Introduction

Abstract Introduction Methods Results Discussion References

Human cytochrome P450 3A isoforms (CYP3A) metabolize a number of widely prescribed, structurally diverse drugs that belong to a variety of therapeutic classes (Guengerich, 1995; Wrighton and Stevens, 1992). CYP3A4 and CYP3A5 are the major isoforms expressed in adults. Although found in many tissues throughout the body, their relative levels of expression are greatest in the liver and villus epithelium of the small intestine. On average, CYP3A composes 25% to 30% of total hepatic cytochromes P450 (Shimada et al., 1994; Wrighton and Stevens, 1992) and an even larger percentage of total small intestinal cytochromes P450 (De Waziers et al., 1990; Watkins et al., 1987). In addition, CYP3A content in both organs is highly variable among individuals. Lown et al. (1994) found CYP3A4 protein to vary >11-fold in S9 fractions prepared from 20 human duodenal biopsies, and Shimada et al. (1994) found a similar variability in liver microsomes prepared from 60 human samples.

Due to the anatomic arrangement of the small intestine and liver, drugs may encounter sequential, CYP3A-mediated first-pass metabolism when taken orally (Thummel et al., 1997). Historically, the liver was considered the major site of CYP3A-dependent first-pass metabolic extraction. Recent in vitro and in vivo studies, however, suggest that mucosal villi of the small intestine can be of equal or greater importance for some drugs such as cyclosporine (Gomez et al., 1995; Hebert et al., 1992; Kolars et al., 1991; Webber et al., 1992) and verapamil (Fromm et al., 1996). We also conducted a series of human studies that indicate extensive intestinal first-pass metabolism of the sedative/hypnotic agent MDZ. MDZ is eliminated entirely (>97%) by oxidative biotransformation reactions catalyzed by the CYP3A subfamily (Fabre et al., 1988; Gorski et al., 1994; Kronbach et al., 1989). A pharmacokinetic analysis of intravenous and oral MDZ disposition in healthy volunteers (Thummel et al., 1996) suggested that the low oral bioavailability observed for this drug (mean, 30 ± 10%) was the result of comparable extraction ratios for the liver (mean, 44 ± 14%) and intestine (mean, 43 ± 24%). Direct measurements of first-pass MDZ extraction during the anhepatic phase of liver transplant operations confirmed an identical mean gut wall extraction fraction (43 ± 18%) (Paine et al., 1996). Both in vivo MDZ studies revealed large interindividual differences in hepatic and intestinal extraction ratios. Hepatic extraction ranged from 22% to 76%, whereas intestinal extraction ranged from ~0% to 77% (healthy volunteer study) and from 14% to 59% (anhepatic study). These interindividual variations are consistent with variably expressed CYP3A content and associated in vitro MDZ 1-hydroxylation activity measured in human liver microsomes (Kronbach et al., 1989) and in S9 fractions prepared from human duodenal biopsies (Lown et al., 1994).

The intravenous formulation of MDZ (Versed), administered orally or directly into the duodenum, is absorbed rapidly from the most proximal region of the small intestine. Other CYP3A substrates that have slow dissolution or administered in sustained-release formulations are absorbed throughout the small intestinal tract. Because mucosal CYP3A4 content is reportedly lower in jejunum and ileum compared with duodenum (De Waziers et al., 1990), first-pass intestinal metabolic extraction may depend on the absorption characteristics of the drug formulation; that is, first-pass intestinal metabolism may be reduced when drug is absorbed at more distal sites of the small intestine. However, very little is known about the extent of interindividual variability in CYP3A expression in epithelium distal to the duodenum as well as the relative catalytic capacities of duodenal, jejunal and ileal CYP3A.

We characterized CYP3A protein content and catalytic activity, as well as two CYP coenzymes (NADPH-dependent cytochrome P450 reductase and cytochrome b₅), along the entire length of six different human donor small intestines. In addition, we determined the relative metabolic capacity (MDZ intrinsic clearance) of the three regions (duodenum, jejunum and ileum) in 15 donor small intestines. For comparison, parallel analyses were performed with 20 human livers. Based on these in vitro results, we estimated whole-organ intrinsic clearances using a conventional flow model (Pond and Tozer, 1984; Wilkinson, 1987) and considered their correlation to gut and liver extraction of MDZ in vivo.

	Methods

Abstract Introduction Methods Results Discussion References

Chemicals. MDZ, 1-OH MDZ and 4-OH MDZ were kindly provided by Drs. William Garland and Bruce Mico (Roche Laboratories, Nutley, NJ). Cytochrome c (horse heart type VI), NADPH (reduced form, tetrasodium salt), NADH, EDTA, PMSF and anti-rabbit IgG alkaline phosphatase conjugate were purchased from Sigma Chemical (St. Louis, MO). N-Methyl-N-(t-butyl-dimethylsilyl)trifluoroacetamide was purchased from Pierce Chemical (Rockford, IL). Acetonitrile and ethyl acetate were purchased from Fisher Scientific (Santa Clara, CA). SDS-PAGE reagents (SDS, acrylamide, ammonium persulfate, TEMED) were purchased from BioRad (Hercules, CA). Nitrocellulose was purchased from Schleicher & Schuell (Keene, NH). BCIP-NBT was purchased from Kirkegaard and Perry (Gaithersburg, MD). All other chemicals were of reagent grade or better.

Organ procurement. The collection and use of human donor tissue for research were approved by the University of Washington Human Subjects Review Board. During procurement, the liver and small intestine were dissected in preparation for removal and then perfused in situ with cold (4°C) University of Wisconsin solution (Viaspan). The organs were kept cold with topical iced saline until removal from the abdominal cavity. After retrieval and transport of each intestine (in "iced saline") to our laboratory, the organ was cut into 1-foot sections, flushed with ~60 ml cold saline and weighed. Sections of mucosa were exposed by a longitudinal cut and carefully scraped free from the remainder of the gut tissue using a glass slide. Mucosal scrapings from each section were then placed separately into 45 ml conical polypropylene tubes, weighed, immediately frozen in liquid nitrogen and stored at 70°C. All processing of tissue was performed on a bed of ice. Full-length small intestines from a total of 20 donors were used for study. The time elapsed between vascular cross-clamp (start of cold ischemia) and the freezing of all mucosal scrapings was usually <3 hr. Two intestines were obtained from out-of-state donors (Alaska and Oregon). Transit and processing time of these organs were <5 hr. Procured livers were kept on ice for a more variable period of time, 12 to 24 hr, before processing. After transfer to our laboratory, the organs were cut into ~10 g pieces and snap-frozen in liquid nitrogen. Tissue from a total of 20 livers was used for study. Six exhibited fatty infiltration in >25% of parenchymal cells. Three livers were obtained from out-of-state donors (Idaho, Montana and Alaska), and eight livers were from the same donors as 8 of the 20 intestines. There were an equal number of male and female intestine donors, but 12 female and 8 male liver donors. The age range for both liver and intestine donors was 10 to 70 years, and nearly all were Caucasian.

Preparation of hepatic and intestinal microsomes. Hepatic microsomes were prepared from 20 livers as described previously (Thummel et al., 1993), except that 0.25 M sucrose containing 1 mM EDTA was used as the storage solution. Intestinal microsomes were prepared from every other 1-foot section of six whole small intestines and from a 1-foot section of each of the three regions (duodenum, jejunum and ileum) of an additional 14 whole small intestines as described previously (Thummel et al., 1996) with some modifications. Briefly, homogenizing buffer (10 mM potassium phosphate, pH 7.4, containing 0.25 M sucrose, 1 mM EDTA and 0.1 mM PMSF) was added directly to the frozen conical tubes containing the mucosal scrapings (5-fold v/v dilution). After the tissue was thawed, the mixture was transferred to and homogenized in a 55-ml glass Wheaton tube using a motor-driven Teflon-tipped pestle (8-10 strokes). The total volume of homogenate was recorded, and two 1-ml aliquots were saved. Homogenate was transferred to 30-ml centrifuge tubes, which were spun at 600 × g for 5 min and then at 11,000 × g for 15 min. The resulting supernatant was filtered through sterile gauze into 60-ml centrifuge tubes, which were spun at 110,000 × g for 70 min. After saving the supernatant for unrelated studies, the remaining pellet was resuspended in wash buffer (10 mM potassium phosphate, pH 7.4, containing 1 mM EDTA and 0.1 mM PMSF) and centrifuged again at 110,000 × g for 70 min. The washed pellet was resuspended in storage solution (0.25 M sucrose, pH 7.4, containing 1 mM EDTA and 0.1 mM PMSF) to a final protein concentration of 10 to 30 mg/ml. Aliquots of the final microsomal suspension (~0.25 ml) were stored at 70°C. Protein concentrations were determined according to the method of Lowry et al. (1951) using bovine serum albumin as the reference standard.

Spectrophotometric assays. Total cytochrome P450 and cytochrome b₅ concentrations in hepatic and intestinal microsomes were measured according to the method of Omura and Sato (1964) using extinction coefficients of 91 and 185 mM¹ cm¹, respectively. Microsomal cytochrome P450 reductase and cytochrome b₅ reductase activities were measured by determining the rates of NADPH- and NADH-dependent cytochrome c reduction, respectively, as described previously (Kurzban and Strobel, 1986). Briefly, 100 µg of microsomal protein, cytochrome c and either NADPH or NADH were mixed with potassium phosphate buffer (0.3 M, pH 7.7) to a final 1-ml volume in a cuvette. The final concentrations of cytochrome c, NADPH and NADH were 40 µM, 1 mM and 0.64 mM, respectively. Immediately after the addition of NADPH or NADH, the change in absorbance (550-538 nm) was measured every 3 sec over 60 sec at 25°C. The rates of cytochrome c reduction were calculated using an extinction coefficient of 21 mM¹ cm¹ (Van Gelder and Slater, 1962).

Western blot analysis of CYP3A4 and CYP3A5. Both homogenate and microsomes were analyzed for these two proteins. Intestinal microsomes, intestinal homogenate and hepatic microsomes were diluted in sample buffer (60 mM Tris · HCl, 25% glycerol, 0.2% Emulgen 911, pH 7.4) to final concentrations of 20 to 50, 50 and 5 µg/20 µl, respectively. After boiling each sample for 2 min, 20 µl was loaded onto 0.1% SDS-9% acrylamide gels, and the proteins were separated as described previously (Favreau et al., 1987). Purified CYP3A4 standards (Kharasch and Thummel, 1993) were also prepared and run in parallel with the microsomal or homogenate samples. At the end of the run (~3.5 hr), the proteins were electrophoretically transferred to nitrocellulose; the sheet was rinsed twice with PBS and stored in blocking buffer (2% nonfat dry milk and 2% Triton X-100 in PBS) overnight. The following morning, the nitrocellulose sheets were incubated with a selective rabbit polyclonal anti-CYP3A4 IgG (Kharasch and Thummel, 1993) at a final concentration of 1 µg/ml in blocking buffer for 2 hr at room temperature. The sheets were washed twice with blocking buffer and then incubated with the secondary antibody, anti-rabbit IgG alkaline phosphatase conjugate (1:1000), for 2 hr. The sheets were washed twice with blocking buffer, twice with PBS and twice with 0.1 M Tris buffer (pH 7.4). The proteins of interest were visualized with the addition of BCIP-NBT according to the manufacturer's instructions. A protein band that comigrated with the CYP3A4 standard was detected in all samples. A second, slightly higher-molecular-weight protein band was also detected in 20% of livers and intestines; this was presumed to be CYP3A5. All blots were scanned (Howtek Scanmaster 3+) and quantified for CYP3A4 and CYP3A5 by densitometry using the software programs Visage and Whole Band Analysis (v4.6M and v2.4, respectively; Millipore BioImage Products, Ann Arbor, MI). Purified CYP3A4 was used as the reference standard.

Hepatic and intestinal microsomal incubations. An internal standard mixture containing ¹⁵N₃-labeled MDZ metabolites (i.e., 1-OH MDZ and 4-OH MDZ) was prepared by incubating 6 nmol of cytochrome P450 (using HL-122 microsomes) with 100 µg of ¹⁵N₃-MDZ and 12 mg of NADPH (final concentration, ~1.5 mM) in potassium phosphate buffer (0.1 M, pH 7.4, in a final volume of 8 ml) at 37°C. After 10 min, the reaction was stopped by the addition of 8 ml of Na₂CO₃ (0.1 M, pH 12). The compounds were extracted twice with 20 ml of ethyl acetate, and the solvent was evaporated to dryness under a stream of nitrogen. The remaining solid was then reconstituted in 20 ml of methanol, split into two 10-ml aliquots and stored at 20°C.

Determination of a correction factor for homogenate protein. With Western blot analysis, we compared the slope obtained from a standard curve using purified CYP3A4 only (i.e., 0, 0.5, 1 and 2 pmol) with that using purified CYP3A4 spiked into 50 µg of jejunal homogenate protein (i.e., 50 µg of protein plus 0.5, 1 or 2 pmol). The slope (change in IOD/change in pmol CYP3A4 added) from the CYP3A4-only curve was higher than that obtained from the standard addition curve. Therefore, to correct for this "matrix effect," five additional jejunal homogenates were randomly chosen and subjected to Western blot analysis as described previously. Again, the slopes from the purified CYP3A4-only standard curves were consistently higher than those from the standard addition curves. The correction factor (slope without homogenate/slope with homogenate) for the six homogenates averaged 1.26 ± 0.29. The median was 1.23 and was used to correct for the median amount of homogenate CYP3A determined from earlier blots. Only corrected medians are reported. We did not find a noticeable matrix effect from microsomal protein and thus did not apply a correction factor to median amounts of microsomal CYP3A.

Calculation of total CYP3A recovery and organ intrinsic clearance. Based on a previous study of the physical characteristics (i.e., weight and length) of human small intestine (Snyder et al., 1975), we assigned the first 1-foot section as the duodenum, sections 2 to 9 as the jejunum and the remaining sections as the ileum. The total mucosal scrapings mass per gram of intestinal tissue was recorded for each section of seven organs, and regional (duodenal, jejunum, ileal) wet weights were determined accordingly. Total CYP3A in each of the three regions was then calculated using the following equation:

(1)

where mucosal CYP3A is homogenate protein (mg/g mucosa) × homogenate CYP3A (pmol/mg of protein). Total regional intrinsic clearance (Cl_in,region) was then calculated from Cl_in,mic, and regional CYP3A was calculated according to the following equation:

(2)

Because mucosal masses were not recorded for every region, from which microsomal CYP3A and intrinsic clearance were determined, median values from equation 1 and for microsomal intrinsic clearance were used for equation 2. From these calculations, we obtained a single estimate of 1-OH MDZ formation clearance for each of the three regions of human small intestine. For comparison, total intrinsic clearance for the liver was calculated using a liver weight of 1500 g (Snyder et al., 1975), the median hepatic Cl_in,mic, and assuming an average microsomal protein recovery of 52.5 mg/g liver (Iwatsubo et al., 1997). This recovery value was obtained from hepatocyte cultures, which we believe is more accurate than that obtained from recovered microsomal protein.

(3)

(4)

Statistical analysis. All statistical analyses were performed using Sigmastat for Windows (v1.0, Jandel Corp., San Rafael, CA). Because several data sets failed the homogeneity-of-variance and normality tests (Levene Median and Kolmogorov-Smirnov tests, respectively), nonparametric methods were used for the majority of analyses. Average values were reported for comparisons of 1-OH MDZ with 4-OH MDZ ratios because the sample sizes were <5. Spearman correlation coefficients (r_s) were considered significant if the P-value was <0.05. One-way ANOVA on ranks (Kruskal-Wallis) was used to determine whether a difference existed in the various kinetic parameter estimates and CYP3A contents among the liver, duodenum, jejunum and ileum (and ignoring that each set of three intestinal regions came from the same donor and that eight livers and intestines were matched; this resulted in a less powerful but more conservative test). For the three intestines from which we could not obtain reliable parameter estimates for one or two regions (due to low CYP3A activity), median regional K_m values were assumed along with V_max and Cl_in,mic values that were ranked lower than the corresponding regional minimum V_max and Cl_in,mic. Likewise, for intestinal regions for which the protein band was too faint to be detected by the densitometer, a ranking lower than the regional minimum was assumed. Further, if ANOVA revealed a significant difference (P < 0.05), then either the Student-Newman-Keuls test (for equal sample sizes) or Dunn's test (for unequal samples sizes) was used to determine which group medians differed from the others. To compare the three intestinal regions, taking into account that each set of three came from the same donor, three pairwise Wilcoxon signed-rank tests were used along with a Bonferroni-corrected level of significance, 0.017 (i.e., 0.05/3).

	Results

Abstract Introduction Methods Results Discussion References

Total CYP, cytochrome b₅ and CYP3A contents. Western blot analysis of microsomes from 20 livers and small intestines showed the presence of CYP3A4 protein in every organ and CYP3A5 protein in four livers and four intestines. A representative blot illustrating the continuous expression of CYP3A5 along the entire length of small intestine is shown in figure 1. Concordant hepatic-intestinal expression of CYP3A4 and CYP3A5 was observed in seven of the eight matched organs (table 1). Six of the seven concordant organs expressed CYP3A4 only, whereas one pair expressed both CYP3A4 and CYP3A5. For the single discordant pair (HL-147/HI-26), both proteins were detected in the liver, but only CYP3A4 was detected in the intestine. Also, although a modest CYP3A4 protein band was detected in the jejunum and ileum of HI-26, no band was detected (by the densitometer) for the duodenum. The CYP3A5-to-CYP3A4 IOD ratio in CYP3A5-positive organs was less than unity in all except one liver and two ileal sections of its intestinal mate. This interpretation may be biased because the detection antibody was raised against purified CYP3A4 protein.

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Western blot of microsomes prepared from HI-30 showing the presence of CYP3A4 and CYP3A5 protein along the entire length of small intestine. The nitrocellulose sheet was developed with a polyclonal anti-CYP3A4 IgG. Lanes 1 to 3, purified human CYP3A4 (0.5, 1 and 2 pmol, respectively); lane 4, duodenum; lanes 5 to 8, proximal to distal jejunum; lanes 9 to 12, proximal to distal ileum. Lanes 4 to 12 were loaded with 50 µg of protein.

MDZ 1-hydroxylation kinetics. Fifteen intestines were examined for regional (duodenal, jejunal and ileal) MDZ 1-hydroxylation kinetics. However, in some regions of three intestines, the amount of product formed was below the limit of quantification at low substrate concentrations and, in some cases, at all substrate concentrations. Thus, we could not obtain reliable kinetic parameter estimates in these cases. Representative Eadie-Hofstee plots of 1-OH MDZ formation kinetics for each of the three intestinal regions of one small intestine are shown in figure 2. All 20 liver microsomal samples produced amounts of product that were readily quantifiable (at all substrate concentrations), so we were able to obtain reliable kinetic parameter estimates. Box plots of K_m, V_max and Cl_in,mic for liver and small intestine are shown in figure 3. Median K_m values were 3.7, 3.8, 3.7 and 4.5 for liver, duodenum, jejunum and ileum, respectively. ANOVA revealed no differences in K_m values among the four tissues (P = .15). Moreover, no statistical difference was found if the liver data were excluded from the analysis (P > .05 for all pairwise comparisons). Median V_max values were 850, 644, 426 and 68 pmol/min/mg for liver, duodenum, jejunum and ileum, respectively. Not surprisingly, ANOVA revealed a statistically significant difference in V_max among the four tissues (P = .0001). With Dunn's test, only the ileum-liver difference was significant. However, Wilcoxon signed-rank tests revealed a significant difference between duodenum and ileum (P = .002) and jejunum and ileum (P = .005), but not between duodenum and jejunum (P = .14). Median intrinsic clearances (Cl_in,mic) for liver, duodenum, jejunum and ileum were 200, 157, 85 and 14 µl/min/mg. ANOVA with subsequent Dunn's test and Wilcoxon signed-rank tests revealed similar contrasts as those described for V_max.

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Microsomal MDZ 1-hydroxylation kinetics for HI-32. Eadie-Hofstee plots are shown for each region of small intestine. Symbols represent observed values (average of duplicate incubations), and solid lines represent regression lines.

View larger version (24K): [in this window] [in a new window]   Fig. 3.   Box plots of Km (top), Vmax (middle) and Clin,mic (bottom) in microsomes prepared from liver (L), duodenum (D), jejunum (J) and ileum (I). Each box represents the 25th to 75th percentile, and the central line represents the median. Upper and lower "whiskers" represent the largest and smallest nonoutlying values, respectively. , Outliers (values between 1.5 and 3 box-lengths from the upper or lower edge). , Extremes (values >3 box-lengths from the upper or lower edge). * Median significantly different from liver median (P < .05, Kruskal-Wallis). ** Median significantly different from duodenal and jejunal medians (P < .01, Wilcoxon signed-rank test). See text for explanation of statistical tests.

Distribution of intestinal CYP3A, cytochrome b₅ and CYP reductase activities. To further define the CYP3A gradient along the entire small intestine and determine the role of coenzymes in the regional variation of CYP3A activity, the following were measured along the entire length of six small intestines (HI-29, -30, -31, -32, -33 and -35): CYP3A catalytic activity (1-OH MDZ formation rate), CYP3A protein content, NADPH- and NADH-dependent cytochrome c reduction rates (as measures of CYP reductase and cytochrome b₅ reductase activity, respectively) and cytochrome b₅ protein content. For four of six intestines (HI-30, -32, -33 and -35), CYP3A catalytic activity increased from duodenum to middle jejunum and then decreased to distal ileum (fig. 4). For HI-31, activity was highest in the first 1-foot section (duodenum) and declined thereafter. For HI-29, activity was low for the first 5 to 6 feet of bowel, increased dramatically in midjejunum and declined thereafter. Fold-differences between the section with the highest activity vs. that with the lowest activity (distal ileum in all cases) were 3-fold (HI-31 and HI-35) to 14-fold (HI-29, excluding duodenum; HI-32) to 25-fold (HI-30 and HI-33). Intraintestinal variability in CYP3A protein content was much less than that for CYP3A activity and ranged from 1.5-fold (HI-31) to 4.6-fold (HI-30). Correlations between CYP3A activity and protein content within an intestine were significant for only four of the six intestines (HI-30, -32, -33 and -35; r_s = .83, .98, .71 and .90, respectively).

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Longitudinal MDZ 1-hydroxylation rates (at 8 µM MDZ) for six different donor intestines. Foot sections 1 to 15 represent duodenal to ileal ends. For simplicity, only the first eight odd sections are shown. For those intestines that had more than eight odd sections (30, 32 and 35), the ensuing distal value was less than or very near the preceding value.

View larger version (42K): [in this window] [in a new window]   Fig. 5.   Median values, by section, for microsomal CYP3A activity, CYP3A protein content, CYP reductase activity (top) and cytochrome b5 and cytochrome b5 reductase activity (bottom) for six different small intestines. Foot section numbers represent the same as described for figure 4. All measurements within each donor intestine were normalized to the peak value for that donor.

1-OH MDZ to 4-OH MDZ ratios. MDZ metabolite ratios (1-OH MDZ/4-OH MDZ) were measured in microsomes prepared from six jejunums and six livers. Three livers and three intestines with the highest CYP3A5/CYP3A4 IOD ratios (HI-30, -31 and -35 and HL-125, -127 and -150) were selected and compared with randomly chosen CYP3A5-negative organs (HI-19, -20 and -28 and HL-107, -122 and -148). Results are summarized in table 2. Ratios for CYP3A5-positive jejunums were higher than CYP3A5-negative jejunums at all four substrate concentrations examined. Interestingly, the jejunum with the highest CYP3A5/CYP3A4 IOD ratio (0.64 for HI-30) had the highest product ratio at all substrate concentrations. Product ratios for CYP3A5-positive livers were also higher than CYP3A5-negative livers at all four substrate concentrations examined (except for HL-125 at 0.25 µM MDZ). Mean product ratios for CYP3A5-negative livers were similar to the corresponding ratios for CYP3A5-negative jejuna. In contrast, the mean product ratios for CYP3A5-positive livers were greater than the corresponding ratios for CYP3A5-positive jejuna.

Total regional intestinal intrinsic clearances and scaling factors. Total intrinsic clearances for duodenum, jejunum and ileum (Cl_in,region) were estimated using microsomal intrinsic clearance (Cl_in,mic), microsomal CYP3A content and total regional CYP3A data. Total regional CYP3A was calculated using total wet weights of each region, mucosal recovery (as percent of tissue wet weight) and mucosal CYP3A content (based on corrected homogenate CYP3A content and homogenate mass/g mucosa). Because we had incomplete data collection, median values for microsomal intrinsic clearance (n = 15), microsomal CYP3A content (n = 20), regional wet weight (n = 7), mucosal recovery (n = 7) and mucosal CYP3A content (n = 10) were used for our calculations. Median regional wet weights were 79, 411 and 319 g for duodenum, jejunum and ileum, respectively. Mucosal masses represented 23%, 16% and 12% of total wet weights. Total median CYP3A contents were 445, 463 and 391 pmol/g mucosa, or 102, 74 and 47 pmol/g wet weight. Total regional CYP3A was thus 9.7, 38.4 and 22.4 nmol for duodenum, jejunum and ileum, respectively (fig. 6). Median Cl_in,mic values were 157, 85 and 14 µl/min/mg. Therefore, from equation 2, Cl_in,region values were calculated to be 50.1, 144.3 and 19.3 ml/min, respectively. The sum of these three values, total gut intrinsic clearance, was 213.7 ml/min. For comparison, we calculated total hepatic CYP3A to be 5490 nmol (69.7 pmol CYP3A/mg × 52.5 mg/g × 1500 g × 1 nmol/1000 pmol) and an intrinsic clearance of 15.8 l/min (0.2 ml/min/mg × 52.5 mg/g × 1500 g × 1 liter/1000 ml).

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Estimated amounts of total CYP3A and total intrinsic clearance for duodenum, jejunum and ileum. For comparison, estimated total hepatic CYP3A and intrinsic clearance were 5490 nmol and 15.8 l/min, respectively.

	Discussion

Abstract Introduction Methods Results Discussion References

Results from this study provide the first comprehensive characterization of intraintestinal and interintestinal variability in CYP3A expression and metabolic function. Although previous studies involved either a limited sample size (De Waziers et al., 1990; McKinnon et al., 1995; Peters and Kremers, 1989; Prueksaritanont et al., 1996; Thummel et al., 1996) or collection of duodenal tissue only (Kivisto et al., 1996; Lown et al., 1994), we examined 20 full-length intestines. Nearly all were obtained from donors who had been hospitalized for no more than 4 days before procurement (one donor, HI-33, was hospitalized for 12 days). The organs were also procured under carefully controlled conditions, with a limited period of cold ischemia time and no warm ischemia time. Thus, although not identical to biopsy tissue from healthy volunteers, we believe that potential differences between donor and biopsy tissue with respect to mucosal CYP3A stability (protein content and catalytic activity) were kept to a minimum.

Our findings that CYP3A is the major cytochrome P450 expressed in all three regions of the human small intestine fully agree with results reported by Watkins et al. (1987) and De Waziers et al. (1990). In contrast, CYP3A represented a smaller percentage of total hepatic CYP compared with the mean percentage reported by Shimada et al. (1994) (17% vs. 29%). For a fairer comparison, we calculated a mean percentage of 20%, which is still lower than the estimate of previous investigators. The discrepancy is due to a slightly higher mean total CYP content and a lower mean CYP3A content observed in the present study compared with those from the earlier study (0.38 vs. 0.34 nmol/mg and 83 vs. 96 pmol/mg, respectively). Nevertheless, small intestinal CYP3A represents a larger percentage of total small intestinal CYP compared with hepatic CYP3A as a percentage of total hepatic CYP.

We found large interindividual variability in CYP3A content for all three regions of the small intestine, supporting the findings reported by Lown et al. (1994), who measured CYP3A content in 20 duodenal pinch biopsies. Moreover, we corroborated the results reported by De Waziers et al. (1990), who observed a progressive decline in microsomal CYP3A content from duodenum to jejunum to ileum. The regional differences we observed, however, were not as great. The previous investigators relied on small and uneven sample sizes (n = 2, 6 and 5 for duodenum, jejunum and ileum, respectively); thus, larger interregional differences and a lower interindividual variability were observed. In contrast, the ranges of CYP3A content we observed for different donors were considerable (3-90, 2-98 and 2-38 pmol/mg for duodenum, jejunum, and ileum, respectively) and are likely to be even larger because we were not able to accurately quantify some very faint protein bands. The source of this interindividual variability is not completely known. It could not be explained by length of hospitalization, prior drug therapy, age or sex of the donor population. Therefore, based on the similar degree of variability in the present study, in which the intestinal mucosae had been largely unexposed to potential dietary modulators for at least 48 hr before procurement, and that reported for duodenal biopsies from healthy volunteers (Lown et al., 1994), we speculate that these interindividual differences are derived largely from homeostatic mechanisms (e.g., enterocyte differentiation, hormonal control of gene expression and enterocyte turnover).

Rates of intestinal NADPH-dependent cytochrome c reduction (range, 12-150 nmol/min/mg across all sections of the six intestines studied) agreed with those reported for 54 liver microsomal preparations (Schmucker et al., 1990; range, 38-135 nmol/min/mg) and six ileal microsomal preparations (Pacifici et al., 1989; range, 32-106 nmol/min/mg). Interestingly, HI-33 had the lowest CYP reductase activity, <30 nmol/min/mg for all sections, which may explain why this intestine had low CYP3A activity despite moderate CYP3A content. Furthermore, the three most proximal and one most distal section of HI-29 and the four ileal sections of HI-30 also had low CYP reductase activity (<30 nmol/min/mg). Again, this may explain why we were not able to obtain kinetic parameter estimates from some regions of these intestines.

With respect to intraintestinal comparisons, a significant correlation between CYP3A content and catalytic activity was observed for four of the six full-length organs examined, indicating that for the majority of individuals, enzyme function parallels protein content. Low intraintestinal variability in CYP3A content and activity (HI-31) or low CYP reductase activity (HI-29) may explain the poor correlations observed for the remaining two intestines. Furthermore, correlations between duodenal and jejunal V_max (r_s = .81) or Cl_in,mic (r_s = .76) for the larger set of 15 donors were highly significant (P < 0.0001), suggesting that for most subjects, an analysis of duodenal pinch biopsies will provide representative metabolic information for the entire proximal intestine.

1-OH MDZ was the dominant metabolite in all intestine and liver microsomal preparations and at all MDZ concentrations examined (0.25-8.0 µM). This is in agreement with previous reports of high 1-/4-OH MDZ ratios for hepatic microsomes when substrate concentrations were <10 µM (Gorski et al., 1994; Kronbach et al., 1989). Collectively, these in vitro findings are consistent with in vivo observations (Heizmann and Ziegler, 1981) that the 4-hydroxylation pathway contributes little to the systemic metabolism of MDZ in vivo because plasma MDZ concentrations are typically well below 1 µM.

The similarity in K_m values for MDZ 1-hydroxylation (~4 µM) for the three intestinal regions and liver, together with strong correlations between V_max and CYP3A content in both organs (except for ileum), suggest that hepatic and proximal gut CYP3A are functionally equivalent. Although not statistically different, ileal K_m values tended to be higher than corresponding duodenal and jejunal values. This, along with a poor correlation between ileal V_max and CYP3A content, implies that ileal CYP3A behaves differently from more proximal CYP3A. Data from our longitudinal studies further support this belief, where median ileal CYP3A activity decreased to a greater extent than did CYP3A content (fig. 5). Median, normalized CYP3A activity decreased 42% from sections 11 to 13 and 47% from sections 13 to 15, whereas CYP3A content decreased 10% and 19%, respectively. Overall, the data indicate a much lower metabolic capacity for the distal compared with the proximal small intestine.

Although we observed a 1-in-5 frequency of CYP3A5 expression in both small intestine and liver, Lown et al. (1994) reported a frequency of 70% for small intestine, and Jounaïdi et al. (1996) reported a frequency of 74% for liver. This discrepancy is likely due to different protein detection methods (the other investigators used the more sensitive enhanced chemiluminescence technique plus prolonged exposure). Nevertheless, in agreement with these earlier studies, CYP3A5 was generally a minor component of total CYP3A, which intimates that for most individuals, CYP3A5 plays a minor role in the intestinal and hepatic first-pass extraction of CYP3A substrates in vivo.

Before our analysis of matched liver and small intestinal tissue, we expected the factors regulating the maintenance of CYP3A5 protein in a given individual to be operative in both organs. This did not always appear to be the case, however. The one discordant pair exhibited a clear immunoreactive CYP3A5 band for liver but not for intestine. Although the reason for this is unclear, it is possible that this intestine received an insult either before or during procurement, leading to enzyme degradation and a decrease in CYP3A5 below the limit of detection. Interestingly, for the two intestines in which the CYP3A5 band was quantifiable (HI-30 and HI-31), the CYP3A5-to-CYP3A4 ratio decreased from duodenum to jejunum and then increased in ileum to values comparable to or greater than those observed for the duodenum. This is consistent with previous reports of predominantly CYP3A5 expression in the stomach and colon (Gervot et al., 1996; Kolars et al., 1994; Peters and Kremers, 1989).

There was no clear trend between the eight matched livers and intestines with respect to MDZ 1-hydroxylation; that is, if the liver had a high Cl_in,mic, the intestine did not necessarily have a high Cl_in,mic. An extreme case is the pair HL-147 and HI-26, where the liver exhibited one of the highest hepatic V_max and Cl_in,mic values and the intestine had virtually no metabolic activity. Another case is the pair HL-148 and HI-27, where the liver had moderately high V_max and Cl_in,mic values and the intestine had low V_max and Cl_in,mic values. Some pairs did exhibit parallel CYP3A activities, such as HL-146/HI-24 and HL-150/HI-30. Collectively, these findings support in vivo observations reported by others (Lown et al., 1994), who found that hepatic and intestinal CYP3A (measured as the erythromycin breath test and 1-OH MDZ formation in duodenal biopsies, respectively) are not coordinately regulated.

We estimated the duodenum to contain almost half the amount of CYP3A compared with the ileum (9.7 vs. 22 nmol), which is striking given that the ileum is ~10 times longer than the duodenum (~10 feet vs. 1 foot; Snyder et al., 1975). The discrepancy lies in the duodenal mucosa being extremely rich in villi, the tips of which are lined with mature CYP3A-containing enterocytes. The jejunum, whose villus density gradually decreases from proximal to distal ends, is ~8 times longer than the duodenum and thus has the greatest total amount of CYP3A. Total intrinsic clearances followed a similar trend except that the duodenal value was higher than the ileal value (50 vs. 19 ml/min). Again, this is reflective of a metabolically active duodenum and a relatively deficient ileum.

Because the total hepatic intrinsic clearance was >70 times that for the small intestine (15.8 and 0.21 l/min for liver and intestine, respectively), it appears that the gut should contribute very little to the metabolism of MDZ in vivo. However, recent studies provide convincing evidence that both organs contribute equally, on average, to the first-pass metabolism of this drug (Paine et al., 1996; Thummel et al., 1996). Although unbound intrinsic clearance is an important determinant of first-pass extraction by both the liver and intestine, there is no a priori reason to believe that the relationships between the extraction ratio and the unbound intrinsic clearance, organ blood flow and plasma protein binding should be the same for all organs of elimination (Wilkinson, 1987). The effect of plasma protein binding and hepatic blood flow on hepatic extraction is fairly well defined, but their impact on gut extraction is unclear. Although exposure of absorbed drug molecules to enterocytic CYP3A might be obligatory for drugs absorbed transcellularly, drug access to hepatic CYP3A depends on the translocation of unbound drug molecules across the sinusoidal membrane into the parenchymal cell. In addition, although blood flow should govern the residence time of drug molecules in the sinusoid and in the intestinal villus, limiting their exposure to CYP3A, the ratio of blood flow to intrinsic clearance may be greater for liver than for the intestinal mucosa.

If we assume a liver plasma flow of 0.78 l/min (MDZ does not partition appreciably into erythrocytes), an unbound fraction of 0.02 (Thummel et al., 1996) and a total hepatic unbound intrinsic clearance of 15.8 l/min and apply these to the well-stirred model for hepatic clearance (Rowland et al., 1973), an in vivo 1-OH MDZ formation clearance of 0.23 l/min was predicted. By performing the same calculations for small intestine and assuming a total wet weight of 809 g, a mucosal plasma flow of 0.16 l/min (Hultén et al., 1977), the same unbound fraction and a total unbound intrinsic clearance of 0.21 l/min, an in vivo systemic clearance of 4.1 × 10³ l/min was predicted. This, together with the predicted hepatic 1-OH MDZ formation clearance (0.23 l/min) being in excellent agreement with the average and median systemic formation clearance we observed in healthy volunteers (0.26 and 0.24 l/min, respectively; Thummel et al., 1996), confirms our finding that the gut contributes little to the systemic clearance of MDZ (Paine et al., 1996).

In vivo organ extraction ratios based on systemic delivery of drug were predicted to be 0.29 and 0.03 for liver and small intestine, respectively. Both are within the range of values observed after intravenous administration (Paine et al., 1996; Thummel et al., 1996). Pond and Tozer (1984) and Mistry and Houston (1987) have suggested that the same relationship for intestinal systemic extraction should apply to intestinal first-pass extraction. However, the gut prediction is >10 times lower than the average first-pass extraction ratio observed after intraduodenal administration (0.43; Paine et al., 1996). If the unbound fraction is excluded from the gut calculation, the extraction ratio increases to 0.57, which is closer to the first-pass gut extraction ratio observed in vivo. This suggests that although plasma protein binding reduces the efficiency of systemic intestinal MDZ extraction, it plays a negligible role in determining the first-pass intestinal extraction of the drug.

In summary, our extensive characterization of intestinal CYP3A metabolic capacity revealed that interorgan variability is much greater than intraorgan variability with respect to CYP3A expression and catalytic activity. This may account for the large interindividual differences observed in the oral bioavailabilities of some CYP3A substrates. Furthermore, our findings indicate that the upper small intestine (duodenum and proximal jejunum) serves as the major site for intestinal first-pass metabolism of midazolam and possibly immediate-release preparations (e.g., solutions, suspensions, uncoated tablets and capsules) of other CYP3A substrates that exhibit poor and unpredictable bioavailabilities. Alternatively, the low metabolic capacity of the ileum implies that slow-release preparations may largely be "spared" an intestinal first-pass effect. Finally, a comparison of predicted extraction ratios for the small intestine and liver suggests that protein binding, a factor that greatly influences the in vivo metabolism of several CYP3A substrates by the liver, does not apply to first-pass metabolic extraction by the intestine.