Liposomes Targeted to Deliver Antisecretory Agents toJejunal Mucosa

R.R.E. Uwiera, D.A. Romancyia, J.P. Wong and G.W. Forsyth

ABSTRACT

The B subunit of cholera toxin hasbeen covalently attached to the surfaceof Uposomes made from a mixture ofphosphatidylethanolamine, phosphati-dylcholine and cholesterol. Adenylatecydase inhibitors and chloride conduc-tance inhibitors were encapsulatedwithin the liposomes. These "targeted"liposomes were used to study the com-bined effects of this novel delivery sys-tem, and a lmited number of possibleantisecretory agents, on net fluid fluxinto the pig jejunum.A state of net secretory fluid flux

was induced in isolated jejunal loopsin weanling pigs by adding theophyflineor cholera toxin to the lumen of theisolated loops. There was no reductionin net fluid secretion when liposomesuspensions without encapsulatedsecretory inhibitors were added to fluidin the lumen of loops treated withtheophyiline. There was also no reduc-tion in net fluid secretion when mico-nazole, a-phenylcinnamate or 5 nitro-2-(3-phenethylamino)benzoate wereencapsulated within targeted liposomesadded to isolated jejunal loops. Thenet fluid flux induced by exposure ofjejunal loops to theophylline was sig-nificantly reduced by adding targetedliposomes containing 2' -deoxy-3' -

AMP. The reduction involved a rever-sal of net secretory fluid flux to anabsorptive value.The net fluid secretory response to

treatment of loops with cholera toxinwas also inhibited by treating loopswith targeted liposomes containing2' -deoxy-3 -AMP. However, thereversal of secretion was less completefor secretion induced by cholera toxinthan for secretion induced by theo-phylline. The reduced antisecretory

efficacy versus cholera toxin was notimproved by encapsulating higher con-centrations of 2 '-deoxy-3' -AMP.A larger dose of the inhibitor deliveredwith increased numbers of liposomescaused a significant reduction in netfluid secretion. Occupancy of mucosalreceptors by native cholera toxin maybe a factor limiting access of liposomeswith surface B subunit, and mayreduce the utility of the B subunit ofcholera toxin as a targeting agent fordelivering antisecretory agents to theintestinal mucosa.

RESUME

La sous-unite B de la toxine du cho-lera a ete fixee a l'aide de liens cova-lents a la surface de liposomes fabriquesa partir d'un melange de phosphatidyl-ethanolamine, phosphatidylcholine etde cholesterol. Des inhibiteurs del'adenylate cyclase et de la conductiondu chlore ont et encapsules a I'inte-rieur des liposomes. Ces liposomes ontete employes pour l'etude de ce nou-vel agent de liberation ainsi que de dif-ferents agents antisecretoires sur desintestins de porc.Une secretion intestinale positive "a

partir d'une anse jejunale de porceletssevrs a 't induite a l'aide detheophylline ou de la toxine du cho-lera. La suspension de liposomes non-charges ainsi que des suspensions deliposomes contenant du miconazole,du 2-phenylcinnamate ou du 5-nitro-2-benzoate n'ont pas modifie l'activitesecretoire de la theophylline sur l'intes-tin. Par contre I'activite secretoire dela theophylline et de la toxine du cho-lera a ete renversee en un processusd'absorption lorsque des liposomescontenant du 2'-desoxy-3'-AMP ontete utilise's. Cependant, le renverse-

ment fut moins marque pour la toxinedu cholera.Une augmentation des concentra-

tions de 2'-desoxy-3'-AMP a l'inte-rieur des liposomes n'a pas augmentel'effet anti-secretoire de la toxine ducholera. Par contre-une augmentationde la dose d'inhibiteur couplee a uneaugmentation du nombre de liposomesa produit une diminution significativede secretion. L'occupation du recep-teur par la toxine naturelle du cholerasemblerait etre un facteur limitantl'acces de la sous-unite B de la toxineliee aux liposomes ce qui indirectementen reduit l'activite des agents qui y sontattaches. (Traduit par Dr PascalDubreuil)

INTRODUCTION

Secretory diarrhea arises from theactivation of regulatory processes thatnormally limit movement of electrolytesand water into the7 intestinal lumen.The main elements of the secretorymechanism include the intracellularsecond messenger, adenosine-3' ,5 '-cyclic monophosphate (cAMP) and theregulated chloride conductance proteinlocated in the apical membrane ofintestinal epithelial cells. Conductivetransport of Cl- into the intestinallumen drives paracellular Na+ andwater movement into the lumen (1).

Antisecretory strategies have targeteda variety of processes including Ca2+-

binding proteins (2,3) and endorphinreceptors (4,5) with varying degrees ofsuccess. Even with partial successes ofthese indirect approaches, the adenylatecyclase that produces cAMP and thechloride conductance protein whichreleases Cl- into the intestinal lumenremain as the major targets for anti-secretory therapy.

Can J Vet Res 1992; 56: 249-255

Veterinary Physiological Sciences, University of Saskatchewan, Saskatoon, Saskatchewan S7N OWO (Uwiera, Romancyia, Forsyth) and BiomedicalDefense Section, Defense Research Establishment Suffield, Box 4000, Medicine Hat, Alberta TIA 8K6 (Wong).Submitted February 10, 1992.

249

Bulk addition of inhibitors ofcAMPproduction or conductive Cl - trans-port to the intestinal lumen has notbeen an effective antisecretory strategy(6,7). The poor efficacy of inhibitorssupplied by this route has been attri-buted to problems in effective and spe-cific delivery of therapeutic agents tosub-populations of intestinal cellswhich are specialized for a role in fluidsecretion. Liposomes might be usefulas a delivery vehicle that could reducethe required dose of inhibitors andminimize the side effects of antidiar-rheal therapy. Liposomal delivery sys-tems may be improved by attachmentof surface ligands that will direct lipo-somes to bind to, and deliver contentsinto the intestinal cells containing sur-face receptors for the binding of bac-terial enterotoxins. Coupling the B sub-unit of cholera toxin to liposomesurfaces significantly increases theapparent binding affinity of the B sub-unit for the jejunal brush border sur-face (8).

Potential antisecretory agents wereencapsulated within liposomes, andliposomes were coupled to a modifiedform of the B subunit of cholera toxin.The targeted liposomes produced bythis procedure were tested for inhibit-ory effects on intestinal secretion by anin situ intestinal loop preparation.

MATERIALS AND METHODS

LIPOSOME PREPARATION ANDCHARACTERIZATION

Egg yolk phosphatidylcholine andcholesterol in an 18:1 molar ratio(140 zimol of total lipid) was dissolvedin chloroform and dried under vacuumin a rotary evaporator to deposit thelipid as a thin film on the wall of aflask. 3H cholesterol (0.1 uCi) wasadded to the lipid mixture to measurelipid content of liposome suspensions.The dried lipid film on the side wall ofthe flask was dispersed by vigorousmixing with an aqueous buffer systemcontaining 10 mM mannitol, 136 mMNaCl, 10 mM K2HPO4 (pH 7.4), aswell as any agent intended for encap-sulation within the liposomes. Theaqueous resuspension system also con-tained 0.28 FCi 1-[14C]mannitol toserve as a marker for liposome volumemeasurements and calculations oftrapping efficiency. The lipid suspen-

sion was treated with five freeze thawcycles to increase interlamellar spac-ings and trapped volumes (9). Lipo-some suspensions produced by vortexmixing were size-sorted by serial pas-sage through two polycarbonate filters(0.4 um pore size) using a high pressurefiltration system (Extruder, LipexBiomembranes Inc., Vancouver, BritishColumbia).

Unincorporated solute was removedby gel filtration of liposomes suspen-sions with Sephadex G-50. Fractionscollected from the column weresampled to measure the content of 3Hand 14C by dual label scintillationcounting. The specific activity of the1-[14C]mannitol was used to convertdpm to nmol of mannitol. The equi-valence of 10 nmol mannitol per gLwas used to calculate the aqueousvolume trapped within a sample ofvesicles. This volume, or trappedspace, was expressed as nL per nmoleof lipid as determined from the amountof 3H cholesterol in the sample.

Dipalmitoylphosphatidylethanol-amine (DPPE) was used as a source ofa functional group for coupling theB-subunit of cholera toxin to liposomesurface. The DPPE was incorporatedby addition into the lipid solution inchloroform. Vesicle space was mea-sured as a function of the percentageof DPPE in the lipid mixture. Mayeret al (9) have reported that repeatedpassage through the extruder apparatusfavors the formation of unilamellarliposomes. Vesicle suspensions werefiltered up to four times to determinethe effect of this treatment on trappedspace within the vesicles.

Trapping efficiency was expressed asthe percent of the total aqueous solutethat could be retained within the vesi-cle space during resuspension of a lipidfilm. Calculations of entrapped spaceand efficiency of solute trapping wereboth based on vesicular content of1-[14C]mannitol. Trapping efficiencywas measured as a function of varyingproportions of DPPE, and total con-centrations of lipid in the resuspensionbuffer.

CHOLERA TOXIN B SUBUNITDERIVATIZATION

The B subunit of cholera toxin wastreated with N-succinimidyl S-thioace-tate (SATA) to introduce reactive sulf-hydryl groups onto the surface of the

protein. The reaction between SATAand cholera toxin B subunit was car-ried out for 30 min at 20°C in a 400 jiLreaction volume containing 1.0 jimolof SATA and 10 nmol of B subunit in50 mM Na2CO3 (pH 9.5). This ratioof reactants gave a tenfold molarexcess of SATA over total lysineresidues in the B subunit. The reactionmixture was separated from unreactedSATA by gel filtration in 10 mM MESbuffer (pH 5.5). Modified B subunitwas collected and deacetylated for twohours in a solution containing 50 mMhydroxylamine-HCl, 5.0 mM NaHPO4and 2.5 mM EDTA (pH 7.5). Thesulfhydryl content of the deacetylatedproduct was determined by reactionwith Ellman's reagent (10).

CHOLERA TOXIN B SUBUNITATTACHMENT TO THE SURFACE OFLIPOSOMES

The DPPE (40 mg) was dissolved indry chloroform, methanol and triethyl-amine (16 mL:2 mL:20 mg) and mixedwith 20 mg m-maleimidobenzoyl-N-hydroxysuccinimide ester (MBS) underanhydrous conditions for 24 h (11).The Rf of the product was measuredby thin layer chromatography on silicagel in a solvent of chloroform,methanol and glacial acetic acid(65:25:13) (11). Maleimide-conjugatedDPPE was added to the chloroformsolution of phosphatidylcholine andcholesterol as 0.25% by weight of thetotal lipid (140 Amol) used for a singleliposome preparation.The modified B subunit with added

SH groups was added immediatelyafter the final passage of the aqueousliposome resuspension system contain-ing maleimide-substituted DPPEthrough the pressure filtration system.The reaction mixture was maintainedat 4°C for 16 h. The liposome suspen-sion was separated from unconjugatedB subunit by passage over a SephadexG-50 desalting column equilibratedwith 10mM MES buffer (pH 5.5). Theefficiency of the coupling reaction wasmonitored with [125I]B subunit.

MEASUREMENT OF INTESTINALSECRETION

Weanling pigs 17 to 20 kg wereanesthetized with halothane and 10 cmjejunal loops were prepared, starting30 cm distal to the ligament of Trietz,by transecting the gut and inserting

250

lucite cannulae. Net fluid flux in theloops was determined by the two com-partment kinetic model of Berger andSteele (12), using [3H]-polyethylene-glycol 4000 (1 ACi per L) as a dilutionmarker. Net and unidirectional chlorideflux measurements were based on netfluid flux, content of Cl- in the fluxsolution and the changes in concentra-tion of 36C1 marker added to the fluxsolutions (13). All measurements werecarried out in an isotonic electrolytemixture formulated to match the nor-mal composition of pig jejunal loopfluid (13).A state of net secretion was induced

by adding 10 mM theophylline tojejunal loop fluid. Theophylline andtargeted liposomes were combined injejunal loop fluid when attempting toinhibit the net secretory effect of thetheophylline.

Cholera toxin (crude lyophilized cul-ture filtrate of Vibrio cholerae, WyethLabs, Philadelphia, Pennsylvania) wasdissolved in 0.15 M NaCl and incu-bated in jejunal loops for 90 min priorto flux measurements. Loops wereemptied and flux measurements were

6-

5-~

'.

00Ec

0

0.ci

co

4-

3-

2-

1I

O 4-2

made in jejunal loop fluid withoutadditional cholera toxin.

POTENTIAL SECRETORY INHIBITORS

Miconazole (1-[2-(2,4-dichlorophe-nyl)-2-[(2,4-dichlorophenyl)-methoxyIethyll-l H-imidazole) (14) and adeno-sine- 2' -deoxy-3' -monophosphate(2' -d-3'-AMP) were used as adenylatecyclase inhibitors (15,16). at-Phenyl-cinnamate (axPC) and 5-nitro-2-(3-phenyl-ethyaminobeate (NPEB)were chosen as potential chloride chan-nel blockers (17,18). A 10mM solutionof each inhibitor was encapsulatedwithin liposomes by including theinhibitor in 1.0 mL of solution used todisperse the dried lipid film. Trappingefficiencies were calculated from the1-[14C]mannitol content of the vesiclesuspension. The suspension was dividedinto six or eight equal fractions foraddition to jejunal loop fluid used influx measurements.

STATISTICAL PROCEDURES

An unpaired Student's t-test wasused to assess the statistical signifi-cance of treatment effects.

TT

iTi

1

UUU00UU

0.00%o DPPE

0.25% DPPE

0.50% DPPE

1.00% DPPE

2.00%h DPPE

3.00% DPPE

5.00%h DPPE

4

Number Of Filtrations

RESULTS

LIPOSOME COMPOSITION,PREPARATION AND PROPERTIES

The effect of vesicle properties ofDPPE addition to the lipid mixtureused for liposome preparation wasmeasured. An optimal level of DPPEwould provide enough sites for B sub-unit attachment without inducing sig-nificant changes in the properties ofthe vesicles. Increasing the amountof DPPE in the lipid mixture reducedthe effective space within liposomes(Fig. 1).Maximizing trapping efficiency is

important when liposomes are used asa delivery vehicle for expensive agents.Trapping efficiency was expressed asthe percentage of the 1-[14C]mannitolaqueous marker that was locatedwithin the liposomes after the normalpreparation steps of dispersion, freeze-thaw cycling and high pressure ultra-filtration. Increasing amounts ofDPPE in the lipid mixture decreasedtrapping efficiencies (Fig. 2). Lipidconcentrations above 100 mg per mLare reported to give higher trappingefficiencies (9), but ultrafiltration wasdifficult with more concentrated lipid.As a compromise, the lipid mixturewas made with 0.25% DPPE, andresuspended for liposome formation(100 mg of lipid per mL of aqueousmedium).

ENCAPSULATION OF INHIBITORS ANDDOSE CALCULATIONS

Standard liposome preparationsinvolved resuspension of 140 umol ofphospholipid and cholesterol in anaqueous volume of 1.0 mL, contain-ing 1-[14C]mannitol and secretoryinhibitor. The internal volume wasdetermined from the 1-[14C]mannitolcontent of a sample of the final lipo-some suspension. Actual doses ofinhibitor delivered in 100 1tL samplesof liposome suspension varied withtrapping efficiency between batches ofvesicles, with a normal range of 5 to20 nmol of inhibitor per 10 cm loopcontained within 1 to 4 1tmol of lipo-some lipid (data not shown).

THEOPHYLLINE AND 2'DEOXY-3'-AMPEFFECTS ON NET FLUID FLUX

Net fluid flux is reported as yL offluid per loop per minute, with nega-tive values representing net absorption.

251

Fig. 1. The effect of dipaimitoylphosphatidylethanolamine on the aqueous volume trapped withinliposomes. Liposome suspensions were size sorted by serial ultrafiltration through 0.4 itm pore sizepolycarbonate ifiters. Endosed space was determined by measuring the amount of 1-[14C]mannitolassociated with vesicle lipid in the void volume elution peak from a Sephadex G-50 column. Barsare averages ± standard deviations from nine separate measurements.

TT

--rl

2

U0

0U

86.9 mg/ml + 0% DPPE

86.9 mg/ml + 5% DPPE

100. mg/ml + 0% DPPE

100. mg/ml + 5% DPPE

200. mg/ml + 0% DPPE

200. mgmI +5% DPPE

-I

4

Number Of FiltrationsFig. 2. The effect of dipahmitoylphosphatidylethanolamine on the efficiency of trapping mannitolwithin liposomes. Liposome suspensions were size sorted by serial ultrafiltration through 0.4 Ampore size polycarbonate filters. Trapping efficiency refers to the percent of the 1-[14C]mannitolpresent in the resuspension buffer that was associated with vesicle lipid in the void volume elutionpeak from a Sephadex G-50 column. Bars are averages ± standard deviations of nine measurements.

TABLE I. Net fluid flux in jejunal loops exposed to theophylline

Experiment 1 Experiment 2Condition IL/loop/min p vs theo AL/loop/min p vs theo

Control -78 ± Ila 0.001 -87 + Ilb 0.001Theophylline 75 ± 13 133 ± 25Theo + Lipo ndC 156 ± 52Theo + Lipo + 2'-d-3' AMP nd 71 ± 53 0.31Theo + Lipo + B subunit nd 14 ± 34 0.019Theo + Lipo + B subunit -45 ± 10 0.001 -6 ± 28 0.008+ 2'-d-3' AMP

aMean ± SEM for flux in 12 loopsbMean ± SEM for flux in 20 loopsCnd = not determined

Positive values represent appearance offluid in the loops (net fluid secretion).The averages of two control loops insix pigs, - 78 ± 11 yL per loop perminute (Table 1), was typical of theexpected normal fluid flux in thisregion of the pig gut (6,7,12). Net fluidsecretion (75 ± 13 jtL per loop perminute) was induced in jejunal loopsby the addition of 10 mM theophyllineto the jejunal loop fluid.The encapsulated adenylate cyclase

inhibitor was partially effective as anantagonist of this secretory response(Table I). Liposomes without surface B

subunit or encapsulated adenylatecyclase inhibitor had no effect on thesecretory response to theophylline, butthere was a significant reduction influid secretion in jejunal loops treatedwith targeted liposomes withoutinhibitor (p = 0.019).

THEOPHYLLINE AND MICONAZOLEEFFECTS ON NET FLUID FLUX

The hydrophobicity of miconazolecreated problems of reduced amountsof enclosed liposome space andreduced trapping efflciencies. Amountsof miconazole delivered in a standard

liposome dose per loop (1 to 3 nmol)were substantially lower than the doseof 2' -d-3' -AMP. Addition of encap-sulated miconazole to flux solutionsdid not reduce the secretory responseto theophylline. The net fluid flux incontrol loops was -68 ± 7 .L/loop/min, in loops treated with theophylline37 ± 23 jiL/ loop/min, and in loopswith theophylline plus liposomes con-taining miconazole 51 ± 23 ,L/loop/min (data not shown).

CHOLERA TOXIN AND CHLORIDECONDUCTANCE INHIBITOR EFFECTSON NET FLUID FLUX

Loop contents were drained follow-ing 90 min of exposure to cholera toxin(50 mg crude filtrate per loop) andreplaced with liposome suspensions ofisotonic saline solutions. After 10 minof exposure to the liposomes contain-ing either aPC or NPEB the loopswere drained and refilled with jejunalloop fluid to start the fluid flux mea-surements. Neither caPC nor NPEBhad any effect of fluid flux induced byprevious exposure to cholera toxin(Table II). The net flux of Cl- intothe loops tended to increase in the pres-ence of the agents that are known toinhibit conductive transport of Cl - inpatch clamp experiments. Encapsu-lated chloride channel blockers werealso ineffective in causing any reduc-tion in the secretory response to theo-phylline (data not shown).

CHOLERA TOXIN AND 2' DEOXY-3 '-AMPEFFECTS ON NET FLUID FLUX

The effect of adding encapsulated2' -d-3 '-AMP during the measurementsof net fluid flux occurring in jejunalloops 90 minutes after addition ofcholera toxin is shown in Table III.The encapsulated inhibitor caused areduction in the fluid secretion occur-ring with the highest and the lowestdose of cholera toxin, but the net fluidflux did not return to control values.The net secretory response to choleratoxin, reported in Table III, was notdependent on the dose of choleratoxin. The antisecretory effect of theadenylate cyclase inhibitor was largestin loops treated with the highest con-centration of cholera toxin.A separate trial was carried out with

the timing protocol described abovefor the conductance inhibitors. After90 min of exposure to 50 mg V. cholerae

40 -

> 30 -

0c0D._

%.-

0) 20-a

cis

0,o 10o-

252

1

TABLE II. Net fluid flux in jejunal loops exposed to cholera toxin foUowed by conductance inhibitors

HOH flux Net C1- flux HOH flux Net Cl- fluxCondition AL/min Aeq/min AL/min /eq/min

Control -17 + 2a -2.0 ± 2 -34 + 8b -3.8 I1Cholera toxin 31 ± 4 4.7 ± 2 76 ± 12 8.0 I1CT + aPC 35 ± 5 6.1 ± 2CT + NPEB 76 ± 11 11.1 1aMean ± SEM n = 12bMean ± SEM n = 12

TABLE III. Net fluid flux in jejunal loops exposed to cholera toxin

200 mg CT/loop 100 mg CT/loop 50 mg CT/loopCondition tL/loop/min uL/loop/min ftL/loop/min

Control -21 ± 12a -28 12b -99 ± 12cCholera toxin 87 ± 20 81 ± 28 17 ± 4Cholera toxin, 2'-d-3'-AMP 23 ± 17d 77 ± 29 -1 ± 12aMean ± SEM n = 12bMean ± SEM n = 6CMean ± SEM n = 15dSignificantly different from cholera toxin flux p = 0.006

filtrate per loop the loop contents were EFFECTS OF CHANGING 2' DEOXY-3' -drained, and liposome suspensions or AMP CONCENTRATION ANDisotonic saline solutions added. The LIPOSOME AMOUNTSloops were drained after 10 min and The delivery protocol for the inhibi-refilled with jejunal loop fluid for tor was modified to test if amounts ofinitiating the fluid flux measurements. 2' -d-3' -AMP in the liposomes wereThere was no substantial improvement sufficient to inhibit the adenylatein inhibitor efficacy with this proce- cyclase activity resulting from priordure for delivery of adenylate cyclase exposure to cholera toxin. The firstinhibitor immediately before measur- modification involved increasing theing fluid flux. Fluid flux in control concentration of inhibitor during lipo-loops was -1 ± 16, in loops treated some formation from 10 to 50 mM.with cholera toxin was 46 ± 10, and The liposome suspensions containingwith cholera toxin followed by 10 min 50 mM 2' -d-3 '-AMP were less effec-with liposomes containing 2'-d-3'- tive than those with lower, 10 mMAMP was 27 ± 7 iL/loop/min (P = concentration. The corresponding net0.18 for a reduced secretory response fluid fluxes for this experiment were:to cholera toxin). control, - 4 ± 21; cholera toxin, 78The timing of treatment with the ± 10; and cholera toxin plus encap-

normal dose of encapsulated 2' -d-3 '- sulated 50 mM 2' -d-3 ' -AMP, 66 ±AMP was changed to expose the loops 15 tL/loop/min, with 12 observationsto the inhibitor prior to addition of for each treatment. The p value for sig-cholera toxin. For this experiment tar- nificance between the two treatmentsgeted liposomes containing inhibitor receiving cholera toxin was 0.65 (datawere added to ligated jejunal loops and not shown).incubated in the loops for 15 min prior Larger amounts of inhibitor wereto emptying the loops and adding also delivered by increasing the dose ofcholera toxin. Loops were drained liposomes containing 10 mM 2'-d-3'-after 90 min and net fluid flux was AMP. The dose per loop was increasedmeasured in a standard 20 minute fourfold, from 2.5 zmol to 10 Amol ofassay. Loops exposed to targeted lipo- liposome lipid. Jejunal loops weresomes before cholera toxin showed a exposed to cholera toxin for 90 min,full secretory response to the toxin (124 drained, and filled with 145 mM NaCl± 11 tL/loop/min) compared to or with liposomes suspended in 145 mMsecretion in loops treated only with NaCl for 10 min prior to measuring netcholera toxin (l14 ± 17 AL/loop/min) fluid flux. This approach was more(data not shown). successful in reducing the secretory

effect of cholera toxin. Fluid fluxvalues with four times the normal doseof targeted liposomes were: control,- 121 + 14; cholera toxin, 74 ± 28;and cholera toxin plus 10mM 2' -d-3 '-AMP encapsulated in 10 iLmol ofliposomal lipid, 12 ± 10 AL/loop/min,with 12 observations for each treat-ment. The p value for a significantantisecretory effect of the inhibitor was0.033.

COMPETITION FOR MUCOSALRECEPTORS BETWEEN B SUBUNITAND INTACT CHOLERA TOXIN

The secretory dose response tocholera toxin reported in Table IIIindicated that even the lowest dose ofcholera toxin used in these experimentsmay be saturating surface receptors onsecretory enterocytes. This possibilitywas investigated via an indirect com-petition experiment. Purified B subunitwas added to jejunal loops and allowedto equilibrate for 10 min prior to drain-ing the loops and adding 50 mg ofcrude V. cholerae culture filtrate to theloops. Preincubation of loops withamounts of isolated B subunit as lowas 10 Ag per 10 cm intestinal loopreduced the net fluid flux response tocholera toxin (Table IV). Previous esti-mates of the potency of the crude cul-ture filtrate of V. cholerae indicatedfairly similar amounts of B subunitprotein in 50 mg of crude culturefiltrate and in 10 iLg of isolatedB subunit (6). Without a clear responseto increased amounts of B subunit thisis incomplete evidence of competitionfor surface receptors between choleratoxin used to induce secretion intointestinal loops and immobilized B sub-unit on the surface of the liposomes.

DISCUSSION

Enclosed liposome space and trap-ping efficiency were both reduced byadding DPPE to the lipid mixture.These results could have been effectsof the pahnitoyl side chains on assemblyor on fluidity of the phospholipidbilayer of the liposomes. There was anecessary trade-off between the advan-tages of a phospholipid that was stableto oxidative attack during liposomepreparation, and one which has acylgroups with lower melting points. Itmay be feasible to use low proportions

253

TABLE IV. Competition between intact cholera toxin and isolated B subunit

Net fluid secretion Net chloride secretionCondition tL/loop/min geq/loop/min

Control -45 ± 4a -5.3 ± 0.7Cholera toxin 86 ± 4 8.8 ± 0.910 ,sg B subunit before cholera toxin 46 ± llb 4.1 ± 1.125 izg B subunit before cholera toxin 42 ± 8b 7.6 ± 1.650 itg B subunit before cholera toxin 42 ± 14b 7.1 ± 2.1

aMean ± SEM n = 6bSignificantly less than flux with cholera toxin (p < 0.05)

of a stable phospholipid when it is only model was approximately 150 AL/10 cmnecessary to have a small number of loop/min. In seven meters of smallfunctional groups available on the sur- intestine this would be equivalent toface of a liposome. Huang and Mason 10 mL of net secretion per minute orreported an average surface area of 600 mL per hour. Assuming that0.74 nm2 per lipid head group in the cholera toxin reduces the normalouter plane of a liposome bilayer (19). absorptive capacity of the colon theseA spherical liposome with a diameter rates may approximate maximalof 100 nm would have a surface area volumes of fecal fluid loss (10 to 15 Lof 3.14 x 104 nm2. Assuming similar per day) measured in adult humansmolecular weights for the phospholipids suffering from natural infections withthere would be 4.25 x 104 head V. cholerae.groups per liposome surface. With The efficacy of antagonists of intes-0.25% of the membrane lipids as tinal secretion must be assessed not justDPPE there would be approximately through preventing net fluid secretion,100 headgroups on the surface of one but in terms of restoring net fluid fluxliposome. This should give sufficient to normal levels. Net fluid flux issites for attachment of the B subunit usually absorptive even in the jejunum,of cholera toxin. ranging from averages of - 17 toWe have shown previously that 10 - 99 AL per 10 cm loop per min in the

to 50 lmol of 2'-d-3 '-AMP added into results summarized in the previous sec-the lumen of isolated jejunal loops did tion. Simultaneous exposure of jejunalnot produce a measurable decrease in loops to levels of theophylline suffi-the net secretory fluid flux induced by cient to produce a maximal secretorytheophylline or cholera toxin (6). In response, and to targeted liposomescomparison to placing soluble agents containing 10 mM 2'-d-3' -AMP,in the intestinal lumen, the liposome reversed a net secretory response to asystem is severely limited in terms of net absorptive response. The reductionthe encapsulated volume available to in the absolute value of the net fluxhold aqueous solute. Delivery of response to 10 mM theophylline wassecretory inhibitor by the targeted approximately 60%Vo for both experi-liposome system could only be effec- ments. This reduction would likely betive if the liposomal vehicle were an sufficient to prevent clinical dehydra-effective delivery system, and if selec- tion and electrolyte losses arising fromtive delivery could be achieved by secretory diarrhea.attachment of a specific ligand to the Chloride conductance inhibitorsliposome surface. The importance of should hold the greatest potential forselective inhibitor delivery to a specific wide spectrum inhibitory effectscell population was supported by the toward secretory diarrhea. Specificeffectiveness of much smaller doses of inhibition of chloride conductanceinhibitor (10-50 nmol) encapsulated should prevent net fluid secretion intowithin targeted liposomes in reducing the gut lumen regardless of the intra-the secretory response to theophylline. cellular second messenger compound

Both theophylline and cholera toxin which is responsible for the activationcaused jejunal loops to change from an of chloride conductance activity. Bulkabsorptive to a net secretory mode. delivery of chloride conductanceMaximal net secretory capacity of inhibitors to the fluid phase in the10 cm jejunal loops in this in situ lumen did not reduce a net fluid secre-

tory response to theophylline or tocholera toxin (7). The targeted liposomedelivery system provided an excellentopportunity to assess the role ofdelivery or accessibility of the conduc-tance inhibitors to the site of the con-ductance protein in secretory entero-cytes. The lack of antisecretory effectsseen with encapsulated aPC andNPEB in the in situ secretory modelmay raise questions about sidedness ofinhibitor actions, or about the relation-ship of chloride channels assayed inin vitro preparations of membranevesicles or cultured cells versus chloridechannels involved in secretory ion fluxin situ.A good secretory dose response has

been produced with V. cholerae culturefiltrate in the range of 50 to 200 mg oflyophilized filtrate per 10 cm jejunalloop (6). The filtrate used in this studyapparently saturated the secretorycapacity of jejunal loops at the lowestdose used, the 50 mg per loop level.These high doses of cholera toxincould have produced a situation thatwas not favorable for reversal of thesecretory response.Reduced efficacy of 2'-d-3'-AMP

against net fluid secretion induced bycholera toxin may be due to problemsin delivery of the inhibitor to the frac-tion of the enterocyte population thathas been targeted previously by choleratoxin. Saturation of secretory capacity(Table III) may imply saturation of thetotal population of ganglioside GM1present on the mucosal brush bordersurface within the lumen of the jejunalloop. In this scenario there could besecretory enterocytes which have allsurface ganglioside binding sites forthe B subunit already occupied byholotoxin. This prior receptor occu-pancy could prevent the delivery ofinhibitors that are encapsulated withinliposomes made with surface B subunit.This effect could occur throughoccupancy of surface ganglioside GM1site with cholera toxin, or throughremoval of these sites from the surfaceafter binding to the B subunit of theholotoxin.

It may be significant that of the sixdifferent experiments reported abovewith cholera toxin as a secretory agentand 2'-d-3'-AMP as the secretoryantagonist there was only one case(50 mg CT dose per loop in Table III)where the net fluid flux after exposure

254

to cholera toxin was actually returnedto a negative or absorptive value. Thiseffect could be related to differentmechanisms of action of cholera toxinon two separate processes: stimulationof unidirectional Cl- secretion versusinhibition of Na+ and Cl- absorption.Unidirectional Cl- secretion may bestimulated primarily by cAMP, whileNaCl absorption may be inhibited byincreased cytosolic concentrations ofCa2+ ion (20,21). This situation couldaccount for a failure of the adenylatecyclase inhibitor to restore the normalabsorptive component of the net fluidflux.To distinguish between the hypoth-

eses presented in the previous para-graphs it may be necessary to measurethe effects of secretory inhibitorsencapsulated in targeted liposomes onnet fluid flux produced by submaximaldoses of cholera toxin. Inhibitor effec-tiveness could be determined wherethere is only partial occupancy of intes-tinal receptors with the native form ofcholera toxin.

REFERENCES

1. FIELD M. Intestinal secretion: effect ofcyclic AMP and its role in cholera. New EnglJ Med 1971; 284: 1137-1144.

2. DONOWITZ M, WICKS J, BATTISTI L.Senokot inhibits Na and Cl transport in ratdescending colon by a Ca2 + dependentmechanism. Gastroenterology 1984; 87:503-5 12.

3. DONOWITZ M, WICKS J, MADARA JL,SHARP GWG. Studies on the role of cal-modulin in Ca2+ regulation of rabbit ilealNa and Cl transport. Am J Physiol 1985;248: G726-G740.

4. DOBBINS J, RACUSEN L, BINDER HJ.Effect of D-alanine methionine enkephalinon ion transport in rabbit ileum. J ClinInvest 1980; 66: 19-28.

5. McKAY JS, LINAKER BD, TURNBERGLA. Influence of opiates on ion transportacross rabbit ileal mucosa. Gastroenterology1981; 80: 279-284.

6. FORSYTH GW, HAMILTON DL, SCOOTA, GOERTZ KE, KAPITANY RA. Failureto reverse cholera toxin-induced intestinalsecretion by agents which decrease mucosalcyclic AMP. Can J Physiol Pharmacol 1979;57: 1004-1010.

7. FORSYTH GW, GABRIEL SE. Effects ofchloride conductance inhibitors on fluidsecretion into ligated ileal and jejunal loopsin pigs. Am J Vet Res 1990; 51: 1551-1555.

8. UWIERA RE, ROMANCYIA DA, WONGJP, FORSYTH GW. Effect of covalentmodification on the binding of cholera toxinB subunit to ileal brush border surfaces.Anal Biochem 1992; (In press).

9. MAYER LD, HOPE MJ, CULLIS PR.Vesicles of variable size produced by a rapidextrusion technique. Biochim Biophys Acta1986; 858: 161-168.

10. ELLMAN GL. Detection of free sulfhydrylgroups with dithio-nitrobenzoic acid. ArchBiochem Biophys 1959; 82: 70-77.

11. HUTCHINSON FJ, JONES MN. Lectin-mediated targeting of liposomes to a modelsurface: an ELISA method. FEBS Lett 1988;234: 493-496.

12. BERGER EY, STEELE JM. The calculationof transfer rates in two compartment systemsnot in dynamic equilibrium. J Gen Physiol1958; 41: 1135-1152.

13. HAMILTON DL, ROE WE, NIELSENNO. Effect of heat stable and heat labile

Escherichia coli entrotoxins, cholera toxinand theophylline on unidirectional sodiumand chloride fluxes in the proximal and distaljejunum of weanling swine. Can J CompMed 1977; 41: 306-317.

14. STALLA GK, STALLA J, VON WERDERK, MULLER OA, GERZER R, HOLLT V,JACOBS KH. Nitroimidazole derivativesinhibit anterior pituitary cell functionapparently by a direct effect on the catalyticsubunit of the adenylate cyclase holoenzyme.Endocrinology 1989; 125: 699-706.

15. ZENSER TV. Inhibition of cholera toxin-stimulated intestinal epithelial cell adenylatecyclase by adenosine analogues. Proc SocExp Biol Med 1976; 152: 126-129.

16. JOHNSON RA, YEUNG SM, STUBNERD, BUSHFIELD M, SHOSHANI I. Catio-nic and structural requirements for P site-mediated inhibition of adenylate cyclase.Mol Pharmacol 1989; 35: 681-688.

17. FORSYTH GW, GABRIEL SE. a-Phenyl-cinnamate inhibits chloride conductance injejunal brush border membrane vesicles.Biochim Biophys Acta 1989; 977: 19-25.

18. WANGEMANN P, WITTNER M, DISTE-FANO A, ENGLERT HC, SCHLATTERE, GREGER R. Cl- channel blockers inthe thick ascending limb of the loop ofHenle: structure activity relationship. Eur JPhysiol 1986; 407: S128-S141.

19. HUANG C, NELSON JT. Geometric pack-ing constraints in egg phosphatidylcholinevesicles. Proc Natl Acad Sci USA 1978; 75:308-310.

20. FAN CC, POWELL DW. Calcium/calmodulin inhibition of coupled NaCltransport in membrane vesicles from rabbitileal brush border. Proc Natl Acad Sci USA1983; 80: 5248-5252.

21. MAENZ DD, GABRIEL SE, FORSYTHGW. Calcium transport affinity, ion com-petition and cholera toxin effects on cytosoliccalcium concentration. J Membrane Biol1987; 96: 243-249.

255