Absolute bioavailability is an important parameter related to the pharmacological efficacy of orally administered drugs, and prediction of oral bioavailability is a key aspect of new drug development. Monkeys are often used in preclinical studies as nonhuman primates because of their genetic similarity to humans. However, the pharmacokinetic properties of drugs in monkeys are not always similar to those in humans. In particular, monkeys tend to exhibit lower bioavailability than humans (Sietsema, 1989; Chiou and Buehler, 2002; Ward et al., 2005), and it is difficult to predict the oral bioavailability in humans from data obtained in monkeys.

Since absolute oral bioavailability is affected by first-pass metabolism and/or excretion in both small intestine and liver, species difference in each process should be carefully examined to understand the factors affecting bioavailability. Chiou and Buehler (2002) showed, by analyzing literature data, including fraction of dose absorbed (F_a) and absolute bioavailability, that monkeys have similar intestinal drug absorption, but larger hepatic (nonrenal) clearance of 43 drugs compared with humans. However, the hepatic blood flow rate in monkeys is larger than that in humans (Davies and Morris, 1993), so that the ratio of hepatic clearance to hepatic blood flow rate is not very different between humans and monkeys (Ward and Smith, 2004), and therefore, the hepatic availability (F_h) may be similar in monkeys and humans. Thus, intestinal first-pass metabolism must be considered as a candidate mechanism to account for the poor oral bioavailability in monkeys.

In contrast to F_a and F_h, there have been few systematic studies of the extent of intestinal metabolism, which directly affects intestinal availability (F_g) following oral administration. Several researchers have pointed out the importance of intestinal metabolism in oral bioavailability, suggesting that it may be a defense mechanism to block passage of xenobiotics across the small intestine. Intestinal expression of CYP3A, a species of cytochrome P450, has been reported in rats (Kolars et al., 1992a; Li et al., 2002; Aiba et al., 2005), monkeys (Hashizume et al., 2001), and humans (Kolars et al., 1992b; Paine et al., 1997; Zhang et al., 1999). Interplay between CYP3A and an efflux transporter, P-glycoprotein (P-gp), the substrate specificities of which partially overlap, has been proposed to be involved in determining the efficiency of intestinal first-pass removal (Watkins, 1997; Benet et al., 2004).

Midazolam, a short-acting benzodiazepine central nervous system depressant, is a typical substrate of CYP3A, its oral bioavailability being 24 to 46% in humans (Sakuda et al., 2006). In humans, F_a · F_g and F_h were estimated to be 57 and 56%, respectively (Thummel et al., 1996), indicating that bioavailability of midazolam is affected by both intestinal and hepatic first-pass effects. Intraduodenally administered midazolam was rapidly absorbed with a T_max of 15 to 30 min, suggesting that midazolam is highly permeable and is mostly absorbed in the upper part of the intestine, where CYP3A expression is higher than in other regions of the small intestine (Paine et al., 1996, 1997).

Unlike humans, cynomolgus monkeys show minimal bioavailability of midazolam (approximately 2–6%), even though F_h was not small (∼66%) (Kanazu et al., 2004; Sakuda et al., 2006). Interestingly, Sakuda et al. (2006) recently showed that midazolam is completely absorbed by intestinal tissue in cynomolgus monkey, implying that intestinal metabolism may have a crucial role in the low bioavailability of midazolam (Sakuda et al., 2006). In rats, on the other hand, bioavailability of midazolam after oral administration and administration into a jejunal loop was only 4.6 and 12%, respectively, but intestinal first-pass removal was at most 25% (Lau et al., 1996; Higashikawa et al., 1999), suggesting that intestinal metabolism plays a relatively minor role in rats, compared with monkeys.

In the present study, we aimed to clarify the nature of the species difference in oral bioavailability of midazolam between monkeys and rats, focusing on the intestinal first pass. As regards membrane transport in the small intestine, several transporters have been suggested to contribute to drug absorption (Tsuji and Tamai, 1996; Katsura and Inui, 2003; Daniel, 2004; Sai and Tsuji, 2004), and in particular, P-gp is involved in regulation of the intestinal permeation of various xenobiotics (Terao et al., 1996). Midazolam shows high membrane permeability and is categorized as BCS class I (Wu and Benet, 2005). It is believed to be absorbed via passive diffusion, but not transported by P-gp (Benet and Cummins, 2001; Polli et al., 2001). On the other hand, another report has suggested that midazolam is a highly permeable P-gp substrate (Tolle-Sander et al., 2003). Therefore, involvement of P-gp in the absorption of midazolam remains controversial. To establish whether an efflux transporter is involved in the intestinal midazolam absorption in monkeys and rats, we evaluated midazolam transport in both apical-to-basal and basal-to-apical directions in various regions of the intestine, using an Ussing-type chamber system.

Materials and Methods

Chemicals. Midazolam was purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan), and 1′- and 4-hydroxymidazolam were purchased from Toronto Research Chemicals Inc. (North York, ON, Canada). Anti-human CYP3A4, anti-rabbit IgG (BD Gentest, Woburn, MA), C219 monoclonal antibody (Dako North America Inc., Carpinteria, CA), and anti-rat CYP3A2 (Daiichi Pure Chemicals, Tokyo, Japan) were commercial products.

Animals. Cynomolgus monkeys (5–6 kg, male) were purchased from China National Scientific Instruments and Materials Import/Export Corporation (Shenzhen, China), and maintained on approximately 108 g of food (Teklad Global 25% Protein Primate Diet; Harlan Teklad, Madison, WI) once a day, with free access to water. Beagle dogs (2–3 years old, 14–19 kg, male) were purchased from Shin Nippon Biomedical Laboratories, Ltd., Japan (Kagoshima, Japan) and maintained on approximately 350 g of food (VE-10; Nippon Pet Food Ltd., Tokyo, Japan) once a day, with free access to water. Sprague-Dawley rats (7 weeks old, male) were purchased from Japan SLC (Hamamatsu, Japan) and maintained with free access to food and water. Animals were deprived of food for 1 day before experiments. Animal studies were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals, Kanazawa University.

Pharmacokinetic Properties of Midazolam in Monkeys, Dogs, and Rats. Midazolam was dissolved in a mixture of dimethylacetamide and saline (1:1 v/v) for intravenous bolus injections at a dose of 0.1 mg/ml/kg for rats, and 0.1 mg/0.2 ml/kg for monkeys and dogs. Intravenous bolus injection was conducted without cannulation into the saphenous vein, cephalic vein, and femoral vein in monkeys, dogs, and rats, respectively. Midazolam was suspended in 0.5% methylcellulose solution for oral administration at a dose of 1 mg/5 ml/kg for rats, and 1 mg/2 ml/kg for monkeys and dogs. Oral administration to rats was done by gavage. Blood samples were collected at 5, 10, 15, and 30 min and 1, 2, 4, 8, and 24 h after intravenous administration, and at 15 and 30 min and 1, 2, 4, 8, and 24 h after oral administration in monkeys, dogs, and rats.

Transport Experiment in Ussing-Type Chamber. The intestinal absorption mechanism of midazolam was addressed by an Ussing-type chamber method, and the scheme of the apparatus is briefly described in Fig. 1A. Segments of upper, middle, and lower small intestine isolated from monkeys and rats were used for the Ussing-type chamber study. The portion approximately 15 and 4 cm from the pylorus was used as the upper small intestine in monkeys and rats, respectively. Segments of approximately 15 and 5 cm around the middle portion of the small intestine were used as middle small intestine, and segments of approximately 15 and 5 cm from the ileocecal junction were used as lower small intestine in monkeys and rats, respectively. Isolated intestinal tissue sheets from which the muscle layer had been removed with fine tweezers were mounted vertically in Ussing-type chambers that provided an exposed area of 0.75 cm² for monkeys and 0.25 cm² for rats. The volume of bathing solution on each side was 1.2 ml, and the solution temperature was maintained at 37°C in a water-jacketed reservoir. The test solution was composed of 128 mM NaCl, 5.1 mM KCl, 1.4 mM CaCl₂, 1.3 mM MgSO₄, 21 mM NaHCO₃, 1.3 mM KH₂PO₄, 10 mM NaH₂PO₄, and 5 mM d-glucose [adjusted at pH 6.0 or 7.4 for the apical (AP) or basal (BL) side, respectively], and gassed with 95% O₂/5% CO₂ before and during the transport experiment. The midazolam concentration on the donor side was set to be 30, 100, or 300 μM for monkeys, and 10, 30, 100, or 300 μM for rats. At the designated times, a 250-μl aliquot of acceptor side buffer was sampled and replaced with an equal volume of fresh buffer. The experiment was continued for 120 min, after which time the tissue and buffer on the donor side were also collected.

Measurement of Midazolam and Its Metabolites by LC with Tandem Mass Spectrometry Analysis. Midazolam, 1′-OH midazolam, and 4-OH midazolam in 100-μl aliquots of plasma were extracted with 10 μl of acetonitrile and 100 μl of internal standard solution (50 ng/ml triazolam in acetonitrile). The mixtures were centrifuged at 3000 rpm to remove precipitated protein; then, 100 μl of supernatant was diluted with 100 μl of 0.01 M ammonium formate (pH 3.0). A 5-μl aliquot was taken and analyzed by means of high-performance liquid chromatography (LC) coupled with tandem mass spectrometry. The LC system was a Shimadzu series 10AD-VP (Shimadzu, Kyoto, Japan) equipped with binary pumps, a degasser, and an SIL-HTc autosampler. The analytical column was an L-column ODS (2.1 × 50 mm, 5-mm particle size) column (Chemicals Evaluation and Research Institute, Tokyo, Japan). The flow rate was set at 0.2 ml/min. Separation was performed at 40°C with a gradient system generated from 0.01 M ammonium formate, pH 3.0 (A) and 0.2% formic acid in methanol (B): B was held at 40% for 1 min, increased linearly to 70% in 0.25 min, held at 70% for another 4.25 min, and then brought back to 40% in 0.1 min, followed by reequilibration for 3.4 min. The total cycle time for one injection was 9 min. Mass spectrometry experiments were conducted on a PE-Sciex API-3000 instrument (Applied Biosystems, Foster City, CA) with positive ionization electrospray. The multiple reaction monitor was set at 325.9 to 291.1 m/z for midazolam, 342.0 to 203.1 m/z for 1′-OH midazolam, 342.0 to 234.2 m/z for 4-OH midazolam, and 342.9 to 308.0 m/z for the internal standard, triazolam. The detection limit was estimated to be 0.5 ng/ml in all cases.

Western Blot Analysis. The mucosa of the monkey intestine was scraped off, immediately frozen in liquid nitrogen, and stored at –30°C until use. The scrapings were thawed in the ice-cold buffer containing 250 mM sucrose, 10 mM Tris, 1 mM EDTA, and Complete Protease Inhibitor Cocktail (Roche Diagnostics, Indianapolis, IN) (adjusted to pH 7.4 with 1 N HCl), and homogenized for 30 s at 25,000 rpm with an Ultra-Turrax T25 (IKA Werke GmbH and Co. KG, Staufen, Germany). Scraped rat intestinal mucosa was directly homogenized for 30 s at 25,000 rpm with an Ultra-Turrax T8 (IKA Werke GmbH and Co. KG) in the same buffer. The homogenates were centrifuged at 3000g at 4°C for 15 min. The supernatants were sonicated for 30 s and centrifuged at 105,000g, at 4°C for 90 min (CP-56G, RPS50-2; Hitachi, Yokohama, Japan). The pellets were resuspended with the same buffer and stored at –30°C until use. Protein concentration was determined by the Bradford method (Bradford, 1976) using a Bio-Rad protein assay kit (Bio-Rad, Hercules, CA). All samples were diluted to the same protein concentration (4.4 mg protein/ml), and 43.4 mM Tris-HCl, pH 6.8, containing 1% SDS and 5% β-mercaptoethanol (final concentrations) was added, followed by urea. Proteins (20 μg per lane) were separated by SDS-polyacrylamide gel electrophoresis (10% polyacrylamide gel) and transferred onto a polyvinylidene difluoride membrane (Millipore Corporation, Billerica, MA) at 2 mA/cm² for 120 min. Ponceau S staining confirmed that each lane was equally efficiently transferred to the membrane (data not shown). For detection of monkey CYP3A protein, the membrane was incubated in Tris-buffered saline (20 mM Tris-HCl, pH 7.5 and 137 mM NaCl) containing 5% skim milk for blocking, and incubated with 0.2% anti-human CYP3A4 antibody (WB-3A4; BD Gentest) in the above buffer containing 0.5% skim milk. For detection of rat CYP3A protein, the membrane was incubated in Tris-buffered saline containing 0.1% Tween 20 and 5% skim milk for blocking, and then incubated with 0.1% anti-rat CYP3A2 antibody (Daiichi Pure Chemicals) in the above buffer containing 0.5% skim milk. The membranes were rinsed with Tris-buffered saline containing 0.1% Tween 20, and reacted with 0.5% horseradish peroxidase-conjugated anti-rabbit IgG (BD Gentest). Bands were visualized using an enhanced chemiluminescence detection method with the ECL Plus Western blotting detection system (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK). Quantitative analysis was done by densitometry using a light-capture apparatus (AE6961FC; ATTO Bioscience, Tokyo, Japan). Precision Plus Protein Standards (#161-0373; Bio-Rad) covering the range of 10 to 250 kDa were used.

Data Analysis. The area under the curve (AUC) and the mean residence time were calculated by moment analysis of plasma concentration-time profile with the trapezoidal rule. Pharmacokinetic parameters were calculated in accordance with their definitions. C_max and T_max are the measured maximum plasma concentration and the time point at which that maximum plasma concentration was observed, respectively.

Transport and metabolic clearances in the Ussing-type chamber were estimated by use of the following equations: In the AP-to-BL direction, In the BL-to-AP direction, where CL_abs, CL_met,a-b, X_basal, and AUC_apical are the transport clearance in the AP-to-BL direction, the metabolic clearance in the AP-to-BL direction, the amount of parent drug appearing in the BL compartment, and the area under the parent drug concentration curve in the AP compartment, respectively, and CL_sec, CL_{met, b-a}, X_apical, and AUC_basal are the corresponding parameters in the opposite direction. The ΣM_{apical, tissue, basal} represents the sum of metabolites (1′-OH and 4-OH midazolam) in the AP, tissue, and BL compartments. The mucosal extraction ratio (ER) was calculated as: This ER is based on the assumption that ER is affected only by CL_abs and CL_met,a-b, and can be defined as the contribution of CL_met,a-b to the overall elimination clearance (CL_abs + CL_met,a-b). Taking eqs. 1 through 4 into consideration, eq. 5 can be transformed as follows: Thus, ER, as defined above by us, is consistent with the definition proposed by Fisher et al. (1999), i.e., the ratio of drug permeation to drug metabolism.

Since metabolite formation in the Ussing-type chamber was much greater in the AP-to-BL direction than in the BL-to-AP direction, a two-step metabolism model was constructed, assuming that first-pass intestinal removal occurs in association with the uptake phase of midazolam from the apical cell surface (see Results, Discussion, and Fig. 1B). Accordingly, CL_abs, CL_met,a-b, CL_sec, and CL_met,b-a are represented by hybrid parameters of plasma membrane permeability, metabolic clearance in intracellular space, and a novel parameter, the fraction of first-pass removal. CL_AT is defined as the membrane permeability of the parent drug (midazolam in the present study) across the apical membrane from outside of the tissue to inside of the tissue (apical-to-tissue). Similarly, CL_TA,CL_TB, and CL_BT represent the tissue-to-apical, tissue-to-basal, and basal-to-tissue membrane permeabilities, respectively. CL_M is metabolic clearance in the intracellular space, and f_met is fraction metabolized during the apical uptake.

When we assume a steady-state condition for the amount of parent drug in the tissue and a sink condition in the acceptor, we have:

In the AP-to-BL direction, In the BL-to-AP direction, where C_tissue,a-b and C_tissue,b-a are the parent drug concentrations in small intestinal tissue, calculated by assuming that the specific gravity of the tissue is unity, at the end of the experiment after drug administration on the apical and basal sides, respectively. Therefore, all the microscopic parameters (CL_AT ∼ CL_M and f_met) can be directly calculated based on the following equations:

Statistical Analysis. Data are presented as mean ± standard error of the mean (S.E.M.) except for Fig. 2 and Table 1. Plasma concentration of midazolam and pharmacokinetic parameters are presented as mean ± standard deviation (S.D.) in Fig. 2 and Table 1. Statistical comparisons were performed by ANOVA with Tukey's honestly significant difference post hoc comparison test. Values of p < 0.05 were considered to be significant. Experimental number refers to the number of animals.

Results

Pharmacokinetics of Midazolam after Intravenous and Oral Administration. To understand the pharmacokinetic properties of midazolam in monkeys, dogs, and rats, we first measured the plasma concentrations of midazolam after oral and intravenous administration. Plasma concentration-time profiles and the estimated pharmacokinetic parameters are shown in Fig. 2 and Table 1, respectively. Monkeys and rats showed minimal bioavailability (2.1% and 1.1%, respectively), whereas the bioavailability in dogs was 15.1% (Table 1). Similarly, C_max after oral administration was higher in dogs than in monkeys or rats (Fig. 2; Table 1). The time to maximum concentration (T_max) and the mean absorption time were within1hin each species (Table 1), indicating that orally administered midazolam was rapidly absorbed. It was also highly distributed in each species, with V_dss of 2∼3 l/kg (Table 1).

Transport of Midazolam across the Upper, Middle, and Lower Small Intestine of Monkeys and Rats. To directly measure the intestinal transport and metabolism, we used an Ussing-type chamber system to examine vectorial transport across upper, middle, and lower small intestinal segments, with 30 μM midazolam in the donor side chamber. One of the advantages of using an Ussing-type chamber for analyzing intestinal drug transport is the ease of assessment of bidirectional transport across intestinal tissues (AP-to-BL and BL-to-AP). Transport clearances (CL_abs and CL_sec) and metabolic clearances (CL_met,a-b and CL_met,b-a) were calculated from the AP-to-BL and BL-to-AP transport data for each segment of the small intestine. Metabolic clearance was much higher than the transport clearance in all segments in monkeys, whereas in rats, transport clearance was higher than the metabolic clearance in all the segments (Fig. 3, A, B, D, and E). There was also clear species difference between monkeys and rats in the absolute values of metabolic clearance (Fig. 3, B and E). ER calculated from eq. 5 was much higher in all segments in monkeys than in rats (Fig. 3, C and F). These results suggest that there is a predominant first-pass effect in the small intestine of monkeys (Fig. 3C), in contrast to a small intestinal first-pass effect in rats (Fig. 3F). It is noteworthy that metabolic clearance and ER in monkey upper small intestine were higher in the AP-to-BL direction than in the BL-to-AP direction (Fig. 3, B and C).

Concentration Dependence of Intestinal Transport Clearance of Midazolam in Monkeys and Rats. To investigate nonlinearity in the intestinal absorption of midazolam, the concentration dependence of CL_abs was examined in the middle small intestine of both species. CL_abs tended to increase with increasing concentrations of midazolam in monkeys, whereas it tended to decrease with increasing concentrations of midazolam in rats (Fig. 4, A and B), suggesting that one or more different mechanisms are involved in the nonlinear absorption of midazolam in the two species. Similar increase and decrease with increasing midazolam concentration were also observed for CL_seq in monkeys and rats, respectively (Fig. 4, C and D). On the other hand, significant saturation of CL_met,a-b and CL_met,b-a was observed in monkey and rat middle small intestine (Fig. 4, E–H). In a nonlinear absorption study with monkey upper small intestine, CL_met,a-b at 300 μM midazolam was much lower than that at 30 μM midazolam and similar to CL_met,b-a (CL_met,a-b and CL_met,b-a at 300 μM were 3.23 ± 0.54 and 2.06 ± 0.41 × 10^–6 cm/s, respectively). This is in contrast to the clear difference between CL_met,a-b and CL_met,b-a at 30 μM midazolam in monkey upper small intestine (Fig. 3B).

Excretion of Midazolam Metabolites. We next measured the appearance of metabolites on the apical and basal sides, and in intestinal tissue (Table 2). When midazolam was added to the apical side, larger amounts of metabolites appeared on the apical side for both monkey and rat small intestinal segments (Table 2). On the other hand, when midazolam was added to the basal side, the appearance of metabolites on the basal side was similar to or higher than that on the apical side, especially at lower substrate concentration (Table 2). Addition of higher concentrations of midazolam to the basal side tended to decrease the appearance of metabolites on the basal side, but to increase that on the apical side (Table 2), suggesting the involvement of saturable efflux system(s) for metabolites in the basal membrane.

Model-Dependent Analysis. To understand in more detail the intestinal transport and metabolism of midazolam, we attempted to estimate intrinsic parameters that directly represent membrane permeability and intracellular metabolism. We used a simple model that included only apical-to-tissue permeability (CL_AT), and tissue-to-apical (CL_TA), tissue-to-basal (CL_TB), basal-to-tissue (CL_BT), and intracellular metabolic clearances (CL_M). In this model, CL_M in both the AP-to-BL and BL-to-AP directions can be calculated from the results of the transport studies. The CL_M values in the two different directions are expected to be similar to each other, if we assume a single compartment inside the tissue. The CL_M values estimated in both directions in rats were actually not so different between each other (CL_M obtained from AP-to-BL and BL-to-AP transport studies at 30 μM midazolam were 1.37 ± 0.20 and 0.775 ± 0.178 in upper small intestine, 1.49 ± 0.22 and 1.65 ± 0.33 in middle small intestine, and 0.907 ± 0.266 and 1.17 ± 0.43 in lower small intestine, respectively (mean ± S.E.M. of 3–4 determinations in units of 10⁶ cm/s), whereas in monkeys, the CL_M values obtained from AP-to-BL and BL-to-AP transport studies at the same concentration of midazolam were 22.6 ± 5.5 and 1.40 ± 0.61 in upper small intestine, 18.1 ± 3.9 and 7.28 ± 1.99 in middle small intestine, and 18.3 ± 4.6 and 11.1 ± 5.6 in lower small intestine, respectively. Thus, disposition of midazolam in monkey small intestine is more complicated than that in rats.

Therefore, it seems that a novel kinetic model is necessary to explain the overall data in rats and monkeys. Since the CYP3A protein distribution has been reported to be biased toward the apical membrane side, at least in rat intestinal epithelial cells (Kolars et al., 1992a), we next considered a novel, two-step metabolism model (Fig. 1B), in which metabolism occurs heterogeneously. Parameters obtained with this model are listed in Table 3. The CL_M was saturable, especially in monkeys, and, in addition, CL_AT and CL_BT were also saturable at higher substrate concentrations in rats. The f_met value, which represents the fraction metabolized during apical uptake, was higher in the upper small intestine than in other segments and tended to decrease at higher substrate concentrations in all segments of monkeys (Table 3). On the other hand, the f_met value was close to zero in all segments of rats (Table 3), suggesting that metabolism of midazolam is heterogeneous in monkeys, but not in rats. The CL_AT in the lower small intestine was higher than that in the upper segments of the intestine in both monkeys and rats (Table 3).

CYP3A Expression in Monkey and Rat Small Intestine. To investigate the correlation between metabolic activity and expression of CYP3A protein in the small intestine, Western blot analysis was performed. In this experiment, anti-CYP3A4 and anti-CYP3A2 antibodies were used to detect immunoreactive proteins in monkeys and rats, respectively, since it has been reported that immunochemical reactivity to anti-CYP3A antibody is similar in monkeys and humans (Shimada et al., 1997), and that CYP3A2 antibody cross-reacts with one or more other proteins that probably belong to the CYP3A family in rats (Aiba et al., 2005). As shown in Fig. 5, CYP3A proteins were detected in all intestinal segments of both monkeys and rats. Furthermore, CYP3A expression tended to be higher in the upper small intestine than in other segments in both monkeys and rats (Fig. 5).

Discussion

Monkeys are sometimes used in preclinical studies because they are expected to show pharmacokinetic characteristics similar to those of humans. Indeed, Ward and Smith (2004) reported that values of total body clearance normalized to hepatic blood flow in humans were better correlated to those in monkeys than to those in rats or dogs, whereas Chiou and Buehler (2002) noted that hepatic clearance per body weight was higher in monkeys than in humans. However, in the case of oral administration of drugs, monkeys often show lower bioavailability than humans (Sietsema, 1989; Chiou and Buehler, 2002; Ward et al., 2005). This could be explained by insufficient intestinal availability in monkeys. To clarify the mechanism of this interspecies difference of oral bioavailability, we chose midazolam as a model drug to analyze intestinal metabolism in monkeys in the present study, since it is useful as a probe for CYP3A, which is functionally expressed in the small intestine, and it also shows a marked interspecies difference in oral bioavailability between human and animals (Sakuda et al., 2006). We confirmed that the bioavailability of midazolam was at most 2% in both monkeys and rats (Table 1). However, we observed extensive intestinal metabolism of midazolam (ER value close to unity) in monkeys, but not in rats (Fig. 3), suggesting that oral bioavailability is affected by intestinal first-pass metabolism in monkeys, whereas in rats, bioavailability of midazolam is mainly determined by hepatic metabolism.

In Ussing chamber experiments, we found that exposure of the apical surface of intestinal segments to midazolam yielded markedly different values of the ratio of transport (represented by CL_abs) to metabolism (represented by CL_met,a-b, Fig. 3) in monkeys and rats. The small intestine in monkeys showed much higher CL_met,a-b than CL_abs, whereas CL_abs was much larger than CL_met,a-b in rats, resulting in a species difference of ER between monkeys and rats (Fig. 3). Thus, most of the midazolam taken up into intestinal tissues is metabolized in monkeys, whereas little is metabolized in rats. This may support the usefulness of cynomolgus monkeys, compared with rats, in predicting drug-drug interactions via CYP3A inhibition and/or induction upon oral administration of midazolam with other drugs, as proposed by Kanazu et al. (2004).

One of the advantages of the Ussing-type chamber as an experimental system for analyzing intestinal drug transport is the ease of assessment of bidirectional transport across intestinal tissues (i.e., in the AP-to-BL and BL-to-AP directions). The present study has provided the first demonstration, using an Ussing-type chamber, that the metabolism of midazolam is strongly dependent on the transport direction in monkey upper small intestine: CL_met-a-b was much higher than CL_met,b-a at 30 μM midazolam (Fig. 3B), but there was little difference between CL_met,a-b and CL_met,b-a at 300 μM midazolam (see Results). Similar “directional metabolism” has been reported in CYP3A4-transfected Caco-2 cells for midazolam and other compounds (Cummins et al., 2002, 2004), and in rat small intestine for verapamil (Johnson et al., 2003), and was theoretically simulated by Tam et al. (2003). Cummins et al. (2002) suggested that the difference in ER of K77 (N-methyl piperazine-Phe-homoPhe-vinylsulfone phenyl), which is a substrate of P-gp and CYP3A, between the AP-to-BL and BL-to-AP directions was caused by a slowing of absorption mediated by P-gp. They also reported that midazolam, which is not transported by P-gp, exhibited a similar ER difference between the AP-to-BL and BL-to-AP directions, and speculated that in this case, the ER difference might be due to a difference in the times required for the intracellular midazolam concentration to reach a steady state (Cummins et al., 2004). Although these studies were performed in transfectant cell lines in vitro, we have shown that such ER differences exist in upper small intestine freshly isolated from monkeys in the present study (Fig. 3). The mechanism(s) of this ER difference has not yet been clarified in detail, but it is noteworthy that the appearance of midazolam metabolites also depended on the direction of transport, with metabolites being found preferentially on the apical side after addition of midazolam to the apical side, but not to the basal side (Table 2). This observation can be explained by asymmetric localization of metabolic activity in the small intestine. CYP3A is known to show inhomogeneous subcellular localization in intestinal epithelial cells, and is more abundant in the vicinity of the apical membrane than the basal membrane, at least in rats (Kolars et al., 1992a). Consequently, we speculate that intestinal metabolic enzymes tend to rapidly metabolize drugs absorbed across the apical membrane, compared with those entering from the basal membrane, thereby providing a barrier against entry of xenobiotics. However, inhomogeneous subcellular localization of CYP3A in monkey small intestine has not yet been demonstrated, and it will be necessary to establish appropriate conditions for detailed immunohistochemical studies to test this hypothesis. If midazolam taken up across the apical membrane into the cell is more rapidly metabolized, the time to reach a steady-state concentration should theoretically be shorter than that when the drug is taken up across the basal membrane, since the time period required to reach a steady state depends on the elimination rate constant inside the cell. By using Caco-2 cells, Fisher and coworkers have shown that midazolam more rapidly reaches a steady state (∼3 min) when added apically than when added basally (>10 min) (Fisher et al., 1999; Cummins et al., 2002). This finding is consistent with our hypothesis that midazolam absorbed from intestinal lumen is rapidly metabolized.

Based on the idea that metabolism of midazolam occurs faster when it is taken up from the apical side, we constructed a novel two-step metabolism model (Fig. 1B). Although the validity of this model should be confirmed by further analyses, this model can well describe the directional metabolism, assuming a single compartment, even though it requires only one additional parameter (f_met) compared with a single-compartment model, together with membrane permeability (CL_AT,CL_TA,CL_BT, and CL_TB) and intracellular metabolism (CL_M). On the basis of this model, ER_abs defined by eq. 4 can be described as follows: Therefore, if f_met is negligible, ER is determined only by basolateral efflux (CL_TB) and intracellular metabolism (CL_M). On the other hand, if f_met is moderate, apical back-efflux (CL_TA) also affects the ER. Thus, this model can describe the function of P-gp as a direct contributor to ER. This is consistent with the idea of slow absorption put forward by Cummins et al. (2002).

Midazolam is classified into BCS class I, with high membrane permeability (Wu and Benet, 2005). Nevertheless, saturable transport of midazolam was observed in apical (CL_AT) and basal (CL_BT) uptake in rat middle and lower small intestine (Table 3), and this was compatible with the saturation of CL_abs and CL_sec in rat middle small intestine (Fig. 4, B and D) and lower small intestine (the values of CL_abs and CL_sec at 30 μM midazolam in the donor side chamber were 28.0 ± 8.1 and 26.2 ± 5.6, and those at 300 μM midazolam were 10.6 ± 2.8 and 12.7 ± 2.0, respectively). Apical uptake (CL_AT) of midazolam also decreased when a concentration of 300 μM was given in the upper small intestine of monkeys (Table 3). Nevertheless, CL_abs increased as the midazolam concentration was increased (CL_abs at 30 μM midazolam in the donor side chamber was 0.454 ± 0.093 and CL_abs at 300 μM was 0.839 ± 0.172). Regarding the saturation of f_met (Table 3), the fraction not metabolized (1 – f_met) was increased approximately 5-fold at the higher midazolam concentration. Thus, the apical net influx clearance [CL_AT × (1 – f_met)] was increased at higher midazolam concentration. In addition to the saturation of CL_AT and f_met, intracellular metabolism, represented by CL_M, also appeared to approach saturation as the midazolam concentration was increased (Table 3), contributing to the enhancement of CL_abs.

Although we have observed saturable intestinal transport of azasetron, which is highly soluble and permeable (Tamai et al., 1997), involvement of influx/efflux transporter(s) in gastrointestinal absorption of BCS class I drugs is believed to be rare (Wu and Benet, 2005). The present report is the first to describe a saturable absorption process for midazolam. Further analyses will be needed to investigate whether a carrier-mediated transport system(s) is involved in the intestinal absorption of midazolam.

Model-dependent analysis of the intestinal absorption of midazolam in monkeys revealed nonlinear behavior of several parameters representing metabolism (CL_M and f_met; Table 3). The K_m values of human CYP3A4 for 1′- and 4-OH midazolam formation were 1.56 and 38.0 μM, respectively (Ghosal et al., 1996). Although the intracellular unbound concentration of midazolam is unknown, saturation could occur at relatively high midazolam concentrations. Actually, CL_M in each intestinal segment of monkeys was decreased at higher concentrations in the donor side buffer (Table 3), indicating saturation of intracellular metabolism. On the other hand, the f_met values were similar at lower (30 μM) and higher (300 μM) concentrations in the donor side medium for monkey middle small intestine (Table 3). Therefore, the characteristics of the enzyme(s) contributing to CL_M and f_met could be different in monkey middle small intestine. The f_met in monkeys was 0.4 to ∼0.9 at the lowest midazolam concentration, whereas in rats, f_met was close to zero in all the segments (Table 3), indicating that rapid and substantial metabolism of midazolam during apical uptake in the small intestine occurs in monkeys, but not in rats. On the other hand, the CL_M was not so different between monkeys and rats, especially in the upper segments of the intestine (Table 3). Thus, the species difference in metabolism of midazolam arises mainly during the apical uptake phase (f_met). It is not surprising that the f_met has not been taken into consideration in estimating intestinal drug absorption, since f_met is monkey-specific (Table 3), and rats, which exhibit minimal f_met, at least for midazolam, are mainly used in preclinical studies. It is generally considered that molecules responsible for uptake and metabolism (e.g., transporters and cytochrome P450 enzymes) are separately localized in cells (e.g., in cell surface membranes and microsomes), so the concept of drug metabolism during uptake from the apical cell-surface may seem to be unphysiological. However, the possible existence of monkey-specific molecules has not yet been excluded. Furthermore, the directional metabolism observed in the current study can be well explained by this model. Therefore, we believe that it is reasonable to conclude that metabolic events are at least tightly linked with uptake from the apical cell surface. In other words, we suggest that first-pass intestinal removal is functionally closely associated with apical uptake in cynomolgus monkeys, even though it is still unclear whether f_met and CL_M are physiologically related to each other, and whether the molecules responsible for f_met and CL_M are different or not. Further studies are necessary to investigate the molecule(s) responsible for f_met and CL_M.

The CYP3A immunoreactivity was higher in the upper small intestine than in other segments of both monkeys and rats (Fig. 5). This may be compatible with the segmental difference in f_met, which was higher in the upper small intestine than in other segments of both species, although its absolute value was quite small in rats (Table 3). However, involvement of other metabolic enzyme(s) in the oxidation of midazolam cannot be ruled out. It is important to realize that oral bioavailability can be affected by many factors, including regional differences in activities of intestinal metabolic enzymes and differences in the residence time of drug molecules in different intestinal segments.

In conclusion, our findings with an Ussing-type chamber system indicate that the minimal oral bioavailability of midazolam in cynomolgus monkeys is a consequence of rapid and extensive intestinal metabolism that is associated with the uptake of midazolam across the apical membrane cell surface. Since midazolam is a typical substrate of CYP3A, many other CYP3A substrate drugs could also be extensively metabolized in monkey intestine and thus show low bioavailability. On the other hand, saturable transport mechanisms are involved in the uptake of midazolam into intestinal tissues, especially in rats. These findings will be helpful for evaluating the results of preclinical studies and predicting their relevance to drug absorption in humans.