Formation of (2R)- and (2S)-8-Prenylnaringenin Glucuronides by Human UDP-Glucuronosyltransferases

Abstract

Occurring in hops (Humulus lupulus) and beer as a racemic mixture, (2R,2S)-8-prenylnaringenin (8-PN) is a potent phytoestrogen in hop dietary supplements used by women as alternatives to conventional hormone therapy. With a half-life exceeding 20 h, 8-PN is excreted primarily as 8-PN-7-O-glucuronide or 8-PN-4’-O-glucuronide. Human liver microsomes and 11 recombinant human UDP-glucuronosyltransferases (UGTs) were used to catalyze formation of the two oxygen-linked glucuronides of purified (2R)-8-PN and (2S)-8-PN, which were subsequently identified using mass spectrometry and NMR spectroscopy. Formation of (2R)- and (2S)-8-PN-7-O-glucuronides predominated over the 8-PN-4’-O-glucuronides except for intestinal UGT1A10, which formed more (2S)-8-PN-4’-O-glucuronide. (2R)-8-PN was a better substrate for all 11 UGTs except for UGT1A1, which formed more of both (2S)-8-PN glucuronides than (2R)-8-PN glucuronides. Although several UGTs conjugated both enantiomers of 8-PN, some conjugated just one enantiomer, suggesting that human phenotypic variation might affect the routes of metabolism of this chiral estrogenic constituent of hops.

Graphical Abstract

Introduction

Hops (Humulus lupulus L.) are used primarily to preserve and flavor beer.¹ Exhibiting considerable estrogenic activity,^2,3 extracts of hops are also under development as dietary supplements for the management of menopausal symptoms in women.^4,5 8-Prenylnaringenin (8-PN) (Figure 1) is the most estrogenic constituent of hops and among the most potent of all phytoestrogens.^6,7 A chiral prenylated flavonoid, 8-PN occurs in hops as a racemate⁸ probably due to non-enzymatic intramolecular cyclization of the corresponding chalcone, desmethylxanthohumol. Both 8-PN enantiomers are estrogenic,⁹ both have higher affinity for estrogen receptor-α than for estrogen receptor-β, and some reports suggest that (2S)-8-PN and (2R)-8-PN might have different ER affinities and estrogenicities.¹⁰

Figure 1.

Chemical structures of (2R)-8-PN, (2S)-8-PN and all theoretical monoglucuronide metabolites. Note that glucuronidation occurred only at the 7-O and 4’-O positions.

8-PN is well absorbed in humans following oral administration but is conjugated rapidly with glucuronic acid.¹¹ UDP-glucuronosyltranferases (UGTs) are expressed in many organs but most abundantly in the liver and intestine. Incubations of 8-PN with human Caco-2 intestinal epithelial cell monolayers formed primarily 8-PN-4’-O-glucuronide (Figure 1), while incubations with human hepatocytes formed 8-PN-7-O-glucuronide as the most abundant metabolite.¹² The isozymes UGT1A1, UGT1A6, UGT1A8, and UGT1A9 were reported to be most responsible for glucuronidation of 8-PN.¹² Although human hepatic metabolism can form a variety of phase I 8-PN metabolites, phase II glucuronidation predominates.

Metabolism of 8-PN enantiomers by human enzymes has not been reported previously. However, Martinez and Davies¹³ administered each enantiomer of the structurally related hop flavonoid isoxanthohumol to rats and observed (2R)-8-PN or (2S)-8-PN and their glucuronides in rat serum and urine. O-demethylation of isoxanthohumol by gut microbiota in the rat, as reported by Bolca, et al.,¹⁴ had resulted in the formation of 8-PN. Martinez and Davies¹³ reported that (2S)-8-PN glucuronides predominated almost 6-fold over (2R)-8-PN glucuronides in rat urine following oral administration of (S)- or (R)-isoxanthohumol, respectively, but they did not identify which glucuronides were formed for each 8-PN enantiomer.

Because women are consuming hop dietary supplements containing estrogenic 8-PN,³ and because of the importance of glucuronidation in 8-PN metabolism and excretion,^11–13 we investigated the glucuronidation of each 8-PN enantiomer by human UGTs. The structures of the regioisomers of (2R)-8-PN glucuronides and (2S)-8-PN glucuronides were determined using high performance liquid chromatography-tandem mass spectrometry (LC-MS/MS), accurate mass measurement, and 1-dimensional and 2-dimensional ¹H- and ¹³C-NMR. Using recombinant human enzymes, the relative contributions of 11 UGT isoforms were determined for the formation of glucuronic acid conjugates of (2R)-8-PN and (2S)-8PN. Significant differences were observed between UGT isoforms concerning the glucuronidation of (2R)-8-PN and (2S)-8-PN.

Materials and Methods

Reagents

Alamethicin and uridine 5′-diphosphoglucuronic acid trisodium salt (UDPGA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Deuterated DMSO and methanol were purchased from Cambridge Isotope Laboratories (Tewksbury, MA, USA). UGT Reaction Mix - Solution B, pooled human liver microsomes and recombinant human UGTs 1A1, 1A3, 1A4, 1A6, 1A8–10, 2B4, 2B7, 2B15 and 2B17 were purchased from BD Gentest (Woburn, MA, USA). High-performance liquid chromatography-grade solvents and analytical-grade reagents were used in all experiments.

Isolation of (2R)-8-PN and (2S)-8-PN

Racemic 8-PN was prepared as described previously,² and the purity was determined to be 95.57% using quantitative NMR. (2R)-8-PN and (2S)-8-PN were separated from racemic 8-PN by chiral chromatography with UV detection at 275 nm. A semi-preparative ChiralPak IA column (10 mm × 250 mm, 5 μM) was eluted with a mobile phase consisting of isocratic n-hexane/ethanol (90:10; v/v) at a flow rate of 2 mL/min. Each 100 μL injection contained 7 mg/mL of racemic 8-PN. Under these conditions, (2R)-8-PN eluted before (2S)-8-PN as determined by circular dichroism polarimetry (Jasco J 710 polarimeter) of each purified enantiomer prepared at 12.5 μM in acetonitrile (see Supporting Information).^8,15 The purities of isolated (2R)-8-PN and (2S)-8-PN were >99.0% based on HPLC-UV and LC-MS analyses.

Enzymatic glucuronidation of 8-PN

To obtain 8-PN glucuronides in ample quantities for NMR analysis, 20 μM (2R,2S)-8-PN was incubated with pooled human liver microsomes (1 mg/mL) in 50 mM Tris-HCl (pH 7.5) containing 8 mM MgCl₂, 25 μg/mL alamethicin, and 2 mM UDPGA as the UDP-glucuronyltransferase cofactor at 37 °C for 60 min with continuous shaking as described previously.^16,17 Negative control incubations were identical except for the omission of UDPGA or human liver microsomes. Reactions were terminated by adding a 3-fold excess (v/v) of ice-cold methanol, followed by centrifugation at 12000×g at 4 °C for 15 min. The supernatants were removed, evaporated to dryness under a stream of nitrogen gas, and re-dissolved in methanol (1/50th of the incubation volume) prior to HPLC-UV separation or LC-MS/MS analysis.

To determine which UGT enzymes were responsible for glucuronidation, 1, 4, 10, 40, or 100 μM of (2R,2S)-8-PN, (2R)-8-PN or (2S)-8-PN were incubated individually with each of 11 recombinant human UGT isoenzymes (1A1, 1A3, 1A4, 1A6, 1A8, 1A9, 1A10, 2B4, 2B7, 2B15, and 2B17). In addition to 8-PN, each incubation mixture contained 125 μg of UGT, 25 μg/mL alamethicin, 8 mM MgCl₂, 50 mM Tris-HCl (pH 7.5), 2 mM UDPGA in a total volume of 500 μL. Note that 8-PN was dissolved in methanol before being added to the incubation mixture such that the final concentration of methanol was 1% of the reaction volume. Each mixture was incubated at 37 °C for 60 min. Control incubations were identical, except for the omission of UDPGA. Reactions were terminated by the addition of 50 μL of ice-cold methanol. After 10 min on ice, samples were centrifuged at 12000×g and 4 °C for 15 min. Supernatants were analyzed by using HPLC-UV and LC-MS/MS.

Semi-preparative HPLC isolation of 8-PN-glucuronides

For the isolation of 8-PN glucuronides, 100 μL aliquots of the reconstituted incubation mixture were injected onto a Shimadzu (Kyoto, Japan) Prominence HPLC-XR system equipped with a UV diode array detector set at 290 nm. Semi-preparative separations were carried out using a YMC (Wilmington, NC, USA) YMC-Pack ODS-A reversed phase HPLC column (10 × 250 mm, 5 μm) with a linear gradient from 10–70% methanol in 0.1% aqueous formic acid as follows: 0–10 min hold at 10% methanol, 10–30 min gradient from 10–70% methanol, and 30–40 min hold at 70% methanol. The flow rate was 5 mL/min.

LC-UV-MS/MS, LC-MS/MS and HPLC-UV analyses

For analytical scale analyses using LC-UV-MS/MS and LC-MS/MS, 10 μL aliquots of the incubation mixture were injected onto a Shimadzu Prominence HPLC-XR system equipped with an Agilent (Santa Clara, CA, USA) XDB C₁₈ column (3.0 × 150 mm, 3.5 μm) maintained at 40 °C in a column oven. 8-PN glucuronides were eluted with a 25 min linear gradient from 5–60% acetonitrile in water containing 0.1% formic acid at a flow rate of 0.2 mL/min. UV absorbance detection was carried out at 290 nm, and mass spectrometric analyses were carried out on a Shimadzu IT-TOF high resolution ion trap/time-of-flight hybrid mass spectrometer using electrospray and polarity switching. Shimadzu LabSolutions software was used for instrument control and data processing. During mass spectrometric analysis, the positive ion and negative ion electrospray conditions were as follows: desolvation line temperature 200 °C; heat block temperature 200 °C; electrospray interface voltage +4.5kV and −3.5 kV; nitrogen nebulizer gas flow 1.5L/min; ion accumulation time 30ms; collision energy 50% for MS/MS; and scan range m/z100–800.

HPLC-UV data were also acquired using a YMC AQ column (2.0 × 100 mm, 3 μm) with a mobile phase consisting of a 15 min linear gradient from 40% to 60% methanol in 0.1% aqueous formic acid. The flow rate was 0.25 mL/min.

1-Dimensional NMR

All 1-dimensional ¹H and ¹³C nuclear magnetic resonance (NMR) spectra were acquired at 900 and 225 MHz, respectively, on a Bruker (Karlsruhe, Germany) Avance-900 NMR spectrometer equipped with a 5-mm triple resonance inverse detection (¹H, ¹³C, ¹⁵N) TCI cryoprobe. The ambient sample temperature at the probe was regulated at 25 °C (298 K) for all experiments. Samples of 8-PN and metabolites were dissolved in 150–175 μL [d₆]-DMSO (Cambridge Isotopes; Tewsbury, MA, USA) contained in 3-mm NMR tubes. The measured NMR chemical shifts (¹H and ¹³C) are expressed in ppm (δ), relative to the solvent (δ DMSO = 2.500 ppm, ¹H; and δ DMSO = 39.50 ppm, ¹³C; which are both computer referenced with respect to tetramethylsilane, TMS = 0.00 ppm).

Survey 1-dimensional ¹H NMR spectra for 8-PN and metabolites were acquired under quantitative proton NMR (qHNMR) conditions: relaxation delay (d1) 60 s; acquisition time (aq) 4.0 s; spectral window (sw) 30 ppm; delay after pulse excitation (de) 26.26 us; dummy scans (ds) 4; zero-filling (si) = 512 k. All 1-dimensional ¹³C data were acquired using DEPT-Q-135 as implemented on Bruker spectrometers, which produced a spectrum in which the phase of the signals reflects the number of protons bonded to each of the carbons: C (non-protonated), CH₂ – down, CH; CH₃ – up. Key acquisition parameters included the relaxation delay (d1) = 1.5 s; acquisition time (aq) 1.0 s; and spectral window (sw) 200 ppm.

1-Dimensional ¹H NMR spectra of the metabolites were additionally acquired with water suppression, which was necessary in order to remove exchange-broadened signal intensities arising from dissolved water, and the carbohydrate and phenolic -OH resonances in the sample which overlapped with signal regions of interest (δ 5.90 – 7.50 ppm). Because of the relatively broad nature of the signals in the metabolite samples, these experiments used the Water Attenuation by Transverse (T₂) Relaxation (WATR).^18,19 technique and were performed using the Carr-Purcell-Meiboom-Gill experiment as implemented on Bruker NMR spectrometers using the pulse sequence “cpmg1d”. The variable echo delay (d20) contained in the sequence was adjusted to suppress the exchange broadened resonances in the spectral regions of interest, which provided crucial information about the glucuronidation sites, but was kept as short as possible in order to avoid or minimize baseline modulation arising from ¹H,¹H- J-coupling. An echo delay (d20) = 6.53 ms for WATR in the experiments described here was found to afford satisfactory suppression in the spectral region of interest.

2-Dimensional NMR

A full complement of 2-D experiments (COSY, HSQC, and HMBC) was acquired on a sample of 8-PN (1.0 mg in 170 uL [d₆]-DMSO) and each of the 8-PN glucuronides. For 8-PN, confirmation of the ¹H and ¹³C NMR assignments was crucial for the subsequent work on the 8-PN metabolites. The original assignments for 8-PN (360 MHz, [d₄]-MeOH) have previously been reported,² but the assignments were reconfirmed for the present work (900 MHz, [d₆]-DMSO). For the glucuronides, a similar complement of 2-D data were acquired. The COSY experiments for 8-PN and 8-PN glucuronides facilitated confirmation of the proton assignments in all cases, which was important for the subsequent determination of the sites of glucuronidation. However, the HSQC/HMBS data for both of the metabolites were disappointing in that they were difficult to acquire with good signal-to-noise due to the limited sample availability. For 8-PN-7-O-glucuronide, which was the more abundant (100–200 ug) of the two glucuronides, the data did allow for assignment of structure of all protons and carbons in the molecule. For the 8-PN-4’-O-glucuronide, even with limited sample availability (< ~100 ug), all protons could be assigned from the COSY data, especially the protons used for the glucuronidation site determination. The DEPT-Q-135 data for both metabolites were used for assignment of all carbons (protonated and non-protonated) for 8-PN-7-O-glucuronide by analogy with data obtained for 8-PN, but only the protonated carbons in 8-PN-4’-O-glucuronide could be observed and assigned. Thus, for the purposes of determining the sites of glucuronidation of 8-PN, ¹H NMR was used exclusively.

Results and Discussion

HPLC-UV analysis of the glucuronides formed during incubation of racemic 8-PN with human liver microsomes showed two abundant peaks eluting at 12.8 and 13.1 min and a smaller peak at 15.3 min (Figure 2A). Almost no 8-PN remained in the mixture (retention time 17.8 min). Partial resolution of the first two peaks suggested that they corresponded to a mixture of (R)-8-PN- and (S)-8-PN-β-D-glucuronides. All three peaks were collected during semi-preparative HPLC-UV (the first two partially resolved peaks were combined), freeze dried and identified as 8-PN-7-O-glucuronide (t_R = 12.8/13.1 min; ~200 μg) and 8-PN-4’-O-glucuronide (t_R = 15.3 min; ~150 μg). The purity of each regioisomer was >90.0% by HPLC-DAD and LC-MS.

Figure 2.

LC-UV-MS/MS analysis of 8-PN-glucuronides formed by incubation of human liver microsomes and cofactor UDPGA with A) racemic 8-PN; B) (2R)-8-PN; or C) (2S)-8-PN. Retention times are longer for the mass chromatogram because it was recorded in tandem to UV detection.

The peaks eluting at 12.8/13.1 min were identified as 8-PN-7-O-glucuronide and the peak eluting at 15.4 min in Figure 2 was determined to correspond to 8-PN-4’-O-glucuronide by using high-resolution tandem mass spectrometry (Figure 3) and¹H-NMR (Table 1). No 8-PN-5-O-glucuronide was observed, which is consistent with previous reports,¹² probably due to hydrogen bonding between the hydroxyl group at carbon-5 and the carbonyl group at position-4. Positive and negative ion electrospray accurate mass measurements confirmed the elemental composition of each glucuronide to be C₂₆H₂₈O₁₁ (measured [M+H]⁺ m/z 517.1704; theoretical 517.1710; ΔM = −1 ppm; and measured [M-H]⁻ m/z 515.1559; theoretical 515.1553; ΔM = −1 ppm).

Figure 3.

Positive ion electrospray product ion tandem mass spectra of 8-PN-glucuronides, [M+H]⁺ m/z 517.1704, obtained using a high resolution IT-ToF mass spectrometer. A) (2R)-8-PN-7-O-glucuronide (t_R = 13.1 min in Figure 2); B) (2S)-8-PN-7-O-glucuronide (t_R = 12.8 min); C) (2R)-8-PN-4’-O-glucuronide (t_R = 15.3 min); and D) (2S)-8-PN-4’-O-glucuronide (t_R = 15.2 min).

Table 1.

¹H-NMR (900 MHz) Chemical Shifts for Racemic 8-PN and its Glucuronides^a

Hydrogen	8-PN δ [ppm], mult (J[Hz])	8-PN-7-O-glucuronide (tR = 12.8/13.1 min; Fig. 3) δ [ppm], mult (J[Hz])	8-PN-4’-O-glucuronide (tR = 15.4 min; Fig. 3) δ [ppm], mult (J[Hz])
H-2	5.416, dd(3.06,12.56)	5.478, dd (3.26, 13.03)b 5.453, dd (3.16, 13.32)	5.494, dd (3.20,13.15)
H-3a	3.203, dd(12.53, −16.96)	3.04–3.08, br mc	3.199, dd (13.07,−17.7)
H-3b	2.713, dd(3.06, −16.96)	2.763, dd (3.06, −17.57) 2.796, dd (3.63, −17.76)	2.752, dd (3.45,−17.7)
H-6	5.961, s	6.260, s 6.245, s	6.052, s
H-2’/6’	7.306 (8.75)d	7.317 (8.49) 7.306 (8.49)	7.441 (8.70)
H-3’/5’	6.786 (8.75)d	6.793 (8.49) 6.798 (8.49)	7.055 (8.70)
H-1”	3.071, br d (7.31)	3.062, br d (7.65) 3.048, br d (6.81)	3.082, br d (7.51)
H-2”	5.078, t.sept (7.31, 2.34)	5.116, t.sept. 5.106, t.sept (7.50, 1.49)	5.093, t .sept. (7.52,1.36)
H-4”	1.581, br s	1.573, br s	1.580, br s
H-5”	1.531, br s	1.539, br s	1.532, br s
5-OH	12.106, s	12.07, br s 12.05, br s	12.08, s
7-OH	10.766, br s	e	e
4’-OH	9.573, s	e	e
H-1’”	/	5.085, d (7.38)	5.075, d(7.70)
H-2’”	/	3.287, t (8.17)	3.268, t(8.66)
H-3’”	/	3.314, t (8.87)	3.319, t(9.10)
H-4’”	/	3.361, t (9.44)	3.418, t(9.41)
H-5’”	/	3.936, t (9.51)	3.902, t(10.16)

^a.Obtained in [d6]-DMSO at 298 K

^b.The presence of two values corresponds to the 2R- and 2S-isomers of the 8-PN-glucuronides

^c.Signals heavily overlapped for both diastereomers. Assignment confirmed from ¹H-¹H COSY spectrum

^d.Displays an AA’MM’/AA’XX’ spin coupling system

^e.The OH resonances are exchange broadened by water + OH groups from the glucuronic acid residue and were not individually observable

The base peak of each positive ion tandem mass spectrum was observed at m/z 341, which represents the protonated aglycone 8-PN formed by elimination of dehydrated glucuronic acid (Figure 3). The regioisomers 8-PN-7-O-glucuronide and 8-PN-4’-O-glucuronide were distinguished by the abundant fragment ions of m/z 397 and m/z 285 in the product ion tandem mass spectra of 8-PN-7-O-glucuronide. The product ion of m/z 397 corresponded to the retro Diels Alder cleavage of the central C-ring with retention of glucuronic acid on the A-ring (Figure 3). The ion of m/z 285 was formed by loss of isobutene from protonated 8-PN.

The sites of 8-PN glucuronidation were confirmed using ¹H NMR analysis of the diasteromers of 8-PN-7-O-glucuronide as well as 8-PN-4’-O-glucuronide in a self-consistent manner. The assignments of all proton and carbon-13 signals for 8-PN and both metabolites are shown in Tables 1 and 2, respectively. Table 3 summarizes a subset of the key ¹H chemical shift data used for determining the sites of glucuronidation of 8-PN. Examination of the ¹H chemical shifts of 8-PN with those of the 8-PN-glucuronide regioisomers indicates that the broad high frequency singlets for the 5-OH remain largely unchanged, clearly indicating that glucuronidation had not occured at this position (Table 3). The signals in the 1-D spectrum of 8-PN and those of the analogous water suppression spectra (spectral region δ 5.90 – 7.50 ppm) of the 7-O- and 4’-O-glucuronides are shown in Figure 4. The telltale chemical shifts used for assigning the sites of glucuronidation are the 5-OH (noted earlier), H-2’/H-6’, H-3’/H-5’ and H-6. The large changes (δΔ) calculated for the specific chemical shifts of H-3’/H’5’, H-2’/H-6’ and H-6 serve to unambiguously confirm the sites of glucuronidation. For example, in the ¹H-NMR spectra of the (2R)-8-PN-7-O-glucuronide and (2S)-8-PN-7-O-glucuronide, the signals for H-6 were shifted 0.284 ppm and 0.299 ppm downfield, respectively, compared with the analogous proton in 8-PN. In the case of (2R)-8-PN-4’-O-glucuronide, glucuronidation of (2R)-8-PN shifted both the H-3’/5’ signal δΔ = +0.269 ppm and the H-2’/6’ signal δΔ = +0.135 ppm downfield (Table 3).

Table 2.

¹³C-NMR (225 MHz) Chemical Shifts for Racemic 8-PN and its Glucuronides^a,b

Carbon	8-PN	7-O-8-PN-glucuronide	4’-O-8-PN-glucuronide
C-2	78.19	78.45/78.37c	76.76
C-3	41.89	42.17/41.94	42.03
C-4	196.73	197.72/197.68
C-4a	101.76	103.28/103.20	101.73
C-5	159.68	159.18
C-6	95.25	94.95/94.89
C-7	161.14	161.09/161.03
C-8	106.89	108.97/108.94
C-8a	164.34	162.60/162.54
C-1’	129.20	129.00/128.94
C-2’/6’	128.05	128.11	127.80
C-3’/5’	115.12	115.18	116.04
C-4’	157.55	157.63
C-1”	21.25	21.47/21.42	21.29
C-2”	122.65	122.43/122.41	122.65
C-3”	130.19	130.45	130.24
C-4”	25.53	25.55
C-5”	17.57	17.63
G-1		99.48/99.65	99.77
G-2		73.0
G-3		75.86
G-4		71.32
G-5		75.13
G-6		170.32

^a.Obtained in [d₆]-DMSO at 298 K

^b.Assignments were confirmed by DEPT-Q-135 and COSY

^c.The presence of two chemical shifts for a particular carbon corresponds to the 2R- and 2S isomers of 8-PN glucuronide, respectively

Table 3.

Key NMR Chemical Shifts (δ, ppm) for Assigning Sites of Glucuronidation of 8-PN

Compound	5-OH	H-2’/H-6’	H-3’/H5’	H-6	C-(CH3)2
8-PN a	12.106	7.306	6.786	5.961	1.581 1.532
8-PN-4’-O-GlcA b	12.085	7.441	7.055	6.052	1.580 1.532
Δδ c	−0.021	+0.135	+0.269	+0.091	−0.001 +0.001
8-PN-7-O-GlcA d	12.070 12.050	7.317 7.306	6.793 6.798	6.260 6.245	1.573 1.539
Δδ c	−0.036 −0.056	+0.011 0.000	+0.007 +0.012	+0.299 e +0.284 e	−0.008 +0.008

^a1.0 mg / 170 μL DMSO-d₆

^b0.1 mg (100 μg) / 170 μL DMSO-d₆

^cΔδ = δ_metabolite - δ_8-PN

^d0.1–0.2 mg (100–200 μg) / 170 μL DMSO-d₆

^eThe diastereomeric ratio was determined by qHNMR: H-6 (6.245 ppm) / H-6 (6.260 ppm) = 1.37 / 1.00

Figure 4.

The expanded region (δ 5.90 – 7.50 ppm) of the 900 MHz ¹H NMR spectra (d6-DMSO) of 8-PN (bottom), the WATR spectrum of diastereomeric 8-PN-7-O-glucuronide (center), and the WATR spectrum of 8-PN-4’-O-glucuronide (top). Note that for 8-PN-7-O-glucuronide (center), the signal for H-6 shifted to higher frequency relative to H-6 in 8-PN, and the pairs of proton signals H-2’/H-6’ and H-3’/H-5’ were essentially unaffected by glucuronidation. In contrast, glucuronidation at the 4’-OH position of 8-PN (top spectrum) shifted the H-6 signal only slightly to higher frequency, but the proton signals for H-2’/H6’ and H-3’/H-5’ were significantly shifted to higher frequency.

In addition to the chiral center at C-2 of 8-PN, glucuronic acid is chiral. Therefore, the (2R)- and (2S)-8-PN-7-O-glucuronides and the (2R)- and (2S)-8-PN-4’-O-glucuronides showed different retention times during HPLC (Figure 2). Although (2R)-8-PN-7-O-glucuronide and (2S)-8-PN-7-O-glucuronide were observed as partially resolved chromatographic peaks following glucuronidation of racemic 8-PN by human liver microsomes, only one peak was detected for 8-PN-4’-O-glucuronide. This isomer was determined to be (2R)-8-PN-4’-O-glucuronide, based on incubation of pure (2R)-8-PN with human liver microsomes (Figure 2B).

The ratio of (2R)-8-PN-7-O-glucuronide to (2S)-8-PN-7-O-glucuronide resulting from incubation of racemic 8-PN with human liver microsomes was determined using quantitative ¹H-NMR. For the ¹H-NMR signals at H-6, the ratio of (2R)-8-PN-7-O-glucuronide (6.245 ppm) to (2S)-8-PN-O-glucuronide (6.260 ppm) was 1.37/1.00. This finding is consistent with the ratio of the corresponding HPLC-UV peak areas (1.33/1.00) eluting at 13.1 min and 12.8 min, respectively (Figure 2B and 2C).

After racemic 8-PN was used to prepare larger quantities of glucuronides for NMR analysis, purified (2R)-8-PN and (2S)-8-PN were incubated in small quantities with individual human recombinant UGT isozymes. The incubation mixtures were analyzed using HPLC-UV or LC-MS/MS to determine which glucuronides were formed and their relative amounts. During preliminary studies, incubations were carried out using a constant amount of UGT enzyme with 5 different concentrations of (2R)-8-PN or (2S)-8-PN. The highest yields of glucuronide metabolites were obtained using 40 μM substrate, and the results of these incubations are summarized in Figure 5.

Figure 5.

Glucuronidation of (2R)-8-PN and (2S)-8-PN (40 μM initial concentration) by recombinant human UGT enzymes. (R.S.D. of mean values, 5 – 15%)

Comparing the relative amounts of glucuronides formed from each UGT isoform incubation, (2R)-8-PN formed more abundant glucuronides than did (2S)-8-PN. Furthermore, 8-PN-7-O-glucuronide predominated over 8-PN-4’-O-glucuronide for both 8-PN enantiomers (Figure 5). Abundant in both liver and intestine, UGT1A1 rapidly catalyzed formation of the 7-O-glucuronide of both 8-PN enantiomers plus a small amount of the 4’-O-glucuronide. Also occurring in both liver and intestine but conjugating only (2R)-8-PN, UGT1A6 was the only enzyme tested to form more (2R)-8-PN-4’-O-glucuronide than (2R)-8-PN-7-O-glucuronide. UGT1A3 and UGT1A9 formed 7-O-glucuronides of (2R)-8-PN and (2S)-8-PN, whereas UGTA1A4 formed only (2S)-8-PN-7-O-glucuronide. UGT1A8 and UGT2B4 conjugated both 8-PN enantiomers to form 7-O-glucuronides but only formed the 4’-O-glucuronide of (2R)-8-PN. The enzymes UGT2B7, UGT2B15 and UGT2B17 did not form any 8-PN glucuronides.

UGT1A1, UGT1A3 and UGT1A6 are expressed in both liver and intestine, UGT1A4, UGT1A9, UGT2B4 are found in the liver but not intestine, and UGT1A8 and UGT1A10 occur in the intestine but not the liver.^20,21 Because UGT1A1 was the only source of (2S)-8-PN-4’-O-glucuronide in the liver and formed just a small amount of this metabolite relative to (2S)-u8-PN-7-O-glucuronide, only a trace amount of (2S)-8-PN-4’-O-glucuronide was detected by HPLC-UV following incubation of (2S)-8-PN with human liver microsomes (Figure 2) and none was detected by NMR. In contrast, UGT1A10, which is expressed in the intestine and colon but not in the liver, formed not only (2R)-8-PN-O-glucuronide but more (2S)-8-PN-4’-O-glucuronide than (2S)-8-PN-7-O-glucuronide. Although the liver can also form (2R)-8-PN-4’-O-glucuronide, an important source of (2S)-8-PN-4’-O-glucuronide in vivo is probably intestinal metabolism during absorption.

Most UGTs catalyzed the formation of 8-PN-7-O-glucuronide, while fewer formed 8-PN-4’-O-glucuronide. Some UGTs conjugated both enantiomers of 8-PN with glucuronic acid, others conjugated just (2R)-8-PN and a few formed no 8-PN glucuronides. (2R)-8-PN was conjugated at both the 7-OH and 4’-OH positions, whereas (2S)-8-PN was conjugated primarily at the 7-OH position. Phenotypic variation of UGTs or inhibition of select UGT enzymes by dietary constituents or pharmaceuticals might also affect the formation of 8-PN glucuronides. Human pharmacokinetics studies indicate that racemic 8-PN has a half-life exceeding 20 h.¹¹ Because (2R)-8-PN forms two oxygen-linked glucuronides while (2S)-8-PN forms only one glucuronide, and because (2R)-8-PN is metabolized by more UGTs, the half-life of (2S)-8-PN is likely to be longer in humans, thereby enhancing its estrogenic effects.