--~--------------------------

Validation of Cell Culture Models forthe Intestine and the Blood-Brain Barrierand Comparison of Drug PermeationUdo Bock, Thomas Flototto and Eleonore HaltnerAcross Barriers GmbH, D-Saarbrticken

SummaryCell culture models are useful tools to study the uptake of drugsacross the barriers of the human body, like the intestine, theskin or the blood-brain barrier. Cell-based in vitro models notonly help to reduce the number of animals used but are alsomuch faster to perform, more cost effective and give morereproducible data than animal studies. Given the increasingnumber of new drugs and chemicals under development, thereis an urgent need for the establishment of such in vitro models.However; the validity of such in vitro models is reflected by itsability to accurately predict the behaviour of a substance at thecorresponding in vivo barrier. Here, we compare a well-established cell culture model for the intestine, based on Caco-2colon carcinoma cells, with a primary cell culture model ofthe blood-brain barrier. We find that Caco-2 cells and cellsof the blood-brain barrier have different barrier properties.Therefore, cells used for cell-based assays should be derivedfrom the corresponding tissue to reflect the in vivo barriercharacteristics.

Zusammenfassung: Validierung von Zellkultur-Modellen furden Darm und die Blut-Him-Schranke und Vergleich derArzneistoffpermeationZellkulturmodelle sind nutrliche Hilfsmittel, um die Aufnahmevon Arzneistoffen iiber natiirliche Barrieren des menschlichenKorpers, wie z.B. den Darm, die Haut oder die Blut-Hirn-Schranke, zu untersuchen. Zellbasierte in vitro Modelle helfennicht nur den Verbrauch von Versuchstieren zu redurieren,sondern sind auch schneller durchrufuhren, kosteneffektiverund liefern besser reproduiierbare Daten als Tierstudien. Auf-grund der steigenden Anzahl neuer Arzneimittel undChemikalien in der Entwicklung besteht ein groj3er Bedarf angeeigneten in vitro Modellen. Die Validitiit eines in vitroModells hdngt allerdings von dessen Fdhigkeit ab, das Ver-halten eines Wirkstoffes an der in vivo Barriere vorherzusagen.In der vorliegenden Studie wird ein gut etabliertes Zellkultur-modell for den Darm, basierend auf Caco-2 Kolon-KarzinomZellen mit einem Primdrzellmodel der Blut-Hirn-Schrankeverglichen. Wir konnten zeigen, dass Caco-2 Zellen undZellen der Blut-Hirn-Schranke unterschiedliche Barriere-Eigenschaften besitzen. Fur ein geeignetes Zellkulturmodellsollten Zellen eingesetzt werden, die ursprungiicb aus dementsprechenden Gewebe stammen, auf dessen Barriere-Eigenschaften sich das Zellkulturmodell bezieht.

Keywords: Caco-2, Porcine Brain Endothelial Cells (PBEC), blood-brain barrier (BBB), Biopharmaceutical ClassificationSystem (BCS)

1 Introduction

The German Federal Institute for Drugsand Medical Devices (BfArm) recentlyannounced a revision of the drug ap-proval process (bioavailability and bio-equivalence) as laid down in article 21 ofthe German Drug Law (AMG). The up-date was based on work carried out by anexpert group at the European Agency forthe Evaluation of Medicinal Products(EMEA) and published as a "Notefor Guidance on the Investigation ofBio-availability and Bioequivalence -CPMPIEWP/QWPIl401l98" (referred to

in the following as "Note for Guidance").After a transitional period of threemonths, which ended on the 26th of June2003, the new announcement came intoforce, replacing the previous 9th an-nouncement on bioavailability and bio-equivalence (CDER, 2000).The first German conference on the

subject of the Biopharmaceutical Classi-fication System (BCS) was held inMarch 2003. The conference, which wasattended by delegates from BfArm, fromthe German Pharmaceutical IndustryAssociation (BPI), from pharmaceuticalcompanies and from contract research

Received 22 December 2003; received in final form and accepted for publication 13 February 2004

ALTEX 21, Suppl. Linz 03/2004

organisations, provided an opportunity tosummarise German experience of theBCS and to formulate common expecta-tions of the system in Germany. The firstbiowaiver was issued by BfArm in 2002for the drug compound sotalol hy-drochloride. Work on this initial pilotscheme involved close collaboration be-tween the Committee on Generic Com-pounds at the BPI and members ofBfArm. BfArm representatives told dele-gates that ten requests for biowaivers hadbeen received over the last two years andthat the requests had been granted in50% of the cases.The BfArm decisions were based on

the EMEA Note for Guidance (EMEA,

57

BOCKETAL. ~----------------~-

1998), which is itself modelled on theguidance published in 2000 by the Foodand Drug Administration (FDA) of theU.S. Department of Health and HumanServices under the name "Waiver of InVitro Bioavailability and BioequivalenceStudies for Immediate-Release SolidOral Dosage Forms Based on a Biophar-maceutics Classification System". Incontrast to the FDA guidance, the EMEAguidance does not specifically exempthighly permeable, highly soluble com-pounds (so called FDA class I com-pounds). In principle, the Note for Guid-ance allows biowaivers to be issued forall classes of compounds provided thatappropriate supporting data is provided.According to the EMEA Note for Guid-ance, a request for a biowaiver that isbased on the use of alternative in vitroexperiments should incorporate a com-plete characterisation of the drug sub-stance (e.g. in vitro permeability, solubil-ity and physicochemical character-isation) and the drug product (content,release, etc.), and should take particularaccount of the effects of excipients usedin the formulation. If the metabolic sta-bility of the drug substance is an issue,the report submitted with the requestmust contain experimental data collectedby the applicant or drawn from the pub-lished literature.The conference agreed that given the

time and costs associated with in vivobioequivalence studies, the number ofrequests for biowaivers is expected torise. It was estimated that the costs of thein vitro methods and solubility measure-ments are about one tenth or one fifteenthof the costs associated with a full in vivobioequivalence study (approx. € 150,000).Furthermore, the time required to acquirethe data and to compile the requisite doc-umentation would decrease from monthsto weeks. Further scientific work leadingto a more differentiated view of drugclassification in terms of solubility andpermeability is desirable.

2 Aim

The US FDA and the European agencyEMEA have issued guidelines underwhich pharmaceutical companies canrequest a waiver from in vivo bioequiva-

58

lence studies based on the BCS (CDER,2000; EMEA, 1998). The theoretical ba-sis of the BCS system was first describedin 1995 (Amidon, 1995). BCS is a scien-tific framework for classification of sub-stances according to their aqueous solu-bility and their intestinal permeability(Yu, 2002). The BCS also takes the dis-solution of the drug product into accountand hence covers the three main factorswhich govern the rate and extent of drugabsorption from immediate-release (IR)solid oral dosage forms (e.g. tablets, cap-sules):-dissolution rate-solubility-permeabilityAccording to the BCS, drug sub-

stances can be classified as belonging toone of four classes:-Class 1: high solubility and highpermeability-Class 2: low solubility and highpermeability-Class 3: high solubility and lowpermeability-Class 4: low solubility and lowpermeabilityAs the pH solubility profile of the test

compound during passage through thegastrointestinal tract may influence drugdissolution rate and permeability, thesolubility profile should be determined inthe pH range of 1-7.5 at 37±1°C. In ad-dition, the stability of the drug compoundat different concentrations of physiologi-cal fluids that mimic gastrointestinalfluid and gastric juice must be taken intoaccount. The drug substance is classifiedas highly soluble when the highest dosestrength is soluble in 250 ml of aqueousmedia over the pH range 1-7.5.Both the FDA and EMEA (CDER,

2002; EMEA, 1998) recommend the useof monolayers of suitable epithelial cellsto classify the permeability of drug com-pounds. One epithelial cell line that hasbeen widely used as a model system formeasuring intestinal permeability isCaco-2 (Fogh et al., 1977; Hidalgo et al.,1989; Madara et al., 1987).In the presented study, a correlation

was established between the drug perme-ability and the fraction of drug absorbedin humans by determining the permeabil-ity using Caco-2 cells. In addition, thedependency of the permeability values

on the following parameters was studied:inter-day precision, the effects of drugapplication on apical-to-basolateral ab-sorption and on basolateral-to-apical ab-sorption, number of days in culture andpassage number.Another goal of the presented study

was the comparison of drug transportacross Caco-2 cells with transport acrossPorcine Brain Endothelial Cells (PBEC),a cell culture model for the blood-brainbarrier (BBB) established by Franke(1999). A special focus of the study wasto measure the activity of the effluxtransporter p-glycoprotein (PGP) andother active transporter proteins, usingthe PGP substrates rhodamine123 anddigoxin in the presence and absence ofthe PGP inhibitor verapamil.The results confirm that the Caco-2

and PBEC models can readily differenti-ate highly permeable compounds, whichare known to be well absorbed, from lesspermeable compounds that are poorlyabsorbed. In addition, we found strikingdifferences in the permeability of PGPsubstrates in the Caco-2 and PBECmodels. Our data indicate that thesedifferences are the result of different sub-strate specificities of different PGP-liketransporters found in the cell membranesof Caco-2 and PBEC. Thus, the cellsused to study drug uptake across aspecific barrier should be derived fromthe corresponding tissue to reflect thepertinent barrier properties.

3 Materials and methods

3.1 Caco-2 cellsCaco-2 cells were obtained from theGerman Cell Culture Collection DSMZ,DSMZ-No.: ACC 169. Aliquots of pas-sage 3 were stored in liquid nitrogen. Allcell cultures tested negative for my-coplasma (PCR testing) prior to cryocon-servation.Dulbecco's Modified Eagle's Medium

(DMEM), non-essential amino acids(NEA) , and gentamycin sulphate werepurchased from Biochrom KG (Berlin,Germany), while the trypsin-EDTA solu-tion was from Sigma Chemicals (Deisen-hofen, Germany). Fetal calf serum (FCS)was supplied by Greiner Labortechnik(Frickenhausen, Germany). The refer-

ALTEX 21, Suppl. Linz 03/2004

m...... BOCK ET AL.--~--------------------------

ence compounds were from local suppli-ers and were of the highest chemical-grade purity.Caco-2 cells were cultured at 37°C,

90% humidity, 10% C02 in cell cultureflasks in DMEM containing 10% FCSand antibiotics. The medium waschanged three times a week, and cellswere passaged once per week until con-fluence was about 80%. Caco-2 cellswere detached by means of trypsin-EDTA and seeded with 104cells/em- onTranswell ™ clear filters (Costar", Wies-baden, Germany) with an area of 1.13 em-and a pore size of 0.4 um, Monolayergrowth was monitored weekly by mea-suring the transepithelial electrical resis-tance (TEER) values. TEER values ofmonolayers were calculated by subtract-ing the TEER of empty filters from themeasured TEER.

3.2 PBECPBEC were prepared as described byGindorf et al. (Gindorf et al., 2001).Cells were cultured at 37°C, 90% humid-ity, 10% CO2 in cell culture flasks inM199 medium with 10% Ox serum (OS)and antibiotics. On the third day afterpreparation, the cells were subcultured.The cells were harvested by trypsinationand seeded on collagen-treated poly-carbonate filters. After three days onTranswell " filters, the M199 mediumwas replaced by DMEM/Ham's F12medium. A volume of 500 III was addedper apical compartment and a volume of1500 IIIwas applied per basolateral com-partment, respectively.

3.3 Quality control ofmonolayer batchesThe quality of a batch of cell monolayerswas monitored by measuring the TEERof a subgroup of randomly selectedmonolayers and by measuring their per-meability with respect to three test com-pounds. Quality control measurementswere performed in triplicate for each celltransport condition (e.g. transport direc-tion, presence or absence of inhibitors).The following quality criteria had to bemet before the Caco-2 and PBEC mono-layer batches were released for permeabil-ity studies with test compounds (Tab. 1).

3.4 TestcompoundsTest compounds were dissolved in DMSOand diluted with KRB (pH 7.4) or cellculture medium (DMEMlHam's F12, forPBEC studies) to a concentration of 5011M,respectively « 1%DMSO final con-centration). The solutions were warmedto 37°C and used as transport solution.Note that PBEC have been shown to beresistant to DMSO up to a concentrationof 1%.

3.5 Caco-2 permeability studiesImmediately before the experiment, thecells were washed twice with KRBbuffer. The buffer was then replaced withKRB buffer containing the test com-pound. After 30-35 minutes pre-incuba-tion, samples were withdrawn from thedonor and acceptor compartments. Thefirst sampling was defined as the start ofthe experiment (t = 0 min). During thetransport study further samples were

withdrawn from the acceptor compart-ments at defined times. The volumeswithdrawn were replaced with freshKRB buffer or KRB buffer with the cor-responding inhibitor. When not beingsampled, the monolayers were incubatedin a CO2 incubator. At the end of thestudy a second sample was taken fromeach donor compartment. Finally, thetransepithelial electrical resistance(TEER) was remeasured; a TEER valueabove 200 Q x em' (not including theTEER of the empty filters) indicated themonolayer was intact.The volumes of the apical and basolat-

eral compartments used in the transportexperiments were 500 j.ll and 1500 j.llrespectively. The recovery of the testcompounds was calculated by comparingpre-incubation samples with sampleswithdrawn at the end of the transportstudy.

3.6 PBECpermeabilitystudiesA PBEC monolayer batch is defined asPBEC isolated, seeded and cultured inparallel under the conditions describedabove on Transwell ™ filters. All trans-port studies were performed on PBECmono layers in triplicate. For the trans-port studies, half of the cell culture medi-um from the donor compartments wasremoved and replaced with transportsolution. The pre-incubation process,sampling and TEER determination wereperformed as described for Caco-2 cells.The acceptance criteria for PBEC aredepicted in Table 1.

Tab. 1: Quality control criteria for Caco-2 and PBEC cell monolayer batches. Qualification criteria have to be met by every monolayer.

Test item Caco-2 PBEC

Number of passages val.idated 10-50 Individual preparations, no passagesfor experiments '"Age of monolayer 14-30 days 4-7 days

Low permeability Pappof fluorescein (ab) < 1-10'6 cm/s Pappof fluorescein (ab) <: 1-10-6 cm/s

High permeability Pappof propranolol (ab) >5.10-6 cm/s Pappof propranolol (ab) >5.10.6 cm/s

Expression of P-glycoprotein ~Papprlioaamine (ba)/apprlioaamTne (ab) >2:(J -Papprlioaamine (ba)/P.,pp rhoaamine (a5) > 2.0 -Tightness of barrier TEER of mono layers TEER of monolayersbefore transport (not including TEER of empty filters) (not including TEER of empty filters)

> 200 Q·cm2 > 500 Q·cm2

Tightness of barrier TEER TEERafter transport (not including TEER of empty filters) (not including TEER of empty filters)

> 200 Q·cm2 > 500Q'cm2

ALTEX 21, Suppl. Linz 03/2004 59

BOCK ET AL. m.....---------------------------------------~---

3.7 Analysis of samplesThe fluorescent compounds were anal-ysed by fluorescence measurement (Vie-tor", Wallac) and radiolabelled com-pounds were measured by liquidscintillation counting (Wallac 1450 Mi-crobeta, Wallac). All other compoundswere analysed by LC-PDA (WatersAlliance 2790). All test methods hadbeen validated for each compound.

3.8 Permeability calculationThe apparent coefficient of permeation(Papp)was calculated using the followingequation:

Eq.l dQ 1 1Papp=_.-.-. VDonor; [cmls]dt IDo A

dQ/dt: permeability rate (steady statetransport rate) obtained from thetransport-time profile of the substrate[e.g. counts/s]A: area of the exposed cell monolayer[em"]mo; the original mass of the markersubstance in the donor compartment[e.g counts]V Donor: volume of donor compartment[cm3

].

4 Results

4.1 Correlation of Caco-2permeability coefficients and"absorbed fraction" data formodel compoundsTo demonstrate that Caco-2 cell mono-layer permeability correlates well within vivo absorption, the Pappvalues for 19model compounds were plotted againsttheir published fractional absorptionvalues (Fa) in humans (Fig. 1). Thesources of human absorption data arelisted in references (Artursson, 1991;Cheong et aI., 1999; Goodman, 1999;Rubas et aI., 1993; Sweetman, 2000; Yee,1997; Zhao et aI., 2001). We found agood correlation between the absorbedfractions in humans and the apparentpermeability co-efficients for most testcompounds (Fig. 1).The reference compounds fluorescein

and rhodaminel23, which were used todemonstrate the intactness of the mono-layer and PGP activity, respectively,were not included in Figure 1, because

60

the relevant Fa data was not available inthe literature. The permeability coeffi-cient of [3H]PEG4000 (compound 11),which has a known low Fa value (zeromarker), was found to be 5.00 x E-06cmls in the Caco-2 system. This high

value may have been due to impurities oflower molecular weight, or to the insta-bility of the radioactive labelling (PEGSupplier, personal communication). Inthe case of radiolabelled digoxin (com-pound 14), the literature values for the

Tab. 2: Comparison of the apparent permeation coefficients of model drugs and theBCS permeability class. The Papp values are mean values from experiments performed intriplicate. Data were determined at cell passage 14.

No. Compound Papp ab [cm/s] BCS permeability class

120

moderate

100

high80

··············11]···················60

40

o

0,1 1 10in vitro permeability [cm/s] x E-06

100

Fig. 1: Comparison of the apparent permeation coefficients of model drugsmeasured in the Caco-2 cell monolayer system with absorbed fractions in humans.Data were determined at cell passage 14.

ALTEX 21, Suppl. Linz 03/2004

~ BOCKETAL.-~--------------

rhodamine 123

propranolol

(14Cjmannitol

(3Hjatenolol

fluorescein

(3HlPEG 4000

(3HjPEG 900

(3HPEG 400

10~ C]abC]ba

10'

4 6 8101214161820

ratio of permeation coefficients [balab I

Fig. 2: Ratios (ba/ab) of the permeation coefficients of modeldrugs in the Caco-2 cell monolayer system. The data are meandata of experiments performed in triplicate. Data were determinedat cell passage 14.

absorbed fraction in humans vary be-cause of its interaction with PGP (Hidal-go et al., 1989). A weak correlation of theabsorbed fraction in humans with Caco-2permeability was also found for atenololand furosemide. For both compounds,the excipients of the oral dosage formhave been demonstrated to increase per-meability greatly (Kanfer, 2002).

4.2 Caco-2 monolayersexhibit apical eHlux due top-glycoproteinTo further validate the suitability of theCaco-2 in vitro model, the monolayerswere tested for PGP activity. Rho-damine123 and digoxin are substrates ofPGP, and verapamil is a PGP inhibitor(Cheong et a1., 1999). The ratio of theP app values from the basolateral side(which corresponds with the blood circu-lation side) to the apical cell side (re-presenting the intestinal lumen) to thetransport rate in the reverse directiondepends on the operating transport mech-anisms. The ratios for selected com-pounds are summarised in Figure 2.

A clear directionality was found fordigoxin and rhodamine123 only. None ofthe other test compounds demonstrateda clear directionality in their transport.

4.3 PGP inhibition by verapamilin Caco-2 cellsAt a specific Caco-2 cell monolayer pas-sage, the apparent permeation coefficients

ALTEX 21, Supp!. Linz 03/2004

without inhibitor 40 flM verapamil

Fig. 3: Comparison of the apparent permeation coefficientsof digoxin in the absence and presence of 40 IJM verapamil.The data are mean data from experiments performed intriplicate. Data were determined at cell passage 11.

of digoxin were measured with and with-out 40 f1M verapamil in the assay buffer.The data are summarised in Figure 3.

Without inhibitor, digoxin exhibits asubstantial Papp (ba) to Papp (ab) transportpolarity ratio of 12.5. In the presence ofthe PGP inhibitor verapamil, the trans-port polarity of digoxin is reduced to avalue of 1.3. These results confirm thatPGP enhances apical efflux activity.

4.4 Caco-2 cell passagenumber and its impact on thepermeability of selectedcompoundsFigure 4 shows the permeability offluorescein and propranolol as a functionof various cell passage numbers inrelation to the quality criteria listed inTable 1. The permeability results donot appear to be dependent on passage

100

D Papp

fluorescein (low permeable)o P

apppropranolol (high permeable)

~ 10.£.f?:0ro(l)ECDa.

g.s;c;

1 ..upper limit forfluorescein (ab)

o 5 W ~ W ~ ~ ~ ~ ~ ~

week (different cell passages)

Fig. 4: Comparison of the apparent permeation coefficients of the model drugspropranolol and fluorescein for different passages. The values are mean values fromexperiments performed in triplicate.

61

B~m~ ~--------------~-

number. The Caco-2 cells gave repro-ducible data over 40 passages.

4~5 Comparison of Caco-2and PBECpermeabilityTo compare Caco-2 and PBEC perme-ability, the permeability data of selectedcompounds in PBEC were plottedagainst the permeability data of the samecompounds in Caco-2 cells (Fig. 5).Our data demonstrates similar per-

meability in both cell types for propra-nolol, clonidin, fluorescein and atenolol.However, the permeability found forrhodamine and digoxin were significant-ly different in the two cell types. Whilethe permeability found for rhodaminewas higher in PBEC, the digoxin perme-ability was higher in Caco-2.

4.6 Apical efflux ofrhodamine in PBECand itsinhibition by verapamilThe difference in rhodamine permeabili-ty in Caco-2 and PBEC may be the resultof different PGP activities in the two celltypes. To test this hypothesis, the PBECmonolayers were tested for PGP activityusing rhodamine. Differences in rho-damine permeability between transportfrom the apical to the basolateral (ab)side and transport from the basolateral toapical side (ba) are a result of polarisedactive transporters involved in the trans-port process. The rhodamine permeabili-ty for three independent PBEC prepara-tions is summarised in Figure 6.Rhodamine permeability was found to

be within the same range for all threepreparations. The permeability for theab transport ranged between 0.5-0.7 xE-6, while the ba permeability wasfound to be in the range of 1.5-1.8 xE-6. Similar rhodamine permeabilitywas found for other preparations (datanot shown) when the qualification crite-ria were met (Table 1). Therefore, thepreparation method used for PBEC cellsreproducibly allows the preparation ofcells with the same barrier properties. Inaddition, the ba transport was repro-ducibly found to be 2-3-fold faster thenthe ab transport, demonstrating thattransporter activity doesn't change withdifferent preparations. The effect wasinhibited by 40 11M of the PGP inhibitorverapamil, demonstrating that PGP-like

62

transporters are involved. Given thelargely reduced apical efflux of rho-damine in PBEC in comparison toCaco-2 cells, the higher rhodamine per-meability in PBECs (Fig. 5) may be theresult of the significantly higher PGPactivity found in Caco-2 cells.

4.7 Comparison of apical effluxof rhodamine and digoxin in PBECTo test whether a higher efflux is foundfor other PGP compounds in Caco-2cells in comparison to PBEC, we alsostudied the efflux of digoxin in PBEC.The ratio of the permeability of rho-

100

10

~£ 1owma..o..i

0,1

Propranolol ....

~.:' ""

Cloni~lR".·'0

RhodornineHI-t •

RuorescelQ .:Atenolol -: cj

~Digoxin

~

0,01 +-..--.•.........,...,.'TTT',.----,.--,.....,..,I'"T'TT'!'T"--,--.-rrrrr,.,--,--,r-r.-r".."0,01 0,1 1

Papp

Caeo-2 [cm/s]

10010

Fig. 5: Comparison of permeation coefficients of model drugs in the PBEC andCaco-2 monolayer system. The data are mean data of experiments performed intriplicate (error bars indicate SD).

2,0

'I'0 1,8

x1,6

~E 1,4.£~ 1,2

:.a1,0co

Q)

E 0.8'-Q)c..0,6

0'-:;::; 0,4>.5:

0,2

0,0Preparation A

[=::J ab transport[=::J ba transportE::J ab transport + 40 fJM Verapamil_ ba transport + 40 fJM Verapamil

Preparation B

Fig. 6: Comparison of the apparent permeation coefficients of rhodamine123 inthe PBEC monolayer system in the absence or presence of PGP inhibitor verapamil(40 I-IM). The data are mean data of experiments performed in triplicate. Data weredetermined on three different PBEC preparations (A, B, C) to demonstrate reproducibilityof results.

ALTEX 21, Suppl. Linz 03/2004

m.... BOCK ET AL.-~-------------

damine and of digoxin in Caco- 2 andPBEC is shown in Figure 7.Interestingly, the ratio of ab to ba

transport for digoxin was found to bewithin the same range for both cell types(Caco-2, 8-fold; PBEC, 7.2-fold). In con-trast to this finding, the rhodamine effluxdiffered greatly for both cell types (Caco-2, 12-fold; PBEC, 2-fold). In conclusion,the differences found in the apical effluxof rhodamine in the two cell modelscannot be the result of different PGP ex-pression alone, but must also be theresult of different substrate specificitiesof active transporters. Note that the per-meability of digoxin and rhodaminewas measured in the same preparation ofPBEC and the same passage of Caco-2cells respectively.

5 Discussion

According to the recommendation of theFDA and EMEA, the Caco-2 cell culturemodel was validated for the classificationof compounds in terms of intestinal per-meability. Several parameters of theCaco-2 cell permeability assay wereevaluated to assess the robustness of thesystem. Caco-2 cell monolayers can beused to study compounds with perme-ability ranging from 1.77 x E-07 cmls(e.g. mannitol) to 3.42 x E-05 cmls (e.g.propranolol), enabling the Caco-2 sys-tem to discriminate between high andlow permeability compounds as classi-fied in the BCS. Caco-2 cell monolayersshow directional transport for the PGP

:0 14~C\le. 12$lc.~ 10if:Q)o 8UCo:;:;6C\lQ)

EQj 4o,

15o'';:::;C\l0:: 0

substrates digoxin and rhodamine 123,with a significantly higher basolateral-to-apical transport than in the reverse di-rection. Directional transport can be in-hibited by the PGP inhibitor verapamil ata concentration of 40 flM. The perme-ability determined for many of the mod-el compounds correlated well with re-ported absorption in humans. A weakcorrelation was found only for the PGPsubstrates digoxin and furosemide. TheCaco-2 cell monolayers demonstrated avery low inter-day variability for mono-layers in culture. The inter-day variabili-ty was always lower than the measuredinter-compound differences.Caco-2 cell monolayers could be used

for more than 40 passages without no-ticeable changes in the permeability ofthe model compounds (Fig. 4). The pre-sent Caco-2 system is therefore a validtool to determine the gastrointestinalpermeability of a test compound in ac-cordance with the FDA's BCS. There-fore, in accordance with the 3R principle(Russell and Burch, 1959), the Caco-2cell assay not only reduces the complex-ity of uptake studies, but also refinestheir reproducibility and can therefore beused to replace animal studies in the drugapproval process.Comparing the in vitro permeability of

different test compounds in Caco-2 cellsand PBEC reveals similarities betweenthese cell culture models. As demonstrat-ed for Caco-2 cells, PBEC can be used tostudy compounds with a wide range ofpermeability ranging from 5.2 x E-08 to3.4 x E-05. Furthermore, the permeabili-

c:::=J Caco-2 cellsC]PBECcelis

\Rhodamine Dlqoxin

Fig. 7: Ratio of permeation coefficients of rhodamine and digoxin in Caco-2 andPBEC cells.

ALTEX 21. Suppl. Linz 03/2004

ty data obtained for non-PGP substrateswas within the same range for both cellculture models (Fig. 5). In addition, asdemonstrated for Caco-2 cells, a cleardirectionality was found for the PGPsubstrate rhodamine123 (Fig. 7),although the PBECs showed only a two-fold higher basal-to-apical transport thanin the reverse direction (Caco-2 cells, 12-fold). Permeability and directionalitywere shown to be within the same rangefor individual preparations. Our datatherefore demonstrates the good repro-ducibility of the preparation method.Furthermore, directionality was inhibitedby adding verapamil, proving the in-volvement of PGP-like transporters.However, comparing the permeability

data obtained from Caco-2 and PBEC al-so revealed some striking differences be-tween Caco-2 and PBEC (Fig. 5). Inter-estingly, these differences were foundexclusively for the PGP substrates rho-damine and digoxin. While the perme-ability ofrhodamine123 was found to beabout twice as high in PBEC, the perme-ability of digoxin was found to be ten-fold lower in Caco-2 cells. Comparingthe PGP activity in Caco-2 and PBEC(Fig. 7) indicated that differing rho-damine permeability might be the resultof different activity of active trans-porters. The different directionalityfound for rhodamine in Caco-2 andPBEC is unlikely to be the result of thetotal amount of receptor found in Caco-2 cells, since a similar degree of direc-tionality was found for digoxin in Caco-2 and PBEC. Therefore, the differencesfound for rhodamine permeability arelikely to be the result of different sub-strate specifics of different PGP-liketransporters in Caco-2 and PBEC. How-ever, digoxin also had different perme-ability in Caco-2 and PBEC (Fig. 5), andsimilar PGP activities were found inboth cell types (Fig. 7). Therefore PGPactivity alone cannot explain the differ-ence in digoxin permeability found forboth cell types. The reason for thediffering digoxin permeability is un-clear, especially since both cell typesdemonstrate similar permeability fornon-PGP compounds. A different degreeof metabolisation may be involved. Fur-ther experiments are needed to explainthis observation.

63

BOCK ET AL. ~-------------------------~--

In conclusion, our data demonstratethat the Caco-2 cell model can be used tostudy drug uptake via the intestine in thedrug approval process. The PBEC cul-ture model gives similar results for non-PGP drugs, but striking differences werefound for the uptake of drugs that aresubstrates of active transporters. Ourdata therefore demonstrate the necessityof specific test systems for differentepithelial barriers.

ReferencesAmidon, G. L., Gordon, L., Lennernas,

H. et al. (1995). A theoretical basis fora biopharmaceutic drug classification:the correlation of in-vitro drug productdissolution and in-vivo bioavailability.Pharmaceut. Res. 12 (3), 413-420.

Artusson, P. and Karlsson, J. (1991).Correlation between oral drug absorp-tion in humans and apparent drugpermeability coefficients in human in-testinal epithelial Caco-2 cells.Biochem. Biophys. Res. Comm. 175,880-85.

Cheong, J., Grove, R., Irvine, J. D. et al.(1999). MDCK (Madin-Darby CanineKidney) Cells: A tool for membranepermeability screening. 1. Pharm. Sci.88,28-33.

Fogh, J., Fogh, J. M. and Orfeo, T.(1977). One hundred and twenty-sevencultured human tumor cell lines pro-ducing tumors in nude mice. Journalof the National Cancer Institute 59,221-226.

Franke, H. (1997). Die Blut-HimSchranke in-vitro: Etablierung einesZellkulturmodells zur Prufung derHirngangigkeit von Pharmaka. Disser-tation, Westfalische Wilhelmsuniver-sitar MUnster.

Gindorf, C., Steimer, A., Lehr, C. M. et

64

al. (2001). Markertransport uber biolo-gische Barrieren in vitro: Vergleichvon Zellkulturmodellen fur dieDunndarmschleimhaut, die Blut-HirnSchranke und das Alveolarepithel derLunge. ALTEX 18, 155-64.

Goodmann, L. S., Gilman, A., Hardman,J. G. et al. (Ed.) (1999). Goodman andGilman's The Pharmacological Basisof Therapeutics. Columbus, New York:McGraw-Hill Professional.

Guidance for Industry (2000). Waiver ofIn Vitro Bioavailability and Bioequiva-lence Studies for Immediate-ReleaseSolid Oral Dosage Forms Based on aBiopharmaceutics Classification Sys-tem. U.S. Department of Health andHuman Services, Food and Drug Ad-ministration, Center for Drug Evalua-tion and Research (CDER). I

Note for Guidance on the Investigationof Bioavailability and Bioequivalence(1998). CPMP/EWP/QWP/~401198(EMEA). I

Hidalgo, 1. J., Raub, T. J. and Borchardt,R. T. (1989). Characterization I of thehuman colon carcinoma cell line(Caco-2) as a model system for intesti-nal epithelial permeability. Gastro-enterology 96 (3),736-749. I

Kanfer, 1. (2002). Report on the inter-national workshop on the biopharma-ceutics classification system /(BCS):Scientific and regula~ory as~ects inpractice. J. Pharm. SCI.5 (1), ~-4.

Madara, J. L., Stafford, J., pharm-sathaphom, K. and Carlson, S. (1987).Structural analysis of a human/ intesti-nal epithelial cell line. Gastroenterol.92, 1133-1145.

Moller, H. (2002). Praktische An-wendung des biopharmazeutischenKlassifizierungssystems. Pharm. Ind.64(4),330-332.

Rubas, W., Jezyk, N. and Grass, G. M.

(1993). Comparison of the Permeabili-ty Characteristics of a Human ColonicEpithelial (Caco-2) Cell Line to Colonof Rabbit, Monkey, and Dog Intestineand Human Drug Absorption. Pharm.Res. 10, 113-118.

Russell, W. M. S. and Burch, R. L.(1959). The Principles of HumaneExperimental Technique. London:Methuen and Co. Ltd.

Sweetman, S. (Ed.) (2000). Martindale33rd edition: The Complete Drug Ref-erence. London: Pharmaceutical Press.Electronic version, Micromedex,Englewood; CO.

Yee, S. (1997). In vitro permeabilityacross Caco-2 cells (colonic) canpredict in vivo (small intestinal) ab-sorption in man - fact or myth?Pharm. Res. 14, 763-66.

Yu, L. X., Amidon, G. L., Polli, J. L. etal. (2002). Biopharmaceutical classifi-cation system: The scientific basis forbiowaiver extensions. Pharm. Res. 19,921-925.

Zhao, Y. H., Le, J., Abraham, M. H. et al.(2001). Evaluation of human intestinalabsorption data and subsequent de-rivation of a quantitative structureactivity relationship (QSAR) withthe Abraham descriptors. 1. Pharm.Sciences 90 (6), 749-84.

Correspondence toDr. Eleonore HaltnerAcross Barriers GmbHScience Park SaarPostfach 15 11 61D-66123 SaarbruckenGermanyphone: +49-(0)-681-95918800fax:+49-(0)-681-95918802e-mail: info@acrossbarriers.dewww.acrossbarriers.de

ALTEX 21, Supp!. Linz 0312004