- Email: [email protected]
Paracellular Porosity and Pore Size of the Human Intestinal Epithelium in Tissue and Cell Culture Models
Paracellular Porosity and Pore Size of the Human Intestinal Epithelium in Tissue and Cell Culture Models ¨ KELA¨,3 JONI PALMGREN,4 TIMO MAURIALA,1,4 CHARLOTTA VEDIN,5 JOHANNA LINNANKOSKI,1,2 JOHANNA MA 5 ANNA-LENA UNGELL, LUCIA LAZOROVA,6 PER ARTURSSON,6 ARTO URTTI,1 MARJO YLIPERTTULA2 1
Centre for Drug Research, University of Helsinki, Helsinki, Finland
2
Division of Biopharmacy and Pharmacokinetics, University of Helsinki, Viikinkaari 5 E, P.O. Box 56, Helsinki 00014, Finland 3
Department of Pharmaceutics, University of Kuopio, Kuopio, Finland
4
Department of Pharmaceutical Chemistry, University of Kuopio, Kuopio, Finland
5
DxDMPK & Bioanalytical Chemistry, AstraZeneca, Mo¨lndal, Sweden
6
Department of Pharmaceutics, University of Uppsala, Uppsala, Sweden
Received 29 April 2009; revised 13 August 2009; accepted 28 August 2009 Published online 13 October 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21961
ABSTRACT: The paracellular space defines the passive permeation of hydrophilic compounds in epithelia. The goal of this study was to characterise the paracellular permeation pathway in the human intestinal wall and differentiated epithelial cell models (MDCKII, Caco-2 and 2/4/A1). The permeabilities of hydrophilic polyethylene glycols (PEG) were investigated in diffusion chambers, and mass spectrometry was used to obtain accurate concentrations for each PEG molecule. The paracellular porosity and the size of the pores in the membranes were estimated from the PEG permeability data using an effusion-based approach. The porosities were found to be low (fraction 107–105 of the epithelial surface) in all investigated membranes. Two ˚ ) were detected in the human intestinal different pore sizes (radii 5–6 and >10 A ˚ ) in the 2/4/ epithelium and the Caco-2 and MDCKII cells, while only one (about 15 A A1 monolayer. The paracellular porosities of the human small intestine and 2/4/A1 monolayers were larger (>107) than that of the MDCKII and Caco-2 cells (<107). We report for the first time the quantitative values describing both porosity and pore size of the paracellular space in the human intestine. The cell models deviate from the small intestine either with respect to porosity (Caco-2, MDCKII) or pore size distribution (2/4/A1). ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:2166–2175, 2010
Keywords: paracellular permeation; drug absorption; human intestine; Caco-2; MDCKII; 2/4/A1; epithelial permeability
INTRODUCTION Johanna Ma¨kela¨’s present address is OrionPharma, Espoo, Finland. Correspondence to: Marjo Yliperttula (Telephone: þ358-4409-35-566; Fax: þ358-91-91-59-580; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 2166–2175 (2010) ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association
2166
Passive drug diffusion across an epithelium takes place either through the hydrophilic pores between the cells or across the lipoidal cell membrane. The route, which a compound prefers, depends on its shape, size and charge.1,2 For instance, hydrophobic compounds are considered
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
to traverse the cell monolayer predominantly by the transcellular route, whereas small hydrophilic solutes prefer the paracellular route.3,4 The negative characteristics of the paracellular space determine that cations permeate across the paracellular route easier than neutral compounds, which, in turn, are more permeable than anions.3,5,6 Paracellular permeation depends not only on the properties of the permeating molecule but naturally also on the morphology of the epithelium. The hydrophilic pores constituting the paracellular route are protein-based structures formed of occludins and claudins.7,8 These structures, also called tight junctions, seal the space between the epithelial cells thereby forming a protective barrier allowing paracellular diffusion of small solutes, but excluding potentially toxic macromolecules and micro-organisms. Epithelial cell monolayers grown on semiporous filters have gained considerable popularity in the study of epithelial drug transport. Previous studies have shown that the commonly used Caco-2 cell model yields even 100 times lower permeability coefficients for low permeability drugs than the small intestine.9 This difference has been proposed to result, at least partly, from the tight junctions of the Caco-2 cells being less permeable than those of the small intestine.9 Another commonly used cell line, the MDCKII cell model, also suffers from the drawback of not predicting accurately the absorption of low permeability drugs.10 Alternative cell lines that would better mimic the human small intestine are under development. The recently introduced 2/4/A1 cell line forms a leakier paracellular pathway than the Caco-2 monolayer and therefore is thought to mimic the human small intestinal epithelium better than the Caco-2 model.11,12 Tavelin et al.12 estimated the average pore radius of the 2/4/A1 ˚ . In the same study, the average cell line to be 9.0 A pore radius of the Caco-2 cell line was found to be ˚ . Fine et al.13 estimated that the average 3.7 A pore radius of the human intestinal epithelium ˚. was 8–13 A Several previous studies, such as the study of Tavelin et al.,12 have estimated the pore radii of the intestinal cell models using an approach called the Renkin method.14 Ha¨ ma¨la¨ inen et al.15 introduced an alternative method, an effusion-based approach, to analyse the paracellular space of epithelial membranes. According to the effusion theory, paracellular drug permeation takes place DOI 10.1002/jps
2167
when randomly colliding drug molecules happen to hit the pores in the membrane. This occurs infrequently, but once a molecule finds a pore, permeation across the pore is assumed to take place rapidly. The effusion approach can only be applied to membranes with short diffusion lengths across the pores, small pore densities and small pore sizes. In addition to predicting the pore size of the investigated membrane, the effusion theory also predicts the porosity of the membrane (the fraction of the paracellular space of the membrane surface), which is not possible with other techniques. Such quantitative information on epithelial cell monolayers and human tissues could provide a basis for improved predictions of drug absorption from cell data. The aim of this study was to characterise the paracellular permeation route of (1) the 2/4/A1, MDCKII and Caco-2 cell lines and (2) excised human intestinal segments. The pore sizes and porosities of these membranes were estimated based on permeabilities of neutral polyethylene glycol (PEG) oligomers and an effusion theorybased data analysis.
MATERIALS AND METHODS Polyethylene Glycol Stock Solution PEGs with mean molecular weights of 200, 400, 600 and 1000 were obtained from Chemical Pressure (Pittsburgh, PA). PEG 200 (final concentration 0.2 mg/mL), PEG 400 (final concentration 4.0 mg/mL), PEG 600 (final concentration 0.6 mg/mL) and PEG 1000 (final concentration 1.0 mg/mL) were dissolved in glutathione bicarbonated Ringer’s (GBR) solution (Caco-2 and MDCKII)/NaCl (2/4/A1 and human jejunum). On the basis of mean molecular weight the concentrations were 0.001 M.
Human Jejunum Human proximal jejunum was obtained from one female patient (age 62, weight 75 kg) undergoing gastro-intestinal bypass surgery (Sahlgrenska University Hospital, Gothenburg, Sweden). A 2 cm 2 cm part of the jejunum was stapled and put in a beaker with cold Krebs-bicarbonate Ringer’s solution (KBR) on ice, which was continuously bubbled with an O2/CO2 (95%/5%) gas mixture. The segment was directly transported to the laboratory under these conditions. The jejunal JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2168
LINNANKOSKI ET AL.
segment was then cut open and the muscularis externa was carefully removed using blunt dissection. During the preparation the segment was submerged in KBR (128C), which was bubbled continuously. Three stripped segments were then mounted in modified Ussing chambers with effective stirring conditions as described earlier.9,16 The effective exposed area of the tissue was 1.00 cm2. The tissues were allowed to recover for about 30 min to reach a temperature of about 36.5–37.58C and to gain physiological stability. The viability of the tissues was continuously monitored by recording of potential difference (PD) and trans-segmental electrical resistance (R). Tissues were only used if the electrical values reached 4 mV and 25 V cm2 before the start of the experiment. The electrical values in the present experiment were: PD SD, 4.9 0.3 mV; SCC, 170 10.5 mA/cm2; R, 29 3.6 V cm2. The variability in electrical values for 50 donors and 220 jejunal segments is represented by the following: PD, 7 1.6 mV; SCC, 240 83 mA/cm2; R, 31.5 8.0 V cm2 (unpublished AstraZeneca inhouse data). The corresponding permeability to the paracellular marker mannitol is stated below.
Ethics All written consents are by writing according to AstraZeneca global human biological sample policy and is regarded nontraceable. The consents are stored in the hospital.
Caco-2 Cell Culture Caco-2 cells were obtained from American Type Culture Collection (Manassas, VA). Growth medium and supplements were purchased from BioWhittaker except foetal bovine serum (Gibco, Invitrogen, Carlsbad, CA). Caco-2 cells were grown in 75 cm2 flasks (Nunc, CityLab, Finland) at 378C and 7% CO2 atmosphere and subcultured twice a week. Growth medium was Dulbecco’s modified Eagle’s medium supplemented with 10% heat inactivated foetal bovine serum, 2 mM L-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin and 1% nonessential amino acids. For transport studies, Caco-2 cells (passages 46–54) were seeded on uncoated Transwell polycarbonate filters (24 mm diameter, pore size 0.4 mm, Corning Costar, CityLab, Finland) at a density of 80,000 cells/cm2. The cells were allowed to differentiate at 378C for 21–23 days. The growth medium was changed JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
two or three times a week and always on the next day after seeding and on the day before transport study.
MDCKII Cell Culture Madin–Darby canine kidney type II cells (MDCKII) were obtained from the Netherlands Cancer Institute (Amsterdam, the Netherlands). Growth medium and supplements were purchased from Gibco. The cells were grown in 75 cm2 flasks (Nunc) at 378C and 5% CO2 atmosphere and subcultured twice a week. Growth medium was Dulbecco’s modified Eagle’s medium supplemented with 10% heat inactivated foetal bovine serum, 50 U/mL penicillin and 50 mg/mL streptomycin. For transport studies, MDCKII cells (passages 32–38) were seeded on uncoated Transwell polycarbonate filters (24 mm diameter, pore size 3.0 mm, Corning Costar) at a density 0.43 106 cells/cm2. The cells were allowed to differentiate at 378C and the permeability study was performed on the fourth day after seeding. The growth medium was changed every day.
2/4/A1 Cell Culture 2/4/A1 cells were expanded at 338C in RPMI1640 medium supplemented with 4% foetal bovine serum (v/v) (Gibco, Invitrogen AB, Lidingo, Sweden), 2 mM L-glutamine, 1 mg/mL BSA, 65 ng/mL dexamethasone, 20 ng/mL epidermal growth factor, 20 ng/mL murine epidermal growth factor, ITS premix containing 10 mg/mL insulin, 5.5 mg/mL transferrin and 5 ng/mL selenic acid according to previously published procedures.17 For transport studies, 2/4/A1 cells (passage 37) were seeded on Transwell polycarbonate filters (12 mm diameter, pore size 0.45 mm, Corning Costar, Sigma–Aldrich Sweden AB, Stockholm, Sweden) coated with ECM gel from Engelbreth– Holm–Swarm murine sarcoma (16 mg/cm2) at a density 0.18 106 cells/cm2. The cells were allowed to differentiate at 398C for 5 days as was described previously11 with small medium modification. The differentiation medium included Optimem (Gibco) supplemented with 2 mM Lglutamine, 1 mg/mL BSA, 65 ng/mL dexamethasone, ITS premix containing 10 mg/mL insulin, 5.5 mg/mL transferrin and 5 ng/mL selenic acid, 1 nM T3 (triiodo-L-thyronine), 2 mM Na-pyruvate, 1% (v/v) 100 penicillin streptomycin and 0.1% DMSO. All media and supplements were DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
2169
purchased from Sigma–Aldrich Sweden AB, unless otherwise stated.
the experiment, and sample volumes were replaced with fresh KBR.
Measurement of Transepithelial Electrical Resistance (TEER)
Analysis of PEGs
The integrity of cell monolayers was confirmed by TEER measurements before permeability studies. The monolayers were rinsed with Hank’s balanced salt solution (Gibco) containing 25 mM HEPES (Sigma Aldrich, Helsinki, Finland), pH 7.4 (HBSS) prewarmed to 378C. TEER of cell monolayers was measured in a chamber with a cap and a base containing voltage-sensing Ag/AgCl electrodes (Endohm, World Precision Instruments, Sarasota, FL) filled with prewarmed HBSS. The actual TEER values were calculated by subtracting the electrical resistance of cell culture inserts without cells.
Permeability Studies in Cell Lines The stock solution of PEGs was diluted in KBR, to a final concentration of 100 mM (donor solution). The pH of the donor solution was checked (7.4). After 20 min, the equilibration solution was replaced with (1.5 mL of Caco-2 and MDCKII, 0.4 mL of 2/4/A1) prewarmed donor solution and the transport of PEGs in apical to basolateral direction was followed for 240 min at 378C incubation in air under constant mild agitation (135 rpm Caco-2 and MDCKII, 100 rpm 2/4/A1). At regular time intervals, samples were withdrawn from acceptor chambers. In the Caco-2 and MDCKII experiments the withdrawn volume was replaced with fresh prewarmed HBSS. In the 2/4/A1 experiments the filters were transferred to new wells containing 1.2 mL prewarmed HBSS. Samples from the donor chambers were also taken at the end of the experiment in order to check for the recovery of the PEGs.
Permeability Studies in Human Jejunum The experiment was carried out in apical to basolateral direction at 378C. Changing the KBR solution on both sides of the membrane to prewarmed KBR and adding the PEG 400 solution to the mucosal side started the experiment. Samples of 100 mL were then withdrawn from the receiver (basolateral) and donor (apical) side at selected time points up to 210 min after start of DOI 10.1002/jps
PEGs were quantified using the combination of reversed-phase HPLC and electrospray ionisation mass spectrometry. The HPLC–ESI–MS method has been described in more detail by Palmgren et al.18 Briefly, the HPLC gradient was 2–31% ACN (containing 10 mM ammonium acetate, pH 6.62) in 4.5 min and the run was performed in 6 min. The chromatographic separation was performed using a Xterra MS C18 reversedphase column (2.1 mm 20 mm, 2.5 mm, Waters, Milford, MA) with a flow rate of 200 mL/min. HPLC–ESI–MS measurements were performed with a LCQ and LTQ quadrupole ion trap mass spectrometers equipped with an ESI ion source (Finnigan, San Jose, CA). Calibration curves were constructed by plotting chromatographic peak ratios of standard area/IS area versus concentration of the standard using linear regression. From these curves the coefficients of correlation (r2) were calculated. Calibration solutions and QC standards were prepared daily and analysed immediately after preparation. Data were processed using the Xcalibur software package version 1.4 SRI.
Calculation of Permeability Values The Papp values (cm/s) were calculated according to Papp ¼
dQ 1 dt A C0
(1)
where dQ/dt is the permeability rate at steadystate flux in the acceptor chamber (mol/s), A is the surface area of the cell monolayer/tissue (cm2) and C0 is the initial concentration in the donor chamber (mol/mL).19 Mass spectrometry was used to measure concentrations of each oligomer both in the donor and receiver phases.
Integrity Studies In order to investigate whether the study procedures influenced the permeability of Caco-2 and 2/4/A1 monolayers, the transport of labelled mannitol (Caco-2 3H-labelled, 2/4/A1 14C-labelled) in apical to basolateral direction was studied. In JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2170
LINNANKOSKI ET AL.
the Caco-2 cell line the labelled mannitol permeability was studied in the absence of the PEGs, whereas in the 2/4/A1 cell line the labelled mannitol permeability was studied in both the absence and presence of the PEGs. As a marker control of epithelial integrity of the human intestine 14C-labelled mannitol (0.7 mCi/mL) was added to the apical side at the start of the experiment. Analysis of the total radioactivity of 14C-mannitol in samples from the transport experiments was performed using liquid scintillation counting. The apparent permeability coefficient ( Papp) for 14C-mannitol was calculated for each segment according to Eq. (1). The mean Papp value for 14C-mannitol for the three segments was calculated to be 8.0(0.3) 106 cm/s. This is within the range of donor variability with respect to human jejunum permeability to mannitol ( Papp SD: mannitol 5.53 2.1 106cm/s, n ¼ 220 segments (¼50 donors) (unpublished AstraZeneca inhouse recordings).
membranes investigated in this work. The theory describes the rate of PEG permeability Jh/C by the following equation: Jh RT" 1 1 ¼ ¼ ½slope 12phNA l rs rs c
(2)
where rs is the radius of PEG oligomers, e the porosity of the membrane, l the jump length ˚ ), h the viscosity of water, R the gas (i.e., 3.1 A law constant, T the temperature and NA the Avogadro’s number. Eq. (2) predicts that the measured permeability, Jh/C, is inversely proportional to the radius of the drug molecule. The porosities of the investigated membranes (i.e., fraction of intercellular spaces) were estimated from the slope of this relation. The pore sizes were estimated by calculating the critical value of the molecular radius still able to penetrate through the paracellular pores. This was achieved by extrapolating f(1/rs) into zero permeability.
RESULTS Estimating the Pore Sizes and Porosity of the Membranes An effusion theory-based equation15 was used to estimate the porosities (e) and pore sizes of the
The results of the permeability experiments with the PEG oligomers are shown in Figure 1. Mass spectrometry allowed the analysis of each oligomer in the polydisperse mixtures of PEGs with
Figure 1. The relationship between permeability and the molecular weight of a PEG oligomers in 2/4/A1 (open squares), human intestine (closed triangles), Caco-2 (closed squares) and MDCKII (open triangles). The values represent mean apparent permeability coefficients SD. The inset is an enlargement of the Caco-2 and MDCKII cell line data. The data are based on permeability experiments with 11–12 time points and 3–6 replicate experiments with each membrane. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
mean molecular weights 200, 400, 600 and 1000. The data show that the overall paracellular permeability is the highest in the 2/4/A1 cell line, followed by the human intestine. The MDCKII and Caco-2 cell lines have the lowest permeabilities. The MDCKII and Caco-2 cell lines, and the human intestine show a biphasic relationship between permeability and molecular weight/ radius of the PEG oligomers, while a linear relationship is seen between permeability and molecular weight/radius in the 2/4/A1 cell model (Fig. 1). The permeability data were further analysed by the effusion-based theory (Eq. 2). This analysis
2171
allows determination of the membrane porosity and pore size(s). The analysis indicates that there are two distinct pore sizes in the MDCKII and Caco-2 cell models and in the human intestine (Fig. 2, biphasic permeability profiles). The radii of the small pores were in the same order of ˚ , Tab. 1). magnitude in the membranes (5.5–6.6 A ˚ The radii of the large pores were 10.1–30.5 A (Tab. 1). The 2/4/A1 monolayer was found to have ˚ ). pores of only one size (14.9 A The membrane porosities are given in Table 1 and Figure 3. The total porosity of the human intestine is 20 times that of the MDCKII cell line, over 10 times that of the Caco-2 cell model and
Figure 2. The relationship between permeability and the radius of a PEG oligomer in (A) MDCKII, (B) Caco-2, (C) 2/4/A1 cell lines and (D) human intestine. The closed squares represent experimental permeability through both small and large pores and the closed triangles through large pores. The open squares represent the calculated permeability through small pores. The porosity is estimated from the slope of the regression line (straight line: large pores; dotted line: small pores) according to the effusion-based theory. The pore size of the large pores is calculated by extrapolation of ˚ ) of the PEG oligomers are as the regression line to zero permeability. The rs values (A follows: 238 (4.51), 282 (4.87), 326 (5.15), 370 (5.47), 414 (5.78), 458 (6.03), 502 (6.27), 546 (6.55), 590 (6.77), 634 (7.03), 678 (7.24), 722 (7.45), 766 (7.69), 810 (7.89), 854 (8.09) and 898 (8.27). DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2172
LINNANKOSKI ET AL.
Table 1. Porosities and Pore Radii in MDCKII, Caco-2, 2/4/A1 Cell Lines and Human Intestine
Membrane Porosity 108 Total Small pores Large pores ˚) Pore size (A Small pores Large pores
MDCKII
Caco-2
2/4/A1
24 16 8
61 54 7
154
5.5 30.5
5.8 10.4
Human Intestine
154
531 442 89
14.9
6.6 10.1
over 3 times that of the 2/4/A1 cell model. The contribution of the small and large pores is shown schematically in Figure 3. The porosity created by small pores is greater than that created by large pores in the MDCKII and Caco-2 cell lines and the human intestine, whereas the total porosity of the 2/4/A1 cell line is constituted by large pores only.
DISCUSSION The paracellular route is considered to have an important role in the oral absorption of many small water-soluble compounds.20–22 Despite this, the dimensions of the intestinal paracellular space have remained poorly defined. In the present study, we profile the paracellular permeation route of the human intestinal epithelium and three commonly used cell models (MDCKII, Caco-2 and 2/4/A1). The study was conducted by measuring the permeabilities of polydispersed PEG oligomers in the membranes and applying an
Figure 3. Porosities constituted by large pores (grey) and small pores (black) in MDCKII, Caco-2, 2/4/A1 cell lines and human intestine. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
effusion-based theory to the PEG permeability data. This approach allows the quantitative characterisation of the paracellular route of the membranes investigated here. Besides the pore size (radius of the pore), the theory also solves the porosity (fraction of the paracellular space on the membrane surface) of the membrane, which is not possible with other techniques. The present results support earlier findings11,23 that the 2/4/A1 cell model predicts the paracellular permeability in human duodenum more reliably than the Caco-2 and MDCKII cell models (Fig. 1). The better prediction power of the 2/4/A1 cell line can be explained by its porosity being similar to that of the human duodenal epithelium: the porosity of the human duodenal epithelium is eight times higher than the porosity of the MDCKII and Caco-2 cell models, but fairly close to the porosity of the 2/4/A1 cell model (Tab. 1). Previous studies have shown that the Caco-2 model yields even 100 times lower permeability coefficients than the small intestine for low permeability drugs.9 This difference has been proposed to result, at least partly, from Caco-2 cells forming less permeable tight junctions than those found in the human small intestine.9 Our analysis indicates, however, that the pores in the Caco-2 monolayers are approximately of the same size as the pores in the human intestinal epithelium (Tab. 1). We conclude that a lower number of paracellular pores per cm2 in the Caco-2 cell monolayer explains the lower PEG permeability in the Caco-2 model (Tab. 1). Although the 2/4/A1 cell line and the human intestinal epithelium have similar porosities, the paracellular permeation properties of these membranes are not the same. The biphasic permeability profile of the human intestine suggests that the human intestinal epithelium has two distinct pore sizes, whereas the monophasic permeability profile of the 2/4/A1 cell line indicates that this membrane has pores of ˚ ). The biphasic one fairly large size (above 10 A permeability profiles observed in the MDCKII and Caco-2 cell lines indicate that the pores in these cell lines are also of two different sizes. In line with our findings, several previous studies suggest that there are two distinct pore sizes in the intestine and cell monolayers.24–27 Using a paracellular sieving model, Watson et al.27 showed that this was the case also for the Caco-2 cell model. The radius of the smaller ˚ , but with the pores was estimated to be 4.3–4.5 A sieving model it was not possible to estimate the DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
radius of the larger pores. Hollander25 suggested that the biphasic permeability profiles observed in intestinal preparations may reflect distinct populations of tight junction pores associated with villus tip and intestinal crypt cells. This theory does not, however, explain the two-component profile observed in intestinal monolayers. Several previous studies1,12,28 have estimated the pore radii of the intestinal cell models with the Renkin method.14 The Renkin method assumes random walk of diffusion in the pore and aims to account for the frictional resistance resulting from the molecules travelling through the pore. However, the validity of the Renkin correction is in doubt when the radius of the solute is over 0.4 times the radius of the pore.15,27 Under these conditions the assumption of random walk is not fulfilled, and the Renkin term exaggerates the diffusional resistance within the pore. With the Renkin method it is not possible to estimate the porosity of membranes. PEGs have been used extensively as a marker of intestinal.29,30 Some investigators have, however, questioned the suitability of them as a paracellular probes.31,32 For example, the intestinal permeability of PEG 400 is significantly higher than the permeability of saccharide markers of comparable molecular weight. Artursson et al.9 found that the permeability of PEG 194 was 6- to 28-fold greater than that of mannitol (molecular weight 182) in cell monolayers. The authors suggested that this could be explained by the molecular structure of PEGs: PEG oligomers are flexible polymers and, therefore, possible changes in their molecular dimensions may affect their permeation through the tight junctions. However, according to Watson et al.27 a relatively small difference in the radii of the probes will have profound effects on permeability particularly where they lie close to the radius of the paracellular pore. Thus, Watson et al. suggest that the higher than expected permeation of PEG 400 compared with other markers is due its smaller hydrodynamic radius. The molecular radii that were used in the calculations are based on molecular diffusivities in the solution assuming spherical shape of these homologous oligomers. This assumption may not be valid, but taking into account the real conformations of the PEG molecules is very difficult, and each PEG anyway has probably several conformations in dynamic motion in time and space. Taking into account molecular conformations is very complicated, and not necessarily reliable. Identical PEGs DOI 10.1002/jps
2173
were used for each membrane and the shape distributions of individual PEGs in the solution are not dependent on the bio-membrane, that is, they should not cause bias in the comparisons between the membranes. In calculating the PEG permeability values of the intestinal epithelium, we used the surface area of the orifice of the diffusion chamber, that is, the surface area of intestinal segment. Since the intestinal epithelium grows as folded structures, the effective surface area of the epithelium might be larger than the area used in our calculations, which would lead to overestimation of the permeability and porosity of the intestinal epithelium. However, the intestinal folds are generally not taken into account in studies of drug permeability in the intestine. Furthermore, since the villous structures are present in vivo, the number of pores per surface area of intestine corresponds to the values obtained with our method of calculation. It should also be noted that the pore size and porosity may vary for cell lines, depending on their origin, evolutionary history and the way they are cultivated. Our cells were cultivated according to the common procedure and the paracellular permeability values obtained in our laboratories are similar to most published data. As stated earlier, the Caco-2 cell model fails in predicting the human oral absorption of many poorly absorbed compounds. This is partly due to their low background permeation by passive diffusion. Quantitative analysis of the paracellular space is expected to provide a link between the passive diffusion in cell models (MDCKII, Caco-2 and 2/4/A1) and the small intestine. Characterising the paracellular space of epithelial cell models helps in bridging the gap in the predictions between human intestinal and cell model permeabilities. It should, however, be noted that the expression of transporters differs in cell models and the human intestine, which adds further uncertainty to cell model predictions. It is known that the Caco-2 model and the MDCKII variants frequently overestimate the importance of active transport and efflux in oral drug absorption.33 In conclusion, the paracellular routes of hydrophilic permeants in the intestine and three widely used cell models have been characterised. Two different pore sizes were detected in the human intestinal epithelium, Caco-2 cells and MDCKII monolayer, while only one in 2/4/A1 cells. The paracellular porosity of the human small intestine JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2174
LINNANKOSKI ET AL.
and 2/4/A1 monolayers was larger than in the MDCKII and Caco-2 cells. 10.
ACKNOWLEDGMENTS We would like to thank Prof. Seppo Auriola for his ˚ sa advice regarding the LC–MS analysis and Ms. A Sjo¨ berg for preparation of the human jejunum segments. The material support of OrionPharma and National Agency of Technology of Finland is gratefully acknowledged.
11.
12.
REFERENCES 1. Knipp GT, Ho NF, Barsuhn CL, Borchardt RT. 1997. Paracellular diffusion in Caco-2 cell monolayers: Effect of perturbation on the transport of hydrophilic compounds that vary in charge and size. J Pharm Sci 86:1105–1110. 2. Pade V, Stavchansky S. 1997. Estimation of the relative contribution of the transcellular and paracellular pathway to the transport of passively absorbed drugs in the Caco-2 cell culture model. Pharm Res 14:1210–1215. 3. Adson A, Burton PS, Raub TJ, Barsuhn CL, Audus KL, Ho NFH. 1995. Passive diffusion of weak organic electrolytes across Caco-2 cell monolayers: Uncoupling the contributions of hydrodynamic, transcellular, and paracellular barriers. J Pharm Sci 84: 1197–1204. 4. He YL, Murby S, Warhurst G, Gifford L, Walker D, Ayrton J, Eastmond R, Rowland M. 1998. Species differences in size discrimination in the paracellular pathway reflected by oral bioavailability of poly(ethylene glycol) and D-peptides. J Pharm Sci 87: 626–633. 5. Kottra G, Fromter E. 1983. Functional properties of the paracellular pathway in some leaky epithelia. J Exp Biol 106:217–229. 6. Nagahara N, Tavelin S, Artursson P. 2004. Contribution of the paracellular route to the pH-dependent epithelial permeability to cationic drugs. J Pharm Sci 93:2972–2984. 7. Furuse M, Hirase T, Itoh M, Nagafuchi A, Yonemura S, Tsukita S, Tsukita S. 1993. Occludin: A novel integral membrane protein localizing at tight junctions. J Cell Biol 123:1777–1788. 8. Furuse M, Fujita K, Hiiragi T, Fujimoto K, Tsukita S. 1998. Claudin-1 and -2: Novel integral membrane proteins localizing at tight junctions with no sequence similarity to occludin. J Cell Biol 141: 1539–1550. 9. Artursson P, Ungell AL, Lofroth JE. 1993. Selective paracellular permeability in two models of intestinal absorption: Cultured monolayers of human
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
intestinal epithelial cells and rat intestinal segments. Pharm Res 10:1123–1129. Irvine JD, Takahashi L, Lockhart K, Cheong J, Tolan JW, Selick HE, Grove JR. 1999. MDCK (Madin–Darby canine kidney) cells: A tool for membrane permeability screening. J Pharm Sci 88:28– 33. Tavelin S, Taipalensuu J, Hallbook F, Vellonen KS, Moore V, Artursson P. 2003. An improved cell culture model based on 2/4/A1 cell monolayers for studies of intestinal drug transport: Characterization of transport routes. Pharm Res 20:373–381. Tavelin S, Taipalensuu J, Soderberg L, Morrison R, Chong S, Artursson P. 2003. Prediction of the oral absorption of low-permeability drugs using small intestine-like 2/4/A1 cell monolayers. Pharm Res 20:397–405. Fine KD, Santa Ana CA, Porter JL, Fordtran JS. 1995. Effect of changing intestinal flow rate on a measurement of intestinal permeability. Gastroenterology 108:983–989. Renkin EM. 1954. Filtration, diffusion, and molecular sieving through porous cellulose membranes. J Gen Physiol 38:225–243. Ha¨ ma¨ la¨ inen K, Kontturi K, Auriola S, Murtoma¨ ki L, Urtti A. 1997. Estimation of pore size and pore density of biomembranes from permeability measurements of polyethylene glycols using an effusion-like approach. J Control Release 49:97–104. Ungell AL, Nylander S, Bergstrand S, Sjoberg A, Lennernas H. 1998. Membrane transport of drugs in different regions of the intestinal tract of the rat. J Pharm Sci 87:360–366. Tavelin S, Milovic V, Ocklind G, Olsson S, Artursson P. 1999. A conditionally immortalized epithelial cell line for studies of intestinal drug transport. J Pharmacol Exp Ther 290:1212–1221. Palmgren JJ, Toropainen E, Auriola S, Urtti A. 2002. Liquid chromatographic-electrospray ionization mass spectrometric analysis of neutral and charged polyethylene glycols. J Chromatogr A 976:165–170. Artursson P. 1990. Epithelial transport of drugs in cell culture. I: A model for studying the passive diffusion of drugs over intestinal absorptive (Caco-2) cells. J Pharm Sci 79:476–482. Flanagan SD, Takahashi LH, Liu X, Benet LZ. 2002. Contributions of saturable active secretion, passive transcellular, and paracellular diffusion to the overall transport of furosemide across adenocarcinoma (Caco-2) cells. J Pharm Sci 91:1169– 1177. Lee K, Thakker DR. 1999. Saturable transport of H2-antagonists ranitidine and famotidine across Caco-2 cell monolayers. J Pharm Sci 88:680–687. Zhou SY, Piyapolrungroj N, Pao L, Li C, Liu G, Zimmermann E, Fleisher D. 1999. Regulation of paracellular absorption of cimetidine and 5-amino-
DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
23.
24.
25.
26.
27.
28.
salicylate in rat intestine. Pharm Res 16:1781– 1785. Linnankoski J, Ranta VP, Yliperttula M, Urtti A. 2008. Passive oral drug absorption can be predicted more reliably by experimental than computational models—Fact or myth. Eur J Pharm Sci 34:129– 139. Fihn BM, Sjoqvist A, Jodal M. 2000. Permeability of the rat small intestinal epithelium along the villuscrypt axis: Effects of glucose transport. Gastroenterology 119:1029–1036. Hollander D. 1992. The intestinal permeability barrier. A hypothesis as to its regulation and involvement in Crohn’s disease. Scand J Gastroenterol 27:721–726. Soderholm JD, Olaison G, Kald A, Tagesson C, Sjodahl R. 1997. Absorption profiles for polyethylene glycols after regional jejunal perfusion and oral load in healthy humans. Dig Dis Sci 42:853– 857. Watson CJ, Rowland M, Warhurst G. 2001. Functional modeling of tight junctions in intestinal cell monolayers using polyethylene glycol oligomers. Am J Physiol Cell Physiol 281:C388–C397. Adson A, Raub TJ, Burton PS, Barsuhn CL, Hilgers AR, Audus KL, Ho NF. 1994. Quantitative
DOI 10.1002/jps
29.
30.
31.
32.
33.
2175
approaches to delineate paracellular diffusion in cultured epithelial cell monolayers. J Pharm Sci 83:1529–1536. Chadwick VS, Phillips SF, Hofmann AF. 1977. Measurements of intestinal permeability using low molecular weight polyethylene glycols (PEG 400). II. Application to normal and abnormal permeability states in man and animals. Gastroenterology 73:247–251. Ma TY, Hollander D, Krugliak P, Katz K. 1990. PEG 400, a hydrophilic molecular probe for measuring intestinal permeability. Gastroenterology 98:39–46. Iqbal TH, Lewis KO, Cooper BT. 1993. Diffusion of polyethylene glycol-400 across lipid barriers in vitro. Clin Sci (Lond) 85:111–115. Maxton DG, Bjarnason I, Reynolds AP, Catt SD, Peters TJ, Menzies IS. 1986. Lactulose, 51Crlabelled ethylenediaminetetra-acetate, L-rhamnose and polyethyleneglycol 400 [corrected] as probe markers for assessment in vivo of human intestinal permeability. Clin Sci (Lond) 71:71–80. Matsson P, Bergstrom CA, Nagahara N, Tavelin S, Norinder U, Artursson P. 2005. Exploring the role of different drug transport routes in permeability screening. J Med Chem 48:604–613.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
Centre for Drug Research, University of Helsinki, Helsinki, Finland
2
Division of Biopharmacy and Pharmacokinetics, University of Helsinki, Viikinkaari 5 E, P.O. Box 56, Helsinki 00014, Finland 3
Department of Pharmaceutics, University of Kuopio, Kuopio, Finland
4
Department of Pharmaceutical Chemistry, University of Kuopio, Kuopio, Finland
5
DxDMPK & Bioanalytical Chemistry, AstraZeneca, Mo¨lndal, Sweden
6
Department of Pharmaceutics, University of Uppsala, Uppsala, Sweden
Received 29 April 2009; revised 13 August 2009; accepted 28 August 2009 Published online 13 October 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21961
ABSTRACT: The paracellular space defines the passive permeation of hydrophilic compounds in epithelia. The goal of this study was to characterise the paracellular permeation pathway in the human intestinal wall and differentiated epithelial cell models (MDCKII, Caco-2 and 2/4/A1). The permeabilities of hydrophilic polyethylene glycols (PEG) were investigated in diffusion chambers, and mass spectrometry was used to obtain accurate concentrations for each PEG molecule. The paracellular porosity and the size of the pores in the membranes were estimated from the PEG permeability data using an effusion-based approach. The porosities were found to be low (fraction 107–105 of the epithelial surface) in all investigated membranes. Two ˚ ) were detected in the human intestinal different pore sizes (radii 5–6 and >10 A ˚ ) in the 2/4/ epithelium and the Caco-2 and MDCKII cells, while only one (about 15 A A1 monolayer. The paracellular porosities of the human small intestine and 2/4/A1 monolayers were larger (>107) than that of the MDCKII and Caco-2 cells (<107). We report for the first time the quantitative values describing both porosity and pore size of the paracellular space in the human intestine. The cell models deviate from the small intestine either with respect to porosity (Caco-2, MDCKII) or pore size distribution (2/4/A1). ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:2166–2175, 2010
Keywords: paracellular permeation; drug absorption; human intestine; Caco-2; MDCKII; 2/4/A1; epithelial permeability
INTRODUCTION Johanna Ma¨kela¨’s present address is OrionPharma, Espoo, Finland. Correspondence to: Marjo Yliperttula (Telephone: þ358-4409-35-566; Fax: þ358-91-91-59-580; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 2166–2175 (2010) ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association
2166
Passive drug diffusion across an epithelium takes place either through the hydrophilic pores between the cells or across the lipoidal cell membrane. The route, which a compound prefers, depends on its shape, size and charge.1,2 For instance, hydrophobic compounds are considered
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
to traverse the cell monolayer predominantly by the transcellular route, whereas small hydrophilic solutes prefer the paracellular route.3,4 The negative characteristics of the paracellular space determine that cations permeate across the paracellular route easier than neutral compounds, which, in turn, are more permeable than anions.3,5,6 Paracellular permeation depends not only on the properties of the permeating molecule but naturally also on the morphology of the epithelium. The hydrophilic pores constituting the paracellular route are protein-based structures formed of occludins and claudins.7,8 These structures, also called tight junctions, seal the space between the epithelial cells thereby forming a protective barrier allowing paracellular diffusion of small solutes, but excluding potentially toxic macromolecules and micro-organisms. Epithelial cell monolayers grown on semiporous filters have gained considerable popularity in the study of epithelial drug transport. Previous studies have shown that the commonly used Caco-2 cell model yields even 100 times lower permeability coefficients for low permeability drugs than the small intestine.9 This difference has been proposed to result, at least partly, from the tight junctions of the Caco-2 cells being less permeable than those of the small intestine.9 Another commonly used cell line, the MDCKII cell model, also suffers from the drawback of not predicting accurately the absorption of low permeability drugs.10 Alternative cell lines that would better mimic the human small intestine are under development. The recently introduced 2/4/A1 cell line forms a leakier paracellular pathway than the Caco-2 monolayer and therefore is thought to mimic the human small intestinal epithelium better than the Caco-2 model.11,12 Tavelin et al.12 estimated the average pore radius of the 2/4/A1 ˚ . In the same study, the average cell line to be 9.0 A pore radius of the Caco-2 cell line was found to be ˚ . Fine et al.13 estimated that the average 3.7 A pore radius of the human intestinal epithelium ˚. was 8–13 A Several previous studies, such as the study of Tavelin et al.,12 have estimated the pore radii of the intestinal cell models using an approach called the Renkin method.14 Ha¨ ma¨la¨ inen et al.15 introduced an alternative method, an effusion-based approach, to analyse the paracellular space of epithelial membranes. According to the effusion theory, paracellular drug permeation takes place DOI 10.1002/jps
2167
when randomly colliding drug molecules happen to hit the pores in the membrane. This occurs infrequently, but once a molecule finds a pore, permeation across the pore is assumed to take place rapidly. The effusion approach can only be applied to membranes with short diffusion lengths across the pores, small pore densities and small pore sizes. In addition to predicting the pore size of the investigated membrane, the effusion theory also predicts the porosity of the membrane (the fraction of the paracellular space of the membrane surface), which is not possible with other techniques. Such quantitative information on epithelial cell monolayers and human tissues could provide a basis for improved predictions of drug absorption from cell data. The aim of this study was to characterise the paracellular permeation route of (1) the 2/4/A1, MDCKII and Caco-2 cell lines and (2) excised human intestinal segments. The pore sizes and porosities of these membranes were estimated based on permeabilities of neutral polyethylene glycol (PEG) oligomers and an effusion theorybased data analysis.
MATERIALS AND METHODS Polyethylene Glycol Stock Solution PEGs with mean molecular weights of 200, 400, 600 and 1000 were obtained from Chemical Pressure (Pittsburgh, PA). PEG 200 (final concentration 0.2 mg/mL), PEG 400 (final concentration 4.0 mg/mL), PEG 600 (final concentration 0.6 mg/mL) and PEG 1000 (final concentration 1.0 mg/mL) were dissolved in glutathione bicarbonated Ringer’s (GBR) solution (Caco-2 and MDCKII)/NaCl (2/4/A1 and human jejunum). On the basis of mean molecular weight the concentrations were 0.001 M.
Human Jejunum Human proximal jejunum was obtained from one female patient (age 62, weight 75 kg) undergoing gastro-intestinal bypass surgery (Sahlgrenska University Hospital, Gothenburg, Sweden). A 2 cm 2 cm part of the jejunum was stapled and put in a beaker with cold Krebs-bicarbonate Ringer’s solution (KBR) on ice, which was continuously bubbled with an O2/CO2 (95%/5%) gas mixture. The segment was directly transported to the laboratory under these conditions. The jejunal JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2168
LINNANKOSKI ET AL.
segment was then cut open and the muscularis externa was carefully removed using blunt dissection. During the preparation the segment was submerged in KBR (128C), which was bubbled continuously. Three stripped segments were then mounted in modified Ussing chambers with effective stirring conditions as described earlier.9,16 The effective exposed area of the tissue was 1.00 cm2. The tissues were allowed to recover for about 30 min to reach a temperature of about 36.5–37.58C and to gain physiological stability. The viability of the tissues was continuously monitored by recording of potential difference (PD) and trans-segmental electrical resistance (R). Tissues were only used if the electrical values reached 4 mV and 25 V cm2 before the start of the experiment. The electrical values in the present experiment were: PD SD, 4.9 0.3 mV; SCC, 170 10.5 mA/cm2; R, 29 3.6 V cm2. The variability in electrical values for 50 donors and 220 jejunal segments is represented by the following: PD, 7 1.6 mV; SCC, 240 83 mA/cm2; R, 31.5 8.0 V cm2 (unpublished AstraZeneca inhouse data). The corresponding permeability to the paracellular marker mannitol is stated below.
Ethics All written consents are by writing according to AstraZeneca global human biological sample policy and is regarded nontraceable. The consents are stored in the hospital.
Caco-2 Cell Culture Caco-2 cells were obtained from American Type Culture Collection (Manassas, VA). Growth medium and supplements were purchased from BioWhittaker except foetal bovine serum (Gibco, Invitrogen, Carlsbad, CA). Caco-2 cells were grown in 75 cm2 flasks (Nunc, CityLab, Finland) at 378C and 7% CO2 atmosphere and subcultured twice a week. Growth medium was Dulbecco’s modified Eagle’s medium supplemented with 10% heat inactivated foetal bovine serum, 2 mM L-glutamine, 100 U/mL penicillin, 100 mg/mL streptomycin and 1% nonessential amino acids. For transport studies, Caco-2 cells (passages 46–54) were seeded on uncoated Transwell polycarbonate filters (24 mm diameter, pore size 0.4 mm, Corning Costar, CityLab, Finland) at a density of 80,000 cells/cm2. The cells were allowed to differentiate at 378C for 21–23 days. The growth medium was changed JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
two or three times a week and always on the next day after seeding and on the day before transport study.
MDCKII Cell Culture Madin–Darby canine kidney type II cells (MDCKII) were obtained from the Netherlands Cancer Institute (Amsterdam, the Netherlands). Growth medium and supplements were purchased from Gibco. The cells were grown in 75 cm2 flasks (Nunc) at 378C and 5% CO2 atmosphere and subcultured twice a week. Growth medium was Dulbecco’s modified Eagle’s medium supplemented with 10% heat inactivated foetal bovine serum, 50 U/mL penicillin and 50 mg/mL streptomycin. For transport studies, MDCKII cells (passages 32–38) were seeded on uncoated Transwell polycarbonate filters (24 mm diameter, pore size 3.0 mm, Corning Costar) at a density 0.43 106 cells/cm2. The cells were allowed to differentiate at 378C and the permeability study was performed on the fourth day after seeding. The growth medium was changed every day.
2/4/A1 Cell Culture 2/4/A1 cells were expanded at 338C in RPMI1640 medium supplemented with 4% foetal bovine serum (v/v) (Gibco, Invitrogen AB, Lidingo, Sweden), 2 mM L-glutamine, 1 mg/mL BSA, 65 ng/mL dexamethasone, 20 ng/mL epidermal growth factor, 20 ng/mL murine epidermal growth factor, ITS premix containing 10 mg/mL insulin, 5.5 mg/mL transferrin and 5 ng/mL selenic acid according to previously published procedures.17 For transport studies, 2/4/A1 cells (passage 37) were seeded on Transwell polycarbonate filters (12 mm diameter, pore size 0.45 mm, Corning Costar, Sigma–Aldrich Sweden AB, Stockholm, Sweden) coated with ECM gel from Engelbreth– Holm–Swarm murine sarcoma (16 mg/cm2) at a density 0.18 106 cells/cm2. The cells were allowed to differentiate at 398C for 5 days as was described previously11 with small medium modification. The differentiation medium included Optimem (Gibco) supplemented with 2 mM Lglutamine, 1 mg/mL BSA, 65 ng/mL dexamethasone, ITS premix containing 10 mg/mL insulin, 5.5 mg/mL transferrin and 5 ng/mL selenic acid, 1 nM T3 (triiodo-L-thyronine), 2 mM Na-pyruvate, 1% (v/v) 100 penicillin streptomycin and 0.1% DMSO. All media and supplements were DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
2169
purchased from Sigma–Aldrich Sweden AB, unless otherwise stated.
the experiment, and sample volumes were replaced with fresh KBR.
Measurement of Transepithelial Electrical Resistance (TEER)
Analysis of PEGs
The integrity of cell monolayers was confirmed by TEER measurements before permeability studies. The monolayers were rinsed with Hank’s balanced salt solution (Gibco) containing 25 mM HEPES (Sigma Aldrich, Helsinki, Finland), pH 7.4 (HBSS) prewarmed to 378C. TEER of cell monolayers was measured in a chamber with a cap and a base containing voltage-sensing Ag/AgCl electrodes (Endohm, World Precision Instruments, Sarasota, FL) filled with prewarmed HBSS. The actual TEER values were calculated by subtracting the electrical resistance of cell culture inserts without cells.
Permeability Studies in Cell Lines The stock solution of PEGs was diluted in KBR, to a final concentration of 100 mM (donor solution). The pH of the donor solution was checked (7.4). After 20 min, the equilibration solution was replaced with (1.5 mL of Caco-2 and MDCKII, 0.4 mL of 2/4/A1) prewarmed donor solution and the transport of PEGs in apical to basolateral direction was followed for 240 min at 378C incubation in air under constant mild agitation (135 rpm Caco-2 and MDCKII, 100 rpm 2/4/A1). At regular time intervals, samples were withdrawn from acceptor chambers. In the Caco-2 and MDCKII experiments the withdrawn volume was replaced with fresh prewarmed HBSS. In the 2/4/A1 experiments the filters were transferred to new wells containing 1.2 mL prewarmed HBSS. Samples from the donor chambers were also taken at the end of the experiment in order to check for the recovery of the PEGs.
Permeability Studies in Human Jejunum The experiment was carried out in apical to basolateral direction at 378C. Changing the KBR solution on both sides of the membrane to prewarmed KBR and adding the PEG 400 solution to the mucosal side started the experiment. Samples of 100 mL were then withdrawn from the receiver (basolateral) and donor (apical) side at selected time points up to 210 min after start of DOI 10.1002/jps
PEGs were quantified using the combination of reversed-phase HPLC and electrospray ionisation mass spectrometry. The HPLC–ESI–MS method has been described in more detail by Palmgren et al.18 Briefly, the HPLC gradient was 2–31% ACN (containing 10 mM ammonium acetate, pH 6.62) in 4.5 min and the run was performed in 6 min. The chromatographic separation was performed using a Xterra MS C18 reversedphase column (2.1 mm 20 mm, 2.5 mm, Waters, Milford, MA) with a flow rate of 200 mL/min. HPLC–ESI–MS measurements were performed with a LCQ and LTQ quadrupole ion trap mass spectrometers equipped with an ESI ion source (Finnigan, San Jose, CA). Calibration curves were constructed by plotting chromatographic peak ratios of standard area/IS area versus concentration of the standard using linear regression. From these curves the coefficients of correlation (r2) were calculated. Calibration solutions and QC standards were prepared daily and analysed immediately after preparation. Data were processed using the Xcalibur software package version 1.4 SRI.
Calculation of Permeability Values The Papp values (cm/s) were calculated according to Papp ¼
dQ 1 dt A C0
(1)
where dQ/dt is the permeability rate at steadystate flux in the acceptor chamber (mol/s), A is the surface area of the cell monolayer/tissue (cm2) and C0 is the initial concentration in the donor chamber (mol/mL).19 Mass spectrometry was used to measure concentrations of each oligomer both in the donor and receiver phases.
Integrity Studies In order to investigate whether the study procedures influenced the permeability of Caco-2 and 2/4/A1 monolayers, the transport of labelled mannitol (Caco-2 3H-labelled, 2/4/A1 14C-labelled) in apical to basolateral direction was studied. In JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2170
LINNANKOSKI ET AL.
the Caco-2 cell line the labelled mannitol permeability was studied in the absence of the PEGs, whereas in the 2/4/A1 cell line the labelled mannitol permeability was studied in both the absence and presence of the PEGs. As a marker control of epithelial integrity of the human intestine 14C-labelled mannitol (0.7 mCi/mL) was added to the apical side at the start of the experiment. Analysis of the total radioactivity of 14C-mannitol in samples from the transport experiments was performed using liquid scintillation counting. The apparent permeability coefficient ( Papp) for 14C-mannitol was calculated for each segment according to Eq. (1). The mean Papp value for 14C-mannitol for the three segments was calculated to be 8.0(0.3) 106 cm/s. This is within the range of donor variability with respect to human jejunum permeability to mannitol ( Papp SD: mannitol 5.53 2.1 106cm/s, n ¼ 220 segments (¼50 donors) (unpublished AstraZeneca inhouse recordings).
membranes investigated in this work. The theory describes the rate of PEG permeability Jh/C by the following equation: Jh RT" 1 1 ¼ ¼ ½slope 12phNA l rs rs c
(2)
where rs is the radius of PEG oligomers, e the porosity of the membrane, l the jump length ˚ ), h the viscosity of water, R the gas (i.e., 3.1 A law constant, T the temperature and NA the Avogadro’s number. Eq. (2) predicts that the measured permeability, Jh/C, is inversely proportional to the radius of the drug molecule. The porosities of the investigated membranes (i.e., fraction of intercellular spaces) were estimated from the slope of this relation. The pore sizes were estimated by calculating the critical value of the molecular radius still able to penetrate through the paracellular pores. This was achieved by extrapolating f(1/rs) into zero permeability.
RESULTS Estimating the Pore Sizes and Porosity of the Membranes An effusion theory-based equation15 was used to estimate the porosities (e) and pore sizes of the
The results of the permeability experiments with the PEG oligomers are shown in Figure 1. Mass spectrometry allowed the analysis of each oligomer in the polydisperse mixtures of PEGs with
Figure 1. The relationship between permeability and the molecular weight of a PEG oligomers in 2/4/A1 (open squares), human intestine (closed triangles), Caco-2 (closed squares) and MDCKII (open triangles). The values represent mean apparent permeability coefficients SD. The inset is an enlargement of the Caco-2 and MDCKII cell line data. The data are based on permeability experiments with 11–12 time points and 3–6 replicate experiments with each membrane. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
mean molecular weights 200, 400, 600 and 1000. The data show that the overall paracellular permeability is the highest in the 2/4/A1 cell line, followed by the human intestine. The MDCKII and Caco-2 cell lines have the lowest permeabilities. The MDCKII and Caco-2 cell lines, and the human intestine show a biphasic relationship between permeability and molecular weight/ radius of the PEG oligomers, while a linear relationship is seen between permeability and molecular weight/radius in the 2/4/A1 cell model (Fig. 1). The permeability data were further analysed by the effusion-based theory (Eq. 2). This analysis
2171
allows determination of the membrane porosity and pore size(s). The analysis indicates that there are two distinct pore sizes in the MDCKII and Caco-2 cell models and in the human intestine (Fig. 2, biphasic permeability profiles). The radii of the small pores were in the same order of ˚ , Tab. 1). magnitude in the membranes (5.5–6.6 A ˚ The radii of the large pores were 10.1–30.5 A (Tab. 1). The 2/4/A1 monolayer was found to have ˚ ). pores of only one size (14.9 A The membrane porosities are given in Table 1 and Figure 3. The total porosity of the human intestine is 20 times that of the MDCKII cell line, over 10 times that of the Caco-2 cell model and
Figure 2. The relationship between permeability and the radius of a PEG oligomer in (A) MDCKII, (B) Caco-2, (C) 2/4/A1 cell lines and (D) human intestine. The closed squares represent experimental permeability through both small and large pores and the closed triangles through large pores. The open squares represent the calculated permeability through small pores. The porosity is estimated from the slope of the regression line (straight line: large pores; dotted line: small pores) according to the effusion-based theory. The pore size of the large pores is calculated by extrapolation of ˚ ) of the PEG oligomers are as the regression line to zero permeability. The rs values (A follows: 238 (4.51), 282 (4.87), 326 (5.15), 370 (5.47), 414 (5.78), 458 (6.03), 502 (6.27), 546 (6.55), 590 (6.77), 634 (7.03), 678 (7.24), 722 (7.45), 766 (7.69), 810 (7.89), 854 (8.09) and 898 (8.27). DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2172
LINNANKOSKI ET AL.
Table 1. Porosities and Pore Radii in MDCKII, Caco-2, 2/4/A1 Cell Lines and Human Intestine
Membrane Porosity 108 Total Small pores Large pores ˚) Pore size (A Small pores Large pores
MDCKII
Caco-2
2/4/A1
24 16 8
61 54 7
154
5.5 30.5
5.8 10.4
Human Intestine
154
531 442 89
14.9
6.6 10.1
over 3 times that of the 2/4/A1 cell model. The contribution of the small and large pores is shown schematically in Figure 3. The porosity created by small pores is greater than that created by large pores in the MDCKII and Caco-2 cell lines and the human intestine, whereas the total porosity of the 2/4/A1 cell line is constituted by large pores only.
DISCUSSION The paracellular route is considered to have an important role in the oral absorption of many small water-soluble compounds.20–22 Despite this, the dimensions of the intestinal paracellular space have remained poorly defined. In the present study, we profile the paracellular permeation route of the human intestinal epithelium and three commonly used cell models (MDCKII, Caco-2 and 2/4/A1). The study was conducted by measuring the permeabilities of polydispersed PEG oligomers in the membranes and applying an
Figure 3. Porosities constituted by large pores (grey) and small pores (black) in MDCKII, Caco-2, 2/4/A1 cell lines and human intestine. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
effusion-based theory to the PEG permeability data. This approach allows the quantitative characterisation of the paracellular route of the membranes investigated here. Besides the pore size (radius of the pore), the theory also solves the porosity (fraction of the paracellular space on the membrane surface) of the membrane, which is not possible with other techniques. The present results support earlier findings11,23 that the 2/4/A1 cell model predicts the paracellular permeability in human duodenum more reliably than the Caco-2 and MDCKII cell models (Fig. 1). The better prediction power of the 2/4/A1 cell line can be explained by its porosity being similar to that of the human duodenal epithelium: the porosity of the human duodenal epithelium is eight times higher than the porosity of the MDCKII and Caco-2 cell models, but fairly close to the porosity of the 2/4/A1 cell model (Tab. 1). Previous studies have shown that the Caco-2 model yields even 100 times lower permeability coefficients than the small intestine for low permeability drugs.9 This difference has been proposed to result, at least partly, from Caco-2 cells forming less permeable tight junctions than those found in the human small intestine.9 Our analysis indicates, however, that the pores in the Caco-2 monolayers are approximately of the same size as the pores in the human intestinal epithelium (Tab. 1). We conclude that a lower number of paracellular pores per cm2 in the Caco-2 cell monolayer explains the lower PEG permeability in the Caco-2 model (Tab. 1). Although the 2/4/A1 cell line and the human intestinal epithelium have similar porosities, the paracellular permeation properties of these membranes are not the same. The biphasic permeability profile of the human intestine suggests that the human intestinal epithelium has two distinct pore sizes, whereas the monophasic permeability profile of the 2/4/A1 cell line indicates that this membrane has pores of ˚ ). The biphasic one fairly large size (above 10 A permeability profiles observed in the MDCKII and Caco-2 cell lines indicate that the pores in these cell lines are also of two different sizes. In line with our findings, several previous studies suggest that there are two distinct pore sizes in the intestine and cell monolayers.24–27 Using a paracellular sieving model, Watson et al.27 showed that this was the case also for the Caco-2 cell model. The radius of the smaller ˚ , but with the pores was estimated to be 4.3–4.5 A sieving model it was not possible to estimate the DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
radius of the larger pores. Hollander25 suggested that the biphasic permeability profiles observed in intestinal preparations may reflect distinct populations of tight junction pores associated with villus tip and intestinal crypt cells. This theory does not, however, explain the two-component profile observed in intestinal monolayers. Several previous studies1,12,28 have estimated the pore radii of the intestinal cell models with the Renkin method.14 The Renkin method assumes random walk of diffusion in the pore and aims to account for the frictional resistance resulting from the molecules travelling through the pore. However, the validity of the Renkin correction is in doubt when the radius of the solute is over 0.4 times the radius of the pore.15,27 Under these conditions the assumption of random walk is not fulfilled, and the Renkin term exaggerates the diffusional resistance within the pore. With the Renkin method it is not possible to estimate the porosity of membranes. PEGs have been used extensively as a marker of intestinal.29,30 Some investigators have, however, questioned the suitability of them as a paracellular probes.31,32 For example, the intestinal permeability of PEG 400 is significantly higher than the permeability of saccharide markers of comparable molecular weight. Artursson et al.9 found that the permeability of PEG 194 was 6- to 28-fold greater than that of mannitol (molecular weight 182) in cell monolayers. The authors suggested that this could be explained by the molecular structure of PEGs: PEG oligomers are flexible polymers and, therefore, possible changes in their molecular dimensions may affect their permeation through the tight junctions. However, according to Watson et al.27 a relatively small difference in the radii of the probes will have profound effects on permeability particularly where they lie close to the radius of the paracellular pore. Thus, Watson et al. suggest that the higher than expected permeation of PEG 400 compared with other markers is due its smaller hydrodynamic radius. The molecular radii that were used in the calculations are based on molecular diffusivities in the solution assuming spherical shape of these homologous oligomers. This assumption may not be valid, but taking into account the real conformations of the PEG molecules is very difficult, and each PEG anyway has probably several conformations in dynamic motion in time and space. Taking into account molecular conformations is very complicated, and not necessarily reliable. Identical PEGs DOI 10.1002/jps
2173
were used for each membrane and the shape distributions of individual PEGs in the solution are not dependent on the bio-membrane, that is, they should not cause bias in the comparisons between the membranes. In calculating the PEG permeability values of the intestinal epithelium, we used the surface area of the orifice of the diffusion chamber, that is, the surface area of intestinal segment. Since the intestinal epithelium grows as folded structures, the effective surface area of the epithelium might be larger than the area used in our calculations, which would lead to overestimation of the permeability and porosity of the intestinal epithelium. However, the intestinal folds are generally not taken into account in studies of drug permeability in the intestine. Furthermore, since the villous structures are present in vivo, the number of pores per surface area of intestine corresponds to the values obtained with our method of calculation. It should also be noted that the pore size and porosity may vary for cell lines, depending on their origin, evolutionary history and the way they are cultivated. Our cells were cultivated according to the common procedure and the paracellular permeability values obtained in our laboratories are similar to most published data. As stated earlier, the Caco-2 cell model fails in predicting the human oral absorption of many poorly absorbed compounds. This is partly due to their low background permeation by passive diffusion. Quantitative analysis of the paracellular space is expected to provide a link between the passive diffusion in cell models (MDCKII, Caco-2 and 2/4/A1) and the small intestine. Characterising the paracellular space of epithelial cell models helps in bridging the gap in the predictions between human intestinal and cell model permeabilities. It should, however, be noted that the expression of transporters differs in cell models and the human intestine, which adds further uncertainty to cell model predictions. It is known that the Caco-2 model and the MDCKII variants frequently overestimate the importance of active transport and efflux in oral drug absorption.33 In conclusion, the paracellular routes of hydrophilic permeants in the intestine and three widely used cell models have been characterised. Two different pore sizes were detected in the human intestinal epithelium, Caco-2 cells and MDCKII monolayer, while only one in 2/4/A1 cells. The paracellular porosity of the human small intestine JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
2174
LINNANKOSKI ET AL.
and 2/4/A1 monolayers was larger than in the MDCKII and Caco-2 cells. 10.
ACKNOWLEDGMENTS We would like to thank Prof. Seppo Auriola for his ˚ sa advice regarding the LC–MS analysis and Ms. A Sjo¨ berg for preparation of the human jejunum segments. The material support of OrionPharma and National Agency of Technology of Finland is gratefully acknowledged.
11.
12.
REFERENCES 1. Knipp GT, Ho NF, Barsuhn CL, Borchardt RT. 1997. Paracellular diffusion in Caco-2 cell monolayers: Effect of perturbation on the transport of hydrophilic compounds that vary in charge and size. J Pharm Sci 86:1105–1110. 2. Pade V, Stavchansky S. 1997. Estimation of the relative contribution of the transcellular and paracellular pathway to the transport of passively absorbed drugs in the Caco-2 cell culture model. Pharm Res 14:1210–1215. 3. Adson A, Burton PS, Raub TJ, Barsuhn CL, Audus KL, Ho NFH. 1995. Passive diffusion of weak organic electrolytes across Caco-2 cell monolayers: Uncoupling the contributions of hydrodynamic, transcellular, and paracellular barriers. J Pharm Sci 84: 1197–1204. 4. He YL, Murby S, Warhurst G, Gifford L, Walker D, Ayrton J, Eastmond R, Rowland M. 1998. Species differences in size discrimination in the paracellular pathway reflected by oral bioavailability of poly(ethylene glycol) and D-peptides. J Pharm Sci 87: 626–633. 5. Kottra G, Fromter E. 1983. Functional properties of the paracellular pathway in some leaky epithelia. J Exp Biol 106:217–229. 6. Nagahara N, Tavelin S, Artursson P. 2004. Contribution of the paracellular route to the pH-dependent epithelial permeability to cationic drugs. J Pharm Sci 93:2972–2984. 7. Furuse M, Hirase T, Itoh M, Nagafuchi A, Yonemura S, Tsukita S, Tsukita S. 1993. Occludin: A novel integral membrane protein localizing at tight junctions. J Cell Biol 123:1777–1788. 8. Furuse M, Fujita K, Hiiragi T, Fujimoto K, Tsukita S. 1998. Claudin-1 and -2: Novel integral membrane proteins localizing at tight junctions with no sequence similarity to occludin. J Cell Biol 141: 1539–1550. 9. Artursson P, Ungell AL, Lofroth JE. 1993. Selective paracellular permeability in two models of intestinal absorption: Cultured monolayers of human
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
intestinal epithelial cells and rat intestinal segments. Pharm Res 10:1123–1129. Irvine JD, Takahashi L, Lockhart K, Cheong J, Tolan JW, Selick HE, Grove JR. 1999. MDCK (Madin–Darby canine kidney) cells: A tool for membrane permeability screening. J Pharm Sci 88:28– 33. Tavelin S, Taipalensuu J, Hallbook F, Vellonen KS, Moore V, Artursson P. 2003. An improved cell culture model based on 2/4/A1 cell monolayers for studies of intestinal drug transport: Characterization of transport routes. Pharm Res 20:373–381. Tavelin S, Taipalensuu J, Soderberg L, Morrison R, Chong S, Artursson P. 2003. Prediction of the oral absorption of low-permeability drugs using small intestine-like 2/4/A1 cell monolayers. Pharm Res 20:397–405. Fine KD, Santa Ana CA, Porter JL, Fordtran JS. 1995. Effect of changing intestinal flow rate on a measurement of intestinal permeability. Gastroenterology 108:983–989. Renkin EM. 1954. Filtration, diffusion, and molecular sieving through porous cellulose membranes. J Gen Physiol 38:225–243. Ha¨ ma¨ la¨ inen K, Kontturi K, Auriola S, Murtoma¨ ki L, Urtti A. 1997. Estimation of pore size and pore density of biomembranes from permeability measurements of polyethylene glycols using an effusion-like approach. J Control Release 49:97–104. Ungell AL, Nylander S, Bergstrand S, Sjoberg A, Lennernas H. 1998. Membrane transport of drugs in different regions of the intestinal tract of the rat. J Pharm Sci 87:360–366. Tavelin S, Milovic V, Ocklind G, Olsson S, Artursson P. 1999. A conditionally immortalized epithelial cell line for studies of intestinal drug transport. J Pharmacol Exp Ther 290:1212–1221. Palmgren JJ, Toropainen E, Auriola S, Urtti A. 2002. Liquid chromatographic-electrospray ionization mass spectrometric analysis of neutral and charged polyethylene glycols. J Chromatogr A 976:165–170. Artursson P. 1990. Epithelial transport of drugs in cell culture. I: A model for studying the passive diffusion of drugs over intestinal absorptive (Caco-2) cells. J Pharm Sci 79:476–482. Flanagan SD, Takahashi LH, Liu X, Benet LZ. 2002. Contributions of saturable active secretion, passive transcellular, and paracellular diffusion to the overall transport of furosemide across adenocarcinoma (Caco-2) cells. J Pharm Sci 91:1169– 1177. Lee K, Thakker DR. 1999. Saturable transport of H2-antagonists ranitidine and famotidine across Caco-2 cell monolayers. J Pharm Sci 88:680–687. Zhou SY, Piyapolrungroj N, Pao L, Li C, Liu G, Zimmermann E, Fleisher D. 1999. Regulation of paracellular absorption of cimetidine and 5-amino-
DOI 10.1002/jps
PARACELLULAR POROSITY OF HUMAN INTESTINAL EPITHELIUM
23.
24.
25.
26.
27.
28.
salicylate in rat intestine. Pharm Res 16:1781– 1785. Linnankoski J, Ranta VP, Yliperttula M, Urtti A. 2008. Passive oral drug absorption can be predicted more reliably by experimental than computational models—Fact or myth. Eur J Pharm Sci 34:129– 139. Fihn BM, Sjoqvist A, Jodal M. 2000. Permeability of the rat small intestinal epithelium along the villuscrypt axis: Effects of glucose transport. Gastroenterology 119:1029–1036. Hollander D. 1992. The intestinal permeability barrier. A hypothesis as to its regulation and involvement in Crohn’s disease. Scand J Gastroenterol 27:721–726. Soderholm JD, Olaison G, Kald A, Tagesson C, Sjodahl R. 1997. Absorption profiles for polyethylene glycols after regional jejunal perfusion and oral load in healthy humans. Dig Dis Sci 42:853– 857. Watson CJ, Rowland M, Warhurst G. 2001. Functional modeling of tight junctions in intestinal cell monolayers using polyethylene glycol oligomers. Am J Physiol Cell Physiol 281:C388–C397. Adson A, Raub TJ, Burton PS, Barsuhn CL, Hilgers AR, Audus KL, Ho NF. 1994. Quantitative
DOI 10.1002/jps
29.
30.
31.
32.
33.
2175
approaches to delineate paracellular diffusion in cultured epithelial cell monolayers. J Pharm Sci 83:1529–1536. Chadwick VS, Phillips SF, Hofmann AF. 1977. Measurements of intestinal permeability using low molecular weight polyethylene glycols (PEG 400). II. Application to normal and abnormal permeability states in man and animals. Gastroenterology 73:247–251. Ma TY, Hollander D, Krugliak P, Katz K. 1990. PEG 400, a hydrophilic molecular probe for measuring intestinal permeability. Gastroenterology 98:39–46. Iqbal TH, Lewis KO, Cooper BT. 1993. Diffusion of polyethylene glycol-400 across lipid barriers in vitro. Clin Sci (Lond) 85:111–115. Maxton DG, Bjarnason I, Reynolds AP, Catt SD, Peters TJ, Menzies IS. 1986. Lactulose, 51Crlabelled ethylenediaminetetra-acetate, L-rhamnose and polyethyleneglycol 400 [corrected] as probe markers for assessment in vivo of human intestinal permeability. Clin Sci (Lond) 71:71–80. Matsson P, Bergstrom CA, Nagahara N, Tavelin S, Norinder U, Artursson P. 2005. Exploring the role of different drug transport routes in permeability screening. J Med Chem 48:604–613.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 4, APRIL 2010