Gut 1996; 39: 416-422

Intestinal paracellular permeability duringmalnutrition in guinea pigs: effect of high dietaryzinc

P Rodriguez, N Darmon, P Chappuis, C Candalh, M A Blaton, C Bouchaud, M Heyman

AbstractBackground-Zinc has been shown tohave beneficial effects in vitro onepithelial barrier function, and in vivo toreduce intestinal permeability in mal-nourished children with diarrhoea.Ains-To determine whether malnutri-tion alters intestinal paracellular perme-ability, and whether zinc prevents suchalterations.Methods-Guinea pigs were fed a normalprotein diet (NP group), a low protein diet(LP group), or a low protein diet enrichedwith 1800 ppm zinc (LPZn group) for threeweeks. Intestinal permeability wasmeasured on jejunal segments mounted inUssing chambers by measuring ionicconductance and mucosal to serosal fluxesof 14C-mannitol, 22Na, and horseradishperoxidase. Tight junction morphologywas assessed on cryofracture replicas.Results-Mannitol and Na fluxes and ionicconductance increased in the LP groupcompared with the NP group butremained normal in the LPZn group.Accordingly, jejunal epithelia from the LPgroup, but not from the LPZn group,showed a small decrease in number oftight junctional strands compared withepithelia from the NP group. Neithermalnutrition nor zinc treatment modifiedhorseradish peroxidase fluxes.Conclusions-Malnutrition is associatedwith increased intestinal paracellularpermeability to small molecules, andpharmacological doses of zinc preventsuch functional abnormality.(Gut 1996; 39: 416-422)

Keywords: malnutrition, zinc, tight junction,paracellular permeability, intestine, guinea pig.

Malnutrition, similar to intestinal diseases suchas inflammatory bowel disease, gastrointestinalallergy, coeliac disease, and toxigenic diar-rhoea, is associated with intestinal dysfunction.In children with diarrhoea, oral zinc supple-mentation was shown to improve intestinalpermeability' and to reduce the severity andduration of diarrhoea.2 The effects of zinc weremore pronounced in the subgroup of mal-nourished children, suggesting that theintestinal dysfunction linked to malnutritionresponds positively to zinc treatment.

In both human and animal malnutrition,intestinal transport of macromolecules in-

creases,3-6 and intestinal ionic conductanceand small solute permeability are alsoenhanced7 8 indicating that function of theintestinal barrier is altered during malnutrition.These findings raise the question ofmorphological integrity of the intestinalepithelium during malnutrition, as intestinalmorphology is strongly implicated in thebarrier function.9 In the intestinal epitheliumthe paracellular permeability is mainlygoverned by tight junctions. They provide acontinuous seal around the apical region ofadjacent cells, restricting the free passage ofmolecules and ions across the paracellularpathway. They are regulated at the cellularlevel by the cytoskeleton'0 and arephysiologically modulated by nutrients."`Cytokines such as interferon -y (IFN-y) 2 ortumour necrosis factor cx (TNFa),'3 andpathological events such as the transmigrationofpolymorphonuclear cells,'4 have been shownto alter the intestinal epithelial barrier in vitro,as indicated by the increase in the paracellularpermeability of filter grown intestinal cells, andby morphological changes in the tight junctionnetwork. Interestingly, in a model ofendothelial monolayers, zinc was mandatory tonormal barrier function'5 and prevented thedisruption of this barrier by TNFa.'6The first objective of this study was to gain

insight into the mechanisms by which theintestinal barrier is altered during malnutrition.We therefore analysed the morphological andfunctional alterations that occurred in thejejunal epithelium of malnourished guineapigs. The second objective was to assess theability of pharmacological doses of zinc toprevent intestinal barrier dysfunction in ourmalnourished guinea pigs.

Methods

ANIMALS AND DIETS

Thirty weanling Dunkin-Hartley guinea pigs(Charles River, Saint-Aubin les Elbeuf,France) weighing 178 (SD 8) g at day 0, wereused.They were maintained at room temperature

under a 12 hour light:12 hour dark cycle, fedad libitum with three different diets for threeweeks, and given free access to water. The dietswere based on the known nutritionalrequirements of these animals'7 and werepurchased from INRA/APAE, Jouy en Josas,France. Table I shows details of theircomposition. As guinea pigs are prone to

Laboratoire deCytologie,Universite Paris VI,Institut deNeurosciences,Paris, FranceP RodriguezC Bouchaud

INSERM U 290,Hopital St Lazare,Paris, FranceN DarmonC CandalhM A BlatonM Heyman

Laboratoire Central deBiochimie,H6pital Lariboisitre,Paris, FranceP ChappuisCorrespondence to:Dr M Heyman,INSERM U 290,H6pital St Lazare,107 rue du Fg St Denis,75010 Paris, France.

Accepted for publication4 April 1996

416

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from

Malnutrition, zinc, and intestinal permeability

TABLE I Composition of the diets

Ingredients (gllOOg diet) NP LP LPZn

Soya proteins 30.0 4.0 4.0L-Methionine 0.3 0.04 0-04Saccharose 14-4 23-15 22-88Corn starch 28-8 46-31 45-78Corn oil 5-0 50 50Cellulose 14-0 14-0 14-0Vitamin mixture* 1-0 1-0 1-0Mineral mixturet 6-5 6-5 6-5Additional zinc (ZnSO4 7H2O) - - 0-8

NP=normal protein diet; LP=low protein diet; LPZn=lowprotein diet with 1800 ppm zinc.*Vitamin mixture (IU or mg/kg diet): retinyl palmitate, 25 000IU/kg; cholecalciferol, 2000 IU/kg; dl-tocopheryl acetate, 135IU/kg; menadione, 20 mg/kg; thiamin HCI, 20 mg/kg;riboflavin, 20 mg/kg; nicotinic acid, 200 mg/kg; calciumD-pantothenate, 40 mg/kg; pyridoxine HCI, 15 mg/kg;inositol, 2000 mg/kg; D-biotin, 0-8 mg/kg; folic acid, 10 mg/kg; cyanocobalamin, 0.04 mg/kg; ascorbic acid, 2 g/kg;paraaminobenzoic acid, 10 mg/kg; saccharose, 2-5 g/kg.tMineral mixture (g or mg/kg diet): CaHPO4-2H20, 8-33 g/kg;CaCO3, 14-56 g/kg; NaCl, 2-81 g/kg; KCI, 4-52 g/kg;CH3COOK, 27- 10 g/kg; FeSO4 7H20, 0.40 g/kg; MnSO4H20, 040 g/kg; MgSO4 7H20, 150 glkg; MgO, 5.00 g/kg;CuSO4-5H20, 0.04 g/kg; NaF, 0.04 g/kg; KI, 0.03 g/kg;(NH4)6Mo7024 4H20, 090 mg/kg; CoSO4 7H20, 003 g/kg;Na2SeO,0390 mg/kg; ZnSO4 7H20, 020 g/kg (45 ppm Zn).

develop sensitisation to milk proteins, soyaproteins were used as the source of dietaryproteins. The guinea pigs were divided intothree dietary groups, fed a normal protein diet(NP, 30% soya protein, n=12), a low proteindiet (LP, 4% soya protein, n= 12), and a lowprotein diet enriched with 0-8% (1800 ppm)ZnSO4 7H20 (LPZn, n=6). The LP and NPdiets contained adequate zinc (45 ppm). Aszinc has been reported to be non-toxic evenat doses of more than 100 times therecommended dietary allowance,'8 especiallyin the guinea pig,17 we chose a dose of 1800ppm zinc on the assumption that its effectswould be apparent in the LPZn group. Thislarge amount of zinc constitutes apharmacological dose that may not apply tosupplementation in human malnutrition.Isocaloric dietary adjustments were made withsaccharose and corn starch (1/3:2/3 w/w). Assoya proteins are deficient in amino acidscontaining sulphur, diets were supplementedwith methionine (1% of the soya content).Because low protein diets are known to induceanorexia,19 animals given such diets probablyhave protein energy malnutrition rather thanstrict protein deficiency.

EXPERIMENTAL DESIGNAfter 18-20 days on their respective diets,guinea pigs were fasted for 18 hours,anaesthetised intraperitoneally with a lethaldose of sodium pentobarbitone (90 mg/kg).The jejunum was removed and carefully rinsedfree of intestinal content with cold Ringer'ssolution containing (in mM) 140 Na+, 5.2 K+,120 Cl-, 25 HCO3-, 1.2 Ca2+, 2-4 HP042-, 0.4H2P04, 1 2 Mg2a, and 2 glutamine. Thejejunum was then stripped of the outer musclelayer and fixed for electron microscopy. Otherpieces of jejunum were opened along themesenteric border and immediately mountedin Ussing chambers for permeability studies.Plasma samples and some jejunal segmentswere also collected and stored at -80°C for zinc

and protein assays by flame atomic absorptionand Folin reagent respectively. All theseexperiments were done according to theprinciples of laboratory animal care.

ANALYSIS OF ZINC STATUS OF GUINEA PIGS

The zinc concentration was measured in theplasma and intestine of guinea pigs from thethree dietary groups. Plasma was diluted 1:5with doubly deionised water and intestinalhomogenates were mineralised in Teflonbombs placed in a microwave enclosure(MDS81D, CEM Corp, Matthews, USA) asfollows: 50 p,I intestinal homogenates weremineralised with 300 gl ultrapure nitric acid at500 W for five minutes and then appropriatelydiluted with doubly deionised water. Sampleswere analysed for zinc by air-acetylene flameatomic absorption (Perkin-Elmer spectro-photometer model 2380, Norwalk, CT, USA)at a wavelength of 213-8 nm.

INTESTINAL PERMEABILITY STUDY

Intestinal permeability was analysed in 24animals only. Four adjacent jejunal segmentsfrom each animal were mounted in Ussingchambers as flat sheets with an exposed areaof 0.1 cm2. They were bathed on both sideswith 1.5 ml Ringer's solution, which wascontinuously circulated at constant tem-perature, oxygenated, and maintained atpH 7.4 with 5% CO2 in 95% 02. The mucosaland serosal bathing solutions were connectedvia agar bridges to calomel electrodes formeasurement of the transepithelial potentialdifference and to Ag-AgCl electrodes forcurrent application. The tissue was kept undershort circuit conditions by an automaticclamping device (World Precision InstrumentsInc, New Haven, CT, USA) that cancelled outfluid resistance. Short circuit current wasconstantly recorded and the tissue was pulsedat 5 mV every 30 seconds. The short circuitcurrent deflection was used to calculate ionicconductance (or its reverse, electricalresistance) according to Ohm's law. Steadystate potential difference, short circuit current,and ionic conductance 30 minutes after tissuemounting were taken as baseline electricalvariables. After equilibration, 5 mM mannitolwas added to both the mucosal and serosalcompartments, and 20 kBq/ml 14C-mannitol,20 kBq/ml 22Na' (both from NEN, Dupont deNemours, France), and 0.4 mg/ml of themacromolecular marker horseradish peroxi-dase (HRP type VI, Sigma) were all addedsimultaneously on the mucosal side only. Theirappearance on the serosal side was monitoredby sampling 800 pl aliquots of thiscompartment at 10, 30, 50, 70, and 90 minuteswith buffer replacement. Radioactivity wasmeasured on 500 gl samples by liquidscintillation photometry (Kontron, SL4000)using double label correction. Mannitol andNa+ fluxes were calculated according to theirspecific activity in the medium and expressedin pimol/h.cm2 and liEq/h.cm2 respectively.Fluxes of intact horseradish peroxidase

417

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from

Rodriguez, Darmon, Chappuis, Candalh, Blaton, Bouchaud, Heyman

(ng/h.cm2) were assessed by measuring theenzymatic activity of peroxidase in 200 p,lserosal samples, as previously described.20Fluxes are given at steady state, as the meanof the fluxes at 70 and 90 minutes.

ELECTRON MICROSCOPY

Studies were carried out on enterocytes of themiddle villous region of the jejunum.

Ultrathin sectionsJejunal segments were fixed for one hour in2.5% glutaraldehyde in 0O1 M phosphatebuffer, pH 7.2. Intestinal segments werepostfixed in 1% OS04 in 0-1 M phosphatebuffer, dehydrated, and embedded in Aralditeresin. Sections 120 nm thick were doublestained with aqueous 0/5% uranyl acetate andlead citrate and observed under a PhilipsEM201 electron microscope operating at80 kV.

Freeze-fracture analysis ofjejunal segmentsMicrovillus morphology and basolateralmembrane plications were the variables used toidentify mature and immature cells, for threereasons: the microvilli of undifferentiated cryptcells are shorter and less abundant than indifferentiated absorptive cells, P face particlesare more numerous in the microvillusmembranes of villus cells than in those of cryptcells, and plications of the basolateralmembrane are present near the tight junctions,especially between cells near the villus tip.2'Here, glutaraldehyde fixed segments werecryoprotected with 25% glycerol in 0 1 Mphosphate buffer for 30 minutes, cut, andmounted on replica hats. Samples were frozenin pasty nitrogen, fractured in a freeze-fracturedevice (Cryofract, Reichert-Jung), shadowedwith platinum, and coated with carbon underultrahigh vacuum. Replicas were examined byelectron microscopy after cleaning. To allowfor variations within and between animals,replicas from two animals per group and twosites per animal were studied. In each group,strand numbers were quantified using 15 to 20electron micrographs. On these micrographsstraight lines were drawn perpendicular to theaxis ofthe tight junction network every 0.2g,m,and the number of intersections with the tightjunctions was counted and represented thenumber of strands. A mean strands numberwas then calculated for each electronmicrograph. Overall, the tight junction lengthmeasured accounted for at least 30 ,um pergroup.

STATISTICAL ANALYSIS

Data were analysed with the SAS package(SAS Institute, Cary, NC, USA). Results areexpressed as means (SD), with n as the numberof animals or electron micrographs. Cor-relation and linear regression analysis weredone to study the relation between numericalvalues. Means and ranges were compared by

analysis of variance (ANOVA) using thegeneral linear model procedure (least squaremeans).

Results

NUTRITIONAL STATUS OF GUINEA PIGSThroughout the three weeks of experimentalfeeding, control guinea pigs (NP group) gainedweight as expected and exhibited normalplasma and intestinal protein concentrations(Table II). By contrast, the malnourishedanimals, in both the LP and LPZn groups,showed weight losses, which after 18 daysresulted in a mean body weight that was belowthe initial level at weaning, and was half themean body weight of the NP group. In the LPand LPZn groups, total plasma proteins weresignificantly reduced but intestinal mucosalproteins remained unchanged. As shown inthe LPZn group, zinc treatment duringmalnutrition (LPZn) raised plasma andintestinal zinc concentrations. However,malnutrition alone did not modify zinc status,because in the LP group it remained close tonormal.

INTESTINAL PERMEABILITY

Basal short circuit current was increased(p=0.06) in the malnourished group (LP:50.49 (16.59) pA/cm2, n=10) compared withthe control group (NP: 39.27 (10.11) pA/cm2,n=9) but remained unchanged whenmalnourished animals were given additionalzinc (LPZn: 42.36 (12.18) puA/cm2, n=5).Ionic conductance, calculated as described inmethods, increased significantly in intestinalsegments from the LP group compared withsegments from the NP and LPZn groups(Fig 1A). This increase in conductance wasconcomitant with an increase in mannitol andNa fluxes (Fig 1B and C). There was a linearrelation between mannitol and Na fluxes(Fig 2); the slope of the relation was 58, whichwas close to the theoretical value of 54 for thefree diffusion of mannitol and Na.22 Mannitoland Na fluxes were closely correlated (r=0.98 1,p<0-0001), and the intercept of this relationwas 0.38 puEq/h.cm2. When malnourishedguinea pigs were given the high zinc diet (LPZngroup), all the permeability indices remainedunchanged compared with control values(p>0.5). Lastly, measurements of intacthorseradish peroxidase fluxes disclosed nosignificant increase in the absorption of thismacromolecular tracer (Fig 1D).

MORPHOLOGICAL FINDINGSInitial examination of the apical region of theplasma membrane of enterocytes from NP, LP,and LPZn groups by thin section electronmicroscopy showed no morphological dif-ferences in this respect between the threedietary groups (not shown). Figure 3 showsfreeze-fracture replicas of jejunal segmentsfrom the three groups. In these micrographs,a tight junction appeared as a network

418

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from

Malnutrition, zinc, and intestinal permeability

TABLE II Weight gain, plasma and mucosal proteins, and zinc concentrations in the threedietary groups ofguinea pigs

No of No of No ofguinea guinea guinea

NP pigs LP pigs LPZn pigs

Weight gain (between 127 (28) 12 -39 (15)* 12 -31 (12)* 6days 0-18)

Plasma proteins (g/l) 42-6 (6.6) 12 30-6 (5.7)* 10 31-8 (5.6)* 6Mucosal proteins 13-9 (1-4) 10 13-4 (3.3) 11 14-9 (2.2) 6

(g/l00g)Plasma zinc (,mol/l) 22-7 (5.6) 12 21 4 (8.9) 10 48.8 (14.6)*t 6Mucosal zinc (nmol/g) 215 (68) 12 298 (81) 10 930 (426)*t 6

Values are means (SD); NP=normal protein diet; LP=low protein diet; LPZn=low protein dietwith 1800 ppm zinc.*p<0.001 v NP; tp<0001 v LP.

composed of strands shown by fracture of thecytoplasmic leaflet of the plasma membrane (Pface), and grooves produced by fracture of theexoplasmic leaflet (E face). The number ofstrands and grooves in the tight junctionnetwork was counted to assess the effect ofmalnutrition and zinc on structure of the tightjunctions. Despite the absence of obviousdifferences between the appearance of the tight

LF

30 F A 0.06 F B

20K

E

E

10K

0

4

3

Ev

rrW~=L

0

0.04 [-

EE

0

E

0.02 F-

G

100C

P = 0.07

75

cm

EU

0'

50

25

0

Jman

D

JNa JHRPFigure 1: Transepithelial ionic conductance and mucosal to serosalfluxes of mannitol, .

and horseradish peroxidase measured in Ussing chambers across jejunal segments fromguinea pigs. Results are means (SD). The number of animals analysed was 9, 10, andthe NP, LP, and LPZn groups respectively. *p<0.05 v NP; tp<0-05 v LP. NP=normprotein diet; LP=1low protein diet; LPZn=low protein diet with 1800 ppm zinc. (A)transepithelial ionic conductance (G); (B) mannitolfluxes (7man); (C) sodium fluxe.(7Na); (D) horseradish peroxidasefluxes (_HRP).

6

5

cl-

-1

z

4

3

2

o0 0.02 0.04 0.06

Jman (,umol/h.cm2)0.08 0.1

Figure 2: Relation between fluxes of mannitol (Jman) andNa (7Na) measured in Ussing chambers across jejunalsegments ofguinea pigs. Each point represents one animal.Circle=normal protein diet (NP); square=low protein diet(LP); star=low protein diet with 1800 ppm zinc (LPZn).The correlation between Na and mannitolfluxes issignificant (r=0-981, p<0 0001) and the slope of thislinear relation is 58. This is close to the theoretical value of54 for the free diffusion of mannitol and Na, indicating thatboth markers use the same paracellular pathway.

junctional network in the three groups studiedn (Fig 3), jejunal epithelia from animals in the

LP group, but not from those in the LPZngroup, exhibited 10% less tight junctionalstrands than epithelia from animals in the NPgroup (Table III). We concede that applicationof ordinary hypothesis testing to these resultswould be flawed on account of non-independence of different sites in the sameanimals; the design of our study followedothers" 23 that also failed to allow for this.Another study in our laboratory,24 usingmalnourished animals fed 4% milk proteins(instead of4% soya proteins in this study) alsoshowed that the number of strands decreasedby 15% in malnourished animals (strandsnumber=4-50) compared with control animals(strands number=5.28).

DiscussionThese results confirm that intestinal ionicconductance-25and unidirectional fluxes ofsmall solutes7 8 increase during experimentalmalnutrition, indicating that paracellularpermeability is altered. They further suggestthat this functional abnormality is associatedwith a reduction in the number of strandsconstituting the tight junctions. Our resultsalso show that treatment with a pharmaco-logical dose of zinc prevents intestinal changesfound in malnourished guinea pigs.

In this study, malnutrition caused functionalchanges in the intestine. Firstly, the totalelectrogenic transport of ions, short circuitcurrent, increased, thereby confirming pre-vious findings6 8 and suggesting that mal-

Na, nutrition induces a hypersecretory state ina the intestine. This increase in short circuit

i5 in current is not due to increased Na+-glucosetal cotransport,26 because no glucose was added ins our experiments. Secondly, ionic conductance

as well as mannitol and Na permeabilities also

419

1

2 .

1

p

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from

Rodigluez, Damron, Chappuis, Canidalh, Blatoni, Bouchaud, Hevlinan

Figure 3: Freeze-fracture micrographs of enterocytes fronii guinea pigs fed with three exper2wimenztal diets. (A) niornial proteinldiet (NP); (B) low protein diet (LP); anid (C) low protein diet with 1800 ppm7t zinlc (IPZn). Tight junctio77 nietworks weresimilar in appearance. Abunzdant inicrovilli (MV) anzd subapical inlterdigitationis of the basolateral mjiemlbranie (asterisk)charactenrsed the jejunal villous epitheliumii. Tight junictionis zvere quanitified by tracinig intersectinig linies perpenidicular to theaxis every 0-2 gm (see A). The nlumber of inztersectionis with the tight jululctioils is counitedfor each linie and represenrts thestrands nium7lber. Arrows indicate the shadow directioni. X 60 000. BarO0.5 gm.

rose. This rise in permeability was probablyparacellular, as we found a consistentcorrelation between mannitol and Na fluxes.Moreover, the slope of the relation betweenthem was 58 - that is, close to the theoreticalvalue of 54 - suggesting that these two markersshared the same paracellular pathway.22 Suchfunctional perturbations might be due tomorphological alteration of the intestine,because malnutrition is generally reported toalter intestinal histology at both the light andelectron microscopy levels.27 28 In our study,the absence of epithelial necrosis indicated thatthe increase in permeability was probably due

to specific perturbations of the paracellularpathway. Slight morphological differenceswere found in freeze-fracture replicas ofintestinal segments. Thus the number of tightjunction strands diminished in the LP group,by about 100%, when compared with the NPand LPZn groups. The same pattern wasobserved in a previous study in ourlaboratory,24 using malnourished guinea pigsfed a low protein diet (4%0 milk proteins insteadof 4% soya proteins in this study), in which thestrands number decreased by 15% inmalnourished animals compared with controlanimals. It has previously been suggested that

420

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from

Malnutrition, zinc, and intestinal permeability 421

TABLE III Quantification of tightjunctions in freeze-fracture replicas ofjejunum from the three dietary groups ofguinea pigs

Strands number

NP (n=20) 5-21 (0 50)LP (n= 15) 4.79 (0.41)LPZn (n=15) 5.07 (0.62)

Values are means (SD); NP=normal protein diet; LP=lowprotein diet; LPZn=low protein diet with 1800 ppm zinc. n isthe number of electron micrographs analysed, taken from twosites per animal and from two animals per group.

due to the nature of the transepithelialresistance, relatively minor alterations injunctional permeability produce substantialalterations in electrical resistance,23 29 and thatthe presence of a small subpopulation ofepithelial cells with permeable junctions mayquickly decrease transepithelial resistancebefore obvious permeability changes aredetected.22

In our model, only the permeability to smallmolecules was modified, without any changesin macromolecular horseradish peroxidasetransport. It has generally been reported thatunder conditions of protein malnutrition, theintestine becomes more permeable tomacromolecules. However, in most animalstudies in which macromolecular absorptionincreased during malnutrition, animals werefed a diet containing milk proteins.' In thisconnection, we showed in a previous in-vestigation that in guinea pigs, the increasein macromolecular (horseradish peroxidase)permeability is linked to the presence of cows'milk proteins in the diet and to subsequentmilk sensitisation.30 This is consistent with theabsence of modification in protein permeabilityreported in mice given a low protein milk freesoya based formula.3'Although the present low protein diet did

not in itself induce hypozincaemia or intestinalzinc depletion, feeding malnourished guineapigs with a large amount of dietary zincprevented the increases in short circuit current,ionic conductance, and Na and mannitol fluxesfound in the LP group without additional zinc(Fig 1). Zinc has been found to be beneficialin clinical trials of zinc supplementation indeveloping countries where the prevalence ofzinc deficiency is high.32 In many of these trials,supplementation consisted of physiologicaldoses of zinc, given during the nutritionalrehabilitation phase. Under such clinicalconditions, the efficacy of zinc may mostly bedue to the covering of the high zincrequirement known to be associated with catchup growth.33 By contrast, in the presentexperiments, zinc was given at high dosesthroughout the period of malnutrition. Underthese conditions, its efficacy in improvingintestinal permeability suggests a specific rolein maintaining intestinal structure andfunction. Ultrastructural lesions of theintestinal mucosa, as well as changes in the nettransport of water, electrolytes, and glucosehave been reported in zinc deficient rats.34 35Zinc is an essential micronutrient presentin biomembranes. It stabilises membrane

structure and is involved in membranefunction by interacting, especially at SH grouplevel, with membrane enzymes, transporters,receptors, and channels.36 Zinc also interactswith intestinal phospholipids, protecting themfrom degradation.37 This might explain whyzinc prevented the activation of short circuitcurrent in our malnourished guinea pigs, asexperimental malnutrition was shown to beassociated with an alteration in lipids in themicrovillus membrane,26 which may beinvolved in intestinal dysfunction, especiallyin alterations affecting active electrolytetransport. The beneficial role of zinc infunction of the epithelial barrier has beenshown both in vitro'5 16 and in vivo. 1 38 In vitrostudies showed that zinc deficiency inducedpartial disruption of the endothelial barrier'5and that adding zinc to the normal culturemedium totally prevented the disruption of thisbarrier induced by TNFa.'6 In vivo studiesshowed that the increased intestinal per-meability in children with diarrhoea was sig-nificantly reduced by zinc supplementation.' 38Previous findings regarding the effects of severezinc deficiency on intestinal permeability inrats have shown that the integrity of junctionalcomplexes, as assessed by lanthanumhydroxide diffusion, did not alter.39 However,this study indicates that a high zinc dietmaintains a normal intestinal permeability inmalnourished guinea pigs. Our morphologicalfindings further suggest that zinc may preventthe loss of barrier function induced by mal-nutrition at the tight junctional level. Suchhypothetical interaction between zinc and tightjunctions may be related to the regulation ofzinc sensitive genes in the intestine40 or to thebinding of zinc to cytoskeletal proteins,4' 42 therole of which in regulating intestinal tightjunctions is well established.'0 22

In conclusion, our findings show thatexperimental malnutrition increases intestinalparacellular permeability to small moleculesand that this increase is fully prevented bypharmacological doses of zinc. They furthersuggest that such intestinal dysfunction mighttake place in the tight junctions, thestabilisation ofwhich may be improved by zinc.The exact mechanism involved in the effects ofzinc remains to be determined, as well as itsclinical relevance to human malnutrition.

We thank the Centre Inter-universitaire de MicroscopieElectronique (CIME-Jussieu) for carrying out the electronmicroscopy studies. We are also grateful to D Raison for herexcellent technical assistance and to M Dreyfus for editorialassistance. PR was supported by a CONICIT-BID Venezuelanscholarship. This work was supported in part by grants from theVOLVIC Research Center on Trace Elements.

1 Roy SK, Behrens RH, Haiders R, Akramuzzaman SM,Mahalanabis D, Wahed MA, et al. Impact of zincsupplementation on intestinal permeability inBangladeshi children with acute diarrhoea and persistentdiarrhoea syndrome. J Pediatr GastroenterolNutr 1992; 15:289-96.

2 Sazawal S, Black RE, Bhan MK, Bhandari N, Sinha A,Jalla S. Zinc supplementation in young children withacute diarrhoea in India. N Engi Jf Med 1995; 333:839-44.

3 Heyman M, Boudraa G, Sarrut 5, Giraud M, Evans L,Touhami M, et al. Macromolecular transport in jejunalmucosa of children with severe mainutrition: aquantitative study. J Pediatr Gastroenterol Nutr 1984; 3:357-363.

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from

422 Rodriguez, Darmon, Chappuis, Candalh, Blaton, Bouchaud, Heyman

4 Rothman D, Lathman MC, Walker WA. Transport ofmacromolecules in malnourished animals: II intravenousclearance after antigen absorption. Nutr Res 1982; 2:475-80.

5 Worthington BS, Syrotuck J. Intestinal permeability to largeparticles in normal and protein-deficient adult rats. JNutr1976; 106: 20-32.

6 Darmon N, Pelissier MA, Heyman M, Albrecht R, DesjeuxJF. Oxidative stress may contribute to the intestinaldysfunction of weanling rats fed a low protein diet. 7 Nutr1993; 123: 1068-75.

7 Carey HV, Hayden UL, Tucker KE. Fasting alters basal andstimulated ion transport in piglet jejunum. Am J7 Physiol1994; 267: R156-63.

8 Butzner JD, Gall DG. Impact of protein-caloriemalnutrition on the developing intestine. A model ofyoung rabbits. Biol Neonate 1988; 54: 151-9.

9 Powell DW. Barrier function of epithelia. AmJrPhysiol 1981;241: G275-88.

10 Madara JL, Stafford J, Barenberg D, Carlston S. Functionalcoupling of tight junctions and microfilaments in T84monolayers. Am Jf Physiol 1988; 254: G416-23.

11 Madara JL, Pappenheimer JR Structural basis forphysiological regulation of paracellular pathways inintestinal epithelia. J Membrane Biol 1987; 100: 149-64.

12 Madara JL, Stafford J. Interferon-y directly affects barrierfunction of cultured intestinal epithelial monolayers. 7Clin Invest 1989; 83: 724-7.

13 Rodriguez P, Heyman M, Candalh C, Blaton MA,Bouchard C. Tumor necrosis factor alpha inducesmorphological and functional alterations of intestinalHT29 cl19A cell monolayers. Cytokine 1995; 7: 441-8.

14 Nash S, Stafford J, Madara JL. The selective and superoxideindependent disruption of intestinal epithelial tightjunctions during leukocyte transmigration. Lab Invest1988; 59: 531-7.

15 Hennig B, Wang Y, Ramasamy S, McClain CJ. Zincdeficiency alters barrier function of cultured porcineendothelial cells. J7Nutr 1992; 122: 1242-7.

16 Hennig B, Wang Y, Ramasamy S, McClain CJ. Zincprotects against tumor necrosis factor-a induceddisruption ofporcine endothelial cell monolayer integrity.JfNutr 1993; 123: 1003-9.

17 Navia JM, Hunt CE. Nutrition, nutritional diseases, andnutrition research applications. In: Wagner JE, ManningPJ, eds. The biology of the guinea-pig. New York: AcademicPress, 1976: 235-67.

18 Walsh CT, Sandstead HH, Prasad AS, Newberne PM,Fraker PJ. Zinc: health effects and research priorities forthe 1990s. Environ Health Perspect 1994; 102 (suppl 2):5-46.

19 McCance RA, Widdowson EM. Protein-deficiencies andcalorie deficiencies. Lancet 1966; ii: 158-9.

20 Maehly AC, Chance B. The assay of catalases andperoxidases. In: Glick D, ed. Methods of biochemicalanalysis, vol 1. New York: Wiley, 1954: 357-98.

21 Madara JI, Trier JS. The functional morphology of themucosa of the small intestine. In: Johnson LR, ed.Physiology of the gastrointestinal tract. Chapter 45. NewYork: Raven Press, 1994: 1577-622.

22 Madara JL, Barenberg D, Carlson S. Effects of cytochalasinD on occluding junctions of intestinal absorptive cells:further evidence that the cytoskeleton may influenceparacellular permeability and junctional chargeselectivity. 7 Cell Biol 1986; 102: 2125-36.

23 Marcial MA, Carlson S, Madara JL. Partitioning ofparacellular conductance along the ileal crypt-villus axis:a hypothesis based on structural analysis with detailed

consideration of tight junction structure-functionrelationships. J Membrane Biol 1984; 80: 59-70.

24 Rodriguez P, Darmon N, Heyman M, Bouchaud C. Effectof protein malnutrition on guinea-pig intestinal tight-junctions. Structure/activity relationship. ElectronMicroscopy, ed. Les editions de Physique. Applications inBiological Sciences 1994; 3: 193-4.

25 Young A, Levin RJ. Diarrhoea of famine and malnutritioninvestigations using a rat model. l-Jejunal hypersecretioninduced by starvation. Gut 1990; 31: 43-53.

26 Butzner JD, Brockway PD, Meddings JB. Effects ofmalnutrition on microvillus membrane glucose transportand physical properties. Am Jf Physiol 1990; 259:G940-6.

27 Shiner M. Immune mechanisms in the small bowel mucosaof children with malnutrition: ultrastructure andcorrelation with function. Front Gastrointest Res 1986; 13:172-86.

28 Guiraldes E, Hamilton J R. Effect of chronic malnutritionon intestinal structure, epithelial renewal and enzymes insuckling rats. Pediatr Res 198 1; 15: 930-4.

29 Gonzalez-Mariscal L, Contreras RG, Bolivar JJ, Ponce A,Chavez de Ramirez B, Cereijido M. Role of calcium intight junction formation between epithelial cells. Am JfPhysil 1990; 259: C978-86.

30 Darmon N, Heyman M, Candalh C, Blaton M, Desjeux JF.Anaphylactic intestinal response to milk proteins duringmalnutrition in guinea-pigs. AmJPhysiol 1996; 270: 442-8.

31 Lamont AG, Gordon M, Ferguson A. Oral tolerance inprotein-deprived mice: II. Evidence of normal 'gutprocessing' of ovalbumin, but suppressor cell deficiency,in deprived mice. Immunology 1987; 61: 339-43.

32 Gibson RS. Zinc nutrition in developing countries. Nutr ResRev 1994; 7: 151-73.

33 Golden MHN, Golden BE. Effect of zinc supplementationon the dietary intake, rate of weight gain, and energy costof tissue deposition in children recovering from severemalnutrition. AmJ7 Clin Nutr 198 1; 34: 900-8.

34 Elmes ME, Jones JG. Ultrastructural changes in thesmall intestine of zinc deficient rats. Jf Pathol 1980; 130:37-43.

35 Ghishan FK. Transport of electrolytes, water, and glucosein zinc deficiency. J Pediatr Gastroenterol Nutr 1984; 3:608-12.

36 Bettger WJ, O'Dell BL. Physiological roles of zinc in theplasma membrane of mammalian cells. Jf Nutr Biochem1993; 4:194-207.

37 Tacnet F, Ripoche P, Roux M, Neumann JM. 31P-NMRstudy of pig intestinal brush-border membrane structure:effect of zinc and cadmium ions. Eur Biophys J 1991; 19:317-22.

38 Alam AN, Sarker SA, Wahed MA, Khatun M, RahamanMM. Enteric protein loss and intestinal permeabilitychanges in children during acute shigellosis and afterrecovery: effect of zinc supplementation. Gut 1994; 35:1707-11.

39 Moran JR, Lewis JC. The effects of severe zinc deficiencyon intestinal permeability: an ultrastructural study.Pediatr Res 1985; 19: 968-73.

40 Shay NF, Cousins RJ. Cloning of rat intestinal mRNAsaffected by zinc deficiency.JtNutr 1993; 123: 35-41.

41 Strzelecka-Golaszewska H, Prochniewicz E, DrabikowskiW. Interaction of actin with divalent cations: 1 the effectof various cations on the physical state of actin. Eur YBiochem 1978; 88: 219-27.

42 Serrano L, Dominguez JE, Avila J. Identification of zinc-binding sites of proteins: zinc binds to the amino-terminalregion of tubulin. Anal Biochem 1988; 172: 210-8.

on February 6, 2020 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.39.3.416 on 1 S

eptember 1996. D

ownloaded from