonstrated that UGT1A9 (fm 0.81) is the primary enzyme responsible for ertugliflozin glucuronidation, whereas the minor route (fm 0.19) was assigned to UGT2B7 and UGT2B4 after incubation with 16β-phenyllongifolol (Bichlmaier et al., 2007; Lapham et al., 2012). Although 16β-phenyllongifolol was previously considered a selective inhibitor of UGT2B7, recent development of a selective UGT2B4 substrate assay in HLMs in our laboratory demonstrated that 16β-phenyllongifolol (3 µM) is a nonselective inhibitor against UGT2B7 and UGT2B4 substrates resulting in >90% inhibition (unpublished data). Quantitative proteomics could provide insight into the potential contribution of UGTs 2B4 and 2B7 toward ertugliflozin minor glucuronidation route (fm 0.19). Expression of UGT2B4 (picomole per milligram protein) in pooled HLMs is approximately 50% lower than UGT2B7 (Fallon et al., 2013b; Achour et al., 2017). Since 16β-phenyllongifolol completely inhibited UGTs 2B4 and 2B7 activity in HLMs and assuming that UGT2B4 and UGT2B7 HLM enzyme kinetic parameters for ertugliflozin glucuronidation are similar, one-third of UGT2B ertugliflozin glucuronidation could be assigned to UGT2B4 (fm 0.06) and two-thirds to UGT2B7 (fm 0.13).
Orthogonal assays were also employed to substantiate or inform UGT chemical inhibition phenotyping based on increasing evidence of successful RAF scaling of rhUGT CLint to HLMs with utility in UGT prototyping (Rouguieg et al., 2010; Zhu et al., 2012). Scaling of rhUGT CLint for UGTs 1A9 and 2B7 provided very similar estimates of relative in vitro fm (0.86 vs. 0.14) compared with chemical inhibition (0.81 vs. 0.19). Although not the focus of this study, it is of interest that the RAF-adjusted in vivo CLint for glucuronidation (22.7 ml/min per kilogram) was within 2-fold or approximately 0.65-fold the HLM glucuronidation CLint (34.9 ml/min per kilogram).
Since UGT1A9 and UGT2B7 are the primary and secondary respective UGTs expressed in human kidney (Fallon et al., 2013a; Zientek and Youdim, 2015), it is important to consider extrahepatic metabolic CL for ertugliflozin. Because ertugliflozin is not subject to significant intestinal first-pass metabolic extraction (Fg 0.98) (Callegari et al., 2020) and UGT1A9 expression level in the intestine is very low or undetectable (Fallon et al., 2013a; Zientek and Youdim, 2015), in vitro intestinal metabolism was not investigated. As anticipated, ertugliflozin in vitro CLint was higher in HKMs compared with HLMs since relative UGT1A9 expression (picomole per milligram tissue) is approximately 2- to 3-fold higher (Mukai et al., 2014; Zientek and Youdim, 2015). However, CLint scaled for tissue CL is approximately 10-fold lower for human kidney relative to liver, primarily due to the differences in organ weight. Only the UGT1A9 metabolite (M5c) was quantifiable in HKMs, in line with relative fm UGT for ertugliflozin and expression levels of UGTs 1A9 and 2B7 (Fallon et al., 2013a; Margaillan et al., 2015), whereas oxidative metabolism was not observed (Table 6). Similar results for UGT1A9 substrate propofol demonstrated Vmax and CLint were 2.1- and 3.7-fold higher in HKMs relative to HLMs (Mukai et al., 2014). It is therefore anticipated that kidney metabolic CL will account for approximately 10% of total systemic ertugliflozin CL, primarily mediated by UGT1A9.
Minor oxidative CL routes (<20%) could prove important during physiologically based pharmacokinetic model development, particularly when assessing CYP3A4-mediated enzyme induction for victim drugs (Gandelman et al., 2011). Chemical inhibition studies demonstrated most of the ertugliflozin oxidative metabolism was mediated by CYP3A (96%) with a small contribution by CYP2C8 (0%–4%). Utilizing chemical inhibition studies in genotyped CYP3A5 tissue fractions (Tseng et al., 2014), it was evident that CYP3A4 (fm 0.86–1.0) was responsible for the majority of CYP3A (fm 0.96) mediated metabolism, relative to CYP3A5 (fm 0–0.1) (Fig. 6). Orthogonal ISEF-adjusted scaling of rhCYP3A4 and CYP2C8 CLint suggested a lessor role of CYP2C8 (fm 0.01) relative to CYP3A4 (fm 0.99). The increased relative contribution of CYP3A is apparently due to the tendency of ISEF-adjusted rhP450 CLint to overestimate CYP3A-mediated CL relative to other CYP enzymes (Chen et al., 2011; Umehara et al., 2017; Lindmark et al., 2018).
Ertugliflozin global in vitro hepatic fm estimated from UGT- and CYP-mediated fCL and fraction of metabolism inhibited (finh) after incubation with isoform-selective chemical inhibitors.
In conclusion, UGT1A9 is the primary enzyme responsible for ertugliflozin metabolic CL, consistent with the major clearance routes identified for structurally related SGLT2 inhibitors such as dapagliflozin (Kasichayanula et al., 2013) and canagliflozin (Mamidi et al., 2014; Devineni and Polidori, 2015). A “bottom-up” in vitro reaction phenotyping approach (Fig. 7) defined UGT1A9 (fm 78%) as the major enzyme responsible for ertugliflozin hepatic CL with minor contributions from UGT2B7/UGT2B4 (fm 18%), CYP3A4 (fm 3.4%), CYP3A5 (fm 0.4%), and CYP2C8 (fm 0.16%). “Top-down” integration of higher observed clinical ertugliflozin oxidative disposition (12% vs. 4%) and renal elimination (Miao et al., 2013) would result in “middle-out” altered in vivo human fm values for UGT1A9 (fm 69.7%), UGT2B7/UGT2B4 (fm 16.3%), CYP3A4 (fm 10.3%), CYP3A5 (fm 1.2%), and CYP2C8 (fm 0.16%). These findings enabled physiologically based pharmacokinetic drug-drug interaction risk assessment with Simcyp using ertugliflozin coadministered with UGT1A9 inhibitor mefenamic acid (Callegari et al., 2020) and could be used for further interpretation of ertugliflozin disposition and drug-drug interaction risk assessment with coadministered drugs.
Integration of “bottom-up” core in vitro phenotyping and orthogonal studies to establish in vitro fm and “middle-out” in vivo fm for ertugliflozin based on “top-down” human mass balance disposition. [Adapted from Figure 7 (Minors et al., 2020)]
Authorship Contributions
Participated in research design: Lapham, Callegari, Cianfrogna, Lin, Goosen.
Conducted experiments: Lapham, Cianfrogna, Lin, Niosi, Orozco, Sharma.
Performed data analysis: Lapham, Cianfrogna, Lin, Niosi, Orozco, Sharma, Goosen
Wrote or contributed to the writing of the manuscript: Lapham, Callegari, Cianfrogna, Lin, Goosen.
Note Added in Proof: During the proof stage, several errors were discovered, the center table heading was cutoff in Figure 6 and a permission statement was accidentally not included in Figure 7 caption of the Fast Forward version that appeared online October 5, 2020. Figure 6 and Figure 7 caption have been corrected.
Footnotes
- Received July 6, 2020.
- Accepted September 22, 2020.
This study was sponsored by Pfizer Inc., New York, NY in collaboration with Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth, NJ.
A portion of this work was presented as a poster at the 22nd North American International Society for the Study of Xenobiotics Meeting; (2018) in Montréal, Canada.
L.P., E.C., J.C., J.L., M.N., C.C.O., R.S., and T.C.G. are employees of Pfizer Inc., New York, NY and may own shares/stock options in Pfizer Inc., New York, NY.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ADME
- absorption, distribution, metabolism, and excretion
- BSA
- bovine serum albumin
- CL
- clearance
- CLint
- apparent intrinsic clearance
- CLint,sc,u
- scaled in vivo whole organ unbound intrinsic clearance
- CLint,u
- unbound in vitro intrinsic clearance
- fCL
- fraction of clearance
- fm
- fraction of metabolism
- fu
- fraction unbound
- HKM
- human kidney microsome
- HLM
- human liver microsome
- IS
- internal standard
- ISEF
- intersystem extrapolation factor
- Km
- apparent substrate concentration at half-maximal velocity
- Ksi
- inhibition constant for substrates exhibiting substrate inhibition kinetics
- LC-MS/MS
- liquid chromatography–tandem mass spectrometry
- M1
- monohydroxylated ertugliflozin
- M2
- des-ethyl ertugliflozin
- M3
- monohydroxylated ertugliflozin
- M5a
- 2-O-β-glucuronide
- M5c
- 3-O-β-glucuronide
- CYP
- cytochrome P450
- PPP
- 2-phenyl-2-(1-piperdinyl)propane
- RAF
- relative activity factor
- rhCLint
- unbound intrinsic apparent clearance in rhUGTs
- rhP450
- recombinantly expressed human P450 enzymes
- rhUGT
- recombinantly expressed human UGT enzymes
- S50
- substrate concentration at half-maximal velocity for substrates exhibiting atypical kinetics
- SGLT
- sodium-dependent glucose cotransporter
- T2DM
- type 2 diabetes mellitus
- UDPGA
- UDP glucuronic acid
- UGT
- UDP-glucuronosyltransferase
- Vmax
- maximal rate of metabolism
- Copyright © 2020 by The American Society for Pharmacology and Experimental Therapeutics