Proc. Nat. Acad. Sci. USAVol. 72, No. 11, pp. 4487-4491, November 1975Cell Biology

Structure of tight junctions in epithelia with different permeability(freeze-fracture/electron microscopy/epithelial transport/transepithelial resistance)

ADOLFO MARTINEZ-PALOMO AND DAVID ERLIJCentro de Investigacion del I.P.N. Apartado Postal 14-740. Mexico, D. F. Mexico

Communicated by George E. Palade, August 21,1975

ABSTRACT Freeze-fracture studies have shown a net-work of intramembrane fibrils in the tight junctions of epi-thelia. A direct correlation between the number of fibrils andjunctional permeability has been suggested by previous stud-ies. However, we have made two groups of observationsshowing that junctional ermeability is not univocally relat-ed to the complexity of the network revealed by freeze-frac-ture. (i) The tight junctions of the rabbit ileum mucosa arepermeable to lanthanum, although they have a complex net-work of fibrils resembling the junctions of toad urinary blad-der, which are impermeable to lanthanum. (ii) The tightjunctions of the toad urinary bladder are normally of lowpermeability; however, when the luminal solution is madehypertonic with Iysine, junctional permeability markedly in-creases and lanthanum permeates through the tight junc-tions. In freeze-fracture replicas, no differences between thefibrils of control and lysine-treated bladders were found. Ourresults indicate that junctional permeability is controlled notonly by the complexity of the fibrilar network, but that somefeatures of the junctions or the fibrils themselves, not yet re-vealed by electron microscopy, play a central role in regulat-ing epithelial permeability.

Tight junctions are specialized cellular contacts that seal lat-eral intercellular spaces at the apical side of lining epithelia.Electron microscopy has shown that the close apposition be-tween adjacent plasma membranes at the tight junction, alsotermed zonula occludens, results in the obliteration of the in-tercellular cleft, forming a continuous sealing belt aroundepithelial cells (1).

Freeze-fracture studies, which expose frozen membranesalong the central plane of the bilayer (2), show that in theregion of the tight junction there is a branching network ofintramembrane "fibrils" (3-6). These junctional fibrils spanthe thickness of the adjacent membranes (4, 7). It is of inter-est to elucidate to what extent the focal obliteration of theextracellular space depends on the organization of the junc-tional fibrils. There are two obvious approaches to the tack-ling of this problem. On the one hand, it has been estab-lished that the permeability of tight junctions as determinedby physiological criteria, such as transepithelial resistanceand permeability to large hydrophylic solutes, and by mor-phological studies, varies considerably among different epi-thelia (8-12). A correlation between normal variations intight junction permeability and the structure and arrange-ment of the junctional fibrils has been reported in a recentfreeze-fracture study of a number of epithelia. A direct cor-relation between the number of fibrils within the membraneand the degree of tight junction permeability was found(13).The second approach is to find whether there are any de-

tectable changes in the arrangement of the fibrils when thepermeability of tight junctions is modified experimentally.One of the most striking cases of alteration of the permeabil-ity of the junctions is found when hypertonic solutions areadded to the apical side of epithelia, such as those of the frogskin (14, 15) or the toad urinary bladder (16-20). The nor-

mally high transmural resistance of these epithelia markedlyfalls when hypertonic solutions are added to the apical side.Several lines of evidence show that this reduction in resis-tance is mainly due to the opening of a paracellular perme-ability pathway localized in the tight junctions (15-17, 19).We describe here results obtained following both lines of

inquiry. In one group of observations, we have comparedthe freeze-fracture morphology of the tight junctions in toadurinary bladder, a low permeability epithelium, with that ofthe rabbit ileum, an epithelium which according to bothphysiological (21) and morphological criteria (9) has perme-able tight junctions (leaky junctions). In other experimentswe have established that solutions made hypertonic with ly-sine induce a reversible opening of the tight junctions of thetoad urinary bladder without gross deformation of the junc-tion, as judged both by thin sections with electron-densetracers and by freeze-fracture studies.

MATERIALS AND METHODSUrinary hemibladders from toads, Bufo marinus, weremounted in normal Ringer's solution for electrical measure-ments, as described (16).The apical side was immersed in Tris/Ringer's solution,

where all the NaCl in the normal Ringer's solution was sub-stituted by Tris'chloride (pH 7.3-7.5). The fine structuraldistribution of lanthanum was studied as described (22). Thedistal 15 cm of the ileum was removed from rabbits anesthe-tized with Nembutal (40 mg/kg) and washed in cold Krebs'solution equilibrated with 95% 02 and 5% CO2.

Samples for freeze-fracture were fixed during 1 hr in2.5% glutaraldehyde, rinsed in 0.1 M cacodylate buffer, andinfiltrated progressively with glycerol up to a 25% concen-tration. The tissues were frozen in Freon 22 and freeze-frac-tured at -115° in a Balzer's 300 device. The replicas wereobserved with a Zeiss EM 10 electron microscope. The mi-crographs are mounted with shadow direction from the bot-tom; shadows are white.To increase the number of tight junction fracture faces

appearing in freeze-fracture replicas, we have used isolatedepithelial sheets in some experiments. The methods used forseparating epithelial sheets from the toad urinary bladderand the small intestine have been described (23-25). Carefulcomparison of replicas from whole tissues and isolated epi-thelial sheets, both of control and urea- or lysine-treated epi-thelia, revealed that the rapid obtention of epithelial sheetsfollowed in a few seconds by fixation with glutaraldehydedid not induce changes in the structure of the tight junction.

RESULTSFreeze-fracture morphology of tight junctionsFigs. 1 and 2 illustrate portions of tight junctions of the rab-bit ileum mucosa and the toad urinary bladder epithelium,

4487

Dow

nloa

ded

by g

uest

on

Dec

embe

r 2,

202

0

4488 Cell Biology: Martinez-Palomo and Erlij

FIGS. 1-3. Fracture faces of tight junctions in normal ileum epithelium (Figs. 1 and 3) and frog urinary bladder epithelium (Fig. 2). Themembrane fibrillar network revealed by the fracture process tends to be similar in both epithelia. Figs. 1 and 2, X55,800; Fig. 3, X21,400.

respectively. Tight junction fibrils appear as ridges on the Aor inner membrane face, and as grooves on the B or outermembrane face. Tight junctions of the rabbit ileum have amean depth of 0.34 gtm 4 0.01 SE (N = 55; range, 0.24-0.50,um) and the mean number of fibrils was 6.72 4 0.21 SE (N= 55; range, 5-10). In the toad urinary bladder, the meandepth of zonulae occludentes was 0.46 ltm + 0.02 SE (N =19; range, 0.32-0.68 ,um) and the mean number of fibrils

was 8.7 I 0.45 SE (N = 19; range, 6-12). Measurementswere made as described earlier (13). In general, the extent ofthe junction exposed by the fracture process in epithelialcells is 2.0 ,m or less in length. Therefore, it is not possible torule out whether there are regions in the zonula occludenswhere the depth and the number of fibrils are different withrespect to other regions examined in the replicas. However,on occasion the fracture plane exposed a large portion of the

Proc. Nat. Acad. Sci. USA 72 (1975)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 2,

202

0

Proc. Nat. Acad. Sci. USA 72 (1975) 4489

3 Rin~ge>(Rh 5OmM 100mM 150mM RInger

0~~~~~~

1.3-*A-0.. 00/

0.9 ** c/

0.5-

0.130 60 90 120 150 180

t(Min)

FIG. 4. Effects induced on the transepithelial resistance ofpaired frog hemibladders by the progressive addition of urea(empty circles) or lysine (filled circles). Tris/Ringer's solution inwhich all sodium had been substituted by Tris-HCl was used onthe mucosal side throughout the experiment.

tight junction. Fig. 3 illustrates a case where 10,m of a tightjunction from the rabbit ileum mucosa were exposed, reveal-ing that the number of fibrils and the depth of the junctionis constant between a given pair of adjacent cells.

Physiological and morphological effects of hypertonicsolutionsIt has been suggested that the opening of tight junctions ofthe toad urinary bladder caused by hypertonic solutions isnecessarily associated with marked deformation of the junc-tion (16, 17). We have surveyed the effects of a number ofsolutes on the resistance and morphology of this epitheliumand have found that lysine causes an opening of the tightjunctions with little structural deformation.

Fig. 4 illustrates an experiment in which we measured theeffects on the transepithelial resistance of progressively in-creasing concentrations of lysine in the apical solution. Aswith other hydrophilic solutes, lysine produces a decrease intransepithelial membrane resistance that was reversed whenisotonic solution was replaced on the apical side of the epi-thelium. It was only necessary to increase osmolarity by rais-ing the lysine concentration to 50-100 mM to observe aclear-cut reduction in resistance. When urea or sucrose isused, the minimal increases in osmolarity that produced adrop in transepithelial resistance corresponded to concentra-tions that varied between 100 and 150 mM for urea and 200and 300 mM for sucrose.To test whether the effect of lysine on transepithelial re-

sistance was due to a chemical action or to the increase in os-molarity, we substituted isosmotically Tris-HCI by lysine insome experiments and found that the transepithelial resis-tance was not altered by this procedure. In other experi-ments, solutions made hypertonic with lysine were placed onthe basal side of the epithelium. When the concentrationwas increased up to 150 mM no effects were detected ontransepithelial resistance. At 200 mM (or more) lysine in theserosal solution a reduction of transepithelial resistance wasobserved. These findings are similar to what is observedwhen other solutes, particularly sucrose and mannitol, areadded to the serosal side of the toad urinary bladder (18).

Although the effects of lysine on the transepithelial resis-tance are similar to those observed when other solutes areused, the effect on the ultrastructure of the tight junctions is

urinary bladder epithelium fixed after 20 min of treatmentwith Ringer's solution I 200 mM urea at the apical side re-vealed the presence of numerous large (0.2-1.0 Itm) vesicles(17-19) within 50-80% of the tight junctions (Fig. 5). Exam-ination of freeze-fracture replicas confirmed the existence ofjunctional blisters induced by hypertonic urea solutions and,in addition, demonstrated the disruption of the fibrillar net-work, particularly at the basal edge of the junction (Fig. 6).When the permeability of the tight junction in urea-treatedurinary bladder was-assessed by the addition of ionic lantha-num before fixation, it was found that a large percentage ofthe junctions, which in control epithelia are not permeatedby lanthanum, were permeable to the tracer whether or nota vesicle appeared at the junction (19).

In sharp contradistinction with the structural modifica-tions induced by urea, the tight junctions of urinary bladderstreated with solutions made hypertonic with lysine (100 or120 mM) and examined in thin sections showed penetrationof lanthanum into 90 to 95% of the junctions (Fig. 7) whileless than 3% of the junctions examined revealed blisters.Tight junctions opened by lysine treatment (100 mM)showed a separation between adjacent membranes whichvaried between 3.0 and 12.0 nm in width. This opening ofthe tight junctions was reversible since in epithelia examinedafter return to isotonicity the junctions were impermeable tolanthanum, as in control samples. When the tonicity of theluminal solution was increased to 300 mM lysine, frequentblisters similar to those observed with urea were found. Infreeze-fracture replicas of toad urinary bladders treatedwith 120 mM lysine for 20 min, the normal pattern of thejunctional intramembrane fibrillar network is preserved(Figs. 8 and 9). In a total of 37 different junctions examined,blisters less than 0.15 Aim in diameter were found only in onecase. In the remaining junctions the number of fibrils, thedepth of the junction, and the general structure of the fibril-lar network remained similar to those seen in control epi-thelia. The mean number of junctional fibrils in bladderstreated with 120 mM lysine was 8.2 I 0.34 SE (N = 37;range, 5-13), and the mean depth of the junctions was 0.41,um + 0.12 SE (N = 37; range, 0.38-0.56 Mm).

DISCUSSION

The first point to emerge from our results is that the relativepermeability of tight junctions is not directly correlated withthe number of junctional fibrils. It has been established byelectron microscopic tracer studies (9, 17, 19) and electro-physiological techniques (21, 26) that the tight junctions ofthe rabbit ileum mucosa are of high permeability, whilethose of the toad urinary bladder are of low permeability.The electrical resistance of the paracellular pathway in thetoad urinary bladder is at least of 12,000 Q-cm2 (26), whilein the rabbit ileum the resistance is 125 Q-cm2 (21). Thefreeze-fracture data shown in this study demonstrate thatthe difference in the number of fibrils between these epi-thelia does not exceed 25%. If we assume that all the fibrilshave uniform resistance properties, it is not possible to postu-late that the differences in junctional permeability betweenrabbit ileum and toad urinary bladder arise solely from vari-ations in the number of junctional fibrils. Further evidencein line with this conclusion arises from observations of thezonulae occludentes at the distal tubules of the rat kidney.This epithelium appears to have a relatively low transmuralresistance (27) and its junctions are permeated by lanthanum

remarkably different. Examination of thin sections of toad

iCell Biology: Martinez-Palomo and Erlij

(28, 29); however, freeze-fracture studies show a relatively

Dow

nloa

ded

by g

uest

on

Dec

embe

r 2,

202

0

4490 Cell Biology: Martinez-Palomo and Erlij

I:

7

FIGS. 5-9. Effects of hypertonic apical solutions on the structure of tight junction of the toad urinary bladder. Figs. 5 and 6 illustratethe junctional blisters induced by urea. Fig. 7 shows the slight separation between adjacent membranes induced by lysine; the opened inter-cellular space is permeable to lanthanum. Lysine-induced hypertonicity does not modify the structure of the tight junctions in freeze-frac-ture replicas, as seen both on fracture face A (Fig. 8) and fracture face B (Fig. 9). Fig. 5, X18,600; Fig. 6, X42,800; Figs. 7 and 8, X61,400; Fig.9, X68,800.

high number of intramembrane fibrils within the tight junc-tions (13, 30, 31). If the number of fibrils within the junc-tional membranes is not the sole determining factor of junc-tional permeability in epithelia, it is necessary to assume thatsome structural and/or chemical property of the junction,not revealed yet by freeze-fracture studies, is involved.

Further support for the notion that the sealing propertiesof the fibrils are not uniform under all circumstances comesfrom our experiments with lysine. This agent increased thepermeability of the tight junctions of the urinary bladderwithout an obvious structural modification of the freeze-fracture morphology of the junctions. This finding is in con-trast with the junctional opening induced by other solutes,which involves an extensive alteration of the fibrillar frame-

work of the zonula occludens, as shown both by thin section(16, 17) and by freeze-fracture (32). Our results show againthat a normal pattern of membrane fibrils of the tight junc-tions, as revealed by freeze-fracture, is compatible with afunctionally leaky junction, as determined both by a de-creased transepithelial electrical resistance and the penetra-tion of lanthanum into the junction.

We thank Dr. E. L. Benedetti for his advice in preparing freeze-fracture samples and Mrs. Amparo Lazaro for technical help. Worksupported in part by a grant from the Hoechst Laboratories, MexicoCity (to D.E. and A.M.-P.), and Grant no. 008 from the Consejo Na-cional de Ciencia y Tecnologia (CONACYT), Mexico (to A.M.-P.).

1. Farquhar, M. & Palade, G. E. (1963) J. Cell Biol. 17,375-412.

Froc. Nat. Acad. Sci. USA 72 (1975)D

ownl

oade

d by

gue

st o

n D

ecem

ber

2, 2

020

Cell Biology: Martinez-Palomo and Erlij

Pinto da Silva, P. & Branton, D. (1970) J. Cell Biol. 45, 598-605.Staehelin, L. A., Mukherjee, T. M. & Williams, A. W. (1969)Protoplasma 67, 165-184.Staehelin, L. A. (1973) J. Cell Sci. 13, 763-786.Chalcroft, J. P. & Bullivant, D. (1970) J. Cell Biol. 47, 49-60.Friend, D. S. & Gilula, N. B. (1972) J. Cell Biol. 53, 758-776.Wade, J. B. & Karnovsky, M. J. (1974) J. Cell Biol. 60, 168-180.Fromter, E. & Diamond, J. (1972) Nature New Biol. 235,9-13.Machen, T. E., Erlij, D. & Wooding, F. B. P. (1972) J. CellBiol. 54,302-312.Ussing, H., Erlij, D. & Lassen, U. (1974) Annu. Rev. Physiol.36, 17-49.Clarkson, T. W. (1967) J. Gen. Physiol. 50,695-727.Rose, R. C. & Schultz, S. G. (1971) J. Gen. Physiol. 57, 639-663.Claude, Ph. & Goodenough, D. A. (1973) J. Cell Biol. 58,390-400.Ussing, H. H. & Windhager, E. E. (1964) Acta Physiol. Scand.61,484-504.Erlij, D. & Martinez-Palomo, A. (1972) J. Membr. Biol. 9,220-240.DiBona, D. R. & Civan, M. M. (1973) J. Membr. Biol. 12,101-128.Wade, J. B., Revel, J. P. & DiScala, V. (1973) Am. J. Physiol.224,407-415.

Proc. Nat. Acad. Sci. USA 72 (1975) 4491

18. Urakabe, S., Handler, J. S. & Orloff, J. (1970) Am. J. Physiol.218, 1179-1187.

19. Erlij, D. & Martinez-Palomo, A. (1973) Fed. Proc. 32,21820. Bindslev, N., Tormey, J. M., Pietras, R. J. & Wright, E. M.

(1974) Biochim. Biophys. Acta 332,286-297.21. Frizzell, R. A. & Schultz, S. G. (1972) J. Gen. Physiol. 59,

318-346.22. Martinez-Palomo, A., Erlij, D. & Bracho, H. (1971) J. Cell

Biol. 50, 277-287.23. Hays, R. M., Singer, B. & Malamed, S. (1965) J. Cell Biol. 25,

195-208.24. Macknight, A. D., DiBona, D. R., Leaf, A. & Civan, M. M.

(1971) J. Membr. Biol. 6, 108-126.25. Miller, D. & Crane, R. K. (1961) Biochim. Biophys. Acta 52,

293-298.26. Reuss, L. & Finn, A. L. (1974) J. Gen. Physiol. 64, 1-25.27. Malnic, G. & Giebisch, G. (1972) Am. J. Physiol. 223, 797-

808.28. Tisher, C. C. & Yarger, W. E. (1973) Kidney Int. 3, 238-250.29. Martinez-Palomo, A. & Erlij, D. (1973) Pflugers Arch. 343,

267-272.30. Pricam, C., Humber, F., Oerrelet, A. & Orci, L. (1974) Lab.

Invest. 30, 286-291.31. Martinez-Palomo, A. & Erlij, D. (1974) Proc. II Latinoamer.

Congr. Elec, Micros., pp. 70-71.32. Wade, J. B. & Karnovsky, M. J. (1974) J. Cell Biol. 62, 344-

350.

2.

3.

4.5.6.7.

8.

9.

10.

11.12.

13.

14.

15.

16.

17.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 2,

202

0