European Journal of Pharmaceutical Sciences 19 (2003) 57–65www.elsevier.com/ locate/ejps

E nantioselective transport and CYP3A4-mediated metabolism ofR/S-verapamil in Caco-2 cell monolayers

*¨Helena Engman, Christer Tannergren, Per Artursson, Hans LennernasDepartment of Pharmacy, Uppsala University, Box 580, SE-751 23 Uppsala, Sweden

Received 18 November 2002; received in revised form 28 February 2003; accepted 10 March 2003

Abstract

We have evaluated the passive and carrier-mediated intestinal transport and CYP3A4-mediated metabolism of R/S-verapamil withrespect to dose dependency and enantioselectivity in modified Caco-2 cells. The present in vitro results were compared to published datafrom human in vivo and rat in situ jejunal perfusions with R/S-verapamil. Caco-2 cell permeability to enantiomers of verapamil andnorverapamil was weakly concentration dependent (2.5–100mM). While Caco-2 permeability to verapamil was 2.6- to 3.7-fold lowerthan in the human jejunum, it was 1.4- to 2.3-fold higher than in rats. However, all three models classified R- and S-verapamil as highpermeability compounds according to the biopharmaceutical classification system. In accordance with human and rat data, R/S-verapamilwas transported to a minor extent by carrier-mediated mechanisms in Caco-2 cells. Neither the passive nor the carrier-mediatedpermeability was enantioselective in any of the three models. CYP3A4-mediated demethylation to R/S-norverapamil was enantioselectivein Caco-2 cells. ApparentV andK values for the conversion of R-verapamil were 3.2 pmol /min/ insert and 0.7mM, respectively, andmax m

for S-verapamil, 5.4 pmol /min/ insert and 0.6mM, respectively. The enantioselectivity in the CYP3A4-metabolism observed in Caco-2cells was in agreement with human data, but not with rat data, indicating that Caco-2 cells better reflect the human small intestine in thisregard. However, all three models suggested that intestinal permeability to verapamil is unaffected by CYP3A4-activity. In summary,modified Caco-2 cells and human jejunum were qualitatively related with respect to R-and S-verapamil transport and CYP3A4-metabolism. 2003 Elsevier Science B.V. All rights reserved.

Keywords: Enantioselective; Permeability; CYP3A4; Caco-2; Human jejunum

1 . Introduction In principle, three approaches have been used to developintestinal cell models for studies of CYP3A4-mediated

The development of rapid and more reliable methods for drug transport and metabolism: (1) use of the CYP3A5predictions of drug transport and metabolism involving expressing Caco-2 clone TC7 as a substitute for CYP3A4intestinal cytochrome P450 (CYP) 3A4, as well as for expression (Raeissi et al., 1997); (2) development ofclassification of drugs according to the biopharmaceutical CYP3A4-expressing Caco-2 cells using heterologous ex-classification system (BCS), is warranted (Amidon et al., pression techniques (Crespi et al., 1996; Hu et al., 1999;1995). Intestinal epithelial cell culture models are attrac- Brimer et al., 2000); and (3) 1a,25-dihydroxy vitamintive for such studies, since they express several transport D -mediated induction of CYP3A4 in the Caco-2 cell line3

mechanisms (passive and carrier-mediated) that can be (Schmiedlin-Ren et al., 1997; Hochman et al., 2000). Themonitored parallel to specific metabolic reactions. How- third approach provides tight cell monolayers exclusivelyever, the cell culture models also have limitations. For expressing the CYP3A4 isoform. Moreover, this techniqueinstance, the expression and activity levels of certain most likely generates a more physiological system, inenzymes and transport proteins have in many cases not which endogenous CYP3A4 is hormonally upregulated andbeen thoroughly evaluated in these otherwise well-studied it was therefore selected for the purpose of our studiescell models. (Engman et al., 2001).

The Caco-2 cell line also expresses appreciable amountsof many efflux proteins, including the MDR1 gene prod-*Corresponding author. Tel.:146-18-471-4371; fax:146-18-471-uct, P-glycoprotein (Pgp) (Hunter et al., 1993; Engman et4223.

¨E-mail address: hans.lennernas@farmaci.uu.se(H. Lennernas). al., 2001; Taipalensuu et al., 2001). The potential impact

0928-0987/03/$ – see front matter 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0928-0987(03)00065-4

58 H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65

of multiple intestinal ATP-binding cassette (ABC) trans- 2 . Materials and methodsporters on the in vivo intestinal uptake of some drugs hasbeen reported (Greiner et al., 1999; Fromm et al., 2000). 2 .1. ChemicalsFurthermore, other transporters, such as members of theorganic cation transporter (OCT) family are potential 1a,25-dihydroxy vitamin D (D3)was purchased from3

mediators of drug transport in both absorptive and secret- Solvay Duphar (Weesp, the Netherlands). R/S-verapamilory directions (Koepsell, 1998; Sun et al., 2002). and R/S-norverapamil were kindly provided by Knoll AG

The enantiomers of the commonly used calcium channel (Darnstadt, Germany). R-methoxyverapamil and tetrahex-blocker verapamil have previously been classified as high ylammonium were purchased from Sigma (St. Louis, MO,permeability compounds according to the biopharmaceuti- USA). GF120918 was a generous gift from Glaxo

14 21cal classification system, with complete intestinal absorp- SmithKline, UK. [ C]-mannitol (52 mCi mmol ) wastion in humans (Sandstrom et al., 1998a, 1999a). It has obtained from New England Nuclear (Boston, MA, USA).also been reported that the first-pass gut wall metabolism Stock solutions of D3 were prepared at 0.1 M in ethanolof verapamil is responsible for as much as 50% of the total and stored at280 8C. Chemicals used for high-perform-first-pass extraction (Fromm et al., 1996; Sandstrom et al., ance liquid chromatography (HPLC) were of analytical or1999a; von Richter et al., 2001). Verapamil, administered HPLC grade.as a racemate, is metabolized by a number of enantio-specific CYP450-enzymes in humans (Kroemer et al.,

2 .2. Cells and culture conditions1993; Busse et al., 1995), and one of the major metabolicpathways, catalyzed by CYP3A4, leads to the formation of

The parental population of Caco-2 cells originating fromnorverapamil (Eichelbaum et al., 1979).

the human colorectal carcinoma Caco-2 (Hidalgo et al.,Only a few drugs have been thoroughly evaluated with

1989) was obtained from American Type Culture Collec-regard to direct measurements of intestinal drug transport

tion (ATCC; Rockville, MD, USA). Caco-2 cells wereand metabolism by means of human in vivo perfusions

cultured as described in detail previously (Artursson, 1990;(Sandstrom et al., 1998a, 1999a; von Richter et al., 2001).

Engman et al., 2001; Tavelin et al., 2002) and used atThus, there is a need for much more in vivo data in order

passage intervals 92-105.to validate cell culture methods for corresponding in vitrostudies (Tucker et al., 2001). In addition, since the rat is ananimal model commonly used in preclinical research and 2 .3. Drug transport experiments in Caco-2 celltoxicity studies, there is an interest in comparisons be- monolayerstween rat and Caco-2 cell data. Verapamil has beenextensively studied by means of human in vivo and rat in In general, R/S-verapamil and R/S-norverapamil weresitu jejunal perfusions (Sandstrom et al., 1998a,b, 1999a; dissolved in Hanks’ Balanced Salt Solution (HBSS, pHSandstrom and Lennernas, 1999; von Richter et al., 2001). 7.4) at a final concentration of 2.5–80 and 5–100mM,Since recently published human and animal data on respectively. The concentration interval for verapamil waspermeability and first-pass metabolism were available in- chosen to cover a range of luminal concentrations expectedhouse, we have used them in this in vitro study to evaluate at the absorption site in the human jejunum (Sandstrom eta modified Caco-2 cell model (Engman et al., 2001). al., 1998a). The drug solutions were sterile filtered andPrevious in vitro studies have focused on verapamil and, in added to the donor side of the monolayers. HBSS withouta few studies, its metabolites as substrate(s) and/or in- the drug was added to the receiver side. The monolayershibitor(s) of Pgp-mediated transport (Makhey et al., 1998; were incubated at 378C in a humidified atmosphere.Pauli-Magnus et al., 2000). Caco-2 cell monolayers have Samples were withdrawn from the receiver chamber at twobeen used frequently as a model of in vivo permeability to to three regular intervals for up to 16 min and stored atmany drugs, including verapamil (Mandagere et al., 2002). 18 8C pending analysis. A calibrated plate shaker wasNevertheless, to our knowledge, this is the first study that used to agitate the monolayers. The integrity of thesimultaneously investigates enantioselective aspects of monolayers was routinely checked with the hydrophilic

14both transport and inherent metabolism in Caco-2 cells, integrity marker [ C]-mannitol and by measuring theapplying clinically relevant concentrations of a commonly transepithelial electrical resistance. The transepithelialused and well-studied drug. electrical resistance was determined in HBSS, pH 7.4 at

The aim of the present study was to evaluate CYP3A4- 378C using an Endohm 24 tissue resistance chamberexpressing Caco-2 cell monolayers as a tool in studies of (World Precision Instruments, Sarasota, FL, USA) and aenantioselective transport and metabolism using R/S-ver- Millicell ERS resistance meter (MilliPore SA, Molshelm,apamil as a marker. The results were compared to recently France). Values ranged from 171 to 310 and from 171 to

2published data on R/S-verapamil obtained from human in 361V cm for D3-treated and untreated cell monolayers,vivo and rat in situ jejunal perfusions. respectively. Thus, the integrity of D3-treated cell mono-

H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65 59

layers did not differ significantly from the integrity of 2 .6. Enantioselective HPLC analysis of verapamil anduntreated cells. norverapamil

2 .4. Evaluation of carrier-mediated transport R/S-verapamil and R/S-norverapamil were quantifiedby enantioselective reversed phase HPLC with fluores-

In order to detect any significant carrier-mediated trans- cence detection (excitation and emission wavelengths 232port of the compound, and to evaluate any possible effects and 310 nm, respectively) using a method modified from

¨of D3 on transport protein functionality, the R/S-ver- Sandstrom et al. (Sandstrom et al., 1998a). The chromato-apamil transport rates were determined in both apical-to- graphic system consisted of a Shimadzu isocratic LC pumpbasolateral (a–b) and basolateral-to-apical (b–a) directions 9A (Shimadzu, Kyoto, Japan), an autoinjector (CMA 200,in the untreated (control) and D3-treated monolayers. CMA/Microdialysis AB, Solna, Sweden), a pre-column

Inhibition of Pgp-, multidrug resistance-associated pro- (Chiral-AGP, 1033.0 mm35 mm, ChromTech, Stockholm,tein, MRP- and OCT-mediated transport of R- and S- Sweden), a column (Chiral-AGP, 15034.0 mm35 mm)verapamil across both control and D3-treated Caco-2 cell and a fluorescence detector (Jasco FP-920, Tokyo, Japan).monolayers was investigated after the addition of the R- and S-verapamil and R- and S-norverapamil werePgp-inhibitor GF120918 (0.2mM), the MRP-inhibitor eluted with phosphate buffer (pH 7.6; ionic strength 0.01) /indomethacin (20mM), and the OCT-inhibitor tetrahexyl- acetonitrile /methanol (74/18/8 v/v) at a flow rate of 1.0ammonium (50mM) to the apical and basolateral compart- ml /min. In order to optimize the separation between S-ments. Prior to the experiments, both sides of the cell verapamil and R-norverapamil, the mobile phase was pre-monolayers were incubated for 30 min with each inhibitor heated and used at 308C. The limits of quantification forin HBSS, pH 7.4. In addition, HBSS used for volume the compounds were 0.13mM (coefficient of variation; CVcompensation contained the relevant inhibitors at the 3%) and 0.11mM (CV 13%) for R-verapamil and S-indicated concentrations. verapamil, respectively, while the limits of quantification

for R-and S-norverapamil were 0.11mM (CV 3%) and2 .5. Metabolism of R /S-verapamil in Caco-2 cell 0.12 mM (CV 1%), respectively. More details on themonolayers enantioselective HPLC method are described elsewhere

(Sandstrom et al., 1999b). Samples from apical andThe level of CYP3A4-mediated metabolism of R/S- basolateral chambers were diluted in HBSS, pH 7.4 and

verapamil was determined in CYP3A4-expressing Caco-2 injected (50ml) on the column. To determine intracellularcell monolayers, using untreated cells as controls. The amounts of verapamil and norverapamil, intact cell mono-CYP3A4-activity in the D3-treated cells in this study was layers were washed in ice-cold PBS, cut and frozen. Forcomparable to that reported previously (Hochman et al., extraction and quantification of intracellular amounts of2000; Engman et al., 2001). R/S-verapamil, frozen monolayers were immersed in

The cell monolayers were washed and equilibrated 15 absolute ethanol and R-methoxyverapamil was added as anmin in pre-warmed HBSS, pH 7.4. Thereafter, 1.8 ml of internal standard (stock solution 2.2mg/ml in water).racemic R/S-verapamil (5–150mM) dissolved in HBSS, Samples were centrifuged for 10 min at 5000 rev. /min,pH 7.4 was added to the apical side of the monolayer, evaporated to dryness and reconstituted in mobile phase.while 1.8 ml HBSS, pH 7.4 without the drug was added to Following another centrifugation step, 50ml of the super-the receiver side. The concentration range was based on natant was injected onto the HPLC column and analyzedK -values for the CYP3A4-mediated conversion of ver- as described above.m

apamil to norverapamil determined in human liver micro-somes (Kroemer et al., 1992). The monolayers were 2 .7. Calculationsincubated at 378C for 0–240 min depending on theexperimental design (time- or concentration-dependent Since verapamil and norverapamil are rapidly trans-metabolic studies). At the end of the incubation, samples ported compounds, sink conditions could not be main-were collected from the apical and basolateral sides. tained during the full duration of all the experiments. Thus,Samples were stored at18 8C pending HPLC analysis of the apparent permeability coefficients (P ) were calcu-appR/S-verapamil and the metabolite, R/S-norverapamil. lated as described previously (Palm et al., 1999) fromExtraction and quantification of intracellular amounts of

C 5 (M /V 1V )R(t) D RR/S-verapamil and R/S-norverapamil were performed2P A(1 /V 11 /V )taccording to the procedure described below. Michaelis– app D R1 (C 2 (M /V 1V )) eR,0 D RMenten kinetics were used to obtain apparentV andKmax m

values for the demethylation of R- and S-verapamil by whereC is the time-dependent drug concentration in theR(t)CYP3A4 using Prism (GraphPad Software Inc., San receiver compartment,M is the total amount of drug in the

Diego, CA, USA). system,V and V are the volumes of the donor andD R

60 H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65

receiver compartments, respectively,C is the drug Studies on directional transport at 8mM racemicR,0

concentration in the receiver compartment at the beginning verapamil resulted in lowP /P efflux ratiosapp (b–a) app(a–b)

of the interval, andt is the time from the start of the in D3-treated cells (1.960.2 and 1.960.2 for R- andinterval. P was obtained from nonlinear regression, S-verapamil, respectively), indicating that efflux has aapp

minimizing the sum of squared residuals (o(C 2 minor quantitative importance in the absorption processR,i,obsd2C ) ), where C is the observed receiver con- (Table 1). The corresponding efflux ratios in untreatedR,i,calcd R,i,obsd

centration at the end of the interval andC is the Caco-2 cells were not significantly different from unity.R,i,calcd

corresponding concentration calculated according to the No net efflux was observed at a 10-fold higher con-equation above. centration (Table 1).

To investigate carrier-mediated processes potentially2 .8. Statistics involved in the efflux ratio at 8mM verapamil, inhibitors

of different influx and efflux systems (GF120918 for Pgp,Values are expressed as mean6one standard deviation indomethacin for MRP, and tetrahexylammonium for

(S.D.) unless otherwise stated. The differences between OCT) were used. TheP /P efflux ratios wereapp (b–a) app(a–b)

mean values were analyzed with unpaired or paired two- brought close to unity by each of the three inhibitors (Fig.tailed Student’st-test as appropriate. Comparisons between 1).more than three groups were performed by using one-wayANOVA followed by Tukey’s post test.P,0.05 was 3 .2. Permeability to enantiomers of norverapamil inconsidered to be significant. Caco-2 cell monolayers

Permeability to R- and S-norverapamil was high in the3 . Results a–b direction in both untreated and D3-treated Caco-2

cells. Using racemic norverapamil at 5, 10 and 20mM in3 .1. Permeability to enantiomers of verapamil in Caco-2 untreated Caco-2 cells,P to R-norverapamil wasapp

26cell monolayers 70.568.0, 98.864.9, and 112.860.4310 cm/s, respec-tively, while P to S-norverapamil was 87.263.1,app

26In both untreated and D3-treated Caco-2 cells, the 101.864.4, and 113.560.6310 cm/s, respectively.permeability to enantiomers of verapamil in a–b direction Thus, within this range (5–20mM), permeability waswas high according to the BCS (Amidon et al., 1995) and concentration dependent. However, at the highest con-showed a weak concentration dependency (P,0.05) in the centration used (100mM), P to both R- and S-nor-app

26studied range (2.5–80mM; Table 1). Further statistical verapamil was decreased to 89.665.1 and 91.665.8310analysis revealed that the concentration dependency was cm/s, respectively. In general, enantioselectivity was notslightly more pronounced in D3-treated cells, in particular observed. Neither untreated nor D3-treated cells showedfor R-verapamil. The permeability was neither enan- signs of carrier-mediated mechanisms in the transport oftioselective, nor affected by the D3-induced CYP3A4- R- and S-norverapamil (efflux ratios close to unity, notactivity in the cells (Table 1). shown).

T able 1Permeability (P ) in absorptive (a–b) direction andP /P efflux ratios determined for R-and S-verapamil in untreated and 1a,25-dihydroxyapp app, b–a app,a–b

vitamin D (D3)-treated Caco-2 cell monolayers3

Racemic Untreated cells D3-treated cellsverapamil(mM) R-verapamil S-verapamil R/S R-verapamil S-verapamil R/S

26 26P (310 , cm/s) P (310 , cm/s)app app

b,c c a,b,c c2.5 68.9613.0 84.6613.1 0.81 66.861.7 70.863.6 0.94c b,c c8 96.3623.7 124.4624.1 0.77 80.165.9 85.765.6 0.94

c c25 123.5612.9 129.0612.2 0.96 97.166.0 87.6624.9 1.0580 156.6619.1 142.6616.7 1.10 129.163.8 127.062.5 1.02

Efflux ratios Efflux ratiosd d8 2.060.6 1.860.4 1.960.2 1.960.2

80 1.060.1 1.060.1 1.160.0 1.060.0a–cDifferences described by superscripts were tested by one-way ANOVA followed by Tukey’s multiple comparison test. Data shown as mean6S.D. of

three to five filters /experimental condition.a Significantly different fromP at 8 mM.appb Significantly different fromP at 25mM.appc Significantly different fromP at 80mM.appd Significantly different from unity,P,0.05.

H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65 61

3 .3. Metabolism of R /S-verapamil in Caco-2 cellmonolayers

Norverapamil was formed in D3-treated Caco-2 cellmonolayers, but not in untreated cells lacking CYP3A4activity (Table 2), thus confirming the importance ofCYP3A4 for this particular metabolic reaction. Evaluationof the time-dependent formation of norverapamil showedthat the initial metabolic process was rapid, reaching aplateau after approximately 60 min incubation (Table 2).CYP3A4-mediated demethylation of R/S-verapamil to R-and S-norverapamil in D3-treated Caco-2 cell monolayerswas enantioselective, i.e. S-norverapamil levels were high-er than those of R-norverapamil (Fig. 2). ApparentVmax

and K values for the conversion of R-verapamil to R-mFig. 1. Vectorial transport of 8mM R/S-verapamil (VL) in 1a,25- norverapamil were 3.2 pmol /min/ insert and 0.7mM,dihydroxy vitamin D (D3)-treated Caco-2 cell monolayers in the absence3 respectively, while the apparentV andK values for themax mand presence of inhibitors of carrier-mediated transport (Pgp/GF120918

formation of S-norverapamil from S-verapamil were 5.40.2 mM; MRP/ indomethacin (Indomet) 20mM; OCT/tetrahexylam-monium (THA) 50mM). Values are shown as mean6S.D. from experi- pmol /min/ insert and 0.6mM, respectively. Levels of R-ments performed in triplicate. The efflux ratios in the control experiment and S-norverapamil in samples gathered from the apical(no inhibitor) was significantly different from unity. * Significantly and basolateral chambers constituted 2.8 and 3.4% of thedifferent from control (P,0.05).

apically applied doses of R- and S-verapamil, respectively.The distribution of verapamil and formed metabolites was

T able 2 comparable between the apical and basolateral chambers,Time-dependent formation of R- and S-norverapamil (NV) from 30mM which is in accordance with the high passive permeabilityracemic R/S-verapamil in 1a,25-dihydroxy vitamin D (D3)-treated3 and a low polarized efflux of the parent drug and itsCaco-2 cell monolayers

metabolites (Table 1). Unfortunately, the intracellularTime (min) R-NV pmol / insert S-NV pmol / insert concentrations of norverapamil were below the limit of

0 0 0 quantification and therefore no extraction ratio could be15 ,LOQ 544.7 calculated (Fisher et al., 1999).30 ,LOQ 567.50 525.7 672.3

120 544.1 691.64 . Discussion180 529.1 729.6

240 585.5 824.2

4 .1. PermeabilityLevels of R-norverapamil at the earliest time points (15 and 30 min)were below the limit of quantification (LOQ). CYP3A4-activity was notdetected in untreated cells, i.e. NV was not formed. Each point represents In the current in vitro study, permeability assessmentsdata obtained from a single culture incubated for the indicated time. were performed at clinically relevant concentrations of

Fig. 2. Concentration-dependent demethylation of R/S-verapamil to (a) R- and (b) S-norverapamil (NV) in 1a,25-dihydroxy vitamin D (D3)-treated3

Caco-2 cells. Samples were drawn separately from the apical (apical) and the basolateral (basolat) compartments. Data from duplicate determinations areshown for each substrate concentration (apical and basolat 1–2, respectively).

62 H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65

verapamil (Sandstrom et al., 1998a). This is clearly an absorption (Fagerholm and Lennernas, 1995). Passiveimportant issue in the design of in vitro studies, which are permeability, especially to rapidly transported compounds,considered to be a step in the classification of new drugs can be significantly reduced by the aqueous boundary layeraccording to the BCS (Amidon et al., 1995; Yu et al., when determined in vitro (Stenberg et al., 2001). Accord-2002). Both verapamil and norverapamil were classified as ingly, to minimize potential effects of in vitro artefacts, wehigh permeability compounds transported predominantly determined the true cellular permeability coefficients un-by a passive, transcellular mechanism. This is in agreement biased by the aqueous boundary layer. We obtained 2- towith previous studies in human and rat jejunum (Sand- 9-fold higher P -values for verapamil compared toapp

strom et al., 1998a,b). Obviously, efflux by Pgp and/or corresponding data obtained without taking the aqueousMRP and transport by OCT were of minor importance. boundary layer into account (Pauli-Magnus et al., 2000;Thus, the efflux ratios for both R-and S-verapamil were Polli et al., 2001). Another reason for the discrepancysmall (|2) and three different inhibitors (Pgp/GF120918, between Caco-2 cells and human jejunum could be relatedMRP/ indomethacin, OCT/ tetrahexylammonium) brought to differences in the effective absorptive surface area.efflux ratios to unity. Together with the observed con- However, since highly permeable drugs are presumed to becentration dependency, it is clear that some transporters are absorbed at the tip of the villi, the areas are likely to be ofinvolved in the absorption of verapamil, although to minor equivalent size.extent. This is supported by other studies showing that The concentration-dependent permeability to R/S-ver-verapamil is effluxed by Pgp (Saitoh and Aungst, 1995; apamil previously found in human and rat jejunal perfu-Sandstrom et al., 1998a; Adachi et al., 2001). In contrast, a sions (Sandstrom et al., 1998a,b) was confirmed in this inrecent study in Caco-2 cells classified verapamil and vitro study. No enantioselectivity was observed in any ofnorverapamil as potent inhibitors of and not substrates for the three models (Caco-2, rat and human jejunum) at thePgp (Pauli-Magnus et al., 2000). These discrepancies in investigated concentrations (Fig. 3). The permeabilities inreported data clearly indicate the complexity in achieving Caco-2 cells and rat jejunum were of similar magnitude,accurate assessments of the transport mechanisms of many although at some concentrations, Caco-2 values were up todrugs. A complete and rapid absorption of verapamil may 2.3-fold higher than those in the rat (Fig. 3). Thisalso involve other processes except for a rapid passive relationship between Caco-2 and rat permeability wasdiffusion, e.g. saturation and/or inhibition of efflux mecha- recently shown to be valid for several other rapidlynisms. transported compounds (Palm et al., 1996). A lower

The passive permeability to verapamil in human je- jejunal permeability in anesthetized rats may be related tojunum was about 3-fold higher than that in Caco-2 cell a reduced systemic blood flow, to the presence of amonolayers (Fig. 3). Several reasons may be advanced to significantly thicker aqueous boundary layer (Fagerholmexplain these observations. One major reason for the veryand Lennernas, 1995), and to differences in the effectivehigh permeability in unanesthetized humans is presumably absorptive surface area (Fagerholm et al., 1996).the presence of the systemic blood flow that makes asignificant contribution to the ‘sink conditions’. In the 4 .2. CYP3A4-mediated metabolism of R /S-verapamilhuman in vivo perfusion model, intestinal motility reducesthe aqueous boundary layer significantly, and thus, the Enantioselective CYP3A4-mediated metabolism of R/S-intestinal epithelium becomes the rate-limiting step for verapamil was observed in Caco-2 cell monolayers, illus-

Fig. 3. Permeability to R-and S-verapamil (VL) in Caco-2 cell monolayers in comparison to published data on human and rat jejunal permeability at (a)8 mM and (b) 80mM racemic R/S-verapamil.Values from Caco-2 experiments are shown as mean6S.D. from experiments performed in triplicate. Human

§and rat permeability data were obtained from human in vivo and rat in situ jejunal perfusions (Sandstrom et al., 1998a,b).

H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65 63

trated by a higher formation of S-norverapamil (Fig. 2). underestimated by solely taking the luminal metaboliteSince norverapamil was not formed in untreated Caco-2 concentration into account, since a large fraction ofcell without CYP3A4-activity (Engman et al., 2001), the metabolites would be distributed to mesenteric bloodconversion of verapamil to norverapamil was regarded as a vessels by passive diffusion. This is supported by theCYP3A4-specific reaction in these cells. Intracellular values of the first-pass gut wall extraction (¯50%)amounts of norverapamil were below the limit of quantifi- (Fromm et al., 1996; Sandstrom et al., 1999a; von Richtercation, which can be interpreted as though norverapamil is et al., 2001). Nevertheless, the enantioselectivity in thea good substrate for efflux proteins (e.g. Pgp), resulting in CYP3A4-catalyzed reactions clearly demonstrated a quali-higher concentrations of norverapamil extracellularly and tative correlation between the modified Caco-2 cell modelpreferentially on the apical side. However, we found only a and the human jejunum.weak tendency towards a polarized efflux of formed A direct comparison between Caco-2 and rat metabolismmetabolites, and low efflux ratios for added, not formed, data was impossible since the only substrate concentrationnorverapamil. At low substrate concentrations, nor- investigated in rats (200mM, Fig. 4) was not well toleratedverapamil concentrations achieved in the apical compart- in the Caco-2 system. The absence of enantioselectivity inment of Caco-2 cell monolayers were comparable to those norverapamil formation in rat jejunum is in contrast to thatobserved on the luminal side of human jejunum (Sand- observed in humans and in Caco-2 cells. The discrepancystrom et al., 1998a) (Fig. 4). Thus, norverapamil seems to was previously suggested to depend on a process withinbe apically effluxed to a similar, low extent in both Caco-2 the enterocytes, since plasma protein binding does notand human jejunum, observations in line with those influence metabolism in the gut wall (Sandstrom andreported by others (Pauli-Magnus et al., 2000). Therefore, Lennernas, 1999). Further investigation of this issue willat present we can only speculate that the low intracellular require studies in a rat intestinal cell line expressing the ratconcentrations of norverapamil in Caco-2 cell monolayers counterpart to CYP3A4.may be related to analytical limitations and requires further Metabolism inside enterocytes was recently suggested toinvestigation. cause an enhanced apical uptake as a consequence of an

Based on the concentrations of the formed metabolite, increased concentration gradient (Li et al., 2002). In theenantioselective CYP3A4-mediated metabolism in the case of verapamil, we disagree with this suggestion, basedhuman jejunum was significantly higher than in Caco-2 on the following facts: (1) results from this in vitro studycells (Fig. 4) (Sandstrom et al., 1998a), which is in show that metabolism of verapamil has no influence onaccordance with a higher level of CYP3A4 expression P , i.e. there were no significant differences inPapp app

(Taipalensuu et al., 2001; Sun et al., 2002) and activity in between untreated and D3-treated cells; (2) in a human invivo (Fromm et al., 1996; von Richter et al., 2001). The vivo study, ketoconazole had no effect on the jejunalmetabolic capacity in the human jejunum is most likely permeability to verapamil (Sandstrom et al., 1999a); (3)

both R- and S-verapamil were metabolized to a greater extent in rats after rifampicin treatment while, on the same

experimental days, permeability was either unchanged ordecreased (Sandstrom and Lennernas, 1999); and (4)permeability to the two enantiomers was the same althoughthe gut wall metabolism was significantly higher for S-verapamil (Sandstrom et al., 1999a).

4 .3. Conclusions

This study shows that Caco-2 cells that expressCYP3A4 activity metabolize R/S-verapamil in an enan-tioselective manner however, less extensively than inhuman and rat jejunum. Verapamil and norverapamil weretransported mainly by passive diffusion in Caco-2 cellmonolayers, suggesting that the apical recycling hypothesis

Fig. 4. Formation of R- and S-norverapamil (NV) observed on themay not be applicable to the two compounds investigatedmucosal side in 1a,25-dihydroxy vitamin D (D3)-treated Caco-2 cell3in the present study. All three intestinal models classifiedmonolayers, compared to the levels observed on the luminal side in

§ §human (Sandstrom et al., 1998a) and rat jejunum (Sandstrom and verapamil and norverapamil as high permeability drugsLennernas, 1999), respectively. In our Caco-2 experiments and in the according to the BCS (Amidon et al., 1995), and they allperfusions of human jejunum (Sandstrom et al., 1998a), racemic ver- indicated that passive permeability to verapamil wasapamil was used at 8 and 80mM as indicated, while 200mM was used in

unchanged by intestinal CYP3A4-activity. Although R-the rat perfusions (Sandstrom and Lennernas, 1999). *Concentration ofand S-verapamil transport and metabolism were lessformed S-norverapamil significantly higher than that of R-norverapamil

(P,0.05). extensive in the Caco-2 cell model, data were qualitatively

64 H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65

ferential induction of prehepatic and hepatic metabolism of verapamilwell related to human jejunal data. The evaluation of thisby rifampin. Hepatology 24, 796–801.cell culture model against human and rat jejunal data

F romm, M.F., Kauffmann, H.M., Fritz, P., Burk, O., Kroemer, H.K.,makes it particularly useful in studies of drug transport andWarzok, R.W., Eichelbaum, M., Siegmund, W., Schrenk, D., 2000. The

metabolism, both in the preclinical and clinical evaluation effect of rifampin treatment on intestinal expression of human MRPof new drugs. transporters. Am. J. Pathol. 157, 1575–1580.

G reiner, B., Eichelbaum, M., Fritz, P., Kreichgauer, H.P., von Richter, O.,Zundler, J., Kroemer, H.K., 1999. The role of intestinal P-glycoproteinin the interaction of digoxin and rifampin. J. Clin. Invest. 104,147–153.A cknowledgements

H idalgo, I.J., Raub, T.J., Borchardt, R.T., 1989. Characterization of thehuman colon carcinoma cell line (Caco-2) as a model system for

We thank Drs Tommy B. Andersson, Ulf Bredberg and intestinal epithelial permeability. Gastroenterology 96, 736–749.¨Anna-Lena Ungell at AstraZeneca R&D Molndal, Sweden, H ochman, J.H., Chiba, M., Nishime, J., Yamazaki, M., Lin, J.H., 2000.

Influence of P-glycoprotein on the transport and metabolism offor valuable discussions. We are also grateful to Dr Andersindinavir in Caco-2 cells expressing cytochrome P-450 3A4. J.¨Karlsson at AstraZeneca R&D Molndal, Sweden, expert onPharmacol. Exp. Ther. 292, 310–318.¨enantioselective analysis, and to Par Mattson for skillful

H u, M., Li, Y., Davitt, C.M., Huang, S.M., Thummel, K., Penman, B.W.,technical assistance. This work was supported by Crespi, C.L., 1999. Transport and metabolic characterization of Caco-2

¨AstraZeneca R&D, Molndal, Sweden, and the Swedish cells expressing CYP3A4 and CYP3A4 plus oxidoreductase. Pharm.Res. 16, 1352–1359.Medical Research Council, grant 9478.

H unter, J., Hirst, B.H., Simmons, N.L., 1993. Drug absorption limited byP-glycoprotein-mediated secretory drug transport in human intestinalepithelial Caco-2 cell layers. Pharm. Res. 10, 743–749.

K oepsell, H., 1998. Organic cation transporters in intestine, kidney, liver,R eferences and brain. Annu. Rev. Physiol. 60, 243–266.

K roemer, H.K., Echizen, H., Heidemann, H., Eichelbaum, M., 1992.A dachi, Y., Suzuki, H., Sugiyama, Y., 2001. Comparative studies on in Predictability of the in vivo metabolism of verapamil from in vitro

vitro methods for evaluating in vivo function of MDR1 P-glycopro- data: contribution of individual metabolic pathways and stereoselectivetein. Pharm Res 18, 1660–1668. aspects. J. Pharmacol. Exp. Ther. 260, 1052–1057.

A midon, G., Lennernas, H., Shah, V., Crison, J., 1995. A theoretical basis K roemer, H.K., Gautier, J.C., Beaune, P., Henderson, C., Wolf, C.R.,for a biopharmaceutic drug classification: the correlation of in vitro Eichelbaum, M., 1993. Identification of P450 enzymes involved indrug product dissolution and in vivo bioavailability. Pharm Res. 12, metabolism of verapamil in humans. Naunyn Schmiedebergs Arch.413–420. Pharmacol. 348, 332–337.

A rtursson, P., 1990. Epithelial transport of drugs in cell culture. I: A L i, L.Y., Amidon, G.L., Kim, J.S., Heimbach, T., Kesisoglou, F., Topliss,model for studying the passive diffusion of drugs over intestinal J.T., Fleisher, D., 2002. Intestinal metabolism promotes regionalabsorptive (Caco-2) cells. J. Pharm. Sci. 79, 476–482. differences in apical uptake of indinavir: coupled effect of P-glycopro-

B rimer, C., Dalton, J.T., Zhu, Z., Schuetz, J., Yasuda, K., Vanin, E., tein and cytochrome P450 3A on indinavir membrane permeability inRelling, M.V., Lu, Y., Schuetz, E.G., 2000. Creation of polarized cells rat. J. Pharmacol. Exp. Ther. 301, 586–593.coexpressing CYP3A4, NADPH cytochrome P450 reductase and M akhey, V.D., Guo, A., Norris, D.A., Hu, P., Yan, J., Sinko, P.J., 1998.MDR1/P-glycoprotein. Pharm. Res. 17, 803–810. Characterization of the regional intestinal kinetics of drug efflux in rat

B usse, D., Cosme, J., Beaune, P., Kroemer, H.K., Eichelbaum, M., 1995. and human intestine and in Caco-2 cells. Pharm. Res. 15, 1160–1167.Cytochromes of the P450 2C subfamily are the major enzymes M andagere, A.K., Thompson, T.N., Hwang, K.K., 2002. Graphical modelinvolved in the O-demethylation of verapamil in humans. Naunyn for estimating oral bioavailability of drugs in humans and other speciesSchmiedebergs Arch. Pharmacol. 353, 116–121. from their Caco-2 permeability and in vitro liver enzyme metabolic

C respi, C.L., Penman, B.W., Hu, M., 1996. Development of Caco-2 cells stability rates. J. Med. Chem. 45, 304–311.expressing high levels of cDNA-derived cytochrome P4503A4. Pharm. P alm, K., Luthman, K., Ungell, A.L., Strandlund, G., Artursson, P., 1996.Res. 13, 1635–1641. Correlation of drug absorption with molecular surface properties. J.

E ichelbaum, M., Ende, M., Remberg, G., Schomerus, M., Dengler, H.J., Pharm. Sci. 85, 32–39.141979. The metabolism ofDL-[ C]verapamil in man. Drug Metab. P alm, K., Luthman, K., Ros, J., Grasjo, J., Artursson, P., 1999. Effect of

Dispos. 7, 145–148. molecular charge on intestinal epithelial drug transport: pH-dependentE ngman, H.A., Lennernas, H., Taipalensuu, J., Otter, C., Leidvik, B., transport of cationic drugs. J. Pharmacol. Exp. Ther. 291, 435–443.

Artursson, P., 2001. CYP3A4, CYP3A5, and MDR1 in human small P auli-Magnus, C., von Richter, O., Burk, O., Ziegler, A., Mettang, T.,and large intestinal cell lines suitable for drug transport studies. J. Eichelbaum, M., Fromm, M.F., 2000. Characterization of the majorPharm. Sci. 90, 1736–1751. metabolites of verapamil as substrates and inhibitors of P-glycoprotein.

F agerholm, U., Lennernas, H., 1995. Experimental estimation of the J. Pharmacol. Exp. Ther. 293, 376–382.effective unstirred water layer thickness in the human jejunum and its P olli, J.W., Wring, S.A., Humphreys, J.E., Huang, L., Morgan, J.B.,importance in oral drug absorption. Eur. J. Pharm. Sci. 3, 247–253. Webster, L.O., Serabjit-Singh, C.S., 2001. Rational use of in vitro

F agerholm, U., Johansson, M., Lennernas, H., 1996. Comparison between P-glycoprotein assays in drug discovery. J. Pharmacol. Exp. Ther. 299,permeability coefficients in rat and human jejunum. Pharm. Res. 13, 620–628.1336–1342. R aeissi, S.D., Guo, Z., Dobson, G.L., Artursson, P., Hidalgo, I.J., 1997.

F isher, J.M., Wrighton, S.A., Watkins, P.B., Schmiedlin-Ren, P., Calamia, Comparison of CYP3A activities in a subclone of Caco-2 cells (TC7)J.C., Shen, D.D., Kunze, K.L., Thummel, K.E., 1999. First-pass and human intestine. Pharm. Res. 14, 1019–1025.midazolam metabolism catalyzed by 1alpha,25-dihydroxy vitamin D3- S aitoh, H., Aungst, B.J., 1995. Possible involvement of multiple P-modified Caco-2 cell monolayers. J. Pharmacol. Exp. Ther. 289, glycoprotein-mediated efflux systems in the transport of verapamil and1134–1142. other organic cations across rat intestine. Pharm. Res. 12, 1304–1310.

F romm, M.F., Busse, D., Kroemer, H.K., Eichelbaum, M., 1996. Dif- S andstrom, R., Lennernas, H., 1999. Repeated oral rifampicin decreases

H. Engman et al. / European Journal of Pharmaceutical Sciences 19 (2003) 57–65 65

the jejunal permeability of R/S-verapamil in rats. Drug Metab. Dispos. of Human Duodenum and Caco-2 Gene Expression Profiles for 12,00027, 951–955. Gene Sequences Tags and Correlation with Permeability of 26 Drugs.

S andstrom, R., Karlsson, A., Lennernas, H., 1998b. The absence of Pharm. Res. 19, 1400–1416.stereoselective P-glycoprotein-mediated transport of R/S-verapamil T aipalensuu, J., Tornblom, H., Lindberg, G., Einarsson, C., Sjoqvist, F.,across the rat jejunum. J. Pharm. Pharmacol. 50, 729–735. Melhus, H., Garberg, P., Sjostrom, B., Lundgren, B., Artursson, P.,

S andstrom, R., Karlsson, A., Knutson, L., Lennernas, H., 1998a. Jejunal 2001. Correlation of gene expression of ten drug efflux proteins of theabsorption and metabolism of R/S-verapamil in humans. Pharm. Res. ATP-binding cassette transporter family in normal human jejunum and15, 856–862. in human intestinal epithelial Caco-2 cell monolayers. J. Pharmacol.

S andstrom, R., Lennernas, H., Ohlen, K., Karlsson, A., 1999b. Enan- Exp. Ther. 299, 164–170.tiometric separation of verapamil and norverapamil using Chiral-AGP T avelin, S., Grasjo, J., Taipalensuu, J., Ocklind, G., Artursson, P., 2002.as the stationary phase. J. Pharm. Biomed. Anal. 21, 43–49. Applications of epithelial cell culture in studies of drug transport.

S andstrom, R., Knutson, T.W., Knutson, L., Jansson, B., Lennernas, H., Methods Mol. Biol. 188, 233–272.1999a. The effect of ketoconazole on the jejunal permeability and T ucker, G.T., Houston, J.B., Huang, S.M., 2001. EUFEPS conferenceCYP3A metabolism of (R/S)-verapamil in humans. Br. J. Clin. report. Optimising drug development: strategies to assess drug metabo-Pharmacol. 48, 180–189. lism/ transporter interaction potential—towards a consensus. Eur. Fed.

S chmiedlin-Ren, P., Thummel, K.E., Fisher, J.M., Paine, M.F., Lown, Pharmaceut. Sci. Eur. J. Pharm. Sci. 13, 417–428.K.S., Watkins, P.B., 1997. Expression of enzymatically active CYP3A4 v on Richter, O., Greiner, B., Fromm, M.F., Fraser, R., Omari, T., Barclay,by Caco-2 cells grown on extracellular matrix-coated permeable M.L., Dent, J., Somogyi, A.A., Eichelbaum, M., 2001. Determinationsupports in the presence of 1alpha,25-dihydroxyvitamin D3. Mol. of in vivo absorption, metabolism, and transport of drugs by thePharmacol. 51, 741–754. human intestinal wall and liver with a novel perfusion technique. Clin.

S tenberg, P., Norinder, U., Luthman, K., Artursson, P., 2001. Experimen- Pharmacol. Ther. 70, 217–227.tal and computational screening models for the prediction of intestinal Y u, L.X., Amidon, G.L., Polli, J.E., Zhao, H., Mehta, M.U., Conner, D.P.,drug absorption. J. Med. Chem. 44, 1927–1937. Shah, V.P., Lesko, L.J., Chen, M.L., Lee, V.H., Hussain, A.S., 2002.

S un, D., Lennernas, H., Welage, L.S., Barnett, J.P., Landowski, C.P., Biopharmaceutics classification system: the scientific basis forFoster, D., Fleisher, D., Lee, K.-D., Amidon, G.L., 2002. Comparison biowaiver extensions. Pharm. Res. 19, 921–925.