Introduction

Top Abstract Introduction Experimental Procedures Results Discussion Appendix References

Hepatic drug clearance is regulated by hepatic blood flow, vascular and tissue binding, transport, metabolism, and biliary excretion. Futile cycling, the metabolic interconversion of two substrates involving different enzymes, is an additional process that influences drug and metabolite clearances. Futile cycling has been noted between 4-methylumbelliferone (4-MU) and 4-methylumbelliferyl sulfate (4-MUS) (Ratna et al., 1993) and methylprednisone and methylprednisolone (Ebling and Jusko, 1986). The futile cycling of estrone sulfate (E₁S) and estrone (E₁), which represents a pharmacologically important biocycle that conserves and regulates endogenous estrogens, however, has not been thoroughly investigated.

Since hepatic processing is a distributed-in-space phenomenon with uptake, metabolism, and efflux occurring repeatedly in hepatocytes along the direction of flow, it is expected that the futile cycling of E₁S and E₁ in liver would be affected by a dual set of transporters and enzymes and their associated zonal heterogeneities. It is expected that E₁ would rapidly diffuse through the cell membrane due to high lipophilicity, as found in recent isolated rat hepatocyte studies (Tan and Pang, 2001). The uptake of E₁S into the rat liver involves transport by members of the organic anion transporting polypeptide superfamily, Oatp1, Oatp2 and Oatp4 (Jacquemin et al., 1994; Noé et al., 1997; Cattori et al., 2000), the multispecific organic anion transporter 3, Oat3 (Kusuhara et al., 1999), and the sodium-dependent taurocholate cotransporting polypeptide, Ntcp (Hagenbuch et al., 1991). However, a lack of acinar heterogeneity was observed for the transport of E₁S (Tan et al., 1999) and E₁ (Tan and Pang, 2001) in rat liver.

In rat hepatocytes, E₁S and E₁ are highly bound to liver tissue (Tan and Pang, 2001). Consequently, the high and nonlinear binding reduces the cellular unbound concentrations of E₁S and E₁ for both metabolism and excretion. Estrone sulfate is mainly deconjugated by arylsulfatase C, a microsomal enzyme, to E₁, which can be further metabolized to estrone glucuronide (E₁G), estradiol (E₂), estriol, their glucuronide and sulfate conjugates, and other minor metabolites (Roy et al., 1987). Although arylsulfatase C is found to be evenly dispersed in the liver acinus (El Mouelhi and Kauffman, 1986), estrogen sulfotransferase, which is responsible for the sulfation of E₁, is predominantly localized in the perivenous region (Tosh et al., 1996; Tan and Pang, 2001). Both UDP-glucuronosyltransferase-1 (UGT1) and 2 (UGT2) are found to glucuronidate estrone (Tukey and Strassburg, 2000), and the UGTs are predominantly localized in the perivenous region (Tosh and Burchill, 1996). The sulfo- and glucuronide conjugates are found in rat bile, with the multidrug resistance-associated protein Mrp2 (or cmoat) mediating the biliary excretion of E₁G (Takikawa et al., 1996). But the transporter(s) involved in E₁S excretion into bile is unknown. Mrp2 does not appear to be involved (personal communications with Dr. Dietrich Keppler, University of Heidelberg, Germany).

Drugs bound to vascular components, namely plasma proteins and red blood cells (RBCs), are expected to reduce drug clearances (Pang and Rowland, 1977; Pang et al., 1995). Estrone is bound to both plasma protein and erythrocytes. Consequently, only 3% estrone exists in the unbound form in human blood (Koefoed and Brahm, 1994). In addition, E₁ is also metabolized by 17-hydroxysteroid dehydrogenase, a cytosolic enzyme in erythrocytes of animals (Challis et al., 1973; Tsang, 1976) and human (Mulder et al., 1972) to E₂. By contrast, E₁S is highly bound to human plasma protein (1.6% unbound in plasma; Rosenthal et al., 1972) but not to erythrocytes. The determination of the vascular binding and metabolism of E₁ and of protein binding of E₁S is another important aspect toward the understanding of factors impacting the hepatic clearances of E₁ and E₁S.

In this communication, the metabolic disposition of simultaneously delivered tracers, [³H]E₁S and [¹⁴C]E₁, was investigated with the recirculating, perfused rat liver preparation. Use of dual radiolabeling of the precursor-product pair allowed for full characterization of the differential metabolism of [³H]E₁S and [¹⁴C]E₁. The strategy was suitable for investigating the effects of vascular and tissue binding, RBC metabolism of E₁, transport, metabolism, and the various zonal aspects on the futile cycling between E₁S and E₁ in the liver, especially when the RBC distribution and metabolism of E₁ were fully characterized. Finally, a series-compartment liver model that embodied zonal and subcellular distribution of metabolic enzymes was developed to interpret the perfusion results.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion Appendix References

Materials. [6,7-³H]E₁S (ammonium salt, specific activity, 53 Ci/mmol), [6,7-³H]E₁ (specific activity, 40.6 Ci/mmol), and [4-¹⁴C]E₁ (specific activity, 56.6 Ci/mol) were purchased from PerkinElmer Life Sciences (Boston, MA). All radiochemical purities found by high performance liquid chromatography (HPLC) or thin-layer chromatography (TLC) were greater than 95%. E₁S, E₁, E₂, and bovine serum albumin (BSA) were obtained from Sigma Chemical Co. (St. Louis, MO). All other reagents were of the highest grade available.

Protein Assay and Hematocrit Count. In all preparations, protein was determined by the method of Lowry et al. (1951), with bovine serum albumin as the standard. The hematocrit was measured by capillary centrifugation in a microhematocrit centrifuge (IEC MB Centrifuge, Damon, Fisher Scientific, Mississauga, ON, Canada).

BSA Binding of tracer E₁, E₂, and E₁S. BSA binding of E₁, E₂, and E₁S was investigated using a commercially available ultrafiltration kit (Centricon 3; Amicon Inc., Beverly, MA). Tracer [³H]E₁ (3.7 ± 0.1 × 10⁶ dpm/ml), [³H]E₁S (1.7 ± 0.1 × 10⁶ dpm/ml), or [³H]E₂ (1.1 ± 0.1 × 10⁶ dpm/ml) was added to 4% BSA (v/v) in Krebs-Henseleit-bicarbonate buffer (pH 7.4). After incubating the mixture for 10 min at 37°C, an aliquot (2 ml) was removed into a Centricon tube and centrifuged at 2500g for 20 min. The radioactivities in the original mixture (0.2 ml) and the resulting ultrafiltrate (0.2 ml) were quantified by liquid scintillation spectrometry (model LS6800; Beckman Instruments Canada, Mississauga, ON). Leakage of BSA into the ultrafiltrate was less than 1% of the original protein concentration.

Distribution and Metabolism of E₁ in Erythrocytes. Bovine erythrocytes (a generous gift from Ryding Regency Meat Packers Ltd., Toronto, ON) were washed three times with saline and twice with lactated Ringer's solution (Baxter Corporation, Toronto, ON). The distribution and metabolism of tracer [³H]E₁ (2.5 ± 0.4 × 10⁵ dpm/ml or 27 ± 4.2 nM) were studied with perfusion media of different compositions: 20 and 60% RBC, in the absence and presence of 4% BSA. Erythrocytes (20 or 60% v/v) and plasma (Krebs-Henseleit-bicarbonate buffer at pH 7.4 containing 0 or 4% BSA and [³H]E₁) were mixed and incubated under oxygenation (carbogen, 95% oxygen and 5% carbon dioxide; Canox Gas, Mississauga, ON) at 37°C in the reservoir of the commercially available TWO-TEN Perfuser. Blood samples were taken at 1, 30, 60, 120, and 180 min, and the hematocrit was measured. The 20 and 60% RBC yielded hematocrits (HCTs) of 0.15 ± 0.01 and 0.5 ± 0.03, respectively.

Recirculating Rat Liver Perfusion. Male Sprague-Dawley rats (290-330 g; Charles River Canada, St. Constant, QC, Canada), which were fed ad libitum, were used for perfusion at 10 ml/min. The temperature of the liver was maintained at 37°C with a heating lamp. Surgery was performed under pentobarbital anesthesia (50 mg/kg, intraperitoneal injection), and the surgical procedure and the perfusion apparatus were identical to those described by deLannoy et al. (1993). The perfusion medium consisted of 20% washed bovine erythrocytes, 4% BSA, and 300 mg/dl glucose (50% dextrose injection USP; Travenol Canada, Mississauga, ON) in Krebs-Henseleit-bicarbonate solution (pH 7.4). The rat liver was recirculated with blank medium for 20 min during the equilibration period, followed by perfusion with medium containing [³H]E₁S (initial concentration of 2.85 ± 0.23 × 10⁵ dpm/ml or 2.4 ± 0.20 nM) and [¹⁴C]E₁ (initial concentration of 1.04 ± 0.12 × 10⁵ dpm/ml or 848 ± 94 nM) from a second reservoir (200 ml). Reservoir perfusate (1-2 ml) was sampled at 0, 2.5, 10, 30, 60, 90, 110, 130, and 150 min. The total volume removed from the reservoir was 7% (14 ml) of the initial volume (200 ml), and no attempt was made to correct for the loss in volume. Bile was collected at 5- and 10-min intervals so the mid-time of the interval coincided with the sampling time of the reservoir.

Extraction and TLC Assays of [³H]E₁ and [³H]E₂ for the RBC Metabolism Studies. The blood and its derived plasma obtained by instantaneous centrifugation (1.5 ml each) were immediately extracted into ethyl acetate (1:2, v/v). One aliquot (1 ml) of the ethyl acetate extract was directly subjected to liquid scintillation counting, and the total count of the sample was determined against a calibration curve constructed of standards containing varying known counts of [³H]E₁ in perfusate and processed in the same fashion. Since [³H]E₁ and [³H]E₂ were completely extracted into ethyl acetate (>99%), the extraction method furnished a mixture of [³H]E₁ and [³H]E₂ in each sample except for time zero, when only [³H]E₁ was present. The ratio of [³H]E₁/[³H]E₂ was further given by TLC described below. A second aliquot (1 ml) of the ethyl acetate was spotted onto the Silica Gel GF (250 µm) TLC plate (Analtech, Newark, DE), which had been preloaded with E₁ and E₂ at the origin to separate [³H]E₁ and [³H]E₂. The plates were developed in a system of toluene:ethanol (9:1, v/v). Regions for E₁ (R_f = 0.76) and E₂ (R_f = 0.57) were visualized under UV light and scraped into minicounting vials. After the addition of water (0.5 ml) and liquid scintillation fluor (5 ml, Ready Safe, Beckman Instruments, Canada) into minicounting vials, the radioactivity was quantified by liquid scintillation spectrometry (model LS6800; Beckman Instruments, Palo Alto, CA). Hence, the amounts of [³H]E₁ and [³H]E₂ in plasma and blood perfusate were quantified by the combined extraction-TLC method. The amounts of E₁ and E₂ in RBC were, however, calculated by difference between the quantities in plasma and blood perfusate of known hematocrit. The radioactivities were expressed as a percentage of the initial concentration of [³H]E₁ used.

HPLC Assays for Quantitation of E₁G, E₁S, and E₁ in the Liver Perfusion Studies. Acetonitrile, which contained 4 µM danazol (the internal standard), was used to terminate any metabolic reactions, with 1:4 (v/v) volume ratio. All perfusate samples (1-2 ml) were immediately transferred to tubes containing acetonitrile (4-8 ml). Contents of the deproteinized samples were dried under nitrogen (Canox Gas) and analyzed by HPLC as described by Tan and Pang (2001). Standards of the calibration curve prepared with samples containing varying known counts of [³H]E₁S and [¹⁴C]E₁ were processed in the same fashion. Bile samples were diluted 1:1 (v/v) with water, and 20-µl aliquots were directly counted. A portion of the diluted bile (20 µl) was subjected to HPLC with internal standardization. The radioactivities in bile and from HPLC radioelution were quantified by liquid scintillation spectrometry. Eluted radioactivities of less than three times the background counts were treated as zeroes. All ³H- and ¹⁴C-radioactivities quantified in the samples were higher than 3000 and 1000 dpm, respectively.

Kinetic Modeling of [³H]E₁ Metabolism in Erythrocytes and Fitting. Various cellular models were tested for their abilities to predict the disposition of [³H]E₁ and [³H]E₂ in erythrocytes. The cellular kinetic model that included plasma protein binding and red cell binding and metabolism of E₁ (Fig. 1) best described the kinetics of E₁ and E₂. Mass balanced rate equations (eqs. 1-4) were written to describe the RBC distribution and metabolism of E₁ and E₂ (Fig. 1). Oxidation of E₂ to E₁ was not included since preliminary study revealed less than 1% metabolism of E₂ to E₁ over 3 h. The same was observed by Tsang (1976).

View larger version (17K): [in this window] [in a new window]   Fig. 1.   The cellular kinetic model for the description of E1 and E2 in blood perfusate. Plasma protein and red cell binding of E1 and E2 and metabolism of E1 were included in the model. The transport of E1 and E2 was denoted by the bidirectional linear transmembrane clearances and the metabolism of E1 was described by the RBC metabolic intrinsic clearance.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

Kinetic Modeling of E₁S and E₁ Disposition in the Recirculating Rat Liver Preparation. A series-compartment, liver model containing two units representing the periportal (PP) and perivenous (PV) regions of the liver, is the minimalized model for purposes of fitting that best predicted the disposition of E₁S and E₁ in the recirculating liver preparation (Fig. 2). In this model, a reservoir compartment was included for recirculation of the perfusate. The flow of substrates occurs unidirectionally from the periportal to the perivenous region, and exchange occurs in the tranverse and not the longitudinal direction. Linear (nonsaturable) transport and metabolic intrinsic clearances prevail in view of the tracer condition studied, and the assumption is justified based on the observed K_m values (in µM) (Tan and Pang, 2001) being in excess of the tracer concentrations (in nM) studied. Species such as E₁S, E₁, and E₁G that were quantified were modeled. Other metabolites formed from E₁ and E₁S (E₂ and estriol [E₃] and their glucuronide and sulfate conjugates such as E₂S, E₂-3S-17G, E₃S, and E₃-3S-16G) were collectively represented by M'.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Schematic representation of the liver by a series-compartment, liver model that embodied acinar and subcompartmentalization and subcellular distribution of metabolic enzymes. The bottom chart conveyed the ascinar distributions of enzymatic activities of estrone sulfotransferase, estrone UDP-glucuronosyltransferase, and estrone sulfatase for metabolism of E1S and E1, expressed as percentages of the total metabolic intrinsic clearances.

Fitting of Data to the Series-Compartment Liver Model. Mass balanced rate equations (see Appendix) were written to describe events of the series-compartment liver model (Fig. 2). The amounts of drug and metabolite in both perfusate and bile were normalized by the dose. Binding to red cell and albumin was assumed to be rapidly equilibrative such that use of on- and off-rate constants was not necessary. Under this instance, the unbound concentrations of E₁S and E₁ in whole blood perfusate equal those in plasma and in RBC. The unbound fraction in blood may then be calculated from either eq. 1 or eq. 2. The unbound fractions of E₁S (f) and E₁ (f) in the liver cytosol were taken from Tan and Pang (2001).

Statistical Analysis. All data were presented as the mean ± standard deviation, and the means were compared by use of ANOVA or the paired t test, with the level of significance set at 0.05. The MSC and the Akaike Information Criteria (Akaike, 1974; Ludden et al., 1994) were used to select the appropriate model(s).

	Results

Top Abstract Introduction Experimental Procedures Results Discussion Appendix References

Plasma Binding of E₁S, E₁, and E₂. The unbound fraction of E₁S in 4% BSA plasma was 0.03 ± 0.01 (n = 3), whereas those for E₁ and E₂ were 0.05 ± 0.01 and 0.04 ± 0.01, respectively (see Table 1, n = 3). The unbound fractions of E₁ and E₂ in perfusate of different compositions as determined by eq. 1 are summarized in Table 1.

Incubation of a Tracer Dose of [³H]E₁ in Blood Perfusate. The time courses for [³H]E₁ and [³H]E₂ in erythrocytes are shown in Fig. 3. Different areas under the concentration-time curves (AUC) for [³H]E₁ were noted in the presence and absence of BSA, and similar observations were found for [³H]E₂ (Table 2). The RBC clearance of E₁ (CL or 0.035 ± 0.02 ml/min) in 60% RBC blood-perfusate was higher than that of the 20% RBC blood-perfusate (0.0092 ± 0.006 ml/min), and these rates for E₂ formation decreased in the presence of 4% BSA (Table 2, Fig. 3). Upon normalization to the hematocrit, normalized values of the RBC clearance of E₁ () in 20 and 60% RBC perfusates became similar (0.061 ± 0.04 and 0.069 ± 0.04 ml/min/HCT, respectively). These were dramatically reduced to 0.0031 ± 0.001 and 0.0024 ± 0.001 ml/min/HCT, respectively, in the presence of 4% BSA. The RBC to plasma partitioning ratios of E₁ and E₂ in the 20 and 60% RBC albumin-free perfusate were higher than those in the presence of 4% BSA. In the presence of 4% albumin in perfusate, values of approximately unity were obtained for the RBC to plasma partitioning ratio for E₁ and E₂ (Fig. 4). Moreover, the ratios reached their equilibrium values almost immediately.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Time-dependent profiles for tracer [3H]E1 incubated with perfusates of various composition at 37°C: Amounts of E1 in blood perfusate (A), E1 in plasma (B), E1 in erythrocytes (C), E2 in blood perfusate (D), E2 in plasma (E), and E2 in erythrocytes (F). The experiment was conducted with perfusate of four different compositions: 20% RBC (), 60% RBC () at 0% BSA, and 20% RBC () and 60% RBC () in 4% BSA. All were means ± S.D. of five experiments, and the lines were the fitted lines based on eqs. 12 to 15 with the optimum weighting scheme of unity.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   RBC to plasma partitioning of E1 (A) and E2 (B) in perfusate of different compositions: 20% RBC () and 60% RBC () in the absence of BSA, and 20% RBC () and 60% RBC () in 4% BSA. All data were means ± S.D. of five experiments.

Fitted Results for the Kinetic Model of E₁ and E₂ in Erythrocytes. Upon examination of the composite data for plasma and RBC, the results showed that E₁ and E₂ rapidly reached equilibrium in less than a minute (Fig. 4). The same observation was found by Koefoed and Brahm (1994). The fitted RBC unbound fractions of E₁ (f) and E₂, obtained upon the simultaneously fitting of the composite data, were 0.073 ± 0.032 and 0.10 ± 0.07, respectively, showing that both E₁ and E₂ were highly bound to erythrocytes. The fitted RBC metabolic intrinsic clearance of E₁ () was 0.11 ± 0.07 ml/min/HCT. Good fits were obtained although high coefficients of variation were found associated with the fitted parameters. The best fit to the model that considered red blood cell and plasma binding E₁ and E₂ and metabolism of E₁ is presented in Fig. 3, and the optimized parameters of five experiments are summarized in Table 3.

Metabolism of the Tracer Dose of [³H]E₁S in the Perfused Rat Liver Preparation. From the plasma unbound fraction (Table 3), values of the unbound fraction of E₁S in the blood perfusate (f) were 0.027 ± 0.004 according to eq. 1. The apparent hepatic clearance of [³H]E₁S was 5.8 ± 0.9 ml/min in the recirculating rat liver perfusion (Table 4). Within 150 min of recirculation, a rapid monoexponential decline of [³H]E₁S (t_1/2 = 27 ± 1 min) to around 1% of its initial concentration was observed (Fig. 5A). The accumulation of [³H]E₁G was higher than that of [³H]E₁ in perfusate within the first hour, followed by a gradual descent (Table 4). The decay half-life of [³H]E₁ paralleled that of its precursor, [³H]E₁S.

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Time-dependent profiles of [3H]E1S and [14C]E1 in the recirculating rat liver preparation. All data {[3H]E1S (), [3H]E1 (), [3H]E1G (), [14C]E1S (), [14C]E1 (), and [14C]E1G ()} were mean ± S.D. of four experiments. The lines were fitted lines based on average parameters shown in Table 6, with mass balanced rate equations described under Appendix and appropriate weighting schemes.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   A, bile flow rates of the rat liver during the recirculating liver preparation. B, the apparent biliary excretion clearances of [3H]E1S (), [3H]E1G (), [14C]E1S (), and [14C]E1G (). The apparent biliary excretion clearances of E1S and E1G were calculated as the biliary excretion rate divided by the midpoint reservoir concentration of each respective species. The data were means ± S.D. of four experiments.

Metabolism of the Tracer Dose of [¹⁴C]E₁ in the Perfused Rat Liver Preparation. Estrone was highly cleared upon the recirculation of [¹⁴C]E₁ with an apparent hepatic clearance of 9.4 ± 2.2 ml/min (Table 4), despite that the unbound fraction of E₁ in the blood (f) was very low. The unbound fraction of E₁ in blood (0.036 ± 0.006) based on the plasma unbound fraction and blood/plasma concentration ratio (eq. 1) agreed well with the value (0.053 ± 0.02) estimated according to eq. 2 that utilized the fitted value for f (0.073). The closeness in the values suggest the soundness in the estimation of f. The concentrations of [¹⁴C]E₁ declined monoexponentially with a slightly shortened half-life of about 20 ± 1.6 min (Fig. 5C). The elimination profile of the metabolite, [¹⁴C]E₁S, was more prolonged and the half-life was eventually similar to that of [³H]E₁S and [³H]E₁ but failed to decay in unison with [¹⁴C]E₁. The accumulation of [¹⁴C]E₁S and [¹⁴C]E₁G in the reservoir was comparable and increased within the first hour, followed by a gradual decline thereafter. During the time course of the experiment, the dose-corrected AUC of [³H]E₁G, when extrapolated to time infinity, was not different from that of [¹⁴C]E₁G (P > 0.05). Large variations were, however, observed (Table 4).

Fitted Results for the Kinetic Model of E₁ and E₁S in the Perfused Liver Preparation. Upon simultaneous fitting of perfusate and bile data consisting of [³H]E₁, [³H]E₁S, [³H]E₁G, [¹⁴C]E₁, [¹⁴C]E₁S, and [¹⁴C]E₁G in each study for the same liver preparation, good fits were obtained although high coefficients of variation were found associated with the fitted parameters. Parameter unidentifiability existed among the fitted parameters due to the remoteness of the endoplasmic reticulum with respect to the sampling compartment and the high correlation among parameters. For example, the endoplasmic reticulum efflux clearance (CL), the glucuronidation intrinsic clearance (CL), and the "pooled" metabolic intrinsic clearance (CL) for E₁ were all highly correlated, as were the bidirectional transmembrane clearance (CL), the desulfation intrinsic clearance (CL), and the pooled metabolic intrinsic clearance (CL) for E₁S. These highly correlated parameters exhibited coefficients of variation larger than one, and their reliability was much reduced. The other parameters exhibited lower coefficients of variation since the parameters were not correlated.

View larger version (17K): [in this window] [in a new window]   Fig. 7.   Simulated profiles of [3H]E1S, [14C]E1, and their metabolites in the recirculating rat liver preparation based on rapid equilibration of all species between the cytosolic and endoplasmic reticulum compartments. A, time-dependent profiles for [3H]E1S (), [3H]E1 (- - -), and [3H]E1G (......). B, time-dependent profiles for [14C]E1S (), [14C]E1 (- - -), and [14C]E1G (......).

Further Simulation for Understanding the Futile Cycling Kinetics of E₁ and E₁S in the Perfused Liver Preparation. Simulations were further performed based on the fitted and assigned parameters shown in Table 6. If rapid equilibration of the species existed between the cytosolic and endoplasmic reticulum compartments (high transport clearance of 1000 ml/min for E₁), similar elimination profile for E₁S and E₁ would result pursuant to [³H]E₁S and [¹⁴C]E₁ dosing as expected of futile cycling (Fig. 7); values for the transport clearances of E₁S and E₁G, when increased to 1000 ml/min, failed to further affect the shapes of the curves. With high exchange of E₁ between the cytosolic and endoplasmic reticulum compartments (inter-compartmental clearance of 1000 ml/min), the observed, discrepant half-lives of E₁ resulting from tracer [¹⁴C]E₁ dose and not the [³H]E₁S dose now disappeared.

Effects of the Reversible Pathway. For understanding the effects of futile cycling on the clearances of estrone and estrone sulfate, the sulfation intrinsic clearance of E₁ (CL) was set to zero to eliminate futile cycling of E₁, the formed metabolite. The result was the accumulation of [³H]E₁ upon elimination of the resulfation pathway after the [³H]E₁S dose (Fig. 8A). The profile of [³H]E₁G remained virtually unchanged because sulfation is a minor pathway. When the desulfation intrinsic clearance of E₁S (CL) was set as zero to prevent futile cycling of E₁S as the formed metabolite of estrone, a greater accumulation of the [¹⁴C]E₁S resulted, and formation of [¹⁴C]E₁G was reduced after the [¹⁴C]E₁ dose (Fig. 8B).

View larger version (15K): [in this window] [in a new window]   Fig. 8.   Simulated profiles of E1S, E1, and E1G following the administration of [3H]E1S in the absence of sulfation, with the sulfation intrinsic clearance of E1 (CL) equal to zero (A) and following the administration of [14C]E1 with the desulfation intrinsic clearance of E1S (CL) equal to zero (B), in the recirculating rat liver preparation. The symbols [3H]E1S (), [3H]E1 (), and [3H]E1G () in (A) are the mean of four experiments, and the lines[3H]E1S (), [3H]E1 (- - -), and [3H]E1G (...) represent the simulated profiles. The symbols[14C]E1S (), [14C]E1 (), and [14C]E1G () in (B) are the mean of four experiments and the lines [14C]E1S (), [14C]E1 (- - -), and [14C]E1G (...) represent the simulated profiles.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion Appendix References

E₁ and E₂ are potent estrogens that are bound tightly to albumin and to red blood cells. The erythrocyte distribution of E₁ and E₂ and metabolism of E₁ in the presence and absence of 4% BSA were characterized in our study. Rapid equilibrium was reached for E₁ and E₂ between plasma and RBC (Fig. 3), justifying use of the unbound fractions instead of the discrete binding association and dissociation rate constants for fitting. The observation was consistent with rapid exchange of E₁ and E₂ into bovine erythrocytes due to their high lipophilicity (log P values of E₁ and E₂ are 3.1 and 4.0, respectively) (Howard and Meylan, 1997) and their rapid dissociation from erythrocytes, as suggested by Koefoed and Brahm (1994). Two conclusions may be made regarding the metabolism of E₁ in the blood perfusate. First, the presence of a higher hematocrit increased the conversion of E₁ to E₂ due to the presence of higher quantities of 17-hydroxysteroid dehydrogenase in the RBCs (Table 2). Upon normalization to hematocrit, the same RBC clearance of E₁ was obtained for both 60% and 20% RBC perfusate. Second, E₁ was bound more strongly to BSA than to bovine erythrocytes (Fig. 4), and albumin greatly reduced both the distribution and metabolism of E₁ in erythrocytes (Table 2). Due to the stronger binding to BSA, the RBC/plasma partition coefficients of E₁ and E₂ decreased in the presence of 4% BSA (Fig. 4). The perfusate of 4% BSA and 20% RBC appeared optimal since the perfusate exhibited minimum RBC metabolism of E₁ but showed adequate oxygenation of the perfused rat liver preparation (Pang et al., 1988). The unbound fractions of E₁S and E₁ in this medium were constant (0.027 and 0.036, respectively).

Our study with the perfused rat liver preparation had entailed the simultaneous delivery of [³H]E₁S and [¹⁴C]E₁. The scheme allowed for the full characterization of the differential metabolism of the interconverting species. The study design was expected to reveal the underlying influence of vascular binding, transport, tissue binding, metabolism, and excretion on the hepatic clearances of E₁S and E₁. The hepatic clearances (0.53 ± 0.08 ml/min/g of liver for E₁S and 0.85 ± 0.2 ml/min/g of liver for E₁ at the blood flow rate of 0.91 ± 0.1 ml/min/g of liver) were lower than values of the sinusoidal transport clearances found previously: 33 ml/min/g of liver for E₁S and 184 ml/min/g of liver for E₁ (Tan and Pang, 2001). The finding suggests that transport is not rate limiting for E₁S and E₁ elimination. An unusual observation was that the half-lives of E₁, as a preformed species and as the metabolite of E₁S, were different. Moreover, parallel decay was observed between E₁S and E₁ after dosing with E₁S as expected of futile cycling but not with E₁.

This led us to postulate that the observed difference in half-lives was due to the sequestration of substrates within subcompartments. An extended, distributed-in-space model (the series-compartment liver model; Fig. 2) that incorporated zonal and subcompartmentalization of metabolic enzymes was indeed needed to interpret data of the recirculating liver perfusion of [³H]E₁S and [¹⁴C]E₁. Since this liver model was very complex, the fitting software allowed a maximum capacity of two zonal units only. Thus, more complicated models with higher zonal units were not examined. However, the use of two zonal units in the liver appeared adequate to describe the perfusion data. Inclusion of the endoplasmic reticulum proved absolutely necessary because its absence furnished an inferior fit, predicting a higher formation of [³H]E₁ (Fig. 7A).

Segregation of the endoplasmic reticulum from the cytosolic space implies not only distinction in metabolic enzymes but of drug partitioning in the endoplasmic reticulum compartment, membranes that are lipoidal in nature (Tirona and Pang, 1996). Creation of the endoplasmic reticulum compartment can be viewed by the following scenario: E₁S is rapidly transported into the endoplasmic reticulum compartment (transport clearance of 740 ± 146 ml/min) and is desulfated rapidly (intrinsic clearance of 332 ± 44 ml/min) to E₁ by estrone sulfatase. The metabolite E₁ is lipophilic and will likely stay within the endoplasmic reticulum space, presumably due to binding to membranes (Rao, 1998), leading to an imbalance in influx (clearance of 86 ± 40 ml/min) and efflux (clearance of 17 ± 2 ml/min). A similarly high partitioning of E₁ into the endoplasmic reticulum was observed by Zakim and Vessey (1977). The close proximity of the membrane bound E₁ and UDP-glucuronosyltransferase readily promotes the sequential glucuronidation of E₁ (intrinsic clearance of 105 ± 30 ml/min). Since UDP-glucuronosyltransferase is facing the luminal side (Iyanagi et al., 1986), the glucuronide conjugate of E₁ is released into the endoplasmic reticulum lumen and transported out to the cytosolic space (0.018 ± 0.001 ml/min) or is excreted into the bile. In addition, E₁ and E₁S in the endoplasmic reticulum space may also be metabolized by the cytochrome P450s (1A1, 3A4, and 2D) (Martucci and Fishman, 1993) to the "pooled metabolites", M', with metabolic intrinsic clearances of 255 ± 60 and 214 ± 57 ml/min, respectively.

The sulfation intrinsic clearance (318 ± 38 ml/min) was found to differ from previous values (11.2 ml/min) (Tan and Pang, 2001) that was estimated in the absence of the endoplasmic reticulum compartment. Rapid equilibration for the species between the cytosolic and endoplasmic reticulum compartments would effectively merge these subcompartments. As shown by the simulations (achieved with high CL_er), the different elimination half-lives of E₁ (Fig. 5) now disappeared in the absence of the endoplasmic reticulum compartment, rendering similar decay profiles for E₁S and E₁ (Fig. 7). Upon further probing of the partitioning species, E₁ and not E₁S nor E₁G was found to be important (simulations not shown). The pattern is characteristic of both drug and metabolite undergoing futile cycling (Ebling and Jusko, 1986; Tan and Pang, 2001). It may thus be concluded that the partitioning of E₁ into the endoplasmic reticulum had resulted in different elimination half-lives for E₁ in the recirculating liver preparation. To understand the influence of the futile cycling of E₁S and E₁ on the hepatic clearance of E₁S and E₁, simulations were further performed by obliterating the reversible pathway in the futile cycle. The simulation showed a greater accumulation of [¹⁴C]E₁S from [¹⁴C]E₁ in the absence desulfation (Fig. 8B), and a greater formation of [³H]E₁ from [³H]E₁S in the absence of sulfation pathway (Fig. 8A).

The fit also revealed that the sinusoidal bidirectional transmembrane clearance of E₁G was similar to that of E₁S (Table 6). Although Kanai et al. (1996) had suggested that E₁G transport was mediated by Oatp1 in a set of inhibition studies, direct data describing E₁G uptake in rat liver was lacking. The biliary intrinsic clearances of E₁S and E₁G were 8.0 ± 0.1 and 1.8 ± 0.2 ml/min, respectively. Upon multiplying the biliary intrinsic clearance with the cytosolic unbound fraction, the "effective" biliary intrinsic clearance of E₁S (0.7 ml/min) becomes smaller than that of E₁G (1.8 ml/min), whose biliary excretion appeared to be mediated by Mrp2 (Takikawa et al., 1996). In addition, a time-dependent decline in the biliary excretion clearances for the metabolites, [³H]E₁G, [¹⁴C]E₁S, and [¹⁴C]E₁G, was observed (Fig. 6B). Since these metabolites were formed in liver and were immediately excreted into the bile, this led to a component of biliary clearance that was not accounted for by the metabolite in circulation. A similar phenomenon was observed for the liver perfusion of enalapril and enalaprilat (deLannoy et al., 1993). However, the biliary excretion profiles of metabolites, E₁S and E₁G, were similar for both [³H]E₁S and [¹⁴C]E₁ doses (Figs. 5 and 6). As for the preformed [³H]E₁S, the excretion clearance reached asymptotic levels at 150 min, after reaching distribution equilibrium in the system. The bile flow rate declined with perfusion time due to the depletion of bile salts in the liver.

In conclusion, E₁ was highly bound to red blood cells and albumin and was metabolized by bovine erythrocytes to E₂. There were at least two factors governing the RBC metabolism of E₁, a higher hematocrit in blood increased the metabolism whereas albumin greatly attenuated RBC metabolism. The hepatic clearances of the simultaneously delivered tracer [³H]E₁S and [¹⁴C]E₁ were high in the rat liver preparation. Moreover, E₁ was highly partitioned into the endoplasmic reticulum compartment, yielding different elimination profiles for E₁S and E₁. The notion was substantiated upon degeneration of endoplasmic reticulum and the cytosolic spaces, obliterating the subcellular partitioning of E₁ in the cell and rendering parallel decay profiles of E₁ and E₁S. In general, futile cycling decreases the formation of the metabolite, which undergoes interconversion but increases formation of other noncycling metabolites. Presently, the complex kinetics of futile cycling are described adequately by a series-compartment (distributed-in-space) model that encompasses transport, metabolic heterogeneity, vascular binding and metabolism, and tissue binding. It is speculated that the above conceptual framework would apply to physiologically relevant conditions in humans if a high partitioning of E₁ exists and the same rate-limiting factors prevail.