paracellular drug transport across intestinal epithelia: influence of charge and induced water flux

European Journal of Pharmaceutical Sciences 9 (1999) 47–56www.elsevier.nl / locate /ejps

Paracellular drug transport across intestinal epithelia: influence of chargeand induced water flux

a , b a a* ˚ ¨Johan Karlsson , Anna-Lena Ungell , Johan Grasjo , Per ArturssonaDepartment of Pharmacy, Division of Pharmaceutics, Uppsala University, Box 580, S-751 23 Uppsala, Sweden

b ¨ ¨Pharmacokinetics and Drug Metabolism, AstraZeneca R and D Molndal, S-431 83 Molndal, Sweden

Received 19 May 1999; accepted 18 June 1999

Abstract

The influence of drug charge and transepithelial water flux on passive paracellular drug transport was investigated in Caco-2 cellmonolayers and rat ileal mucosa in vitro. Three small hydrophilic compounds with different net charges (creatinine, erythritol andfoscarnet) were used as model drugs. A hypotonic glucose-rich solution was applied apically to induce epithelial absorption of water. Inthe Caco-2 monolayers, permeability to creatinine (positively charged) was 25-fold greater than to foscarnet (negatively charged),indicating a pronounced cation selective paracellular permeability. During apical exposure to the hypotonic glucose-rich solution,transport of all model drugs increased in both the absorptive and secretory directions. This enhanced transport coincided with a decreasein transepithelial resistance. Further, fluorescence and transmission electron microscopy indicated dilatations of the paracellular spaces butno damage to the cell membranes. These findings suggested that the enhancement in drug transport was attributable to increasedparacellular tight junction permeability rather than to ‘‘solvent drag’’. In the ileal segments, mucosal exposure to the hypotonicglucose-rich solution had no effect on transepithelial resistance and only a marginal increase in drug transport was observed. Takentogether, the modest absorption enhancement demonstrated in the in vitro models agrees with results obtained in vivo, supporting theconclusion that a more pronounced disruption of the tight junction barrier than that obtained through stimulation of epithelial absorptionof water is required for efficient enhancement of paracellular intestinal drug absorption. 1999 Elsevier Science B.V. All rightsreserved.

Keywords: Paracellular drug transport; Caco-2 cells; Rat ileum; Hypotonic glucose solution; Epithelial water transport; Epithelial permeability; Tightjunctions; Solvent drag

1. Introduction magnitude of the flux through each route is not known, ithas been suggested that a significant fraction is paracellular

Attempts to enhance the absorption of hydrophilic drugs (Chang and Rao, 1994). Intestinal absorption of smalland peptides have focused on ways to increase the paracellular marker molecules has been shown to increaseparacellular permeability of the intestinal epithelium, e.g., in direct proportion to the net water absorption, suggestingby the use of pharmaceutical additives. However, this that absorption could be mediated by ‘‘solvent drag’’ in theapproach has so far been unsuccessful because of other fluid absorbed through the paracellular pathway. (Pap-nonspecific effects of these additives on the intestinal penheimer and Reiss, 1987; Krugliak et al., 1989). Paracel-barrier (Hochman and Artursson, 1994). A more physio- lular absorption by ‘‘solvent drag’’ has been suggested tological approach may be to exploit the natural coupling be an important component for intestinal absorption ofbetween water and passive solute transport in the intestine. many hydrophilic drugs and peptides (Nellans, 1991;Water flux across the intestinal epithelium occurs via both Leahy et al., 1994; Pappenheimer et al., 1997).transcellular and paracellular routes. Although the relative Absorption of water across the intestinal epithelium can

be increased by hypotonicity or sodium-coupled transportof nutrients such as glucose. Both these mechanisms have

*Corresponding author. Present address: AstraZeneca R and Dbeen put to use in the formulation of oral rehydration¨ ¨Molndal, S-431 83 Molndal, Sweden. Tel.: 146-31-776-2677; fax: 146-solutions (Hunt et al., 1992). Hypotonic or glucose-rich31-776-3834.

E-mail address: [email protected] (J. Karlsson) solutions have increased the intestinal absorption of some

0928-0987/99/$ – see front matter 1999 Elsevier Science B.V. All rights reserved.PI I : S0928-0987( 99 )00041-X

48 J. Karlsson et al. / European Journal of Pharmaceutical Sciences 9 (1999) 47 –56

14but not all drugs in studies carried out in experimental [ C]-foscarnet sodium (trisodium phosphonoformic acid;¨animals and humans [reviewed by Lennernas (1995)], and 52 mCi/mmol) from Moravec (Brea, CA, USA).

in vitro in isolated intestinal tissues and cell monolayers Creatinine, foscarnet, erythritol and buffer reagents were(Atisook et al., 1990; Fricker and Drewe, 1995). The purchased from Sigma (St. Louis, MO, USA). Hoechstvariable results could be related to variations in the 33258 was purchased from Molecular Probes (Eugene,physicochemical properties of the drug such as size and/or OR, USA). Human colon carcinoma Caco-2 cells werecharge, different experimental methodologies, or local, obtained from the American Type Culture Collectionregional or interspecies differences in the intestinal tissues. (Rockville, MD, USA). Cell culture media and supple-

In vitro models for intestinal drug absorption, such as ments were obtained from Gibco through Life Tech-¨excised intestinal segments and monolayers of intestinal nologies (Taby, Sweden). Transwell and Snapwell filters

epithelial cells, provide the opportunity to investigate were from Corning Costar (Cambridge, MA, USA). Themechanisms of drug absorption at the epithelial level. The orbital shaker was an IKA-Shuttler model MTS 4 frommost popular human intestinal cell line for drug absorption Janke and Kunkel (Germany). The LabView programstudies is Caco-2 (Artursson et al., 1996a; Delie and (v2.2) was obtained from National Instruments (Austin,Rubas, 1997). The Caco-2 cell line has previously been TX, USA), the diffusion chambers were from Precisionused to study water and solute coupling in epithelial cells Instrument Design (Los Altos, CA, USA) and Ag/AgCl-without interference from supracellular structures (villae, electrodes were obtained from Radiometer (Copenhagen,crypts) and subepithelial components (Parisi et al., 1993), Denmark).and to study the transport of hydrophilic model compoundsunder anisotonic conditions (Noach et al., 1994). The 2.2. Cell culturepurpose of this study was to investigate how enhancedtransepithelial water flow (stimulated by hypotonic Caco-2 cells were maintained in culture according toglucose-rich solution) affects the paracellular transport of previous descriptions (Artursson et al., 1996b), and weredrugs of different charge across the intestinal epithelium used at passage numbers 90–105. Cells were seeded onunder well-controlled conditions in vitro. To investigate 24.5 mm diameter Transwell or 12 mm diameter Snapwell

5 2this, the transepithelial transport of three small hydrophilic filters (pore size 0.4 mm) at a density of 4.2?10 cells /cmmodel drugs (creatinine, erythritol and foscarnet) was and were cultured in a medium comprising Dulbecco’smeasured in Caco-2 monolayers and excised rat ileal Modified Eagle’s medium (4.5 g/ l D-glucose and 3.7 g/ lmucosa during experimentally stimulated water flux. A NaHCO ) supplemented with 10% heat-inactivated foetal3

hypotonic (¯200 mosmol /kg) glucose-rich solution was bovine serum, 1% nonessential amino acids, 100 U/mlapplied apically to obtain maximum stimulation of absorp- penicillin and 100 mg/ml streptomycin. Cell monolayerstive water flux without damage of the epithelial barrier were maintained at 378C in an atmosphere of 10% CO2

¨(Hunt et al., 1992; Lennernas et al., 1994). The compounds and 95% relative humidity. The culture medium wasused in this study were smaller than most conventional replaced every second day until use of the monolayersdrugs. The choice of small model drugs was motivated by between days 25 and 28.the low paracellular permeability of the Caco-2 mono-layers to hydrophilic marker molecules of ‘‘normal size’’ 2.3. Solutions(Artursson et al., 1993), by the observation that chargeselectivity is most pronounced for small molecules in The control Krebs–Ringer’s [4-(2-hydroxyethyl)-1-Caco-2 monolayers (Adson et al., 1994), and by studies on piperazine-ethanesulphonic acid (HEPES)-buffered] solu-the effect of induced water absorption in the human tion used in experiments with Caco-2 cell monolayers hadjejunum where the permeability to the larger drugs, the following composition (in mM): 115.0 NaCl, 11.5antipyrine, atenolol and enalaprilat were unchanged despite D-glucose, 20.0 HEPES, 4.7 KCl, 1.8 NaH PO , 0.42 4

an increase in net water absorption during perfusion with KH PO , 1.2 MgSO , 1.25 CaCl , 4.9 Na-glutamate, 5.42 4 4 2

¨hypotonic glucose-rich solution (Lennernas et al., 1994). Na -fumarate and 4.9 Na-pyruvate; adjusted to 7.4 with2

1.0 M NaOH; ¯310 mosmol /kg. The hypotonic glucosesolution had the same composition as the control Krebs–

2. Experimental procedures Ringer’s except for 15 mM NaCl and 80 mM D-glucose;¯205 mosmol /kg.

2.1. Materials The control Krebs–Ringer’s (bicarbonate-buffered) so-lution used in experiments with excised rat ileal segments

14[ C]-Creatinine (55 mCi /mmol) was obtained from had the following composition (in mM): 108.0 NaCl, 11.5American Radiolabeled Chemicals (St. Louis, MO, USA), D-glucose, 15 NaHCO , 4.7 KCl, 1.8 NaH PO , 0.43 2 4

14[ C]-erythritol (54 mCi /mmol) from Amersham (Arling- KH PO , 1.2 MgSO , 1.25 CaCl , 4.9 Na-glutamate, 5.42 4 4 214ton Heights, IL, USA), [ C]-mannitol (55 mCi /mmol) Na -fumarate and 4.9 Na-pyruvate; pH 7.4 with 95%2

from New England Nuclear (Boston, MA, USA) and O /5% CO ; ¯310 mosmol /kg. The hypotonic glucose2 2

J. Karlsson et al. / European Journal of Pharmaceutical Sciences 9 (1999) 47 –56 49

2solution had the same composition as the control Krebs– onset of each pulse. The R (V cm ) and PD (mV) weret

Ringer’s except for 8.0 mM NaCl and 80 mM D-glucose; determined from the linear relationship:¯205 mosmol /kg. The hypertonic Krebs–Ringer’s solu- V 5 (PD 1V ) 1 (R 1 R ) ? I (1)x s t s xtion (¯510 mosmol /kg) was prepared by the addition of200 mM sorbitol. The total osmolalities of all solutions where correction is made for the offset voltage between thewere measured with a vapour pressure osmometer (model electrodes (V ) and the composite series resistance of thes

5500 Wescor, Logan, UT, USA). bathing solutions and the supporting filter (R ). The Is sc2(mA/cm ) was calculated by:

2.4. Electron microscopy I 5 2 PD/R (2)sc t

Transmission electron microscopy was performed on The monolayers were equilibrated in Krebs–Ringer’sCaco-2 cell monolayers exposed to the hypotonic glucose solution for 30–40 min. Hypotonic glucose solution wassolution on the apical side for 5, 30 and 120 min at 378C. then added to the apical side, and variations of R , I andt scGlutaraldehyde-fixed monolayers were immersed consecu- PD were recorded every 5 min over the experimentaltively in 1% osmium tetroxide and 1% uranyl acetate, period of 120 min. To obtain corrected R , PD and It scdehydrated and embedded in Epon. Thin sections, stained values, measurements of the offset voltage between thewith uranyl acetate and lead citrate, were examined with a electrodes and series fluid resistance were performed priorPhilips 420 electron microscope operated at 60 kV. to each monolayer experiment using the same bathing

solutions but with empty filters mounted in the chambers.2.5. Fluorescence microscopy Thus, note that the reported values of R of the Caco-2t

monolayers during exposure to the hypotonic glucoseThe DNA staining dye, Hoechst 33258, was used to solution are corrected for the decreased conductivity of this

discern cells with damaged cell membranes, since it does solution.not permeate intact cell membranes (Lindmark et al.,1995). The monolayers were rinsed twice with Krebs– 2.7. Transport studies in Caco-2 monolayersRinger’s solution, incubated with 1 mg/ml Hoechst 33258in either Krebs–Ringer’s solution or hypotonic glucose 2.7.1. Drug transport experimentssolution on the apical side for 5, 30 and 120 min at 378C, The effect of the hypotonic glucose solution on passiverinsed twice with phosphate-buffered saline (PBS), and paracellular drug transport was investigated using threethen fixed for 10 min in 4% formaldehyde in PBS on ice hydrophilic compounds varying in molecular charge:before being rinsed again in PBS. The monolayer and the creatinine, erythritol, foscarnet (Table 1). Theoreticalsupporting polycarbonate filter were carefully cut out and estimates of the octanol /water partition coefficients (Logmounted in glycerol /PBS under a cover glass. The prepa- P) were determined using CLOGP v4.51 (Daylight C.I.S.,rations were examined using a Zeiss Axioskop fluores- Irvine, CA, USA) for creatinine and erythritol and forcence microscope (Oberkochen, Germany) fitted with a foscarnet using ProLogP v4.2 (CompuDrug NA, Roches-340 objective for water. ter, NY, USA) since appropriate fragment data for this drug

were not available.2.6. Electrical measurements in Caco-2 monolayers All measurements of drug and water fluxes were per-

formed in ambient atmosphere at 378C with Caco-2 cell2Measurements of transepithelial resistance (R ), sponta- monolayers grown on large Transwell inserts (4.71 cm )t

neous potential difference (PD) and short-circuit current placed directly in six-well culture plates rather than(I ) of the Caco-2 monolayers were performed using ansc

Table 1in-house computer-based system. Automatic data record-Physicochemical properties and oral absorption in humans of the investi-ing, analysis and display was provided by an applicationgated compoundsdeveloped from the LabView program. Caco-2 monolayers

a b cCompound MW Net charge Log P % Absorbed(grown on Snapwell filters) were mounted in verticaldiffusion chambers which were maintained at 378C by a Creatinine 113 11 21.8 80

Erythritol 122 0 23.0 90heating block. The chambers were equipped with Pt-elec-Foscarnet 126 22.8 21.8 17trodes for passing of current and Ag/AgCl-electrodes

aconnected to the chambers via salt bridges (3 M KCl/2% Net charge at pH 7.4. Creatinine has pK values of 4.8 and 9.2;a

foscarnet has pK values of 0.5, 3.4 and 7.3 (Dawson et al., 1986; Swaanagar) for measurement of the transepithelial voltage. Direct a

and Tukker, 1995).current pulses (I ) of 0, 615, and 630 mA with a 235 ms bx Log octanol /water partition coefficients (Log P) were calculated asduration were sent across the monolayer. The voltage described in Drug transport experiments.

cresponse (V ), taken as the mean of eight recordings Average percentage of each drug absorbed after oral administration tox

¨sampled at 5 ms intervals, was measured 200 ms after the humans (Sjovall et al., 1988; Pappenheimer, 1990; Noda et al., 1994).


mounted in diffusion chambers. The monolayers were 2.8. Transport studies and electrical measurements inagitated during the experiments using an orbital plate rat ileal segmentsshaker that induced a planar rotary motion. Agitation of thecell monolayers during the transport experiments mini- Segments of the distal part of the ileum from femalemises the influence of the aqueous boundary layer, as Sprague-Dawley rats (250–300 g) were excised, strippeddescribed previously (Karlsson and Artursson, 1991). The of serosal layers and mounted in modified Ussing cham-14C-labelled compounds together with unlabelled drug bers as described previously (Ungell et al., 1992). Tissueswere dissolved in Krebs–Ringer’s solution or hypotonic were bathed with isotonic Krebs–Ringer’s solution (378C)glucose solution to a final concentration of 0.5 mM. The and gassed with 95% O /5% CO (pH57.4), and were2 2

monolayers were equilibrated in prewarmed Krebs–Rin- allowed to equilibrate in the chambers for 30–40 min. PDger’s solution for 20–30 min prior to the transport experi- and I were recorded simultaneously and R was calcu-sc t

ment. Two ml of fresh prewarmed drug solution was then lated as described (Ungell et al., 1992). Any tissue with a2added to either the apical or the basolateral side of the R ,30 V cm and a PD ,3 mV was omitted before thet

monolayers and 2.00 ml of drug-free solution was added to start of the experiment. The experiments were started bythe opposite side. The hypotonic glucose solution was replacing the mucosal solution with prewarmed hypotonicadded only to the apical side. Samples (200 ml, receiver; glucose solution containing the radiolabelled marker to be20 ml, donor side) were removed from both sides at 30-min tested and replacing the serosal solution with hypertonicintervals for up to 120 min. The volume removed was Krebs–Ringer’s solution. The latter solution was used toreplaced with appropriate fresh, prewarmed buffer solu- establish a hyperosmolar serosal compartment so as totion. Samples of the drug solutions were taken prior to the produce a similar condition to the in vivo multiplier systemexperiment. Drug transport was studied in both the apical- in the capillary network of the villi (Jodal et al., 1978).to-basolateral (AP-to-BL) and basolateral-to-apical (BL-to- The amount of sorbitol added (200 mM) to obtain 510

14 mosmol /kg did not induce any changes in the electricalAP) directions. The C-activities of the samples wereresistance or morphology of the investigated ileal seg-determined using liquid scintillation counting. Apparentments. Samples were withdrawn from the serosal chamberpermeability coefficients (P ; cm/s) in the AP-to-BLapp

at regular intervals for up to 150 min. The radioactivity of(P ) and BL-to-AP (P ) directions were calculateda–b b–a

the samples was determined using liquid scintillationaccording to Eq. (3):counting. P values were calculated according to Eq. (3).app

dQ 1 Control experiments were performed with isotonic Krebs–] ]]P 5 ? (3)app dt A ? C Ringer’s solution bathing both sides of the tissue prepara-0

tions. The electrical parameters were monitored every 5where dQ /dt is the rate at which the compound appeared min during the course of the experiments. To obtain R , PDtin the receiver solution (steady-state flux), C is the initial0 and I values corrected for the contribution of the offsetscconcentration of the compound in the donor solution and A voltage between the electrodes and series fluid resistance,is the surface area of the membrane. The initial donor measurements were carried out before each experimentconcentration was assumed to be constant during the time using the same bathing solutions but with no tissuescourse of these experiments. mounted in the chambers.

2.7.2. Assessment of net water flux 2.9. Statistical analysisThe effect of the hypotonic glucose solution on net

water flux was measured gravimetrically. Firstly, 2.00 ml All results are expressed as means6SD. The differencesof Krebs–Ringer’s solution was added to a well in the between mean values were analysed using the unpairedculture plate and weighed. A Transwell cell monolayer Student’s t-test (two-tailed) or one-way analysis of vari-plus 2.00 ml of hypotonic glucose or Krebs–Ringer’s ance when appropriate. Scheffe’s F-test was used forsolution were weighed, placed in the basolateral well and multiple comparisons. Values of P,0.05 were consideredthe total weight measured. After 2 h of incubation at 378C, statistically significant.the total weight and the weight of the basolateral andapical solutions were separately determined. By this pro-cedure it was possible to assess the change in volumes of 3. Resultsthe basolateral and apical compartments as well as the totalevaporated volume. The net water flux (expressed as ml / 3.1. Effects of apical exposure to hypotonic glucose

24.71 cm /2 h) was calculated as the average of the solution in Caco-2 cell monolayersobserved changes in the apical and basolateral fluidvolumes taking into account the evaporation during the 3.1.1. Morphology and membrane permeabilityexperiment. The evaporation from each compartment was Electron microscopic examination of Caco-2 cell mono-assumed to be proportional to the exposed surface area. layers showed that the control cells had the characteristics


Fig. 1. Transmission electron micrographs showing the effect of hypo-tonic glucose solution on morphology of the Caco-2 cell monolayers. (A)Control monolayers. The cells form well-differentiated polarized mono-layers with tight junctions and typical apical microvilli. (B) Monolayersexposed to hypotonic glucose solution apically for 120 min. The cellmembranes are intact. The intercellular spaces are markedly dilated Fig. 2. Time course of the effect of the hypotonic glucose solution on the(asterisk) and numerous vacuoles appear in the cytoplasm (arrows). The transepithelial resistance of Caco-2 cell monolayers. Control monolayersbars indicate 2 mm. (s) and monolayers exposed to hypotonic glucose solution on the apical

side (d) for 120 min. Data shown are resistance values expressed as apercentage of readings taken at time zero (mean6SD, n54–5).

of well-differentiated columnar cells with tight junctionsand apical microvilli (Fig. 1A). The lateral intercellularpathways varied from tortuous to straight and were gener- was observed under control conditions. Apical addition ofally without dilatations. Vacuoles were rarely observed in the hypotonic glucose solution induced an absorptive netthe cytoplasm. When the cell monolayers were exposed to water flux. Concurrent with this observation was anthe hypotonic glucose solution on the apical side, several increase in the osmolality of the applied solution bymodifications of the cell morphology, including large approximately 50 mosmol /kg over the experimental perioddilatations of the intercellular spaces and numerous vac- of 120 min (Table 2).uoles in the cytoplasm were observed (Fig. 1B). However,the cell membranes appeared to be intact and no cell 3.1.3. Transepithelial electrical parameterssloughing was observed. No significant accumulation in The time-dependent changes in transepithelial resistancethe cell nuclei of the membrane-impermeable DNA stain- after apical addition of the hypotonic glucose solution toing dye, Hoechst 33258, was detected following 120 min the Caco-2 monolayers are shown in Fig. 2 and Table 3.of exposure to the hypotonic glucose solution (data not The resistance of untreated monolayers remained stableshown). These results indicate that the cell membranes during the 120 min experiments. In contrast, the resistancewere not damaged by the hypotonic glucose solution, in was gradually reduced to ,60% of the initial value overagreement with the electron microscopic examination. 120 min in the presence of the hypotonic glucose solution.

These results indicate that the hypotonic glucose solution3.1.2. Net water flux induced reduction of the tight junction integrity, since the

The net water flux across Caco-2 monolayers under overall Caco-2 monolayer resistance is mainly determinedcontrol conditions (no osmotic gradient) and in the pres- by the resistance of tight junctions (Karlsson et al., 1995).ence of the hypotonic glucose solution was determined In the presence of apical hypotonic glucose solution, theregravimetrically (Table 2). A small secretory net water flux was a concentration gradient (DNaCl5100 mM) favouring

Table 2Net water flux and change in gradient of osmolality across Caco-2 cell monolayers following 2 h of apical exposure to hypotonic glucose solution

a b cCondition Net water flux Osmolality gradient (mosmol /kg)2(ml /4.71 cm /2 h)

t50 h t52 h

Control 78613 (secretion) 21 (307/308) 23 (317/320)*Treatment 2227628 (absorption) 2104 (204/308) 242 (252/294)

a Control: isotonic Krebs–Ringer’s solution on both sides of the cell monolayers. Treatment: apical exposure to hypotonic glucose solution.b 2Net water flux across monolayers grown on 4.71 cm Transwell filters. Values are mean6SD, n53–4.c Values in parentheses represent mean osmolalities of the apical /basolateral solutions. The relative SD was ,2% in all instances.* Significantly different from control (P,0.001).


Table 3 1). These results indicate selectivity for cations in theEffect of apical exposure to hypotonic glucose solution on transepithelial paracellular permeability of Caco-2 monolayers underresistance (R ), short-circuit current (I ) and potential difference (PD) int sc control conditions.Caco-2 cell monolayers; results are mean6SD, n54–5

Apical addition of hypotonic glucose solution enhanceda 2 2Condition R (V cm ) I (mA/cm ) PD (mV)t sc the transport of all three test compounds in both directions

Control across the Caco-2 monolayers (Fig. 3A–C). The increasest50 min 203615 1662 23.260.5 in P values ranged from 2.2- to 2.8-fold (P,0.001;a–bt55 min 203615 1662 23.160.5

Table 4). The corresponding increases in P valuesb–at5120 min 206615 1462 22.860.3ranged from 1.6- to 8.1-fold (P,0.05; Table 4). For

Treatment erythritol and foscarnet, the P /P ratios in the pres-a–b b–at50 min 217621 1763 23.760.8ence of hypotonic glucose solution were not statistically* *t55 min 220610 283613 18.262.4

† † different from the ratios observed under control conditions* * *t5120 min 114614 276611 8.660.6a (Table 4). Only in the case of creatinine was an increasedControl: isotonic Krebs–Ringer’s solution on both sides of the cell

P /P ratio in the presence of hypotonic glucosemonolayers. Treatment: apical exposure to hypotonic glucose solution. a–b b–a†* P,0.001 compared with control; P,0.001 for 120 min versus 5 solution observed (P,0.05, Table 4).

min.

passive diffusion of NaCl in the BL-to-AP direction. This 3.2. Effects of mucosally applied hypotonic glucosecondition caused apically directed I and PD to become solution on electrical parameters and drug transport insc

apical-side electropositive (Table 3), presumably due to rat ileal segmentscation selective permeability (P 4P ) of the paracellu-Na Cl

lar pathway (Davis et al., 1982; Reuss, 1992). The control value of transepithelial resistance in the rat2ileal segments was 43610 V cm (n524) and this value

remained stable over the 150 min experiments under3.1.4. Paracellular drug transport control conditions (Fig. 4). Exposure of the ileal segments

Marked differences in the P values for creatinine to mucosal hypotonic glucose solution did not induce anyapp

(positively charged), erythritol (neutral) and foscarnet change in the resistance (Fig. 4). The control values for Isc2(negatively charged) were observed in the Caco-2 mono- and PD in the ileal segments were 140640 mA/cm and

layers (Table 4). The P for creatinine was 2.5-fold 26.162.1 mV (mucosal-side negative), respectively. Pres-app

higher than that for erythritol and 25-fold higher than that ence of mucosal hypotonic glucose solution caused rapidfor foscarnet. The P values in the AP-to-BL and BL-to- reduction in I , with an average value of 218636 mA/app sc

2 2AP directions were comparable under control conditions, cm (n510) at 10 min and 254643 mA/cm at 150 mingiving P /P ratios close to 1. Thus, it can be assumed (P,0.001 compared to control). Accordingly, PD becamea–b b–a

that the mechanism of transport was predominantly passive slightly mucosal-side positive or near zero, with an aver-diffusion. Moreover, the hydrophilic properties of the age value of 0.861.5 mV (n510) at 10 min and 2.361.9compounds (Table 1) and their overall low P values, mV at 150 min (P,0.001 compared to control).app

27 27ranging from 0.3?10 to 7.7?10 cm/s, support the As in the Caco-2 monolayers, the P values of the testapp

notion that they were restricted to transepithelial diffusion compounds increased in the order foscarnet,erythritol,via the paracellular pathway (Artursson and Magnusson, creatinine (Table 5). The permeability to creatinine was1990; Artursson et al., 1993). By comparison, the per- however only 2.3-fold higher than to foscarnet (P,0.05),meability to mannitol, a well-established marker of indicating that effect of the molecular charge on the

27paracellular diffusion, was 2.1260.96?10 cm/s (n58). permeability was less pronounced in the ileal segmentsSince the molecular weights of the test compounds were than in the Caco-2 monolayers. Mucosal application of thealmost identical, it can be concluded that the differences in hypotonic glucose solution resulted in a small increase inP values were related to their different charges (Table P for erythritol but not for foscarnet and creatinineapp app

Table 4Permeability coefficients of small hydrophilic compounds across Caco-2 monolayers in the apical-to-basolateral (P ) and basolateral-to-apical (P )a–b b–a

directions under control conditions and during apical exposure to hypotonic glucose solution; values are mean6SD, n53–6

Compound Control Hypotonic glucose solution27 27 27 27P (10 cm/s) P (10 cm/s) P /P P (10 cm/s) P (10 cm/s) P /Pa–b b–a a–b b–a a–b b–a a–b b–a

** * *Creatinine 7.7060.34 9.8160.13 0.7860.04 21.6662.99 15.3262.23 1.4160.28** **Erythritol 3.1260.42 3.9160.40 0.8060.14 6.7260.59 8.2260.96 0.8260.12** **Foscarnet 0.3560.11 0.3260.12 1.0960.53 0.9260.08 2.6060.33 0.3560.05

* **P,0.05 compared with control; P,0.001 compared with control.


Fig. 4. Effect of mucosal exposure to hypotonic glucose solution ontransepithelial resistance in rat ileal segments. Control segments (s) andsegments exposed to hypotonic glucose solution (d) for 150 min. Datashown are resistance values expressed as a percentage of readings takenat time zero (mean6SD). Resistance values at time zero were 43611 V

2 2cm (n514) for the control segments and 4367 V cm (n510) for thesegments exposed to hypotonic glucose solution.

Table 5Permeability coefficients (P ) of small hydrophilic compounds in ratapp

ileal segments under control conditions and during mucosal exposure tohypotonic glucose solution; values are mean6SD; n53–6

26Compound Mucosal-to-serosal P (10 cm/s)app

Control Hypotonic glucose solution

Creatinine 8.761.1 9.863.2*Erythritol 6.562.1 11.361.8

Foscarnet 3.860.7 3.560.9

* Significantly different from control; P,0.05.

(Table 5). The limited permeability-enhancement wasconsistent with the retained transepithelial resistance.

4. Discussion

In this study, the effect of enhanced transepithelial waterflux (stimulated by a hypotonic glucose-rich solution) onthe paracellular transport of three small hydrophilic modeldrugs was investigated in Caco-2 cell monolayers and ratileal segments. The presence of the hypotonic glucosesolution decreased the transepithelial resistance of theCaco-2 cell monolayers and increased paracellular drugtransport to a limited extent. The underlying mechanismfor the enhancement effect appears to be attributable to anincrease in the permeability of the tight junctions ratherthan to ‘‘solvent drag’’. We conclude that a more pro-

Fig. 3. Effect of apical exposure to hypotonic glucose solution on nounced disruption of the tight junction permeabilitytransport of creatinine (A) erythritol (B) and foscarnet (C) across Caco-2 barrier than that induced by enhanced epithelial watercell monolayers. Apical-to-basolateral transport in control monolayers absorption is required to increase the paracellular transport(s) and monolayers exposed to hypotonic glucose solution (d).

of incompletely absorbed drugs in clinically effectiveBasolateral-to-apical transport in control monolayers (h) and monolayersamounts. This conclusion is supported by recent resultsexposed to hypotonic glucose solution (j). Data represent mean6SD,

n53–6. obtained in the perfused jejunum of humans, which


suggested that enhanced intestinal water absorption is of BL-to-AP direction (P /P ,1). Secondly, the increasea–b b–a

little importance for quantitative oral drug absorption in in Caco-2 monolayer permeability coincided with a clearhumans in vivo (Fagerholm et al., 1999). decrease in transepithelial resistance, which has been

A small secretory net water flux was observed in the extensively used as a marker for tight junction integrity inCaco-2 monolayers under control conditions. In contrast, a this model (Delie and Rubas, 1997). Thirdly, the paracellu-considerable absorptive net water flux was generated after lar spaces became structurally dilated but the cell mem-apical addition of the hypotonic glucose solution. These branes remained intact after exposure to hypotonic glucoseresults are in agreement with previous observations on solution. Fourthly, the drug concentration gradient duringwater transport in Caco-2 monolayers (Parisi et al., 1993). the time course of the experiments (DC5C 2C )donor receiver

The water flux in the rat ileal segments was not measured did not in any case increase by more than 15% over that ofin the present study since a large number of studies using the initial concentration gradient. Thus, the induced trans-different intestinal preparations indicate that luminal glu- epithelial water flow did not produce any significantcose and hypotonic solutions stimulate water absorption change in the drug diffusion-driving concentration gradient(e.g. Krugliak et al., 1989; Lu et al., 1992). across the epithelium (Fine et al., 1994). Moreover, the

The 25-fold higher permeability of untreated Caco-2 experiments were performed under well-stirred conditionsmonolayers to creatinine as compared to foscarnet indi- which effectively obviated any unstirred layer effects forcates that the paracellular pathway in this epithelium is the hydrophilic compounds used in this study (Karlssonstrongly selective for positively charged molecules. This and Artursson, 1991). Taken together, these results indi-conclusion is in agreement with previous reports (Adson et cated that the enhancement in transepithelial drug transportal., 1994), and supported by our observation that the in the presence of the hypotonic glucose solution waspresence of apical hypotonic glucose solution (BL-to-AP caused by an increase in the paracellular (diffusive) tightNaCl gradient) established a salt diffusion PD of approxi- junction permeability and by a change in the transepithelialmately 18 mV (apical-side electropositive). Such a response electrical gradient rather than ‘‘solvent drag’’.is indicative of paracellular cation selectivity (Reuss, Results published during the course of this study suggest1992). The more pronounced cation selectivity in Caco-2 that a substantial part of epithelial water transport ismonolayers than in rat ileal tissues could be a result either transcellular (Loike et al., 1996; Loo et al., 1996). Thus,of the lower paracellular permeability of this cell line or, although it has been hypothesised that alterations in themore likely, of the simplified nature of the homogeneous tight junction integrity could increase the hydraulic con-monolayers (Artursson et al., 1993). Furthermore, the ductance of the tight junctions and so facilitate osmoticpossibility that foscarnet was transported across the rat flow and ‘‘solvent drag’’ of small hydrophilic compoundsileal segments to some extent by a transcellular carrier- via the paracellular pathway, our data provide little supportmediated mechanism cannot be entirely excluded (Swaan for this suggestion (Pappenheimer and Reiss, 1987;and Tukker, 1995). If this was the case, the observed Atisook et al., 1990). The reason for the limited effect ofparacellular selectivity for charge may have been reduced. the hypotonic glucose solution on transepithelial resistance

The most likely explanation for the enhanced passive and transport of the hydrophilic model drugs in the rat ilealtransport of the model drugs across the Caco-2 monolayers segments is unclear but may be related to a more compen-in the presence of the hypotonic glucose solution is that the satory osmoregulatory response and leakier paracellularpermeability of the tight junctions was increased. Firstly, pathway in this tissue than in the Caco-2 monolayersfor the uncharged compound erythritol, the ratio between (Spring, 1991; Noach et al., 1994).the permeability coefficients in the AP-to-BL and BL-to- The magnitude of the increase in transepithelial drugAP directions was close to unity (P /P 51) and was transport in the presence of hypotonic glucose solution thata–b b–a

unchanged in the presence of the hypotonic glucose was observed here compares well with other studies. Forsolution. This indicated that only passive diffusion was instance, a 2.9-fold increase in the clearance of creatinine,responsible for the enhanced transepithelial transport. In associated with a 2-fold increase in water absorption, wasother words, although the hypotonic glucose solution demonstrated when rat small intestine was perfused withstimulated AP-to-BL net water flux, net drug transport by 25 mM glucose (Pappenheimer and Reiss, 1987). During‘‘solvent drag’’ could not be observed. The fact that the stimulation of water absorption by perfusion of rat jejunalhypotonic glucose solution caused the transepithelial PD to intestine with 112 mM glucose, the effective permeabilitybecome highly apical-side positive probably influenced the to the low molecular weight, hydrophilic drug acetamino-passive diffusion of the charged compounds (Davis et al., phen increased 2.3-fold (Lu et al., 1992). Studies in vivo in1982; Jackson, 1987). This might explain the divergent humans have demonstrated very small or no increases ineffects of the hypotonic glucose solution on the enhance- the intestinal absorption of small hydrophilic moleculesment of AP-to-BL and BL-to-AP fluxes of creatinine and under conditions of glucose- or hypotonicity-induced water

¨foscarnet. It is likely that the induced transepithelial PD absorption (Fine et al., 1993, 1994; Lennernas et al., 1994;favoured transport of creatinine in the AP-to-BL direction Bijlsma et al., 1995). In agreement with the results from(P /P .1) and favoured transport of foscarnet in the the present study using Caco-2 monolayers, the effectivea–b b–a


permeability in two models of intestinal absorption: cultured mono-permeability of human jejunum to creatinine was found tolayers of human intestinal epithelial cells and rat intestinal segments.increase 2.6-fold during perfusion with a 170 mosmol /kgPharm. Res. 10, 1123–1129.

hypotonic solution (Fagerholm et al., 1999). A limitation Artursson, P., Karlsson, J., Ocklind, G., Schipper, N., 1996b. Studyingfor using hypotonic, glucose-rich solutions as an effective transport processes in absorptive epithelia. In: Shaw, A. (Ed.),

Epithelial Cell Culture — A Practical Approach, Oxford Universitymethod for increasing the absorption of orally administeredPress, Oxford, pp. 111–133.drugs is that equilibration of osmotic gradients in the small

Atisook, K., Carlson, S., Madara, J.L., 1990. Effects of phlorizin andintestine is presumed to be quite rapid (Miller et al., 1979), sodium on glucose-elicited alterations of cell junctions in intestinalthus the duration of the effects observed are likely to be epithelia. Am. J. Physiol. 258, C77–C85.

Bijlsma, P.B., Peeters, R.A., Groot, J.A., Dekker, P.R., Taminiau, J.A.,short-lived in the in vivo situation.Van Der Meer, R., 1995. Differential in vivo and in vitro intestinalIn conclusion, the improvement in transepithelial trans-permeability to lactulose and mannitol in animals and humans: aport of paracellular marker molecules during conditions ofhypothesis. Gastroenterology 108, 687–696.

enhanced water absorption demonstrated here and in Chang, E.B., Rao, M.C., 1994. Intestinal water and electrolyte transport.previous studies may well be explained by increases in the Mechanisms of physiological and adaptive responses. In: Johnson,

L.R. (Ed.), Physiology of the Gastrointestinal Tract, third ed., Ravenpassive diffusional permeability of the tight junctionsPress, New York, pp. 2027–2081.rather than paracellular ‘‘solvent drag’’. Our results indi-

Davis, G.H., Santa Ana, C.A., Morawski, S.G., Fordtran, J.S., 1982.cate that stimulation of transepithelial water transport by Permeability characteristics of human jejunum, ileum, proximal colonusing a hypotonic glucose-rich solution induces increases and distal colon: results of potential difference measurements and

unidirectional fluxes. Gastroenterology 83, 844–850.in the paracellular tight junction permeability. However,Dawson, R.M.C., Elliot, D.C., Elliot, W.H., Jones, K.M., 1986. Data forthe level of enhancement of paracellular drug transport by

Biochemical Research, third ed., Oxford University Press, New York.this physiological mechanism is modest and probably Delie, F., Rubas, W., 1997. A human colonic cell line sharing similaritieslimited to molecules smaller than conventional drugs. with enterocytes as a model to examine oral absorption: advantages

and limitations of the Caco-2 model. Crit. Rev. Ther. Drug CarrierThus, in order to obtain quantitatively and clinicallySyst. 14, 221–286.significant enhancement of the paracellular intestinal ab-

¨Fagerholm, U., Nilsson, D., Knutson, L., Lennernas, H., 1999. Jejunalsorption of poorly absorbed hydrophilic drugs and pep-permeability in humans in vivo and rats in situ: investigation of

tides, a more pronounced disruption of the tight junction molecular size selectivity and solvent drag. Acta Physiol. Scand. 165,barrier than that obtained through enhanced transepithelial 315–324.

Fine, K.D., Santa Ana, C.A., Porter, J.L., Fordtran, J.S., 1993. Effect ofwater flux is required.D-glucose on intestinal permeability and its passive absorption inhuman small intestine in vivo. Gastroenterology 105, 1117–1125.

Fine, K.D., Santa Ana, C.A., Porter, J.L., Fordtran, J.S., 1994. MechanismAcknowledgements by which glucose stimulates the passive absorption of small solutes by

the human jejunum in vivo. Gastroenterology 107, 389–395.Fricker, G., Drewe, J., 1995. Enteral absorption of octreotide: modulationThis work was supported in part by The Swedish

of intestinal permeability by distinct carbohydrates. J. Pharmacol. Exp.¨ ¨Medical Research Council Grant 9478, Centrala Forsoksd-Ther. 274, 826–832.

¨jursnamnden Grant 93-11, The Swedish Fund for Scientific Hochman, J., Artursson, P., 1994. Mechanisms of absorption enhance-¨Research without Animals and Astra Hassle AB. The ment and tight junction regulation. J. Control. Release 29, 253–267.

authors would like to thank Anna Boklund and Lena Utter Hunt, J.B., Elliott, E.J., Fairclough, P.D., Clark, M.L., Farthing, M.J.,1992. Water and solute absorption from hypotonic glucose–electrolyte¨for excellent technical assistance, Goran Ocklind for helpsolutions in human jejunum. Gut 33, 479–483.¨with fluorescence microscopy, Tapio Nikkila for electron

¨Jodal, M., Hallback, D.-A., Lundgren, O., 1978. Tissue osmolality in¨microscopic studies and Filip Heijkenskjold for advice intestinal villi during luminal perfusion with isotonic electrolyte

concerning construction of equipment for electrophysio- solutions. Acta Physiol. Scand. 102, 94–107.logical measurements on Caco-2 monolayers. Jackson, M.J., 1987. Drug transport across gastrointestinal epithelia. In:

Johnson, L.R. (Ed.), Physiology of the Gastrointestinal Tract, seconded., Raven Press, New York, pp. 1597–1621.

Karlsson, J., Artursson, P., 1991. A method for the determination ofReferences cellular permeability coefficients and aqueous boundary layer thick-

ness in monolayers of intestinal epithelial (Caco-2) cells grown inAdson, A., Raub, T.J., Burton, P.S., Barsuhn, C.L., Hilgers, A.R., Audus, permeable filter chambers. Int. J. Pharm. 71, 55–64.

˚ ¨K.L., Ho, N.F.H., 1994. Quantitative approaches to delineate paracel- Karlsson, J., Grasjo, J., Artursson, P., 1995. Simultaneous determinationlular diffusion in cultured epithelial cell monolayers. J. Pharm. Sci. 83, of membrane and tight junction effects of absorption enhancers in1529–1536. Caco-2 cell monolayers. Pharm. Res. 12, S198.

Artursson, P., Magnusson, C., 1990. Epithelial transport of drugs in cell Krugliak, P., Hollander, D., Ma, T.Y., Tran, D., Dadufalza, V.D., Katz,culture. II. Effect of extracellular calcium concentration on the K.D., 1989. Mechanisms of polyethylene glycol 400 permeability ofparacellular transport of drugs of different lipophilicities across perfused rat intestine. Gastroenterology 97, 1164–1170.monolayers of intestinal epithelial (Caco-2) cells. J. Pharm. Sci. 79, Leahy, D.E., Lynch, J., Finney, R.E., Taylor, D.C., 1994. Estimation of595–600. sieving coefficients of convective absorption of drugs in perfused rat

Artursson, P., Palm, K., Luthman, K., 1996a. Caco-2 monolayers in jejunum. J. Pharmacokinet. Biopharm. 22, 411–429.¨experimental and theoretical predictions of drug transport. Adv. Drug Lennernas, H., 1995. Does fluid flow across the intestinal mucosa affect

Deliv. Rev. 22, 67–84. quantitative oral drug absorption? Is it time for a reevaluation? Pharm.¨Artursson, P., Ungell, A.-L., Lofroth, J.-E., 1993. Selective paracellular Res. 12, 1573–1582.


¨¨Lennernas, H., Ahrenstedt, O., Ungell, A.-L., 1994. Intestinal drug Pappenheimer, J.R., 1990. Paracellular intestinal absorption of glucose,absorption during induced net water absorption in man; a mechanistic creatinine, and mannitol in normal animals: relation to body size. Am.study using antipyrine, atenolol and enalaprilat. Br. J. Clin. Pharmacol. J. Physiol. 259, G290–G299.37, 589–596. Pappenheimer, J.R., Reiss, K.Z., 1987. Contribution of solvent drag

¨Lindmark, T., Nikkila, T., Artursson, P., 1995. Mechanism of absorption through intercellular junctions to absorption of nutrients by the smallenhancement by medium chain fatty acids in intestinal epithelial intestine of the rat. J. Membr. Biol. 100, 123–136.Caco-2 cell monolayers. J. Pharmacol. Exp. Ther. 275, 958–964. Pappenheimer, J.R., Karnovsky, M.L., Maggio, J.E., 1997. Absorption

Loike, J.D., Hickman, S., Kuang, K., Xu, M., Cao, L., Vera, J.C., and excretion of undegradable peptides: role of lipid solubility and netSilverstein, S.C., Fischbarg, J., 1996. Sodium–glucose cotransporters charge. J. Pharmacol. Exp. Ther. 280, 292–300.display sodium- and phlorizin-dependent water permeability. Am. J. Parisi, M., Escobar, E., Huet, C., Ripoche, P., Louvard, D., Bourguet, J.,Physiol. 271, C1774–C1779. 1993. Water handling in Caco-2 cells: effects of acidification of the

¨Loo, D.D., Zeuthen, T., Chandy, G., Wright, E.M., 1996. Cotransport of medium. Pflugers Arch. Eur. J. Physiol. 423, 1–6.1water by the Na /glucose cotransporter. Proc. Natl. Acad. Sci. 93, Reuss, L., 1992. Tight junction permeability to ions and water. In:

13367–13370. Cereijido, M. (Ed.), Tight Junctions, CRC Press, Boca Raton, FL, pp.Lu, H.-H., Thomas, J., Fleischer, D., 1992. Influence of D-glucose- 49–76.

¨ ¨ ¨induced water absorption on rat jejunal uptake of two passively Sjovall, J., Karlsson, A., Ogenstad, S., Sandstrom, E., Saarimaki, M.,absorbed drugs. J. Pharm. Sci. 81, 21–25. 1988. Pharmacokinetics and absorption of foscarnet after intravenous

Miller, D.L., Hamburger, S.T., Schedl, H.P., 1979. Effects of osmotic and oral administration to patients with human immunodefficiencygradients on water and solute transport: in vivo studies in rat virus. Clin. Pharmacol. Ther. 44, 65–73.duodenum and ileum. Am. J. Physiol. 237, E389–E396. Spring, K.R., 1991. Mechanism of fluid transport by epithelia. In: Field,

Nellans, H.N., 1991. Mechanisms of peptide and protein absorption. (1) M. and Frizzel, R.A. (Eds.), Handbook of Physiology, Section 6: TheParacellular intestinal transport: modulation of absorption. Adv. Drug Gastrointestinal System, vol. IV. Intestinal Absorption and Secretion.Deliv. Rev. 7, 339–364. American Physiological Society, Bethesda, pp. 195–207.

Noach, A.B.J., Sakai, M., Bloom-Roosemalen, M.C.M., de Jonge, H.R., Swaan, P.W., Tukker, J.J., 1995. Carrier-mediated transport mechanism ofde Boer, A.G., Breimer, D.D., 1994. Effect of anisotonic conditions on foscarnet (trisodium phosphonoformate hexahydrate) in rat intestinalthe transport of hydrophilic model compounds across monolayers of tissue. J. Pharmacol. Exp. Ther. 272, 242–247.human colonic cell lines. J. Pharmacol. Exp. Ther. 270, 1373–1380. Ungell, A.-L., Andreasson, A., Lundin, K., Utter, L., 1992. Effects of

Noda, K., Nakayama, K., Oku, T., 1994. Serum glucose and insulin levels enzymatic inhibition and increased paracellular shunting on theand erythritol balance after oral administration of erythritol in healthy transport of vasopressin analogues in the rat. J. Pharm. Sci. 81,subjects. Eur. J. Clin. Nutr. 48, 286–292. 640–645.

paracellular drug transport across intestinal epithelia: influence of charge and induced water flux

Documents