Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Hydrophilic and ionic compounds generally possess low membrane permeability and therefore are poorly absorbed after oral administration. For some drugs, metabolism has also been identified as an important intestinal barrier (Krishma and Klotz, 1994; Friedman and Amidon, 1991; Prueksaritanont et al., 1996a). More recently, it has been increasingly recognized that the intestinal absorption of a compound could also be limited in part by intestinal efflux systems (Hunter et al., 1993a; Karlsson et al., 1993; Terao et al., 1996; Fricker et al., 1996; Cavet et al., 1996; Lang et al., 1997). P-glycoprotein is a well-recognized secretory transporter localized on epithelia of animal and human tissues, such as intestine, kidney, liver, and adrenal glands, as well as the blood-brain barrier (Thiebaut et al., 1987; Croop et al., 1989; Hsing et al., 1992). Features common to most P-glycoprotein substrates recognized to date include hydrophobicity and an amino group (Ford and Hait, 1990). Active secretory transport has also been implicated for some polar compounds, including anions and small peptides (Saitoh et al., 1996; Burton et al., 1993; Langguth et al., 1997).

L-767,679 (I), a potent fibrinogen receptor antagonist, is a highly polar (log P < 3) and zwitterionic compound containing peptide linkages (fig. 1). In animal models, the low oral bioavailability of I was attributed to poor absorption and not to first-pass metabolism (Prueksaritanont et al., 1997). Although the low intestinal permeation of I was likely the result of low lipophilicity, the possible involvement of intestinal efflux could not be completely ruled out. Nevertheless, a prodrug approach, aiming to improve intestinal permeability by increasing lipophilicity, was undertaken (Prueksaritanont et al., 1997; Hutchinson et al., 1996). For several prodrugs in this structural series, extensive hepatic/intestinal first-pass metabolism to metabolites other than the corresponding active drugs, and not poor absorption of the prodrugs, was demonstrated to be primarily responsible for the low oral bioavailability of both the prodrugs and their active drugs after administration of the prodrugs to animals (Prueksaritanont et al., 1996a, 1997). However, our preliminary studies in Caco-2 cells indicated that absorption of some of the prodrugs also was limited. Structurally, the cationic and lipophilic prodrugs are potential substrates for P-glycoprotein.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Chemical structures of I and its prodrug II.

In view of the above discussion, it was desirable to gain some insight into intestinal barriers for the active drug I, as well as its prodrug. The benzyl ester prodrug L-775,318 (II) (fig. 1), a relatively lipophilic prodrug (log P = 0.7), was chosen for the study. In the present study, the intestinal transport mechanism for and metabolism of I and II were examined using an in vitro Caco-2 cell model, in conjunction with known transport modifiers (verapamil and quinidine) and an esterase inhibitor (PMSF¹) (Ford and Hait, 1990; Wacher et al., 1995; Morgan et al., 1994). The Caco-2 cell system offers considerable advantages, including its human origin and a simple epithelial monolayer structure that enables directional transport studies. This human intestinal cell line also expresses functional transporters, such as dipeptide carriers and P-glycoprotein, and some drug-metabolizing enzymes, which are all known to be present in human small intestines (Fricker et al., 1996; Prueksaritanont et al., 1996b; Hunter et al., 1993b; Adibi, 1997). Although good agreement was observed between Caco-2 cell and in vivo findings, with respect to P-glycoprotein involvement, in several studies (Terao et al., 1996; Fricker et al., 1996; Leu and Huang, 1995), recent results showed that Caco-2 cells underestimated the absorption of P-glycoprotein substrates, compared with in vivo observations (Yee, 1997). Apparently, there are some discrepancies between in vitro and in vivo conditions that are relevant to the contribution of P-glycoprotein to the transport of compounds. Therefore, in this study, the intestinal absorption of I and II was also investigated using an in situ rat intestinal loop technique. Because the intestine is known to contain drug-metabolizing enzymes, including esterases (Friedman and Amidon, 1991; Prueksaritanont et al., 1996a; Kaminsky and Fasco, 1992; Heymann and Mentlein, 1988), the in vitro metabolism of I and II was also examined using rat intestinal S9 preparations, in the absence and presence of verapamil and PMSF. Because of the presence of esterases in the rat lumen (Friedman and Amidon, 1991; Prueksaritanont et al., 1996a; Campbell et al., 1987), similar in vitro metabolism studies were also conducted for II using rat intestinal lumen washes.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Chemicals. I [N-{[7-(piperazin-1-yl)-3,4-dihydro-1(1H)-isoquinolinone-2-yl]acetyl}-3(S)-ethynyl--alanine] and its ester prodrug II (fig. 1) were synthesized at Merck Research Laboratories (West Point, PA), as described (Hutchinson et al., 1996). PMSF, verapamil, and NADPH were purchased from Sigma Chemical Co. (St. Louis, MO), whereas quinidine was obtained from Aldrich (St. Louis, MO). Solvents used for analysis were of analytical or HPLC grade (Fisher Scientific, Pittsburgh, PA). The human intestinal cell line Caco-2 was obtained from the American Type Culture Collection (Rockville, MD). The cells were maintained in Opti-MEM medium (Gibco BRL, Grand Island, NY) supplemented with 10% fetal calf serum, nonessential amino acids, and L-glutamine and were used at passages 20-50.

In Vitro Caco-2 Cell Studies. Caco-2 cells, grown on 12-mm Millicel polycarbonate filter inserts (Millipore, Bedford, MA) at 100,000 cells/cm² for 3 weeks, were used throughout. The integrity of the monolayer was monitored by determining transepithelial electrical resistance (320-490 ·cm²) and lucifer yellow permeation. In addition, the transport of caffeine, a compound known to be well transported transcellularly and well absorbed in vivo, was also monitored. The transport studies with I and II were carried out in triplicate or quadruplicate at 37°C using 1-1000 µM I or II, with an apical compartment pH of 7.4 [10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid buffer] or 5.5 [10 mM 2-(N-morpholino)ethanesulfonic acid buffer] and a basal compartment pH of 7.4 [10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid buffer]. Samples were taken from the recipient compartment at designated times and were replaced with an equal volume of fresh buffer. Calculation of cumulative transport included correction for dilution effects. For studies of the inhibitory effects of verapamil, PMSF, and quinidine, the incubations were performed at pH 7.4 for both donor and recipient compartments, with inhibitor concentrations of 200 µM verapamil, 250 µM quinidine, or 500 µM PMSF. Solutions from the two compartments were analyzed for I and II by the HPLC method described below. Both compounds were adsorbed negligibly to the wells or filters, as judged by virtually 100% recovery.

In Situ Rat Intestinal Loop Studies. The studies were reviewed and approved by the Merck Research Laboratories Institutional Animal Care and Use Committee. Male Sprague-Dawley rats (230-320 g) were prepared for the intestinal absorption study according to the procedure of Barr and Riegelman (1970). In brief, a segment of proximal jejunum (about 20 cm long) was isolated and tied off at both ends to form a closed loop. The mesenteric vein that collected blood from the ligated segment of the intestine was cannulated for blood collection. The animal was placed under a heating lamp to help maintain the body temperature at 37°C. Dosing solution (0.5 mg of I or II/2 ml of normal saline), with or without verapamil (500 µM), was injected directly into the lumen of the intestinal loop. Blood samples were collected at appropriate times, and the blood lost from the mesenteric vein was continuously replaced with an equal volume of heparinized fresh rat blood, through the cannulated jugular vein. At the end of the experiment (1 hr), the remaining contents in the intestinal loop were collected. The loop was rinsed with 2 × 3 ml of saline, and the washes were combined. An aliquot of the intestinal wash was immediately mixed with 3 ml of acetone to stop the hydrolytic reaction. The intestine was homogenized using 4 volumes of saline, and an aliquot was taken and immediately extracted with 3 ml of acetone to terminate the hydrolytic reaction. Blood samples (25-75 µl) were also extracted using 3 ml of acetone. The acetone extracts were evaporated, reconstituted with the mobile phase, and analyzed for I, II, and verapamil by the HPLC methods described below.

In Vitro Metabolism Studies. Subcellular fractions of rat intestine (N = 3) were prepared as described previously (Prueksaritanont et al., 1996b). Studies on the metabolism of I or II were performed, in 0.2-ml incubation mixtures, using 0.2 mg of intestinal S9 protein, 0.2 µmol of NADPH, 2.5-100 nmol of I or II, 2 µmol of MgCl₂, and 20 µmol of phosphate buffer, pH 7.4. The reaction was terminated, after incubation at 37°C for various times up to 60 min, by the addition of 0.2 ml of acetonitrile. After centrifugation, the supernatant was analyzed by HPLC. The inhibitory effects of verapamil, PMSF, and quinidine on the metabolism of II were investigated using similar conditions but in the presence of verapamil (25-500 µM), PMSF (100-1500 µM), or quinidine (100-2500 µM) and with an incubation time of 10 min. A preliminary study indicated that the hydrolytic reaction was linear during the 10-min incubation.

Analytical Procedures. An HPLC method for simultaneous determination of I and its prodrug II (Prueksaritanont et al., 1997) was used with minor modifications. The system consisted of a Waters 600E multisolvent delivery system, a Waters 717 Plus autosampler, and a Jasco 821-FP fluorescence detector. The sample analysis of I and II was conducted with a Spherisorb SCX column (4.6 × 250 mm, 5 µm), with the mobile phase (solvent A, acetonitrile; solvent B, 0.04 M phosphate buffer, pH 4.2) being delivered at a flow rate of 1 ml/min (maintenance at 27% solvent A for 5 min, gradient to 60% solvent A in 1 min, and maintenance at 60% solvent A for 6 min). The effluent was monitored at excitation and emission wavelengths of 245 nm and 440 nm, respectively.

Data Analysis. Apparent K_M and V_max values were estimated using a nonlinear regression program (Enzfit; Biosoft, Ferguson, MO). The intrinsic clearance was estimated by dividing V_max by K_M. Determination of the type of inhibition was based on visual inspection of 1) double-reciprocal plots of the data and 2) the patterns of changes in K_M and V_max values in the presence and absence of inhibitors. K_i values for competitive inhibition were then estimated by fitting nontransformed data to the following equation, using a nonlinear regression program (PCnonlin; Scientific Consulting, Cary, NC): V = (V_max × S)/[S + [K_M(1 + I/K_i)]], where S and I represent substrate (II) and inhibitor (verapamil, PMSF, or quinidine) concentrations, respectively.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

In Vitro Caco-2 Cell Studies. For all Caco-2 cell monolayers used, the apical-to-basal transport of caffeine, a positive control for transcellular transport, was >18%/hr. In most cases, the apical-to-basal or basal-to-apical transport of lucifer yellow, a compound known to be transported paracellularly, was <1%/hr. Transport of I across Caco-2 cell monolayers was comparable at the apical pH values of 5.5 and 7.4 (data not shown), and therefore subsequent studies were conducted only at pH 7.4. Over the concentration range of 10-100 µM, the transport of I in both the apical-to-basal and basal-to-apical directions was independent of the initial concentration (fig. 2A). The apical-to-basal permeation was slightly but not statistically greater than the permeation in the opposite direction (fig. 2A). There appeared to be a correlation between the transport of I and that of lucifer yellow (fig. 2B), although the correlation was higher for the apical-to-basal direction (r² = 0.998) than for the basal-to-apical direction (r² = 0.74). Verapamil, a known P-glycoprotein inhibitor, did not affect the transport of I in either direction (fig. 2C). No metabolism of I was observed in experiments with Caco-2 cells.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   A, Bidirectional transport of I as a function of concentration; B, correlation between the transport of I and that of lucifer yellow; C, effects of verapamil on the apical-to-basal (A-TO-B) and basal-to-apical (B-TO-A) transport of I. Results are means ± SD (N = 3). For A and B, all incubations were performed at 37°C for 60 min, using pH 7.4 for both compartments. For C, all incubations were performed at 37°C, using 10 µM I and pH 7.4 for both compartments, in the absence or presence of verapamil (200 µM).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   A, Bidirectional transport of II as a function of concentration; B, correlation between the transport of II and that of lucifer yellow. Results are means ± SD (N = 3). All incubations were performed at 37°C for 60 min, using pH 7.4 for both compartments. *, Statistically significant differences from the apical-to-basal permeation.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   Effects of verapamil, quinidine, and PMSF on the transport of II in the apical-to-basal (A) and basal-to-apical (B) directions. Results are means ± SD (N = 4). All incubations were performed at 37°C and pH 7.4, using 50 µM II, in the absence (control) or presence of verapamil (200 µM), quinidine (250 µM), or PMSF (500 µM). *, p < 0.05; **, p < 0.01, statistically significant differences from control.

In Situ Rat Intestinal Loop Studies. The intestinal absorption of I was limited, in both the absence and presence of verapamil. At the end of the experiment (60 min), the total dose absorbed, as reflected by the sum of the amounts of I recovered in blood and intestine, was 3.9 ± 2.0% in the absence of verapamil and 3.2 ± 0.7% in the presence of verapamil (table 1). In blood, the amount of I recovered over the experimental time course was essentially the same with or without verapamil (fig. 5A). Verapamil also did not appear to affect the total recovery of I in the intestinal tissue or intestinal lumen (table 1). Total recovery of I in all sections (blood, intestine, and lumen) was >85% (table 1), suggesting minimal metabolism of I in rats.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Recovery of I in rat blood as a function of sampling time after administration of I (A) or II (B). Results are means ± SD (N = 4) obtained after administration of I or II (0.5 mg/2 ml) in the rat intestinal loop, in the absence (control) or presence of verapamil (500 µM). *, Statistically significant differences from control.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Recovery of verapamil in rat blood, as a function of sampling time (A), and in gut and lumen (B). Results are means ± SD (N = 4) obtained after administration of verapamil (500 µM) in the rat intestinal loop, with I or II (0.5 mg/2 ml).

In Vitro Metabolism Studies. Intestinal Metabolism. I was resistant to metabolism, whereas the prodrug II was metabolized extensively by rat intestinal S9 fractions, in agreement with the in situ observations. The metabolism of II, both in the presence and in the absence of NADPH, yielded exclusively I, indicating the primary involvement of esterases. The hydrolysis of II displayed monophasic kinetics, with an apparent K_M of 720 µM and a V_max of 36 nmol/min/mg S9 protein. In vitro metabolism was also assessed, to examine a potential inhibitory effect of verapamil on the hydrolysis of II. Verapamil appeared to be a competitive inhibitor of a II-hydrolyzing enzyme system (fig. 7A), with an apparent K_i value of 90 µM. Under similar conditions, PMSF competitively inhibited hydrolysis, with a K_i value of 20 µM (fig. 7B). These results suggested that verapamil was a relatively potent inhibitor of the intestinal hydrolysis of II.

View larger version (21K): [in this window] [in a new window]   Fig. 7.   Effects of verapamil (A) and PMSF (B) on the hydrolysis of II by rat intestinal S9 fractions. The data points represent averages of duplicate determinations. All incubations were carried out in the absence or presence of verapamil (25-500 µM) or PMSF (100-750 µM), at 37°C for 10 min, using rat intestinal S9 protein (1 mg/ml) and II (25-1000 µM) with NADPH (1 mM). V, I formation (nmol/min/mg); S, II concentration (µM).

Intestinal Luminal Metabolism. Hydrolysis of II was also examined using the intestinal lumen wash obtained from untreated rats. II was hydrolyzed considerably (fig. 8), presumably because of the presence of esterases in the rat lumen; hydrolysis was minimal in the absence of the lumen wash. Verapamil, in the concentration range examined, did not alter the rate of hydrolysis of II (fig. 8), in agreement with the in situ observations. The findings also suggested that the hydrolysis of II in the intestine and that produced by the lumen wash resulted from different hydrolytic enzymes and that verapamil was a relatively specific inhibitor of the intracellular intestinal esterases.

View larger version (25K): [in this window] [in a new window]   Fig. 8.   Effects of verapamil on the hydrolysis of II by rat lumen wash. Results are means ± SD (N = 3) obtained using 250 µM II.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This study used both in vitro Caco-2 cell and in situ rat intestinal loop models to characterize the absorption barriers for I and its prodrug II. The results with Caco-2 cells suggested that the low levels of absorption of I were not the result of polarized efflux systems. The primary pathway for the transport of I appeared to be paracellular. These conclusions were based on the findings that the transport of I in both the apical-to-basal and basal-to-apical directions was concentration independent and was correlated with that of lucifer yellow. In addition, the transport of I was not inhibited by verapamil. Considering that the apical-to-basal permeation was slightly greater than the basal-to-apical permeation and that I was detected on the basal side when II was initially placed in the apical chamber, I could also be transported (albeit to a much lesser extent) transcellularly. In contrast, the prodrug II was transported predominantly by the transcellular pathway and modestly by the paracellular route. The net transcellular transport of II was attenuated by polarized and saturable efflux systems. These conclusions were made based on the following findings: 1) basal-to-apical transport in Caco-2 cells was up to 5-fold greater than apical-to-basal permeation, 2) the transport of II only in the apical-to-basal (and not the basal-to-apical) direction was correlated with that of lucifer yellow in the corresponding directions, 3) verapamil and quinidine increased the apical-to-basal transport and decreased the basal-to-apical transport of II, and 4) the secretory transport displayed saturable kinetics. The apparent K_M value for II was ~400 µM, which is considerably higher than values reported for cyclosporine and vinblastine, known P-glycoprotein substrates (Fricker et al., 1996; Hunter et al., 1993b), but comparable to those noted for digoxin and pristinamycin, also P-glycoprotein substrates (Cavet et al., 1996; Phung-Ba et al., 1995). The results suggested that the polarized efflux of II might be moderate and would be less favorable in the presence of a compound with much higher affinity.

To examine the in vivo contribution of the intestinal efflux system, the transport of I and II was further investigated using the in situ rat intestinal loop model. Qualitatively, results obtained from the in situ experiment, for both I and II, agreed well with those from Caco-2 cells. Quantitatively, however, there appeared to be a discrepancy. The apparently greater effects of verapamil on the transport of II in Caco-2 cells, compared with those in rats, might be a result of species differences in the amount and/or function of the responsible secretory transporter expressed in the two models. In Caco-2 cells, P-glycoprotein could be expressed at very high and variable levels, depending on several factors, including the passage number and culture medium (Burton et al., 1997). In the present study, the jejunum section of rat intestine, which is known to contain P-glycoprotein and possibly other secretory transporters (Hsing et al., 1992; Saitoh and Aungst, 1995), was used. It is not known whether the transporters present in the Caco-2 cells and rat intestine exhibited different affinities for the substrate II (K_M) or the inhibitor verapamil (K_i). The disparity observed could have resulted from differences in the concentrations of II and verapamil in these two systems, relative to their respective K_i/K_M values. Further studies on the quantitative contribution of the intestinal efflux system in vivo appear warranted.

Considering that verapamil and quinidine are P-glycoprotein inhibitors (Ford and Hait, 1990; Wacher et al., 1995), that verapamil is not a potent inhibitor of the multidrug resistance-associated protein (another efflux carrier) (Lautier et al., 1996; Flens et al., 1996), and that P-glycoprotein is expressed substantially in Caco-2 cells and rat intestine (Hsing et al., 1992; Hunter et al., 1993b), it is likely that II was a substrate for P-glycoprotein. Because P-glycoprotein is also expressed in the human gastrointestinal tract (Thiebaut et al., 1987), it is conceivable that P-glycoprotein also would play a role in the intestinal absorption of II in humans. Interestingly, at physiological pH, II is a cation with reasonable lipophilicity, in common with most P-glycoprotein substrates (Ford and Hait, 1990). To our knowledge, the involvement of P-glycoprotein has not been demonstrated previously to be an absorption barrier for an ester prodrug, although similar findings were noted for peptides with a masked carboxyl terminus (Lang et al., 1997). Also, not all cationic prodrugs screened in our laboratory exhibited preferential secretory transport in Caco-2 cells. Apparently, more studies are needed for a better understanding of the substrate specificity of intestinal efflux systems.

The present study also indicated that metabolism was not a barrier to intestinal absorption of I in Caco-2 cells or rats. A similar conclusion is anticipated for humans, based on an earlier metabolism study of I using human intestinal S9 fractions (Prueksaritanont et al., 1997). In the case of II, intestinal metabolism also was not a major barrier in Caco-2 cells, because metabolism was not extensive. In rats, however, metabolism by esterases in the intestinal lumen served as an additional, although not principal, barrier to the absorption of II. Based on the available data, it is unclear whether esterases in the intestinal tissue also played an important role in limiting the absorption of II in rats. If II were absorbed substantially before hydrolysis, this metabolism would not pose a major barrier to the intestinal availability of I after administration of II, because I was not a substrate for an efflux system. However, the systemic availability of I after administration of II could still be low, because I, once formed inside the cells, is not expected to be readily transported across the basal membrane into the bloodstream. In fact, this is consistent with the greater accumulation of I observed in intestinal tissue after administration of II (table 2), compared with administration of I (table 1). In view of the low recoveries of I plus II in intestinal tissue and blood and the high levels of unchanged II in the rat lumen in the presence of verapamil (table 2), mechanisms other than intestinal efflux and metabolism might also be responsible for the low levels of absorption of II in this animal model. In humans, intestinal metabolism would be unlikely to play a significant role, because II was found to be minimally metabolized by human intestinal S9 fractions in a preliminary study (data not shown).

In addition to being known as P-glycoprotein inhibitors, verapamil and quinidine are cytochrome P450 substrates/inhibitors (Wacher et al., 1995; Hovgaard et al., 1995; Newton et al., 1995). The present study demonstrated, for the first time, that verapamil is also a potent and relatively specific intestinal esterase inhibitor. Quinidine was a very weak inhibitor of esterases (K_i > 2 mM, data not shown). Conceivably, the effect of verapamil on the net absorption of II in rats resulted from its ability to inhibit P-glycoprotein. However, the basis of the effect of intestinal esterase inhibition on the absorption of II is less clear. Theoretically, inhibition of the esterases would result in increased intracellular concentrations of II, which, in the presence of functional P-glycoprotein, should lead to a decrease in the net transport of II across the intestine. This is because more II would be available for intestinal efflux. However, if the intracellular concentration of II was increased and rapidly exceeded the capacity of the efflux system, the transport of II could also be increased. It is possible that inhibition of both efflux and esterase activity was responsible for the increased absorption of II in the presence of verapamil in rats.

In summary, by means of in vitro Caco-2 cell transport, in situ rat intestinal absorption, and in vitro metabolism studies, this report demonstrated that secretory transport served as an intestinal barrier only for the more lipophilic and cationic prodrug II and not for the hydrophilic zwitterion I. Metabolism also contributed, in part, to poor absorption of II in rats. The absorption of II was increased substantially after inhibition of the efflux system by P-glycoprotein inhibitors. Thus, increasing the hydrophobicity of compounds using a prodrug approach may not always result in increased absorption. On the contrary, as was demonstrated in this study, poor absorption resulting from active intestinal secretion of prodrugs is also possible. The potential involvement of intestinal efflux should be examined in the evaluation of cationic prodrugs.