Brief communication

Optimizing Caco-2 cell monolayers to increase throughput in drug

intestinal absorption analysis

Maria Markowskaa,*, Rebecca Oberlea, Steve Juzwina, Cheng-Pang Hsua,Margaret Gryszkiewiczb, Anthony J. Streetera

aDepartment of Drug Metabolism, The R.W. Johnson Pharmaceutical Research Institute, Route 202, P.O. BOX 300, Raritan, NJ 08869, USAbRutgers, The State University of New Jersey College of Pharmacy, 160 Frelinghuysen Road, Piscataway, NJ 08854-8020, USA

Received 11 October 2001; accepted 19 November 2001

Abstract

Introduction: The aim of this investigation was to evaluate methods for increasing Caco-2 cell throughput for assessing drug intestinal

absorption. The use of 6-, 12-, and 24-well membranes and the effect of membrane size on permeability and the integrity of the Caco-2 cell

monolayer were assessed. In an effort to optimize the assessment of drug permeability, increased throughput was investigated by testing

compounds singly or as mixtures of analytes. Method: The transepithelial electrical resistance (TEER) of cell monolayers was measured on

0.33, 1.0, and 4.7 cm2 polycarbonate membranes using EVOM, over a 25-day period. Absorptive transport was determined on all compounds

tested using LC-MS/MS assays, or liquid scintillation spectrometry. Results: The effect of multiple compounds in one well compared to

single compounds was assessed with atenolol, nadolol, metoprolol, and propranolol for mixtures of four compounds and with RWJ-53308,

atenolol, terbutaline, propranolol, naproxen, piroxicam, topiramate, and furosemide for mixtures of eight compounds. The apparent

permeability (Papp) values correlated well between single analytes and mixtures of four and eight analytes in each well. Drug permeability

decreased slightly with an increase in well size. The TEER value increased with the number of days in culture for each of the 6-, 12-, and 24-

well sizes. Discussion: It was demonstrated that the 24-well format system is ideal for high-throughput assessment. Furthermore, the

approach of mixing four or eight analytes in each well to further increase throughput was also demonstrated to be valid. D 2001 Elsevier

Science Inc. All rights reserved.

Keywords: Caco-2 monolayers; High-throughput screening; Intestinal transport; LC-MS/MS analysis; Methods; Permeability; Polycarbonate membrane

1. Introduction

Today, Caco-2 cell monolayers have gained enormous

popularity as a reliable and high-throughput in vitro model

system for the evaluation of a large number of drug candi-

dates for their intestinal absorption potential (Artursson &

Karlsson, 1991; Chong, Dando, Soucek, & Morrison, 1996).

For orally administered compounds, permeability through

Caco-2 cell monolayers correlates well with in vivo absorp-

tion in man (Artursson & Karlsson, 1991; Artursson, Palm,

& Luthman, 1996). Experimentally, the Caco-2 screening

assay includes two consecutive procedures, a transport ele-

ment and a subsequent sample analysis. The transport

experiment for new compounds is extensive and requires

optimization. The preparation of a fully differentiated con-

fluent Caco-2 cell monolayer, however, generally requires a

3-week cell culture period with 9–10 intensive cell feedings.

In order to expedite the discovery of orally available drugs,

several approaches to increase the efficiency of this tech-

nique for drug screening have been attempted. Our study

describes unique approaches for increasing productivity by

evaluating Caco-2 performance under the following condi-

tions: (i) single analytes and mixtures of four and eight

analytes per well tested, (ii) permeability values determined

in 6-, 12-, and 24-well formats, (iii) the integrity of cells

evaluated in 6-, 12-, and 24-well formats by the measure-

ments of transepithelial electrical resistance (TEER), and (iv)

the permeability properties evaluated in 6-, 12-, and 24-well

formats with the paracellular marker, mannitol, and several

passively transported compounds. Results from this study

demonstrate that these approaches can be used to optimize

1056-8719/01/$ – see front matter D 2001 Elsevier Science Inc. All rights reserved.

PII: S1056 -8719 (01 )00161 -7

* Corresponding author. Tel. +1-908-704-4765; fax: +1-908-704-8412.

E-mail address: mmarkows@prius.jnj.com (M. Markowska).

Journal of Pharmacological and Toxicological Methods 46 (2001) 51–55

permeability assessment throughput of certain types of drugs

in the early phases of drug discovery.

2. Materials and methods

2.1. Materials

14C-Mannitol, 3H-propranolol, 14C-testosterone were ob-

tained from NEN (Boston, MA). 3H-Cimetidine and 14C-

PEG 4000were obtained fromAmersham Pharmacia Biotech

(Little Chalfont, Buckinghamshire, England).

RWJ-53308 ([S-(R*,S*)]-b-[[[1-[1-oxo-3-(4-piperidinyl)-propyl]-3-piperidinyl]carbonyl]amino]-3-pyridinepropanoic

acid and topiramate were obtained from The R.W. Johnson

Pharmaceutical Research Institute (RWJPRI), Raritan, NJ.

Atenolol, nadolol, furosemide,metoprolol, propranolol, cime-

tidine, naproxen, piroxicam, and terbutaline were obtained

from Sigma (St. Louis, MO).

All tissue culture reagents and media were obtained from

Gibco BRL Life Technology (Grand Island, NY). All flasks

were obtained from VWR Scientific Products (Bridgewater,

NJ). The transwell polycarbonate membranes inserts with

areas of 4.7 cm2, 1.0 cm2, and 0.33 cm2 (0.4 mm pore size)

were obtained from Corning Science Product Division

(King of Prussia, PA). The epithelial tissue voltohmmeter

(EVOM) was obtained from World Precision Instruments

(Sarasota, FL).

2.2. Cell culture

The Caco-2 cell line was obtained from American Type

Culture Collection (Manassas, VA). Cells were seeded at

6.3� 104 cells/cm2 and grown in a medium comprised of

Dulbecco’s Modified Eagle’s Medium (DMEM) containing

4.5 g/l glucose and supplemented with 10% (v/v) fetal bovine

serum (FBS), 1% (v/v) glutamine, penicillin (100 U/ml),

streptomycin (100 mg/ml), and 1% Minimum Essential

Medium (MEM) nonessential amino acids. Cultures were

maintained at 37 �C in an atmosphere of 95% air and 5%CO2.

The medium was changed every 2–3 days. Routine passing

of cell stocks was carried out in 75-cm2 flasks. Studies were

conducted on Caco-2, passages #20–35.

2.3. TEER measurements during growth

The TEER of the cell monolayers was measured over a

25-day period. Cells were seeded at 6.3� 104 cells/cm2 on

0.33 cm2, 1.0 cm2, and 4.7 cm2 polycarbonate membranes

with a pore size of 0.4 mm. Prior to TEER measurements,

Hanks’ Balanced Salt Solution (HBSS) with 25 mM HEPES

pH 7.4 was added to the apical and basolateral membranes

and allowed to equilibrate to 37 �C for 25 min. The TEER

of each monolayer was measured at Days 4, 8, 10, 16, 21,

and 25 using an EVOM. The intrinsic resistance of the

system (insert alone) was subtracted from the total resist-

ance (cell monolayer plus insert) to yield the monolayer

resistance. The resistance was expressed as V� cm2.

2.4. Transport study

All transport studies were conducted at 37 �C, in a

medium of HBSS, pH 7.4, containing 25 mM HEPES. Prior

to the transport study, cell monolayers were washed with the

transport buffer and preincubated for 25 min. Test com-

pounds at final concentrations of 10 mM in HBSS were

added to the apical compartment. The basolateral compart-

ment contained only HBSS. All transport studies were

conducted in the apical (A) to basolateral (B) direction. At

selected times, ranging from 10 to 60 min, 100–500 ml ofsamples were removed from the basolateral compartment

for analysis and replaced with an equal volume of HBSS.

The test compound (50–500 ml) (donor solution) was re-

moved for analysis. Samples were analyzed using liquid

chromatography mass spectrometry (LC-MS/MS) and liquid

scintillation spectroscopy.

The initial flux of the drug ( J ) was determined from the

slope of the linear plot of the cumulative amount of drug

transported versus time. Permeability was estimated by

calculating the apparent permeability coefficient (Papp)

according to (Artursson & Karlsson, 1991):

Papp ¼ J=ASC0

where C0 is the initial concentration in the donor compart-

ment and AS is the surface area of the monolayer.

Potential depletion of the donor drug concentration during

the course of the study was monitored by analysis of the

donor solution at the termination of each study. Two con-

ditions were required to be satisfied for the equation to be

accurately applied: (i) sink conditions in the receiver com-

partment, i.e., the accumulated concentration must be less

than 10% of the donor concentration at all times within one

hour after experiment initiation, and (ii) the donor concen-

tration must remain within ± 10% of the initial donor con-

centration throughout the experiment.

Standards were prepared in HBSS containing 25 mM

HEPES. To minimize ion suppression due to the matrix

effects from HBSS, and to eliminate manual preparation, an

on-line extraction method was developed involving column

switching. This served to remove the high salt concentra-

tions in the biological buffer. After a 1-min desalting period,

the LC column eluate was switched to an analytical column.

Valve switching prevented the early eluting salts from

entering and contaminating the LC/MS interface. Detection

was accomplished on a PE Sciex API-365 mass spectro-

meter (Toronto, Ontario, Canada). The data collected from

the analytes were monitored in the multiple reaction mode

(MRM) using turboionspray in the positive and negative ion

mode. All standards were assayed in duplicate. Analyte

concentrations were calculated from standard curves of the

peak area versus standard concentration. Radiolabeled com-

M. Markowska et al. / Journal of Pharmacological and Toxicological Methods 46 (2001) 51–5552

pounds tested, containing either tritium or carbon-14, were

analyzed with a Packard liquid scintillation spectrometer

(Downers Grove, IL). After correction for background, dpm

values were converted to nanograms per milliliter.

2.5. Statistical analysis

The Student’s t test was used to compare the data of the

single analytes and mixture of analytes. P values < .05 were

considered to be statistically significant. Results are ex-

pressed as mean values ± S.D. The value of n is defined as

the number of wells per compound.

3. Results

3.1. Single and multiple compound assessment

The Papp values across Caco-2 cell monolayers of het-

erologous series of single analytes and mixtures of analytes

per well using 1 cm2 membrane (Table 1) correlated closely.

Two control markers, mannitol and propranolol, for para-

cellular and transcellular transport, respectively, were

included in all of the studies (although mannitol data were

not included in Table 1). In this study, the Papp values

determined for a single compound per well are consistent

with values determined for mixtures of four analytes as well

as with literature values (Artursson & Karlsson, 1991;

Chong et al., 1996; Fagerholm et al., 1996; Gres & Julian,

1998). Based on rank-order comparison of permeabilities

obtained for several reference compounds (Table 2), these

compounds fall within the same rank order (low, medium,

and high permeability of compounds).

Due to the sensitivity and selectivity provided by the LC-

MS/MS analytical method, inclusion of more analytes in a

mixture was studied. To minimize the cost and enhance HTS

using Caco-2 cells, a mixture of eight analytes was utilized to

perform the transport of drugs across the Caco-2 cell mono-

layers. Similarly, the Papp values determined for a single

compounds per well using 1 cm2 membrane were compar-

able with values determined from the mixture of eight

compounds per well (Table 1), and are also in agreement

with literature values (Artursson & Karlsson, 1991; Chong et

al., 1996; Fagerholm et al., 1996; Gres & Julian, 1998; Irvine

et al., 1999; Lennernaes et al., 1996; Polli & Ginski, 1998).

Although increases in the Papp value were observed with

atenolol (0.7 ± 0.1 vs. 1.4 ± 0.5, P < .05) and with piroxicam

(38.1 ± 0.5 vs. 53.1 ± 1.9, P < .05, Table 1), nevertheless,

based on rank-order comparison of permeabilities obtained

for several reference compounds (Table 2), these compounds

also fall within the same rank order (low, medium, and high

permeability of compounds).

3.2. Single compound using different size membrane

Permeability values of PEG-4000, mannitol, cimetidine,

propranolol, and testosterone, conducted in 6-, 12-, and 24-

Table 1

Papp values for compounds conducted either singly or in mixtures of four or eight analytesa across Caco-2 monolayers

Papp� 10� 6 (cm/s)

Compound Single Mixture-4 Single Mixture-8 Fa%b References

Atenolol 1.4 ± 0.4c 1.1 ± 0.2 50 Artursson & Karlsson, 1991

Nadolol 0.7 ± 0.1 0.8 ± 0.1 35 Chong et al., 1996

Metoprolol 32.3 ± 0.1 34.1 ± 1.4 95 Artursson & Karlsson, 1991; Fagerholm, Johansson, & Lennernaes, 1996

Propranolol 32.5 ± 0.3c 42.4 ± 0.8 90 Artursson & Karlsson, 1991; Gres & Julian, 1998

RWJ-53308 0.6 ± 0.3 0.6 ± 0.1

Atenolol 0.7 ± 0.1c 1.4 ± 0.5* 50 Artursson & Karlsson, 1991; Chong et al., 1996

Terbutaline 1.4 ± 0.8 0.9 ± 0.2 73 Artursson & Karlsson, 1991; Chong et al., 1996

Propranolol 23.9 ± 0.5c 29.3 ± 0.9 90 Artursson & Karlsson, 1991; Gres & Julian, 1998

Naproxen 54.2 ± 2.3 52.9 ± 3.6 100 Lennernaes, Palm, Fagerholm, & Artursson, 1996

Piroxicam 38.1 ± 0.5 53.1 ± 1.9* 100 Polli & Ginski, 1998

Topiramate 28.4 ± 2.2 30.8 ± 1.2 81 Wu et al., 1994

Furosemide 0.6 ± 0.1 0.6 ± 0.1 61 Fagerholm et al., 1996; Irvine et al., 1999

a The apparent permeability ( Papp) was conducted at 10 mM donor drug concentration in the A to B direction. This experiment was performed at pH value

of 7.4 across Caco-2 cell monolayers using a 1.0 cm2 membrane. Results are the means ± S.D. (n= 4).b Fraction of dose absorbed in humans is the absorption of the drugs after oral administration in humans (Artursson & Karlsson, 1991).c P < .05 compared in two settings of single determination.

* P < .05 compared to a single analyte.

Table 2

Papp values for reference compounds across Caco-2 cell monolayers using

three differently sized membranesa

Papp� 10 � 6 (cm/s)

Compound

0.33 cm2

membrane

1.0 cm2

membrane

4.7 cm2

membrane

[14C] PEG-4000 0.2 ± 0.1 0.13 ± 0.02 0.14 ± 0.01

[14C]-Mannitol 1.4 ± 0.2 1.3 ± 0.1 0.90 ± 0.1

[3H]-Cimetidine 1.4 ± 0.1 1.3 ± 0.1 1.04 ± 0.1

[3H]-Propranolol 38.9 ± 1.9 36.6 ± 0.9 24.2 ± 0.3

[14C]-Testosterone 53.5 ± 1.0 48.1 ± 1.6 36.4 ± 2.3

a The apparent permeability ( Papp) was conducted at 10 mM donor drug

concentration in the A to B direction. This experiment was performed at a

pH value of 7.4 across Caco-2 cell monolayers. Results are the

means ± S.D. (n= 4).

M. Markowska et al. / Journal of Pharmacological and Toxicological Methods 46 (2001) 51–55 53

well format (Table 2), showed slight declines in drug

permeability with an increase in well size (Table 2).

3.3. Membrane integrity

To optimize the high-throughput screening, assessment

of the integrity of the Caco-2 cell monolayers was deter-

mined by measuring the TEER in 6-, 12-, and 24-well

format for 25 days. These wells had membrane sizes of

0.33, 1.0, and 4.7 cm2, respectively. The TEER of Caco-2

monolayers in the three different size membrane formats

increased significantly with days in culture (Fig. 1), as

expected. Little increase in TEER was observed between

Days 16 and 25.

4. Discussion

To the best of our knowledge, this is the first report of an

in vitro system that compares the transport of single analytes

(atenolol, nadolol, metoprolol, or propranolol) with a mix-

ture of the four analytes across Caco-2 cell monolayers.

There have been a number of studies performed utilizing

high-throughput approaches for rapid pharmacokinetic

screening (Bayliss & Frick, 1999; Bu, Poglod, Micetich,

& Khan, 2000; Frick, Higton, Wring, & Wells-Knecht,

1998; Garberg, Eriksson, Schipper, & Sjostrom, 1999;

Liang, Chi, Wright, Timby, & Unger, 1999; Tabit & Ber-

man, 1998; Taylor, Gibbons, & Braeckman, 1997) and high-

throughput Caco-2 permeability screening (Bu et al., 2000),

but none of them differentiate the permeability of analytes

singly versus in a mixture of analytes.

To our knowledge, this is also the first report of an in

vitro system that compares the transport of single analytes

(RWJ-53308, atenolol, terbutaline, propranolol, naproxen,

piroxicam, topiramate, furosemide) with a mixture of the

eight analytes across Caco-2 monolayers. One should

always bear in mind the possibility of potential drug–drug

interactions involving transporters or metabolism when mix-

tures of compounds are used, and this may explain the

difference seen for atenolol and piroxicam. In our studies,

the permeability values measured by this approach (mix-

tures of four or eight analytes) suggest that using either

mixture of analytes is a reliable approach for assessing high-

throughput screening early in drug discovery. In the present

study, it has to be noted that Papp values of atenolol and

propranolol (P < .05, Table 1), assessed singly in separate

experiments, are different. However, these differences may

be due to the use of a heterogeneous cell line.

To enhance the HTS assessment and to lower the cost,

more studies were conducted to compare and select the best

conditions to study drug transport across Caco-2 monolayers.

Five reference compounds, PEG-4000, mannitol and cime-

tidine, propranolol, and testosterone with very low, low,

medium, and high permeability, respectively, were selected.

Transport was measured as described above, except that three

different membrane sizes were tested. The permeability

decline observed was attributable to the heterogeneous

Fig. 1. The TEER of Caco-2 monolayers in three differently sized transwell polycarbonate membranes measured over a 25-day period. Results are the

means ± S.D. (n= 4).

M. Markowska et al. / Journal of Pharmacological and Toxicological Methods 46 (2001) 51–5554

nature of the cell line in the Caco-2 monolayer. Studies have

shown that more permeable monolayers form poorly de-

veloped actin rings, suggesting differences in passive trans-

port caused by variable development of tight junctions

(Walter & Kissel, 1995), and our data are consistent with

this observation. This suggests that permeability of an

impermeable marker like mannitol is a more sensitive indic-

ator of the size of the tight junctions than the TEER assay

(Table 2 and Fig. 1). If studies were to be performed on a

cloned cell line, one ought not to see this discrepancy.

To use this model successfully, it is necessary to validate

the use of Caco-2 monolayers using marker compounds

under the defined culturing and experimental conditions that

will be used.

Caco-2 monolayers reached an optimal stable state for

study after 16–26 days, regardless of membrane size format.

However, a number of studies have reported a wide range of

TEER value measurements, indicating that in different

laboratories, under different culture conditions, Caco-2

monolayers can display electrical properties typical of either

small intestinal or colonic enterocytes (Yee, 1997). There-

fore, as stated above, an impermeable marker such as man-

nitol is more reliable and sensitive than the resistance

measurement as an indicator of tight junctions.

References

Artursson, P., & Karlsson, J. (1991). Correlation between oral drug absorp-

tion in humans and apparent drug permeability coefficients in human

intestinal epithelial (Caco-2) cells. Biochemical and Biophysical Re-

search Communications, 175, 880–885.

Artursson, P., Palm, K., & Luthman, K. (1996). Caco-2 monolayers in

experimental and theoretical predictions of drug transport. Advanced

Drug Delivery Reviews, 22, 67–84.

Bayliss, M. K., & Frick, L. W. (1999). High-throughput pharmacokinetics:

cassette dosing. Current Opinion in Drug Discovery & Development, 2,

20–25.

Bu, H. Z., Poglod, M., Micetich, R. G., & Khan, J. K. (2000). High-

throughput Caco-2 cell permeability screening by cassette dosing and

sample pooling approaches using direct injection/on-line guard car-

tridge extraction/tandem mass spectrometry. Rapid Communications

in Mass Spectrometry, 14, 523–528.

Chong, S., Dando, S. A., Soucek, K. M., & Morrison, R. A. (1996). In vitro

permeability through Caco-2 cells is not quantitatively predictive of in

vivo absorption for peptide-like drugs absorbed via the dipeptide trans-

porter system. Pharmaceutical Research, 13, 120–123.

Fagerholm, U., Johansson, M., & Lennernaes, H. (1996). Comparison be-

tween permeability coefficients in rat and human jejunum. Pharmaceut-

ical Research, 13, 1336–1342.

Frick, L. W., Higton, D. M., Wring, S. A., & Wells-Knecht, K. J. (1998).

Cassette dosing: rapid estimation of in vivo pharmacokinetics. Medic-

inal Chemistry Research, 8, 472–477.

Garberg, P., Eriksson, P., Schipper, N., & Sjostrom, B. (1999). Automated

absorption assessment using Caco-2 cells cultured on both sides of

polycarbonate membranes. Pharmaceutical Research, 16, 441–445.

Gres, M. C., & Julian, B. (1998). Correlation between oral drug absorption

in humans, and apparent drug permeability in TC-7 cells, a human

epithelial intestinal cell line: comparison with the parental Caco-2 cell

line. Pharmaceutical Research, 15, 726–733.

Irvine, J. D., Takahashi, L., Lockhart, K., Cheong, J., Tolan, J. W., Selick,

H. E., & Grove, J. R. (1999). MDCK (Madin–Darby canine kidney)

cells: a tool for membrane permeability screening. Journal of Pharma-

ceutical Sciences, 88, 28–33.

Lennernaes, H., Palm, K., Fagerholm, U., & Artursson, P. (1996). Compar-

ison between active and passive drug transport in human intestinal

epithelial (Caco-2) cells in vitro and human jejunum in vivo. Interna-

tional Journal of Pharmaceutics, 127, 103–107.

Liang, L., Chi, C., Wright, M., Timby, D., & Unger, S. (1999). Use of the

N-in-1 approach for rapid pharmacokinetic screening of drug candidates

using LC-MS–MS. American Biotechnology Laboratory, 17, 8–10.

Polli, J. E., & Ginski, M. J. (1998). Human drug absorption kinetics and

comparison to Caco-2 monolayer permeabilities. Pharmaceutical Re-

search, 15, 47–52.

Tarbit, M. H., & Berman, J. (1998). High-throughput approaches for eval-

uation absorption, distribution, metabolism and excretion properties of

lead compounds. Current Opinion in Chemical Biology, 2, 411–416.

Taylor, E. W., Gibbons, J. A., & Braeckman, R. A. (1997). Intestinal

absorption screening of mixtures from combinatorial libraries in the

Caco-2 model. Pharmaceutical Research, 14, 572–577.

Walter, E., & Kissel, T. (1995). Heterogeneity in the human intestinal cell

line Caco-2 leads to differences in transepithelial transport. European

Journal of Pharmaceutical Sciences, 3, 215–230.

Wu, W.-N., Heebner, J. B., Streeter, A. J., Moyer, M. D., Takacs, A. R.,

Doose, R. D., & Ferraiolo, B. L. (1994). Evaluation of the absorption,

excretion, pharmacokinetics and metabolism of the anticonvulsant, top-

iramate in healthy men. Pharmaceutical Research, 11, S-336.

Yee, Y. (1997). In vitro permeability across Caco-2 cells (colonic) can

predict in vivo (small intestinal) absorption in man—fact or myth.

Pharmaceutical Research, 14, 763–766.

M. Markowska et al. / Journal of Pharmacological and Toxicological Methods 46 (2001) 51–55 55