biorelevant refinement of the caco-2 cell culture model to assess efficacy of paracellular...

Biorelevant Refinement of the Caco-2 Cell Culture Model toAssess Efficacy of Paracellular Permeability Enhancers

TIMOTHY K. TIPPIN,1,2 DHIREN R. THAKKER1

1Division of Molecular Pharmaceutics, School of Pharmacy, The University of North Carolina at Chapel Hill,Chapel Hill, North Carolina

2Drug Metabolism and Pharmacokinetics Department, Metabolic Diseases Center of Excellence in Drug Discovery,GlaxoSmithKline, Research Triangle Park, North Carolina 27709

Received 17 January 2007; revised 26 April 2007; accepted 4 June 2007

Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21118

Corresponde962-0092, Fax:E-mail: dhiren_

Journal of Pharm

� 2007 Wiley-Liss

ABSTRACT: Epithelial cell monolayers are routinely used to evaluate efficacy of para-cellular permeability enhancers (PPEs). The purpose of the present work was toinvestigate how biorelevant refinements to the Caco-2 cell model impact in vitro efficacy(decrease in transepithelial electrical resistance and increase in mannitol permeability)of PPEs. Standard transport buffer was replaced by fasted-state simulated intestinalfluid (FaSSIF) or serum; or stirring was performed to decrease the unstirred water layerthickness. Apical FaSSIF significantly reduced the efficacy of amphiphilic PPEs palmi-toylcarnitine and hexadecylphosphocholine and reduced the amount of these PPEsassociated with cells. In contrast, FaSSIF did not affect efficacy of nonamphiphilicPPEs, ethylenediaminetetraacetic acid or 3-nitrocoumarin. Basolateral serum increasedthe transepithelial flux of PPEs, but did not lessen their potency. Stirring increased theflux of all PPEs, and also enhanced the potency of the amphiphilic PPEs. These resultsshow that inclusion of FaSSIF and agitation in the cellular models significantly alter theefficacy of amphiphilic PPEs but not of hydrophilic or lipophilic PPEs. Future studiesshould be directed at evaluating the ability to these refined in vitro systems to predictin vivo effects of PPEs. � 2007 Wiley-Liss, Inc. and the American Pharmacists Association J

Pharm Sci 97:1977–1993, 2008

Keywords: intestinal absorption; abso
rption enhancer; Caco-2 cells; paracellulartransport; tight junction; paracellular permeability enhancer; in vivo/in vitro correla-
tions (IVIVC); palmitoylcarnitine chloride; hexadecylphosphocholine; sodium caprate

INTRODUCTION

The oral route is the preferred route of adminis-tering drugs. However, orally administered drugsmay exhibit poor bioavailability because they arenot able to traverse the intestinal epithelium. Forexample, hydrophilic drugs, containing too many

nce to: Dhiren R. Thakker (Telephone: (919)(919) 966-3525;[email protected])

aceutical Sciences, Vol. 97, 1977–1993 (2008)

, Inc. and the American Pharmacists Association

JOURNAL O

hydrogen bond donor or acceptor moieties, pre-sence of a charged moiety, or large polar surfacearea, may not partition well into the lipophilic cellmembranes. Therefore, they are relegated to crossthe intestinal epithelium via the aqueous para-cellular pathway, which is inefficient due to therelatively low surface area and restricted passageacross the tight junctions.

Among the approaches to improve oral absorp-tion of poorly absorbed compounds is to modify theintestinal barrier by coadministration of a para-cellular permeability enhancer (PPE). In thismethod, cell-to-cell tight junctions are transiently

F PHARMACEUTICAL SCIENCES, VOL. 97, NO. 5, MAY 2008 1977

1978 TIPPIN AND THAKKER

and selectively loosened by coadministration ofa PPE, which opens the paracellular space toallow greater absorption of the poorly absorbedmolecule. This technique has advantages overapproaches that modify the compound, such asmaking a lipophilic prodrug, since it could beapplied generically to diverse hydrophilic com-pounds, including small molecules, as well aslarger peptide-containing molecules. Furthermore,since molecules remain in the paracellular spaces,they would bypass the intracellular metabolizingenzymes and the mucosal membrane efflux trans-porters that can pose a formidable biochemicalbarrier to transcellularly absorbed molecules.1

The search for a safe and effective PPE has beenunderway for the past 15 years. This search hasbeen greatly aided by the discovery and use ofin vitro models of the intestinal epithelium, suchas Caco-2 cells, which has allowed the rapidassessment of structure–efficacy relationships formultiple PPE analogs, as well as elucidation ofPPE mechanisms of action. However, the effectiveconcentration of a PPE in cell culture modelscan be much lower compared to that found toincrease absorption of hydrophilic moleculesin vivo in animal models. For example, palmi-toylcarnitine chloride (PCC) was effective in vitroat a concentration of 200mM,2 however, 10- to 100-fold higher concentrations were required toachieve significant in vivo absorption improve-ment.3 Similarly, sodium caprate increased thepermeability of hydrophilic solutes in vitro atconcentrations of 10–13 mM,4,5 but required 30- to50-fold higher concentrations to achieve signifi-

Table 1. Effective Concentration and Putative Mechanism

PPE Effective In Vitro Concentration

Amphiphilic PPEsHPC 0.03 mM increased mannitol

flux 10-foldPCC 0.2 mM increased Lucifer

Yellow flux 9-fold

Sodium caprate 10 mM increased mannitolflux 8-fold

Nonamphiphilic PPEsEDTA 1 mM increased FITC-Dextran4

flux 6-fold

3-NC 0.03 mM increased mannitolflux 5-fold

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 5, MAY 2008

cant in vivo absorption improvement.6,7 Thereasons for much higher concentrations that areneeded to achieve absorption improvement in vivohave not been thoroughly investigated.

We hypothesize that the in vitro epithelial cellmodel may be overestimating the potency of theseand other PPEs of the amphiphilic structuralclass, because this system lacks certain in vivocomponents. Indeed, others have shown thatimplementing biorelevant conditions, such asthe use of simulated intestinal fluid or serumcomponents in place of balanced salt solutions,has changed the permeability of certain drugmolecules across cell monolayers.8,9 Therefore, wesought to assess the impact of certain physiologicparameters on the potency of PPEs, particularlyamphiphilic PPEs, by implementing modifiedconditions in the cell-based systems. Threemodifications were examined: standard bufferswere replaced with simulated intestinal fluid onthe apical side of monolayers or serum on thebasolateral side of monolayers, or monolayerswere agitated at different rates to change theunstirred water layer thickness (Hw). In additionto PCC and sodium caprate, the amphiphilic PPEhexadecylphosphocholine (HPC), as well as twononamphiphilic PPEs, ethylenediaminetetraace-tic acid (EDTA) and 3-nitrocoumarin (3-NC) werealso included in the test set. These five PPEs havewell-established paracellular permeability enhan-cing properties, and represent the diverse chemi-cal space of PPEs studied to date as well as themultiple mechanisms by which PPEs act to opentight junctions (Tab. 1, Fig. 1).

of Action for Selected PPEs

Putative Mechanism of Action References

PLCb inhibition leading toactin disorganization

22

Membrane perturbation;no effect on actin organization,Ca2þ independent mechanism

2,35

Activation of PLC, increasedintracellular Ca2þ

4,5

Extracellular chelation ofCa2þ, leading to disruption ofextracellular E-cadherin

23

PLCg inhibition leading tohyperphosphorylation of ZO-2

10

DOI 10.1002/jps

Figure 1. Structures of selected PPEs.

IN VITRO PPE EFFICACY IN BIORELEVANT FLUIDS 1979

METHODS

Materials

The synthesis of 3-NC was as describedpreviously.10

D-(1-14C)-mannitol (55 mCi/mmol) waspurchased from American Radiolabeled Chemi-cals, Inc. (St Louis, MO). EDTA was obtainedfrom USB Corp. (Cleveland, OH). Other chemicalsand reagents, including PCC, HPC, sodiumcaprate, phosphatidylcholine, taurocholate, Hanks’balanced salt solution, HEPES buffer, and glucosewere purchased from Sigma Chemical Co. (St

DOI 10.1002/jps J

Louis, MO). Cell culture reagents were purchasedfrom Invitrogen (Carlsbad, CA). TranswellsTM

were purchased from Corning Costar (Cambridge,MA). Fasted-state simulated intestinal fluid(FaSSIF) was prepared as previously reported.11

The mixture contained 3 mM sodium taurocholateand 0.75 mM phosphatidylcholine in a buffered,isosmotic sodium salt solution (pH 7.1). Malehuman serum was obtained from Bioreclamation,Inc. (Hicksville, NY). The thawed serum wasfiltered through a 0.22 mM filter and adjusted topH 7.1 immediately prior to use.

OURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 5, MAY 2008


Cell Culture

The Caco-2 cell line was obtained from AmericanType Culture Collection. Caco-2 cells were grownin 75 cm2 tissue culture flasks in the presence ofcell culture medium (minimum essential mediumwith Earle’s salts, 10% (v/v) fetal bovine serum,and 1:100 dilutions of 100� stock solutions ofnonessential amino acids and antibiotic–antimy-cotic agents (penicillin G, streptomycin, ampho-tericin B)). Cells were passaged approximatelyevery 4 days. At approximately 90% confluency,the cells were detached using 0.05% trypsincontaining EDTA. Cells between passage num-bers of 26 and 40 were used for transport studies,except when noted otherwise. Cells were seeded ata density of 120000 cells/cm2 on polycarbonatemembranes of TranswellTM cell culture inserts(pore size 0.4 mm, diameter 1 cm2). Cell culturemedium was replaced every second day. Cellswere considered fully differentiated and ready forPPE experiments after 21 days. TEER values ofmature cell monolayers were between 300 and600 V cm2, as measured using Endohm Voltohm-meter (World Precision Instruments, Sarasota,FL) after correction for filter resistance.

Evaluation of Paracellular PermeabilityEnhancement

The ability of the selected PPEs to modulateparacellular permeability was measured aftertreatment of cell monolayers on the apical sideof the TranswellTM compartment with increasingconcentrations of PPEs. Two parameters weremeasured: (1) a decrease in TEER, measured atthe end of the treatment period, and (2) anincrease in the amount of the paracellular markercompound, mannitol, appearing in the basolateralside during the PPE treatment period.

Prior to the addition of PPEs, cell monolayerswere equilibrated for approximately 60 minin transport buffer (Hanks’ balanced salt solutionsupplemented with 25 mM D-glucose and 25 mMHEPES, pH 7.1) at 378C. At the end of theequilibration period, TEER was measured usingEndohm Voltohmmeter maintained at 378C withslide warmer. This measurement was defined asinitial TEER. Measurements of changes to cellmonolayers due to PPE treatment were initiat-ed upon replacement of the apical transportbuffer with dose solutions (0.4 mL) containing14C-mannitol and the PPE dissolved in transportbuffer, which was modified slightly to improve


stability as follows: For amphiphilic compounds,1% FaSSIF (v/v) in transport buffer was used tostabilize dose preparations. Dose solutions ofsodium caprate were prepared in Ca2þ-freetransport buffer. Dose solutions of 3-NC wereprepared in transport buffer adjusted to pH 5.6.These modifications were also included in controlwells and did not adversely affect the integrity ofthe monolayers. The basolateral compartmentcontained 1.5 mL of prewarmed transport buffer.14C-mannitol that had accumulated in the baso-lateral compartment over the entire treatmentperiod was quantified by mixing 0.5 mL aliquotswith 5 mL of UniverSolTM ES scintillation fluid(ICN, Costa Mesa, CA). Radioactivity in eachsample was measured in a Tri-Carb 2500TRscintillation spectrometer (Perkin-Elmer Lifeand Analytical Sciences, Boston, MA).

The effect of FaSSIF on the ability of the PPEsto increase paracellular permeability of mannitolor decrease TEER was determined by preparingdosing solutions of each PPE in FaSSIF ratherthan standard transport buffer. Control mono-layers for this condition contained FaSSIF with-out PPE in the apical compartment. Otherprocedures were the same as those describedabove for standard conditions.

The effect of human serum on the ability ofthe selected PPEs to increase paracellular per-meability of mannitol or decrease TEER wasdetermined by putting 1.5 mL of serum in thebasolateral compartment rather than standardtransport buffer. Control monolayers for thiscondition contained standard transport bufferwithout the PPE in the apical compartment andserum in the basolateral compartment. Dosingsolutions of PPEs as well as other procedures werethe same as those described above for standardconditions.

The effect of stirring and the associated changesto the unstirred water layer (Hw) on the para-cellular enhancing ability of the selected PPEswere determined in separate experiments in thepresence of standard transport buffer. The Hw wasreduced by agitating monolayers on a rotaryshaker (Lab-line Model 3520 Orbit Shaker) at low(80 rpm), moderate (120 rpm) and high (160 rpm)speeds in the presence of increasing con-centrations of PPEs. The marker compound 3H-testosterone (100 mM) was used to calibrate thechange in Hw at each stirring rate as describedpreviously,12,13 using the equations shown below.In order to simulate the transient exposure to aparticular intestinal segment, the duration of PPE

DOI 10.1002/jps


treatment was reduced to 15 min. Controlmonolayers for this condition contained standardtransport buffer without PPE in the apicalcompartment and were agitated at the samespeeds as PPE-treated monolayers. Dosing solu-tions of PPEs were prepared in the same manneras described for standard conditions.

PPE Disposition Across Cell Monolayers

The disposition of the PPEs across the cellmonolayer system was assessed under standardand modified conditions. At the end of the treat-ment period, aliquots of apical and basolateralcompartments were mixed with ethanol and werecentrifuged at 14000 rpm (20800g) for 3 min asnecessary in order to remove precipitated compo-nents prior to analysis by LC–MS as describedbelow.

Measurement of the PPE that was associatedwith cell monolayers was accomplished by rinsingthe cell monolayer once with cold transport buffer,followed by excising the filter-monolayer from theplastic support and immersing in 1 mL of ethanol.The sample was vortexed and sonicated briefly tolyse cells. Aliquots of the cell extract were furtherdiluted with ethanol and analyzed by LC–MS asdescribed below.

LC–MS Analysis of PPEs

Separation of PPEs from buffer components wasaccomplished with an Agilent high performanceliquid chromatograph (LC, Palo Alto, CA) using aPhenomenex C18 Luna column (2 mm� 100 mm,3 mm particle size), and a mobile phase consistingof acetonitrile and 13 mM ammonium formate(pH 3.5) (or ammonium acetate (pH 6.8) for sodi-um caprate analysis) at a flow rate 0.3 mL/min.The percentage of acetonitrile was changed overa 2–3 min linear gradient so that the PPEswere retained on the LC column for 4–6 min.This retention allowed the first 3 min of eluateto be diverted to waste. PPEs were detected usinga Sciex API 100 mass spectrometer (AppliedBiosystems, Foster City, CA) equipped with aTurboionspray interface operated in the positiveionization mode, except for sodium caprate, whichwas detected in negative ionization mode. Nitro-gen was used as both the sheath and drying gas ata pressure of 10 arbitrary units and a flow-rate7 L/min, respectively. The spray voltage was set at5.0 kV and the drying gas temperature was set at

DOI 10.1002/jps J

4258C. Declustering potential and focusing poten-tial were optimized at 25 and 225 V, respectively.For some analyses, an Agilent mass-selectivedetector equipped with an electrospray interfacewas used, with nitrogen as the drying gas at aflow-rate 12 L/min and temperature of 3508C.The nebulizing pressure was set at 35 psig. Thecapillary voltage was set at 4.0 kV. The fragmen-tor voltage was set at 100 V and the dwell time was1000 ms. The elution of each PPE was monitoredby single ion monitoring for the MþHþ ions at m/z400.5 (PCC), 408.5 (HPC), 192 (3-NC), and M�Hions for sodium caprate at 171. Mass/charge peakarea was quantified using calibration standardswith known PPE concentrations prepared inblank matrix. LC–MS analysis was successfullyconducted on PCC, HPC, sodium caprate and3-NC. EDTA was not suited to LC–MS measure-ments, and therefore not included in the PPEdisposition studies. 3-NC samples were acidifiedafter collection to pH 2 in order to preventdegradation of the compound. Basolateral PPEconcentrations were input into Eq. (1) to calculatePapp values for each PPE.

Calculations

TEER was measured across control and PPE-treated monolayers at the end of the treatmentperiod. The change in TEER due to PPE-treatmentwas normalized to the change in TEER measuredacross control monolayers (% control). The %control TEER values for individual wells wereplotted versus the PPE concentration, and fit to afour parameter Hill model with power-of-the-mean variance. The median EC50 TEER, which isdefined as the PPE concentration which decreasesTEER to 50% of the control TEER value, wasestimated from the fitted curves.

The amount of 14C-mannitol appearing in thebasolateral compartment at 378C from t¼ 0 tot¼T after each treatment condition was used tocalculate the mannitol apparent permeabilitycoefficient (Papp) according to Eq. (1),

Papp ðnm=s ð1 � 10�7 cm=sÞÞ ¼ dQ

dt

1

A

1

C0(1)

where dQ/dt is the amount (Q) of mannitolmeasured in basolateral compartment after timeT, A is the membrane surface area and C0 is theinitial concentration of compound.

The marker compound 3H-testosterone wasused to calculate Hw at different stirring speeds.



In order to insure that sink conditions weremaintained over the experimental period, theapical compartment was transferred to naı̈vebasolateral solutions at 5 min intervals up to20 min. The testosterone flux was obtained from alinear fit of the testosterone amount versus timeplot and input into Eq. (1) to obtain a Papp value fortestosterone at each stirring speed. The testoster-one Papp values at stirring rates of 40, 80, 120, 160,and 180 rpm were input into Eqs. (2)–(5) todetermine Hw,

Rapp ¼ 1

Papp¼ 1

Pwþ 1

Pcþ 1

Pf(2)

1

Pf¼ Hf

npr2pDaq

(3)

1

Pw¼ 1

KV0:8¼ Hw

Daq(4)

1

Papp¼ 1

KV0:8þ ð1=Pc þ 1=Pf Þ (5)

where the apparent resistance (Rapp) to testoster-one flux across the monolayer system is equal tothe reciprocal of apparent permeability (Papp) oftestosterone. Papp is comprised of sum of thepermeability across the unstirred water layer(Pw), the cell (Pc), and the filter (Pf). The 1/Pf maybe determined from Eq. (3), where Hf is thethickness of the filter (10 mm), n is the number ofpores per filter area (1� 108/cm2), rp is the poreradius (0.2 mm), and Daq is the aqueous diffusioncoefficient for testosterone (7.8� 10�6 cm2/s).12

The parameter 1/Pw is equal to 1 divided byconstant (K) comprised of the Daq, buffer viscosityand geometrical factors of the Caco-2 cell modeltimes the rate of stirring, V, in terms of rpm,raised to the 0.8 power.13 Substitution of Eq. (4)into Eq. (2) yields Eq. (5) which yields the term Pc

from a plot of 1/Papp versus V0.8. Substitution of Pc

and Pf into Eq. (2) allows calculation of Pw. Usingthe relationship between Daq and Pw in Eq. (4), thedesired term Hw may be obtained at each stirringrate.

Data Analysis

Data are expressed as mean� standard deviation(SD) from at least three measurements. Statisticalsignificance for all studies, except EC50 TEER

comparisons, was evaluated using a two sample


Student’s t-test. The samples were assumed tohave unequal variance. Significant differenceswere assigned at p< 0.05. For comparisons ofEC50 TEER in standard versus modified conditions,Bayesian methodology was employed. A four-parameter Hill model was fit to the dose–responsedata. Because there was no prior information(a necessary component for Bayesian modeling),normally distributed priors with extremely largevariances were used, which is nearly non infor-mative. The model was used to back-calculate thePPE concentration, that yields a relative TEERvalue of 0.5 (i.e., EC50 TEER). The Bayesianmodeling method produces a posterior distribu-tion for the set of parameters, which can be used todraw inference about the EC50 TEER. The medianfold difference between the EC50 TEER values forstandard and modified conditions were determin-ed along with 95% Bayesian credible intervals(i.e., the fold difference is inside the interval with95% probability). If the median fold difference forno effect (i.e., fold difference¼ 1) was not withinthe 95% Bayesian credible intervals for fold dif-ferences, the fold difference result was consideredsignificant.

RESULTS

PPE Potency in Caco-2 Cells

The effect of the amphiphilic PPEs, HPC, PCC,and sodium caprate, as well as the two non-amphiphilic PPEs, EDTA and 3-NC, on TEER andmannitol permeability across Caco-2 cell mono-layers is shown in Figure 2. After 60 minincubation with HPC, monolayer resistance wasreduced and the apparent permeability coefficient(Papp) of 14C-mannitol was increased in a dosedependent manner. A much larger increase inmannitol Papp was observed over the 60–120 minperiod after HPC treatment (data not shown),indicating a lag time in the onset of substantialimprovement of mannitol flux versus the changein TEER. This was due in part to the slow onset ofTEER decrease after treatment with HPC underthese conditions, although others have reportedthat differences in baseline TEER, or decreasesin TEER due to PPEs, do not always result inchanges in permeability of hydrophilic solutes.14,15

Over the same concentration range, PCC caused adecrease in TEER as well as a greater increasein mannitol Papp compared to HPC-treated cells.The maximum increase in mannitol Papp caused

DOI 10.1002/jps

Figure 2. Effect of PPEs on TEER (solid circles) and mannitol permeability (opencircles) across Caco-2 Cells. TEER was measured across Caco-2 cell monolayers 60 minafter apical treatment with solutions of PPEs; mannitol permeability coefficient (Papp)values were calculated from the amount of 14C-mannitol appearing in the basolateralcompartment during the 0–60 min period after cotreatment with PPEs. Solutions forHPC, PCC and sodium caprate (NaCap) were prepared in 1% (v/v) FaSSIF, and solutionsfor EDTA and 3-NC were prepared in unmodified transport buffer. The dose solution forsodium caprate was free of Ca2þ ions to avoid precipitation. The dose solution for 3-NCwas adjusted to pH 5.6. EDTA was added to apical and basolateral chambers.Mean values from at least three monolayers are plotted. Error bars indicate standarddeviation.

DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 97, NO. 5, MAY 2008



by 500 mM PCC was approximately 10-fold abovethat in control monolayers. Sodium capraterequired much higher concentrations to producedecreases in TEER and increases in mannitolPapp, with significant changes occurring after anapical concentration of approximately 10 mM. Theeffective concentration for sodium caprate is closeto the reported critical micelle concentration(CMC) for sodium caprate of 13 mM (determinedin standard transport buffer16). However, theeffective concentrations of PCC and HPC werewell above their reported CMCs (PCC¼ 50–75mM17; HPC¼ 10mM18), indicating that micellesmay play a role, although are not essential, in theparacellular permeability enhancing activity forthese compounds. The rank order of potency forthe amphiphilic PPEs in Caco-2 cell monolayersunder standard conditions was PCC>HPC>sodium caprate.

The nonamphiphilic compound EDTA, whenapplied to both the apical and basolateral com-partments for 60 min, caused increases in man-nitol Papp beginning at 2 mM, with a maximumincrease of approximately fivefold. Higher con-centrations of EDTA up to 20 mM did not causeany further increase in mannitol Papp or decreasein TEER (data not shown). When 3-NC wasapplied in modified transport buffer (pH 5.6), itcaused a threefold increase in mannitol Papp at thehighest concentration tested (600 mM), with adecrease in TEER to approximately 35% of TEERin control monolayers. Thus, 3-NC is slightly morepotent than the other nonamphiphilic compoundEDTA under standard conditions.

PPE Potency in Caco-2 Cell ModelUsing Biorelevant Conditions

Fasted-State Simulated Intestinal Fluid (FaSSIF)in the Apical Compartment

In order to simulate the GI environment presentin the lumen of the upper small intestine, anisosmotic phosphate buffer solution that con-tained 3 mM taurocholate and 0.75 mM phospha-tidylcholine was used in the apical compartmentof Caco-2 cell TranswellTM model system in placeof standard transport buffer. This proportion oftaurocholate and phosphatidylcholine has pre-viously been reported to be a good mimic of fasted-state intestinal fluid,11 and has also been shown tobe nontoxic to Caco-2 cell monolayers.19 WhenFaSSIF alone was applied to the apical side of


Caco-2 cell monolayers, a drop in TEER ofapproximately 20% was observed (data notshown). This drop in TEER due to FaSSIF hasbeen observed by others and has been explained tobe due to differences in the salt composition ofFaSSIF compared to that of the transport buffersolution on the basolateral side of the mono-layer.19

When solutions of PCC were prepared inFaSSIF rather than in the standard dosingvehicle, much higher PCC concentrations wererequired to cause TEER decrease of the samemagnitude (Fig. 3a). The EC50 TEER, which isdefined as the enhancer concentration thatdecreases the TEER to 50% of the TEER measuredin control monolayer, increased from approxi-mately 160 mM in the standard dosing vehicle to2000 mM in the presence of 100% FaSSIF in theapical compartment. When the effect of PCC onTEER was examined at 250 mM in differentdilutions of FaSSIF, it was clear that FaSSIF,even at 25% (v/v) concentration, did not allow PCCto exert any effect on Caco-2 cell monolayers(Fig. 3b). Partial effect of PCC was observed at10% (v/v) FaSSIF concentration, and only at 1%(v/v) FaSSIF concentrations did PCC exert its fulleffect. The data on the percentage of PCC dosethat was associated with the cells show thatFaSSIF, at 10% (v/v) and higher concentrations,sequesters PCC, thus reducing its cell-associatedconcentrations by as much as threefold at 100%FaSSIF (Fig. 3b).

The effect of FaSSIF on the EC50 TEER of all thePPEs examined in this study is shown in Table 2.FaSSIF had no effect on the two nonamphiphilicPPEs, EDTA and 3-NC. It had a modest effect onthe efficacy of HPC, while for the structurallysimilar PCC, FaSSIF reduced efficacy by >10-fold.Consequently, HPC, which was approximatelyhalf as potent as PCC in the standard buffer, was>3-fold more potent than PCC in 100% FaSSIF.Although sodium caprate is amphiphilic, FaSSIFhad no effect on its potency.

At concentrations approximately equivalent toEC50 TEER, the PPEs caused significant increasesin the permeability of mannitol, a paracellularmarker (Fig. 4). The enhancement of the mannitolpermeability caused by HPC and PCC, preparedin standard transport buffer was completelyeliminated when dose solutions were preparedin 100% FaSSIF. Sodium caprate and the non-amphiphilic compounds EDTA and 3-NC, on theother hand, remained fully effective as enhancersof mannitol permeability even in the presence of

DOI 10.1002/jps

Figure 3. Impact of fasted-state simulated intestinal fluid (FaSSIF) on the ability ofPCC to cause a decrease in TEER across Caco-2 cell monolayers. (a) Effect of 1% (no effectFaSSIF concentration) versus 100% FaSSIF on potency of PCC to cause TEER decrease.TEER was measured across Caco-2 cell monolayers 60 min after apical treatment withsolutions of PCC as described in Methods. PCC solutions were prepared in eithertransport buffer containing 1% (v/v) FaSSIF (solid circles) or in 100% FaSSIF (opencircles). Mean values from at least three monolayers are plotted. Error bars indicatestandard deviation. Arrow indicates EC50 TEER for PCC in 100% FaSSIF. (b) Effect ofFaSSIF concentrations on the efficacy of PCC (250 mM) to cause a decrease in TEERacross Caco-2 cell monolayers and on PCC cell association. PCC solutions (250 mM) wereprepared in different dilutions of FaSSIF (v/v in transport buffer). Cell monolayers wererinsed 1� following TEER measurement at 60 min endpoint and extracted with ethanol.The amount of PCC in the cell extract was determined by LC–MS as described inMethods, and normalized to the amount of PCC initially added (% dose). Mean TEERvalues (vertical bars) and PCC % dose remaining associated with cells (triangles) from atleast three monolayers are plotted. Error bars indicate standard deviation. Asterisksindicate a significant difference (p< 0.05) between parameters after PCC treatment indiluted FaSSIF versus PCC treatment in transport buffer alone.


100% FaSSIF. In fact, the increase in mannitolPapp resulting from apical and basolateral treat-ment with 2 mM EDTA seemed to be enhancedthreefold in the presence of FaSSIF. The reasonfor this FaSSIF-induced enhancement of mannitolPapp by EDTA is unknown.

Consistent with the observed changes inpotency, the amount of HPC and PCC in cellextracts was significantly reduced in the presenceof FaSSIF (Fig. 4, inset). However, no change inthe amount of 3-NC or sodium caprate in cellextracts was observed. These data suggest thatFaSSIF decreased the potency of HPC and PCCby their incorporation into the taurocholate–phosphatidylcholine mixed micelles of FaSSIFthereby reducing their access to the cell.

DOI 10.1002/jps J

Serum in the Basolateral Compartment

In order to simulate the in vivo environment onthe serosal side of the small intestine, humanserum was used in the basolateral compartment ofCaco-2 monolayers grown on TranswellsTM inplace of aqueous transport buffer. This conditionby itself did not change the TEER or mannitolpermeability across the cell monolayers.

Since many of the PPEs tested in this study arelipophilic and could potentially be bound by serumcomponents, it was important to first examinewhether serum in the basolateral compartmentwould impact the disposition of the PPEs. Asshown in Table 3, the Papp values of HPC and PCCin the presence of standard buffer were quite low


Table 2. EC50 TEER Evaluated in Caco-2 CellMonolayers After PPE Treatment Using StandardBuffer or FaSSIF in the Apical Compartment

PPE

EC50 TEER (mM)a

FoldIncreaseb

StandardBufferc FaSSIFd

HPC 350 800 2.3�

PCC 160 2000 12.5�

Sodium caprate 11000 8700 0.8EDTA 2200 1800 0.83-NC 510 350 0.7

aTEER was measured across Caco-2 cell monolayers 60 minafter apical treatment with solutions of PPEs prepared ineither standard transport buffer or FaSSIF (n¼3 per concen-tration, per condition) as described in Methods.

bThe median fold-difference between the EC50 TEER curvesfor standard and modified conditions.

cStandard buffer for HPC, PCC and sodium caprate was 1%(v/v) FaSSIF in transport buffer. Ca2þ-free transport bufferwas used for sodium caprate solutions. Standard buffer forEDTA and 3-NC was not modified, except that pH was adjustedto 5.6 in 3-NC solutions. EDTA was added to both the apicaland basolateral compartments.

dFasted-state simulated intestinal fluid was used in theapical compartment with PPEs instead of standard transportbuffer.

�The fold difference was significant based on Bayesiananalysis, as described in Methods.

Figure 4. Mannitol Papp and PPE cell associationevaluated in Caco-2 cell monolayers after PPE treat-ment with standard buffer or FaSSIF in the apicalcompartment. Mannitol Papp was calculated from theamount of 14C-mannitol appearing in the basolateralcompartment during the 0–60 min period (0–90 min forHPC) after cotreatment with a single concentration ofPPE (250 mM HPC, 250 mM PCC, 13 mM sodiumcaprate (NaCap), 2 mM EDTA and 600 mM 3-NC). PPEswere prepared in either transport buffer containing 1%(v/v) FaSSIF (open bars) or in 100% FaSSIF (solid bars).For EDTA experiments, a second dose solution wasprepared in transport buffer and added to the basolat-eral compartment. Inset: Cell monolayers were rinsed1� following the end of the treatment period andextracted with ethanol. The amount of each PPE inthe cell extract was determined by LC–MS as describedin Methods, and normalized to the amount of PPEinitially added. Values are mean % dose (SD) for threemonolayers. Asterisks indicate significant difference(p< 0.05) in parameters between dose solutions pre-pared in 100% FaSSIF versus standard buffer.


compared to those of sodium caprate and 3-NC.However, in the presence of serum in thebasolateral compartment Papp values of HPC,PCC and sodium caprate were increased severalfold. The increase in sodium caprate Papp shouldbe interpreted with caution since it precipitatedon the basolateral side of the filter membrane instandard transport buffer, which could haveartificially lowered the Papp value in standardbuffer. Sodium caprate crystal deposits were notobserved on the filter membrane in the presence ofserum.

The disposition of PPEs in the TranswellTM

system was further examined by measuring theamount of compound associated with the cellmonolayers. Despite the increase in Papp causedby serum in the basolateral compartment, therewas no change in the amount of HPC and PCCthat was associated with the cells (Tab. 3). Thecell association of sodium caprate and 3-NC wassignificantly reduced by the presence of serum.However, the decrease in cell-association ofsodium caprate could be accounted for by thecompound that was associated with the bottom ofthe filter itself due to precipitation in calcium-containing transport buffer (data not shown).


Despite the increase in Papp for HPC and PCC,or the decrease in cell association for 3-NC, thesePPEs remained fully effective in the presence ofserum as evidenced by their ability to increase thePapp of mannitol (Tab. 3). In addition, serum didnot significantly impact the ability of EDTA orsodium caprate to increase the mannitol Papp.Thus, the paracellular permeability enhancingpotency for all five of the PPEs tested wereunaffected by adding the serum in the basolateralcompartment.

DOI 10.1002/jps

Table 3. PPE Disposition and Effect on Mannitol Papp Evaluated in Caco-2 Cell Monolayers After PPE Treatment inthe Presence or Absence of Human Serum in the Basolateral Compartment

Treatment

Mannitol Papp (nm/s)a PPE Papp (nm/s)a PPE % Cell Associationb

Standard Bufferc Serum Standard Bufferc Serum Standard Bufferc Serum

Control 2.5 (0.3) 1.9 (0.6) — — — —HPC 9.0 (1.1) 11 (2) 1.8 (0.4) 22 (9) 31.2 (2.0) 33.0 (1.1)PCC 14 (4) 12 (3) 1.4 (0.4) 30 (1)� 10.2 (0.3) 10.7 (0.4)Sodium caprate 13 (6) 15 (7) 230 (21) 622 (34)� 2.0 (0.2) 0.5 (0.1)�

EDTA 20 (4) 6 (8) — — — —3-NC 10 (2) 7 (1) 500 (6) 670 (95) 7.4 (1.2) 2.8 (0.5)

aPermeability coefficients (Papp) for each PPE or mannitol were calculated from the amount of test compound appearing in thebasolateral compartment during the 0–60 min period (0–90 min for HPC) after cotreatment with a single concentration of PPE (250mM HPC and PCC, 13 mM sodium caprate, 2 mM EDTA and 600 mM 3-NC). The basolateral compartment contained either transportbuffer or human serum. Values are mean (SD) for three monolayers.

bThe amount of each PPE in the cell extract was determined by LC–MS as described in Methods, and normalized to the amount ofPPE initially added. Values are mean % dose (SD) for three monolayers.

cStandard buffer was used in the basolateral compartment for all PPEs. EDTA was added to both the apical and basolateralcompartments.

�A significant difference (p< 0.05) between parameters determined in the presence of serum compared to those obtained withstandard buffer in the basolateral compartment.


Modulation of the Unstirred Water Layer Thickness

GI motility can decrease the unstirred water layerthickness (Hw) and contact time of a compound ata given intestinal absorption site. Both theseparameters can affect the transfer of lipophiliccompounds from bulk solution to the lipid bilayerof the cell membrane, which in turn, could affectthe efficacy of the PPEs that are being examinedin this study. In order to examine the effect ofGI motility on the potency of PPEs, the cellmonolayers grown on TranswellsTM were agitatedon a rotary shaker at different rates. The markercompound 3H-testosterone was used to calibratethe change in Hw at each stirring rate. As stirringrates were increased from 40 to 180 rpm, the Papp

of 3H-testosterone increased from 500 to 1600 nm/s.Based on these changes in flux, the decrease in Hw

was calculated to change from 1400 to 320 mm.These data for testosterone are consistent withthose reported previously by Adson et al.13 Asexpected, the Papp of 14C-mannitol was unchangedby stirring (data not shown).

Having established the change in Hw with themarker compound testosterone, the effect ofstirring rate on the potency of the selected PPEswas determined. Several doses of each PPE weretested at low, moderate, and high stirring ratesand their ability to decrease TEER was comparedto that under unstirred conditions. In order tomimic the limited contact time that a PPE wouldhave in a particular intestinal segment due to GI

DOI 10.1002/jps J

motility, the experimental treatment time for theagitation experiments was limited to 15 min.

As the stirring rate increased, the TEER-decrease versus PPE concentration curves gradu-ally shifted to the left for PCC and HPC, indicatingenhanced potency of these PPEs (shown for HPCin Fig. 5). At the highest stirring rate tested,the EC50 TEER of PCC and HPC decreased severalfold relative to unstirred conditions (Tab. 4). Dueto a greater shift in EC50 TEER for HPC withagitation, the less potent HPC became equipotentto PCC when Hw was decreased by stirring.Agitation had no effect on the ability of amphi-philic compound sodium caprate, nor of thenonamphiphilic compounds EDTA and 3-NC todecrease TEER (Tab. 4).

The Papp values of all PPEs increased understirred conditions (Tab. 4). Thus, the transepithe-lial flux of both the amphiphilic and the non-amphiphilic PPEs is controlled to some extent bythe Hw. However, the improved access to the cellmonolayers due to reduction in Hw did not resultin improved efficacy for sodium caprate and 3-NC,indicating that access to the cells was not the ratelimiting step for these compounds to exert theireffect. Indeed sodium caprate and 3-NC both hadhigh baseline permeability under standard con-ditions in the absence of stirring compared to HPCand PCC (Tab. 4).

At the abbreviated exposure time of 15 min, arelevant intestinal contact time,20 HPC, PCC andsodium caprate were fully effective. However, 3-NC


Figure 5. EC50 TEER for HPC evaluated in Caco-2 cellmonolayers with or without agitation. TEER wasmeasured across Caco-2 cell monolayers 15 min aftertreatment with solutions of HPC prepared in 1% (v/v)FaSSIF in transport buffer without agitation (standardcondition) or with agitation at three different rates on arotary shaker. Arrows indicate approximate EC50 TEER

under standard conditions and with stirring at 160 rpm.Mean values from at least three monolayers are plotted.Error bars indicate standard.


was not effective during the shorter 15 mintreatment period but required at least 30 minto cause a decrease in TEER across cell mono-layers (data not shown). If the 3-NC dosingsolution was removed at the 15 min time point

Table 4. EC50 TEER and PPE Papp Evaluated in Caco-2 CeAgitation

Treatment15 min EC50 TEER

a (mM)Fold Dec

Standard condition Agitation With Agi

HPC 370 110 3.4PCC 220 83 2.7Sodium caprate 12000 14000 0.9EDTA 1100 1200 0.93-NC No efficacy No efficacy —

aTEER was measured across Caco-2 cell monolayers 15 min apreviously without agitation (standard condition) or with agitation

bThe median fold-difference between the EC50 TEER curves for scPPE permeability coefficients (Papp) were calculated from the am

the 0–15 min period after treatment with a single concentration of PPsodium caprate, 2 mM EDTA and 600 mM 3-NC) with or without agithree monolayers.

�The fold difference was significant based on Bayesian analysis,


and replaced with blank transport buffer, adecrease in TEER was still observed at latertimes, which was of similar magnitude to thatachieved when 3-NC dose solution was left incontact with the monolayers for the duration ofthe experiment (data not shown). Thus for 3-NC,in contrast to HPC, a lag time in onset of PPE-induced TEER decrease is observed even after Hw

is reduced by agitation of monolayers.

DISCUSSION

Epithelial cell monolayers grown on semiperme-able membranes are well-established models forthe intestinal epithelium, and have been used toevaluate the paracellular permeability enhancingpotency, cytotoxicity and mechanisms of action ofPPEs.21 In the present study, three biorelevantmodifications have been applied to the Caco-2 cellmonolayer system, and the impact on the potencyof five structurally diverse PPEs was examined.

The selected PPEs had previously been exam-ined by others for their potency using in vitroconditions.2,4,10,22,23 The in vitro potency found inthe current study for PCC, sodium caprate andEDTA using standard conditions was comparableto that found previously, with effective concentra-tions in the range of 160–250 mM, 11–13 mM,and 1–2 mM, respectively. However, for HPC and3-NC approximately 20-fold higher concentra-tions were required to cause a decrease in TEERor an increase in mannitol Papp compared to

ll Monolayers After PPE Treatment With or Without

reasebPPE Papp (nm/s)c

Fold Increasetation Standard condition Agitation With Agitation

� 1.6 (0.6) 17 (6) 11�

� 6.1(2.5) 29 (28) 4.8230 (20) 705 (20) 3.1�

— — —500 (80) 1100 (20) 2.2�

fter treatment with solutions of PPEs prepared as describedon a rotary shaker at 160 rpm.tandard and modified conditions.ount of PPE appearing in the basolateral compartment duringE in standard transport buffer (250 mM HPC and PCC, 13 mM

tation on a rotary shaker at 160 rpm. Values are mean (SD) for

as described in Methods.

DOI 10.1002/jps


previous determinations in cell monolayers. Sig-nificant differences in the experimental conditionsin the current study are likely to have contributedto higher 3-NC and HPC effective concentrations,including the use of Caco-2 cells rather thanMDCK cells, lack of a 30 min pretreatment periodprior to commencement of measurements, and useof pH-adjusted transport buffer (pH 5.6) in theapical compartment of 3-NC studies rather thanthe typical pH 7.4 used in previous studies.

The use of FaSSIF in the apical chamberinstead of typical balanced salt solutions produceda large negative change on PPE potency, and aconcomitant decrease in cell-association of PCCand HPC (Figs. 2–4 and Tab. 2). Both of thenegatively affected PPEs were in the amphiphilicstructural class, a class which contains a largenumber of the PPEs studied to date. Each had acharged head group and one lipophilic carbonchain of considerable length making their incor-poration into the mixed micelles of FaSSIF highlyprobable. The three- to fourfold decrease in theamount of these PPEs recovered in cell extracts inthe presence of FaSSIF provides strong evidencethat PCC and HPC are incorporated into mixedmicelles of FaSSIF, resulting in a reduced cellularamount of the compounds.

Interestingly, one of the amphiphilic com-pounds, sodium caprate, was not affected byFaSSIF. Sodium caprate has a carbon chain thatis shorter than that of PCC and HPC (10 carbonsvs. 16 carbons), which may allow sodium caprateto partition into and out of the mixed micelles ofFaSSIF with relative ease compared to the longerchain PPEs. Alternatively, the relatively higheffective concentration of sodium caprate (10 mM)may exceed the capacity of the micellar compo-nents of FaSSIF (3 mM taurocholate and 0.75 mMphosphatidylcholine), allowing a sufficient amountof free sodium caprate to interact with the cells.In the presence of higher concentrations of tauro-cholate and phosphatidylcholine, which are pre-dicted to occur in the fed-state (Fed-state simulatedintestinal fluid; 15 mM taurocholate and 4 mMphosphatidylcholine),11 sodium caprate may bemore effectively sequestered. However, theseconditions were not tested in the present studiesbecause they have been reported to damage cellmonolayers.19

The PPE 3-NC was also not affected by thepresence of FaSSIF in the apical compartment.This result, however, is consistent with the lack ofamphiphilicity of 3-NC. Ingels et al.8 have alsoobserved minimal impact of FaSSIF in the donor

DOI 10.1002/jps J

compartment of Caco-2 cells on the transepithelialpermeability for some drugs. However, for com-pounds with appreciable lipophilicity, improvedrecovery from the apical compartment wasobserved in the presence of FaSSIF. Thus, thepresence of FaSSIF in the donor compartmentmay lead to changes in the disposition of lipophilicor amphiphilic compounds across cell monolayersin vitro. Whether this also occurs in the dynamicand more complex setting of the GI tract is stilllargely unknown and will require further inves-tigation. However, it is interesting to note thatthe in vitro potency for PCC determined in thisstudy in the presence of FaSSIF was found to beapproximately 2 mM (Tab. 2), a shift of approxi-mately 10-fold relative to that determined understandard in vitro conditions. This value moreclosely approximates in vivo PCC concentrations(12.5–46 mM) that have been effective in rats toimprove the absorption of poorly absorbed com-pounds.3 Furthermore, PCC is reportedly noteffective when administered into the duodenum ofrats, yet is effective when administered in thejejunum, ileum and colon.24 Since the duodenumis the intestinal region with the highest con-centration of bile, a plausible explanation forthe lack of in vivo efficacy of PCC in the duode-num is the sequestration of PCC in bile micelles.Based on the potential for this type of adverseinteraction of certain amphiphilic PPEs withintestinal fluid, amphiphilicity may be a struc-tural feature to avoid in the design of future PPEs.

The PPEs in this test set were fully effective inthe presence of serum in the basolateral compart-ment of Caco-2 cell monolayers grown on Trans-wellsTM (Tab. 3). Interestingly, the transepithelialflux of all amphiphilic PPEs was increased severalfold by basolateral serum (Tab. 3). Similarincreases in transepithelial flux by serum compo-nents have been reported for lipophilic, highlyplasma protein bound drug candidates. Forexample, flux was increased for several drugsby as much as 5- to 40-fold across Caco-2 cellmonolayers in the presence basolateral albumincompared to typical transport buffer.25 While theincrease in flux did not result in a large change inthe cellular amount of these PPEs in Caco-2 cellmonolayers, we speculate that high PPE fluxin vivo due to serosal blood components couldresult in rapid depletion of certain PPEs from theintestine. Indeed, rapid absorption of sodiumcaprate has been reported in pigs,26 and in thisstudy sodium caprate was found to have highin vitro permeability in the presence of serum



(600–700 nm/s). Thus, for sodium caprate, a likelycontributing factor to the requirement for higherin vivo concentrations is its rapid depletion fromthe intestinal lumen due to absorption. Thus,the biorelevant in vitro permeability coefficientsof PPEs determined in these studies provideimportant information regarding the rate of PPEabsorption/depletion that is likely to occur in vivo,and therefore, can serve as a key piece of infor-mation to predict whether controlled-releaseformulations could be successfully employed toimprove in vivo efficacy of PPEs. Indeed, asubstantial improvement in the sodium caprate-induced absorption of a poorly absorbed tradi-tional Chinese natural product was recentlydemonstrated through the use of controlled-release formulations.27

Also a likely contributing factor to lack ofadequate in vivo PPE potency after oral adminis-tration is the transient PPE exposure to intestinalcells in a given intestinal segment due to GImotility. In the present study, we simulatedtransient intestinal exposure by treating cellmonolayers for 15 min, instead of the typical60 min treatment period, and included agitationto lessen the unstirred water layer thickness (Hw).The in vitro potency (EC50 TEER) after shortertreatment time was not dramatically different formost PPEs (with agitation) as many of the selectedcompounds exerted their effect within the first fewminutes after treatment (Tabs. 2 and 4). Thus,even brief exposure of PPEs to cells resulted inparacellular permeability enhancement. Theseresults suggest that conceivably, a reduction inthe height of the unstirred water layer (Hw) couldimprove the efficiency with which the amphiphilic(lipophilic) PPEs can interact with the cellmembrane, and thus compensate for the reducedcontact time available due to intestinal motility.For 3-NC, however, no changes to TEER wereobserved during the 15 min treatment period.This PPE required at least 30 min for the onsetof paracellular permeability enhancement. Thedelay in the onset of paracellular permeabilityeffect for this PPE is not related to cell accessof 3-NC since it has high permeability (Papp¼500 nm/s), but is likely related to its mechanismof action that involves inhibition of PLC10 andsubsequent secondary events that require timefor changes to be manifested. This represents apotential disadvantage for the in vivo use of 3-NC,as it would require PPE administration at least30 min prior to administration of the poorly absorb-ed drug in order to achieve maximum efficacy.


Another aspect of GI motility investigated inthese studies was how changes to Hw impactedthe ability of the PPEs to decrease TEER. Theminimum Hw achieved by placing the Trans-wellTM-grown Caco-2 monolayers on a rotaryshaker was 320 mm. This value is similar to whatothers obtained using the same mechanicalagitation method to modulate Hw,13 but stillseveral fold higher than the reported in vivo Hw of40 mm.28 This relatively thick Hw could be relatedto the geometry of the in vitro system and themechanical agitation method employed in thesemeasurements.29

When Hw was decreased by agitation of mono-layers, the potency of PCC and HPC increased byseveral fold, while that of sodium caprate, 3-NCand EDTA was not affected (Fig. 5, Tab. 4).Examination of the disposition of the PPEs byLC–MS revealed that transepithelial flux (Papp)of all four quantifiable PPEs increased severalfold when Hw decreased by stirring (Tab. 4). Theincrease in flux with a decrease in Hw for thesecompounds is consistent with their amphiphilicity(or lipophilicity, in the case of 3-NC).30,31 Thus,transepithelial movement of amphiphilic or lipo-philic PPEs across cell monolayers is controlled tosome extent by the Hw. By implication, the accessof these PPEs to the apical surface of the cells, andsubsequently to the site of action inside the cell, isalso controlled by Hw and could be underestimatedin vitro if evaluated in the absence of sufficientstirring. The lack of a significant change inefficacy for sodium caprate and 3-NC with stirringmay suggest that Hw is not rate limiting in theiraccess to the cell membrane.

The qualitative impact of the cell monolayerrefinements on the measured potency of the fivePPEs is summarized in Table 5. The use of FaSSIFas a donor vehicle on the apical side of the cells hada significant negative effect on the potency of thetwo amphiphilic compounds, PCC and HPC, yetthe shift in potency of these PPEs is significantlydifferent. These PPEs were also affected by adecrease in Hw due to monolayer agitation, but inthe opposite direction—potency was increased byagitation of monolayers. In order to test whetherthese opposing effects on potency would beadditive resulting in a net effect of zero, the PPEswere tested with FaSSIF in the apical compart-ment using agitation of the monolayers. Theresults indicated that the overriding factor withthese PPEs is prevention of cell access by FaSSIF,even when Hw is reduced by stirring. It is impor-tant to evaluate the potency of amphiphilic/

DOI 10.1002/jps

Table 5. Impact of Biorelevant Refinements on PPEPotency

FaSSIF(AP)

Serum(BL) Agitation

Amphiphilic PPEsHPC # — ""PCC ### — ""Sodium caprate — — —

Nonamphiphilic PPEsEDTA " # —3-NC — — —

Key: " �2-fold; "" �2- to 10-fold; """ 10-fold.


lipophilic PPEs in an in vitro model thatincorporates the effect of Hw because a staticmodel would not only underestimate the potencyof many PPEs, but it would also not be able toassess if a PPE would be effective over a shortcontact period (with intestinal epithelium)encountered in vivo due to intestinal motility.Another refinement that was not evaluated in thisstudy, but may also be important for the evalua-tion of PPE potency in the cell culture models, isthe use of models that possess a mucus layer.Indeed, the presence of mucus is a barrier to thepassive diffusion of lipophilic compounds liketestosterone,32 and appears to be the reason forthe reduced in vivo potency of certain amphiphilicPPEs,33 as well as cationic polymers such aschitosans.34

The results in this study demonstrate thatintestinal fluid, contact time, and agitation areimportant parameters in defining the potency ofPPEs using cell-based in vitro models. Limitedin vivo data on sodium caprate and PCC suggestthat these refinements yield a better correspon-dence between the in vitro and in vivo potency ofPPEs, and provide an impetus for future studies todetermine if these refinements are sufficient toyield better prediction of in vivo efficacy of wideranging PPEs.

ACKNOWLEDGMENTS

The authors would like to acknowledge the impor-tant contributions of several individuals; DanielMorgan for assistance with LC–MS analyses;Steven Novick for statistical analyses; andZhiyang Zhao and Jeff Krise for helpful discus-sions. This work was financially supported byGlaxoSmithKline.

DOI 10.1002/jps J

REFERENCES

1. Tang F, Borchardt RT. 2002. Characterization ofthe efflux transporter(s) responsible for restrictingintestinal mucosa permeation of an acyloxyalkoxy-based cyclic prodrug of the opioid peptide DADLE.Pharm Res 19:780–786.

2. Hochman JH, Fix JA, LeCluyse EL. 1994. In vitroand in vivo analysis of the mechanism of absorptionenhancement by palmitoylcarnitine. J PharmacolExp Ther 269:813–822.

3. Fix JA, Engle K, Porter PA, Leppert PS, Selk SJ,Gardner CR, Alexander J. 1986. Acylcarnitines:Drug absorption-enhancing agents in the gastroin-testinal tract. Am J Physiol 251:G332–G340.

4. Lindmark T, Kimura Y, Artursson P. 1998. Absorp-tion enhancement through intracellular regulationof tight junction permeability by medium chainfatty acids in Caco-2 cells. J Pharmacol Exp Ther284:362–369.

5. Lindmark T, Nikkila T, Artursson P. 1995.Mechanisms of absorption enhancement by med-ium chain fatty acids in intestinal epithelial Caco-2cell monolayers. J Pharmacol Exp Ther 275:958–964.

6. Motlekar NA, Srivenugopal KS, Wachtel MS,Youan BB. 2005. Oral delivery of low-molecular-weight heparin using sodium caprate as absorptionenhancer reaches therapeutic levels. J Drug Target13:573–583.

7. Chao AC, Nguyen JV, Broughall M, Griffin A, FixJA, Daddona PE. 1999. In vitro and in vivo evalua-tion of effects of sodium caprate on enteral peptideabsorption and on mucosal morphology. Int JPharm 191:15–24.

8. Ingels F, Beck B, Oth M, Augustijns P. 2004. Effectof simulated intestinal fluid on drug permeabilityestimation across Caco-2 monolayers. Int J Pharm274:221–232.

9. Sawada GA, Barsuhn CL, Lutzke BS, HoughtonME, Padbury GE, Ho NF, Raub TJ. 1999. Increasedlipophilicity and subsequent cell partitioningdecrease passive transcellular diffusion of novel,highly lipophilic antioxidants. J Pharmacol ExpTher 288:1317–1326.

10. Ward PD, Klein RR, Troutman MD, Desai S, Thak-ker DR. 2002. Phospholipase C-gamma modulatesepithelial tight junction permeability throughhyperphosphorylation of tight junction proteins.J Biol Chem 277:35760–35765.

11. Galia E, Nicolaides E, Horter D, Lobenberg R,Reppas C, Dressman JB. 1998. Evaluation of var-ious dissolution media for predicting in vivo perfor-mance of class I and II drugs. Pharm Res 15:698–705.

12. Karlsson J, Artursson P. 1991. A method for thedetermination of cellular permeability coefficientsand aqueous boundary layer thickness in monolayers



of intestinal epithelial (Caco-2) cells grown inpermeable filter chambers. Int J Pharm 71:55–64.

13. Adson A, Burton PS, Raub TJ, Barsuhn CL, AudusKL, Ho NF. 1995. Passive diffusion of weak organicelectrolytes across Caco-2 cell monolayers: Uncou-pling the contributions of hydrodynamic, transcel-lular, and paracellular barriers. J Pharm Sci 84:1197–1204.

14. Artursson P, Ungell AL, Lofroth JE. 1993. Selectiveparacellular permeability in two models of intest-inal absorption: Cultured monolayers of humanintestinal epithelial cells and rat intestinal seg-ments. Pharm Res 10:1123–1129.

15. Van Itallie C, Rahner C, Anderson JM. 2001. Regu-lated expression of claudin-4 decreases paracellularconductance through a selective decrease in sodiumpermeability. J Clin Invest 107:1319–1327.

16. Anderberg EK, Lindmark T, Artursson P. 1993.Sodium caprate elicits dilatations in human intest-inal tight junctions and enhances drug absorptionby the paracellular route. Pharm Res 10:857–864.

17. Knipp GT, Ho NF, Barsuhn CL, Borchardt RT.1997. Paracellular diffusion in Caco-2 cell mono-layers: Effect of perturbation on the transport ofhydrophilic compounds that vary in charge andsize. J Pharm Sci 86:1105–1110.

18. Ouyang H. 2001. Enhanced oral delivery of hydro-philic drugs [microform]: Design and evaluationof tight junction modulators and cyclic peptideprodugs, Hui Ouyang (ed.).

19. Ingels F, Deferme S, Destexhe E, Oth M, Van denMooter G, Augustijns P. 2002. Simulated intestinalfluid as transport medium in the Caco-2 cell culturemodel. Int J Pharm 232:183–192.

20. Assuming that the PPE-drug dosage form is con-tained in a glass of water (240 mL), and that itmoves a cylindrical bolus with a diameter equal tothat of human intestine (5 cm) along the entirelength of the intestine (625 cm) during a 3.3 hperiod, the contact time at a particular referencepoint on the intestine would be approximately4 min. Human small intestinal parameters found inKararli TT, 1995. Comparison of the gastrointest-inal anatomy, physiology, and biochemistry ofhumans and commonly used laboratory animals.Biopharm Drug Dispos 16:351–380.

21. Ward PD, Tippin TK, Thakker DR. 2000. Enhan-cing paracellular permeability by modulatingepithelial tight junctions. Pharm Sci Technol Today3:346–358.

22. Ward PD, Ouyang H, Thakker DR. 2003. Role ofphospholipase C-beta in the modulation of epithe-lial tight junction permeability. J Pharmacol ExpTher 304:689–698.

23. Quan YS, Hattori K, Lundborg E, Fujita T,Murakami M, Muranishi S, Yamamoto A. 1998.Effectiveness and toxicity screening of various


absorption enhancers using Caco-2 cell monolayers.Biol Pharm Bull 21:615–620.

24. Sutton SC, LeCluyse EL, Cammack L, Fix JA. 1992.Enhanced bioavailability of cefoxitin using palmi-toyl L-carnitine. I. Enhancer activity in differentintestinal regions. Pharm Res 9:191–194.

25. Aungst BJ, Nguyen NH, Bulgarelli JP, Oates-LenzK. 2000. The influence of donor and reservoir addi-tives on Caco-2 permeability and secretory trans-port of HIV protease inhibitors and other lipophiliccompounds. Pharm Res 17:1175–1180.

26. Raoof AA, Ramtoola Z, McKenna B, Yu RZ, HardeeG, Geary RS. 2002. Effect of sodium caprate on theintestinal absorption of two modified antisense oli-gonucleotides in pigs. Eur J Pharm Sci 17:131–138.

27. Lin X, Xu DS, Feng Y, Li SM, Lu ZL, Shen L. 2006.Release-controlling absorption enhancement ofenterally administered Ophiopogon japonicus poly-saccharide by sodium caprate in rats. J Pharm Sci95:2534–2542.

28. Strocchi A, Levitt MD. 1991. A reappraisal ofthe magnitude and implications of the intestinalunstirred layer. Gastroenterology 101:843–847.

29. Karlsson J, Artursson P. 1992. A new diffusionchamber system for the determination of drug per-meability coefficients across the human intestinalepithelium that are independent of the unstirredwater layer. Biochim Biophys Acta 1111:204–210.

30. Cho MJ, Adson A, Kezdy FJ. 1990. Transepithelialtransport of aliphatic carboxylic acids studied inMadin Darby canine kidney (MDCK) cell mono-layers. Pharm Res 7:325–331.

31. Lennernas H, Palm K, Fagerholm U, Artursson P.1996. Comparison between active and passive drugtransport in human intestinal epithelial (Caco-2)cells in vitro and human jejunum in vivo. Int JPharm 127:103–107.

32. Wikman A, Karlsson J, Carlstedt I, Artursson P.1993. A drug absorption model based on the mucuslayer producing human intestinal goblet cell lineHT29-H. Pharm Res 10:843–852.

33. Anderberg EK, Nystrom C, Artursson P. 1992.Epithelial transport of drugs in cell culture. VII.Effects of pharmaceutical surfactant excipients andbile acids on transepithelial permeability in mono-layers of human intestinal epithelial (Caco-2) cells.J Pharm Sci 81:879–887.

34. Schipper NG, Varum KM, Stenberg P, Ocklind G,Lennernas H, Artursson P. 1999. Chitosans asabsorption enhancers of poorly absorbable drugs.3. Influence of mucus on absorption enhancement.Eur J Pharm Sci 8:335–343.

35. LeCluyse EL, Appel LE, Sutton SC. 1991. Relation-ship between drug absorption enhancing activityand membrane perturbing effects of acylcarnitines.Pharm Res 8:84–87.

DOI 10.1002/jps

biorelevant refinement of the caco-2 cell culture model to assess efficacy of paracellular...

Documents