Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Orally administered drugs often exhibit incomplete availability to the systemic circulation. In addition to the physicochemical properties of a drug that can limit its net absorption from lumen to portal blood, exposure to enzymes in the mucosal epithelium of the gastrointestinal tract and in the liver can result in significant first-pass metabolic elimination. The majority of the drug metabolizing enzymes, specifically the cytochrome P-450s (CYPs), are located in the liver. However, it has recently been shown that intestinal CYP3A4, the dominant mucosal P-450 (Watkins et al., 1987; de Waziers et al., 1990; Paine et al., 1997), also participates in first-pass drug elimination. Indeed, the obligatory passage of transcellularly absorbed drug through the villous epithelial barrier with its relatively high CYP3A4 content (Kolars et al., 1994) results in significant intestinal first-pass metabolism of at least two CYP3A substrates, cyclosporine (Watkins et al., 1987) and midazolam (Paine et al., 1996).

It has been proposed that mucosal metabolic extraction can be described by the same well-stirred model that is commonly invoked for hepatic extraction (Klippert et al., 1982; Pond and Tozer, 1984; Mistry and Houston, 1987). Assuming that metabolic clearance is perfusion-limited and that the rate of elimination is first-order, the extraction ratio will be a function of the unbound intrinsic clearance of the eliminating organ, plasma protein binding of drug, and blood flow to the organ (Klippert et al., 1982). Mucosal clearance (Cl_m) can be expressed as the product of the mucosal blood flow (Q_m) and the mucosal extraction ratio (E_m). Gillette and Pang (1977) proposed that the contribution of the gut wall to total splanchnic clearance is represented by the product of the hepatic bioavailability and mucosal clearance. For a well-stirred model, the mucosal extraction ratio can be re-expressed as the following:

(1)

where

(2)

and Cl'_in,m is the unbound intrinsic clearance, and f_u is the plasma-free fraction. It has been suggested but not proven that plasma protein binding will have an identical modulating effect on the intrinsic clearance and mucosal extraction ratio when drug is delivered to the mucosal epithelium via arterial blood or the gut lumen (Pond and Tozer 1984; Mistry and Houston, 1987). However, because of obvious experimental limitations, this hypothesis has not been rigorously tested in humans.

The effect of plasma protein binding on the extraction of a drug by the human mucosal epithelium might be more easily evaluated with an in vitro model system. Caco-2 cells are routinely used for the prediction of in vivo intestinal drug absorption in humans (Artursson and Karlsson, 1991; Gres et al., 1998). Differentiated Caco-2 cells have also been used to study specific transporters such as P-glycoprotein (Hunter et al., 1993; Gan et al., 1996; Alsenz et al., 1998) and certain drug-metabolizing enzymes, including CYP1A1, CYP3A5, and glutathione S-transferase (Meunier et al., 1995; Gan et al., 1996; Boulenc, 1997). Unfortunately, under standard culturing conditions, Caco-2 cells do not express appreciable levels of CYP3A4, the dominant drug-metabolizing enzyme of the human small intestine. Thus, these cells have found limited utility in the study of intestinal first-pass drug metabolism.

It was shown recently that CYP3A4 expression and catalytic activity can be increased dramatically in the Caco-2 monolayer by treating the cells for 2 weeks postconfluence with 1,25-dihydroxy vitamin D₃ [1,25-(OH)₂-D₃] (Schmiedlin-Ren et al., 1997). Immunoblot analysis of homogenate prepared from modified Caco-2 monolayer homogenate revealed that each 4.2-cm² culture insert contained ~4 pmol of CYP3A (Fisher et al., 1999). Using midazolam (MDZ) as a well-absorbed (Heizmann and Ziegler, 1981; Smith et al., 1981) and specific probe for CYP3A activity (Kronbach et al., 1989; Gorski et al., 1994), conditions were defined that permitted the calculation of a first-pass MDZ extraction ratio (14.5 ± 3.1%; Fisher et al., 1999). Although significantly lower than the 43% mean in vivo first-pass MDZ extraction ratio (Paine et al., 1996), the Caco-2 extraction ratio was deemed more than sufficient for the experimental objectives of the present work. Those objectives were to characterize the effect of extracellular protein binding on the accumulation of MDZ within the cell monolayer, the extent of MDZ metabolism, and the first-pass metabolic extraction ratio, after apical and basolateral MDZ administration. We also examined the distribution of intracellularly formed MDZ metabolites in the culture system in the absence or presence of extracellular protein.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. Caco-2 cells (American Type Culture Collection HTB37) were cloned by limiting dilution as described previously (Schmiedlin-Ren et al., 1997). Dulbecco's modified Eagle medium (DMEM), nonessential amino acids, penicillin, streptomycin, and Hanks' balanced salt solution were obtained from Life Technologies, Inc. (Grand Island, NY). Fetal bovine serum (FBS) was purchased from Hyclone (Logan, UT). Uncoated track-etched polyethylene terephthalate inserts and mouse laminin were obtained from Collaborative Biomedical Products (Bedford, MA). The hormone 1,25-(OH)₂-D₃ was obtained from Calbiochem (La Jolla, CA). N-Methyl-N-(t-butyl-dimethylsilyl)trifluoroacetamide was purchased from Pierce Chemical (Rockford, IL). The following chemical standards, MDZ, ¹⁵N₃-MDZ, 1'-OH MDZ, 4-hydroxymidazolam (4-OH MDZ) and 1'-[²H₂]1'-OH MDZ, were gifts from Roche Laboratories (Nutley, NJ). Acetonitrile and ethyl acetate were purchased from Fisher Scientific (Santa Clara, CA). Purified human albumin (fraction V) and dimethyl sulfoxide (DMSO) were purchased from Sigma Chemical Co. (St. Louis, MO). Stock solutions of MDZ and 1,25-(OH)₂-D₃ were prepared in DMSO and absolute ethanol, respectively.

Culture Conditions. The Caco-2 subclone, P27.7 (Schmiedlin-Ren et al., 1997), was obtained at passage 12 and grown on culture dishes in passage medium as described previously (Fisher et al., 1999). All experiments were performed with cells at passage no. 19 or 20. Cells were seeded onto hand-applied, laminin-coated polyethylene terephthalate inserts as described previously (Fisher et al., 1999) at 5.2 × 10⁶ cells/cm² and grown in complete growth medium (passage medium supplemented with 45 nM DL--tocopherol) until confluent. Upon achieving confluence, cell monolayers were fed for 2 weeks, every 2 to 3 days, with complete differentiation medium (DM), which contained 0.25 µM 1,25-(OH)₂-D₃ and 5% FBS (Fisher et al., 1999).

Monolayer Integrity. Measurements of TEER were used to assess cell monolayer integrity. Resistance (ohms) was measured on each culture with a Millipore Millicell electrical resistance system (Bedford, MA) immediately before an experiment. For a given set of cell cultures, a single insert that did not contain cells was measured for background resistance. The product of the background-corrected resistance and the surface area of the insert (4.2 cm²) was defined as TEER.

Determination of MDZ Free Fraction. Free fractions of MDZ in experimental medium containing either 5% FBS (v/v) or 4 g/dl HSA were determined by equilibrium dialysis as described previously (Paine et al., 1996). Briefly, MDZ was added to protein-supplemented DM to achieve a final concentration of 2 µM. Spiked medium (300 µl) was placed on one side of a M_r 12,000 to 14,000 molecular weight cut-off dialysis membrane (Spectrapor; VWR Scientific) and allowed to equilibrate for 4 h in a 37°C water bath against an equal volume of modified DM. At the end of the incubation period, each fraction was analyzed for MDZ. The ratio of MDZ concentration in dialysate to MDZ concentration in supplemented DM was assumed to be the free fraction of MDZ in HSA-supplemented culture medium during an incubation period.

MDZ, 1'-OH MDZ, and 4-OH MDZ Assays. Measurement of 1'-OH MDZ was the primary means of assessing the extent of MDZ metabolism by Caco-2 cell cultures (Schmiedlin-Ren et al., 1997). The clinically minor metabolite, 4-OH MDZ (Dundee et al., 1984), was also quantitated when enzyme-saturating concentrations of MDZ were investigated. The formation of 4-OH MDZ was not measured routinely because of a limited supply of stable isotope-labeled 4-OH MDZ internal standard. Levels of MDZ were used to determine the extent of drug absorption into and across the cell monolayer. For each experimental condition (variable time or MDZ concentration), apical medium, basolateral medium, and cell homogenate from a single culture were assayed in duplicate for parent and metabolite(s) and quantified by gas chromatography-mass spectrometry as described previously (Fisher et al. 1999).

Effect of Extracellular Protein Binding on the Cell Uptake and Metabolism of MDZ. The time course of uptake and metabolism in the Caco-2 monolayer was determined for both apical and basolateral routes of MDZ administration. The MDZ concentration was 3 µM in a 1.5-ml volume, a concentration below the median K_m for 1'-hydroxylation (3.8 µM) found in human duodenal microsomes (Paine et al., 1997) and below the apparent K_m in apically dosed Caco-2 cell monolayer cultures (9.1 µM; Fisher et al., 1998). For "treated cultures", HSA (4 g/dl) was added to the modified DM in the basolateral compartment. Extracellular medium and cell monolayers were collected from treated and control cultures after 0, 10, 20, 30, 45, 60, 90, and 120 min of incubation and analyzed for MDZ and 1'-OH MDZ.

Determination of First-Pass Metabolic Extraction Ratio. Results from the time-course study allowed us to define conditions for measurement of a first-pass metabolic extraction across the Caco-2 cell monolayer in the presence and absence of HSA or FBS in the basolateral compartment. For determination of the extraction ratio, MDZ (3 µM) was incubated under sink conditions (10-20 min in duration, when <10% of the apical dose had crossed into the basolateral compartment). The amount of 1'-OH MDZ found in the apical, basolateral compartments and cell monolayer at the end of the incubation was summed and entered into eq. 3, with the amount of MDZ in the basolateral compartment, to obtain the first-pass extraction ratio (ER):

(3)

When 4-OH MDZ was also measured (enzyme saturation experiment), the sum of 1'-OH MDZ and 4-OH MDZ produced in the incubation interval was used for the ER calculation.

Effect of HSA on the Saturation Curve for MDZ 1'-Hydroxylation. Cell monolayers were incubated with apically administered MDZ (3, 8, 25, and 100 µM) for 10 min, with 4 g/dl HSA in the basolateral medium. Total formation of 1'-OH MDZ and 4-OH MDZ were quantitated as described above. Because apical dosing concentrations may not be identical with the unbound concentration of drug at the enzyme active site, particularly when HSA is added to the basolateral medium, kinetic analysis was performed using either the initial apical MDZ concentration or the predicted intracellular MDZ concentrations at the end of the 10-min incubation period. The mean cell wet weight of an individual monolayer grown on a 4.2-cm² insert was found previously to be 65 mg (Schmiedlin-Ren et al., 1997). Assuming a density of 1.0, intracellular MDZ concentrations were calculated as the ratio of the total amount of MDZ (pmol) measured in the Caco-2 monolayer over a volume of 0.065 ml. Application of a single-enzyme, Michaelis-Menten equation to the resulting apical concentration versus metabolite formation rate data and the predicted cell concentration versus metabolite formation rate data yielded apparent K_m and V_max values for the HSA-treated monolayer. Michaelis-Menten parameters were estimated using WinNonlin version 1.0 with a constant coefficient of variance error model.

Statistics. All statistical analyses were performed with the statistical software SPSS version 7.5. For Table 1, ANOVA was used to determine whether there was a significant difference in the mean ERs and the mean rates of 1'-OH MDZ formation between the three treatment groups (control, FBS-treated, and HSA-treated). ANOVA was also used for data in Table 2 to test for a dose-dependent difference in the apical/basolateral (A/B) concentration ratio of the MDZ metabolites, the 1'-OH MDZ/4-OH MDZ product ratio, and the calculated first-pass extraction ratio. When a significant difference was found (p < .05), multiple comparisons were performed with a Bonferroni adjustment (/3 = 0.017) to determine which groups were statistically different.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

TEER. TEER (ohm · cm²) was measured on all cultures at 2 weeks after confluence. The mean TEER was 574 ohm · cm² with a range between 378 and 760 ohm · cm². These values were consistent with TEER measurements reported previously by our laboratory (Fisher et al., 1999) and those of our collaborators (Schmiedlin-Ren et al., 1997) for 1,25-(OH)₂-D₃-treated Caco-2 monolayers.

MDZ Free Fraction. The free fraction of MDZ in DM containing either 5% FBS or 4 g/dl HSA was found to be 30.8 and 3.3%, respectively. The MDZ free fraction obtained for 5% FBS was 3-fold higher than our previous measurement of 10.2% (Schmiedlin-Ren et al., 1997), possibly due to the lot-to-lot variability in MDZ-binding FBS proteins.

Effect of Basolateral HSA on MDZ Distribution Kinetics After Apical Administration. When Caco-2 monolayers were treated apically with 3 µM MDZ, the flux of MDZ across the monolayer (AB) was 2.6-fold faster during the first 20 min of incubation when basolateral HSA was present, compared with control (Fig. 1A). Uptake of MDZ into the cell monolayer was rapid under both control and treated conditions (Fig. 1B), but its disappearance (0-60 min) from the apical compartment was ~2-fold faster when HSA was present in the basolateral compartment than when it was absent (Fig. 1A). As a consequence of these changes, the transcellular MDZ concentration equilibrium at 120 min of incubation was shifted dramatically by HSA, such that the A/B concentration ratio was decreased from 1.69 (control) to 0.23 (HSA).

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Time course of MDZ distribution in Caco-2 monolayers dosed with 3 µM apically applied MDZ in the presence (closed symbols) or absence (open symbols) of basolateral HSA. Apical (squares) and basolateral (triangles) MDZ levels (A) and intracellular MDZ levels (circles) (B) are shown. Data for each time point represent results from a single culture.

Effect of Basolateral HSA on MDZ Distribution Kinetics After Basolateral Administration. The effect of HSA on MDZ uptake from the basolateral compartment was also monitored over a 2-h incubation period. Compared with control cultures, the presence of basolateral HSA greatly slowed MDZ diffusion from the basolateral to apical (BA) compartment (Fig. 2A). In fact, the transcellular flux (0-30 min) was reduced by 96%. The rate of loss of MDZ from the basolateral compartment in treated monolayers was negligible. In close agreement with these observations, the peak intracellular MDZ level was ~14-fold lower in HSA-treated monolayers compared with control, and this differential remained largely unchanged over the 2-h incubation period (Fig. 2B).

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Time course of MDZ distribution in Caco-2 monolayers dosed with 3 µM basolaterally applied MDZ in the presence (closed symbols) or absence (open circles) of basolateral HSA. Apical (squares) and basolateral (triangles) MDZ levels (A) and intracellular MDZ levels (circles) (B) are shown. Data for each time point represent results from a single culture.

Effect of Basolateral HSA on 1'-OH MDZ Formation and Extracellular Distribution. The formation and distribution of 1'-OH MDZ was assessed in Caco-2 monolayers treated apically or basolaterally with 3 µM MDZ. As seen in Fig. 3A, total 1'-OH MDZ formation was similar in control and HSA-treated monolayers for the first 30 min of incubation with apically applied MDZ, despite the ~35% reduction in peak intracellular MDZ accumulation. Beyond 30 min, the rate of 1'-OH MDZ formation in HSA-treated cells appeared to slow, and after 2 h, total 1'-OH MDZ formed was ~25% lower than control.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Time course of total 1'-OH MDZ formation in Caco-2 monolayers in the presence (closed circles) or absence (open circles) of basolateral HSA. Apically applied 3 µM MDZ (A) and basolaterally applied 3 µM MDZ (B) are shown. Data for each time point represent results from a single culture.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Dependence of 1'-OH MDZ A/B concentration ratios on the presence (closed squares) or absence (open squares) of basolateral HSA. Data for each time point represent results from a single culture administered 3 µM MDZ into the apical compartment.

Effect of Basolateral Protein on MDZ First-Pass ER. The effect of HSA in the basolateral medium on first-pass MDZ extraction after apical administration was determined under conditions when <10% of the apical dose had reached the basolateral compartment. Data are summarized in Table 1. Extraction ratios calculated according to eq. 3 were found to be 13.3 ± 0.1% and 5.2 ± 1.0% for control and HSA-treated cultures, respectively. The reduction in first-pass ER by HSA was significant (p = .002). Parallel incubations were performed where 5% FBS was added to the basolateral culture medium. Although the percentage of the apical MDZ dose reaching the basolateral compartment exceeded 10% for a few of the FBS-treated cultures, the mean ER was found to be 11.5 ± 2.4%. There was no difference in ER between FBS-treated and control monolayers (p = .582), but a difference was found between the ER in FBS-treated and HSA-treated monolayers (p = .006). The total amount of MDZ that had accumulated in the cell monolayer was significantly reduced, in comparison to control, by both FBS addition to the basolateral medium (p = .026) and HSA addition (p < .001); 389 ± 8.5 pmol, 330 ± 9.5 pmol, and 225 ± 30 pmol for control, FBS-treated, and HSA-treated monolayers, respectively. However, there was no significant difference in the mean absolute rates of 1'-OH MDZ formation between any of the three groups (p = .063).

Effect of Basolateral Protein on the Saturation of First-Pass MDZ Metabolism. The amount of MDZ found within the Caco-2 monolayer after 10 min of incubation was directly proportional to the initial apical MDZ concentration (Fig. 5A). However, at each concentration (3-100 µM), the accumulation of MDZ within the monolayer was substantially reduced by the presence of HSA, in comparison to the accumulation of MDZ in monolayers that were incubated without basolateral HSA (Fisher et al., 1999). In contrast to the concentration-proportional cell uptake of MDZ, a plot of the 1'-OH MDZ and 4-OH MDZ formation rates versus the initial apical MDZ dosing concentration was nonlinear (Fig. 5B). Application of a single-enzyme, Michaelis-Menten model to the concentration-velocity data yielded V_max and K_m,app values of 22.6 pmol/min/culture and 33.8 µM for 1'-OH MDZ formation. Saturation of the 4-hydroxylation pathway was minimal and did not permit an accurate determination of Michaelis-Menten parameters (K_m,app presumed to be greater than 100 µM). However, the mean 1'-OH/4-OH MDZ product ratio decreased with increasing MDZ dose, from 8.81 ± 0.33 to 3.35 ± 0.25, p < .001 (Table 2). At a MDZ dose concentration of 3 µM, formation of 4-OH MDZ represented 10.2% of the combined 1'-OH MDZ + 4-OH MDZ produced. At 100 µM MDZ, a much greater fraction of MDZ metabolism (24%) proceeded through the 4-hydroxylation pathway. These data are consistent with the published difference in the K_m for 1'- and 4-hydroxylation of MDZ with purified or cDNA expressed CYP3A4 incubations (Gorski et al., 1994; Ghosal et al., 1996).

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Concentration-dependent 1'-OH MDZ and 4-OH MDZ formation kinetics in the presence of basolateral HSA. Increasing concentrations of MDZ (3 to 100 µM) were applied to the apical compartment. Total cellular MDZ as a function of initial apical MDZ concentration (A) and 1'-OH MDZ (open squares) and 4-OH MDZ (closed squares) formation as a function of initial apical MDZ concentration (B) are shown. Data for each time point represent the mean and S.D. for three replicate cultures. Error bars for some of the rates were too small to be seen on the plots. The dotted lines represent the fit of a single-enzyme, Michaelis-Menten model to the data: Km,app = 33.8 µM, Vmax = 22.6 pmol/min/culture (1'-OH MDZ).

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

It is generally accepted that mesenteric blood flow promotes the absorption of drug from the intestinal lumen by maintaining a high concentration gradient across the mucosal epithelium. According to current theory, blood flow will have an effect on the organ extraction efficiency in opposition to the intrinsic metabolic clearance (eq. 1). Effectively, intracellular to vascular diffusion and metabolic elimination compete for drug that has crossed the apical plasma membrane. It has been suggested but not proven that, in comparison to a basal condition where the plasma free fraction is unity, extensive binding of drug to plasma proteins will facilitate absorption and reduce the first-pass metabolic extraction ratio (Pond and Tozer, 1984; Mistry and Houston, 1987). Furthermore, plasma protein binding will have a quantitatively identical effect on the intestinal mucosal metabolic extraction of a drug dosed orally or i.v. This hypothesis was tested by incubation of the CYP3A-probe substrate MDZ with 1,25-(OH)₂-D₃ modified Caco-2 cell monolayers. The major findings from this work suggest that, in vivo, binding of a drug to plasma proteins will impact both first-pass and systemic MDZ intestinal extraction efficiency, but that the effect will be more pronounced for a drug that is delivered to mucosal enterocytes by way of arterial blood, compared with oral drug delivery.

The inclusion of a physiologically relevant concentration of human serum albumin (4 g/dl; Wilkinson, 1983) into the basolateral (vascular) medium of the Caco-2 monolayer model reduced the apical to basolateral (AB) first-pass MDZ extraction ratio. However, the effect, a 61% decrease, was less than might be expected by a 97% reduction in the basolateral unbound MDZ concentration. The substantial binding of MDZ to HSA enhanced the initial AB MDZ flux and decreased intracellular MDZ levels. However, despite a consistent 40 to 50% reduction in the total intracellular MDZ level during the first 30 min of incubation (Fig. 1B and Table 1), total 1'-OH MDZ formation during this same time interval was only modestly affected. There was no change or a 20% reduction in total 1'-OH MDZ formation for the respective 0 to 30-min and 0 to 120-min periods of incubation (Fig. 3A) and a 21% reduction for the replicate 0 to 30-min incubation data presented in Table 1. Overall, the protein binding sink in the basolateral compartment caused MDZ to pass through the cell more rapidly, thereby reducing the first-pass extraction efficiency.

We had previously considered the possibility that the extensive binding of MDZ to plasma proteins would not be an important determinant in the first-pass extraction of MDZ after oral administration (Thummel et al., 1997). By ignoring the parameter (excluding f_u from a modified well-stirred model for gut wall extraction), it was possible to obtain reasonable agreement between the observed intestinal first-pass extraction ratio (Paine et al., 1996) and that predicted from in vitro kinetic data (Paine et al., 1997). Results from the present series of experiments suggest that the plasma free fraction is a relevant factor, but that the well-stirred model of mucosal metabolic extraction may be overly simplified. Perhaps a model that incorporates a diffusional clearance (as suggested by Gwilt et al. 1988), where the diffusional clearance is some function of blood flow, the plasma free fraction as well as other terms, would be more appropriate.

It is also generally believed that with respect to the systemic circulation only unbound drug can pass biological membranes and be subject to metabolism. In our experiments where MDZ was administered into the basolateral compartment with HSA, we observed a dramatic decrease in both intracellular and apical MDZ accumulation, effects attributed to a decrease in unbound drug concentration in the basolateral compartment (f_u = 3%). The decrease (HSA-treated compared with control) in the intracellular MDZ area under the amount versus time curve (93%), the apical MDZ concentration versus time curve (96%), and total 1'-OH MDZ formation after 120 min of incubation (86%) were all roughly proportional to the decrease in basolateral MDZ free fraction (97%). Unfortunately, the static Caco-2 monolayer model described here was not suitable for the calculation of a MDZ extraction ratio analogous to the extraction of drug from mucosal arterial blood. Continuous medium flow through the basolateral chamber and sampling from both the in-flow and out-flow under steady-state conditions would be needed for an estimation of this ER. Despite the limitation of the present approach, the data presented indicate that the effect of basolateral HSA on MDZ BA flux, intracellular MDZ accumulation, and total 1'-OH MDZ formation after basolateral MDZ administration was much greater than its effect on the same parameters associated with apical MDZ administration. Thus, one could expect that the effect of basolateral HSA on the in vivo first-pass (luminalvascular) intestinal extraction ratio would also be quantitatively different from the systemic (arterialvenous) intestinal extraction ratio.

The intracellular binding and k_cat constants for CYP3A4-catalyzed MDZ hydroxylation would not be expected to change by the presence of HSA in the extracellular, basolateral compartment. However, our K_m estimates for 1'-hydroxylation based on initial apical MDZ dosing concentrations in the presence of HSA were found to be 3-fold higher than previous estimates obtained in the absence of basolateral HSA (K_m = 33.8 versus 9.1 µM, respectively; Fisher et al. 1999). This discrepancy may reflect a difference between the ratio of the unbound apical MDZ concentration to the unbound MDZ concentration at the enzyme active site for the two different incubation conditions. A more accurate predictor of the active-site concentration would be the unbound intracellular concentration. Although we had no way of determining this concentration, we could apply the total intracellular concentration to the Michaelis-Menten model. For Caco-2 monolayers incubated in the absence or presence of basolateral HSA, nonlinear regression of the 1'-hydroxylation data using estimated intracellular MDZ concentrations yielded K_m,app values that were relatively similar [12.7 and 14.8 µM for ()HSA and (+)HSA, respectively]. This analysis suggests that the unbound MDZ concentration at the enzyme active site was reduced by HSA, and that in comparison to ()HSA control, a higher MDZ dose was needed to achieve a comparable reaction velocity.

It is unclear whether MDZ first-pass metabolism is saturable in vivo. An oral dose of MDZ can be as high as 10 mg (~31 µmol); thus, luminal concentrations could exceed the K_m,app, depending on the rate of dissolution, luminal fluid volume, and site of absorption. In a previous study of intestinal MDZ metabolism in healthy volunteers, a 2-mg oral dose solution (~6 µmol) was administered with 50 ml of apple juice (Thummel et al., 1996), producing a nominal luminal concentration of 120 µM MDZ. This would be expected to undergo further dilution with gastric and intestinal fluids (~100 ml in a fasting state), yielding an initial luminal concentration of 40 µM. Thus, according to our in vitro estimates with Caco-2 monolayers (K_m,app ~34 µM), some saturation of first-pass intestinal metabolism of MDZ might occur during the initial absorption phase when a 2-mg oral MDZ dose is administered under the described conditions. However, it is unlikely to be of much overall significance since luminal concentrations will decline as the absorption process proceeds.

The effect of basolateral HSA on the distribution of 1'-OH MDZ after its intracellular formation was similar after an apical or basolateral MDZ dose. In the absence of basolateral serum, 1'-OH MDZ preferentially sorted to the apical compartment over the entire incubation interval. As mentioned previously (Fisher et al. 1998), it is unclear whether this sorting was due to an apically directed efflux pump that would maintain the concentration gradient even at equilibrium (A/B ratio, ~1.5-2.0) or whether it was due to differences in the diffusional flux of 1'-OH MDZ from the site of formation to the apical or basolateral compartments. Inclusion of HSA into the basolateral medium altered the 1'-OH MDZ A/B concentration ratio, presumably by decreasing the unbound concentration of 1'-OH MDZ in the basolateral compartment and increasing the intracellular to basolateral metabolite flux. Thus, inclusion of HSA changed the preferred sorting mechanism and established a new pseudoequilibrium where the A/B metabolite ratio was approximately one-third of that seen in the absence of basolateral protein. This finding is fully consistent with in vivo observations that show that the majority of an oral MDZ dose (~72%) can be recovered in urine as 1'-OH MDZ or its conjugate (Smith et al., 1981; Thummel et al., 1996) rather than as fecal metabolite.

In conclusion, the modified Caco-2 monolayer provides a unique in vitro model for identifying biochemical and physiological factors that influence CYP3A-mediated first-pass and systemic drug metabolism by the intestinal epithelium. Our results suggest that, in vivo, the binding of MDZ to plasma proteins found on the vascular side of the intestinal epithelium influences both systemic and first-pass metabolic MDZ extraction, but that the effects are quantitatively different. Additional substrates with a range of pharmacokinetic characteristics, including unbound intrinsic clearance, plasma protein binding, and susceptibility to active transport processes, must be studied before a useful paradigm for the prediction of gut wall drug disposition from in vitro data can be established.