Formation of tight junctions in epithelial cells

Experimental Cell Research 156 (1985) 103-l 16

Formation 1. Induction

of Tight Junctions

in Epithelial

Cells

by Proteases in a Human Co/on Carcinoma Cell Line

ESTHER COHEN,’ AHUVA TALMON,’ ADELBERT BACHER2 and YEHUDA

ORTWIN FAFF,2 BEN-SHAUL’, *

The George S. Wise Life Sciences Faculty, Tel-Aviv University, Tel-Aviv 69978, Israel, and ‘Lehrstuhl fiir Organische Chemie und Biochemie, Technische Universitiit Miinchen, D-8046 Garching, Germany

The experimental modulation of tight junctions (TJ) was studied in the human adenocarcinema cell line HT29 by freeze-fracture electron microscopy. The cell line has virtually no TJ when grown in culture. TJ could be induced by mild treatment with a variety of endopeptidases (trypsin, chymotrypsin, collagenase, elastase, plasmin, thrombin, papain, and pronase). Pronase induced the formation of TJ at low (but not at high) concentrations. All exopeptidases studied were unable to induce the formation of TJ. At 0°C the trypsininduced formation of TJ was greatly slowed down although not entirely inhibited. However, when cells were briefly treated with trypsin at 0°C and subsequently transferred to 37°C in the presence of protease inhibitors, TJ were rapidly assembled. Thus an induction phase at low temperature and an assembly phase at high temperature could be experimentally separated. When cells were briefly trypsinized at 0°C and subsequently kept at 0°C without trypsin for several hours, TJ still formed abundantly upon incubation at 37°C. It appears therefore that the effect produced by the protease is retained for long periods in the cold. IQ 1985 Academic Press, Inc.

Tight junctions (TJ) are specialized cell membrane domains at contact regions of neighbouring epithelial cells, in which they form a belt of intimate contacts at the most luminal region of the junctional complex [ 1, 21. TJ can be observed by electron microscopy of thin sections or by freeze-fracturing in which they appear as ridges on P-face and complementary furrows on E-face in glutaraldehyde-fixed specimens [3, 41. More recent studies using unfixed specimens and rapid-freezing procedures showed fib& on the E-face and furrows on the P-face [5]. It has been proposed that the fibrils observed in freeze-fracture electron microscopy consist of proteins [3, 61. This assumption was challenged on the basis of rapid-freezing freeze-fracture studies from which it was claimed that TJ consist exclusively or predominantly of lipids [7, 81. It was proposed that the structure observed in replicas of freeze-fractured specimens correspond to two laterally displaced inverted lipid micelles. Recent studies have shown that inhibitors of protein synthesis prevent the formation of TJ in freshly plated MDCK cells [9]. This indicates the involvement of proteins in the formation of TJ. However, it does not prove that proteins are structural components of TJ. It was * To whom offprint requests should be sent. Copyright @ 1985 by Academic Press, Inc. All rights of reproduction in any form resewed 001448827/85$03.00

104 Cohen et al. also shown recently that TJ fibrils are not dissolved by deoxycholate [lo]. So far, the question whether TJ consist predominantly of lipids or proteins has not found a conclusive answer. It is widely accepted that TJ are dynamic structures subject to rapid modulation [7, 111. Their formation can be induced by a variety of experimental agents such as proteases [12-141, vitamin A [15], cyclic nucleotides [16], treatment with a hypertonic solution of mannitol [17] or even incubation of tissue at 37°C [ 181. These studies were performed either with pieces of tissue or with cultured cells. The formation of TJ subsequent to trypsin treatment of cultured HT 29 cells has been studied in considerable detail [14, 191.This cell line, which has been derived from a human colon adenocarcinoma, develops virtually no TJ under standard tissue culture conditions. However, junctions appeared abundantly after treatment of the cells with trypsin. This could be observed as early as 1 min after the onset of treatment. Extensive and complex meshworks appeared following prolonged treatment. Under the same experimental conditions, the spot desmosomes which were originally present were split into hemidesmosomes. Cytochalasin but not colchicine interfered with the formation of TJ in this experimental system 1191. This paper describes the effect of different proteases and treatment protocols on the formation of TJ in HT 29 cells. The favorable properties of this experimental systems facilitated quantitative evaluation of TJ formed. We have found that TJ can be formed by mild proteolytic treatment with a variety of endopeptidases and that pretreatment with cycloheximide had no effect [14]. The induction, i.e., the treatment of the cells with proteases which leads to the assembly of TJ, could be experimentally separated from the assembly process. A preliminary report on some of these results has been published [20].

MATERIALS

AND METHODS

Media and Cell Cultures HT29 cells were cultured in Dulbecco’s modified Eagle medium (DMEM) (Gibco) with 0.37% glucose and 10% fetal calf serum (FCS). Cells were kept at 37°C in a humidified atmosphere containing 5% CO*. They were transferred every 5 days by treatment with trypsin (0.05 mg/ml) in PBS containing 0.53 mM EDTA. Experiments were performed on cultures grown for 4 days. Viability of cells subsequent to treatment with enzymes was monitored by trypan blue exclusion and/or subculturing.

Enzymes Enzymes were from the following commercial sources: Chymotrypsin type I and type VII, trypsin papain type III, collagenase type II, carboxypeptidase A, aminopeptidase C and plasmin from Sigma, Munich, aminopeptidase M from Boehringer, Mannheim, thrombin from Calbiothem, Giessen, pronase, elastase and collagenase from Serva, Heidelberg. Enzyme activities were monitored according to the following publications: aminopeptidase C and aminopeptidase M [21], carboxypeptidase (221, plasmin 1231,chymotrypsin 1241,trypsin 1251,collagenase [26], elastase [27], papain 1281,pronase 1291,thrombin 1301.Enzyme units are as defined in each respective publication. 2x crystallized,

Exp Cell Res 156 (1985)

Tight junction formation in epithelial cells. Z 105 Table 1. Induction of TJ in HT29 cells by trypsin Membranes Bypsin hh-4

Total”

With TJ’

% TJ’

0 2.5 5 25 50 250 2500

523 87 48 240 72 57 154

5 0 1 73 45 25 87

0.9 0 2 30 62 43 56

Cells were treated for 15 min with solutions of trypsin in DMEM at 37°C. Q No. of membranes observed. ’ No. of membranes with TJ. c % membranes with TJ.

Freeze-fracturing Cells after treatment were fixed in suspension by 1% glutaraldehyde in 0.1 M cacodylate buffer or in phosphate-buffered saline (PBS) pH 7.4 for 1 h at room temperature. They were then washed and suspended in 30% glycerol in 0.1 M cacodylate buffer or PBS pH 7.4 at 4°C overnight. Freezefracturing was performed in a Balzers freeze-etching unit. Replicas were examined with a Jeol 100CX electron microscope. Whenever cells were still attached to the culture dishes after treatment with proteases (e.g., aminopeptidases, carboxypeptidase, collagenase, papain, plasmin, thrombin), they were fixed in situ and subsequently scraped off with a rubber policeman. They were then processed for freeze-fracturing as described above.

RESULTS It has been shown earlier that untreated HT29 cells have very few tight junctions (TJ) [14]. The number of TJ formed could therefore be estimated quantitatively by counting cell membranes with and without junctions in replicas of freeze-fractured specimens. Each exposed contiguous membrane surface which could be unequivocally identified as part of the cell membrane (topological aspects or presence of microvilli) was accepted as one unit, irrespective of size. It is of course possible that an exposed membrane without junctions may belong to a cell which does have TJ on other parts of its surface which were not exposed by freeze-fracturing. Thus, this method of evaluation has a tendency to underestimate the percentage of cells with TJ (because the total surface of any given cell cannot be observed). However, it provides a quantitative estimate which has been found to be highly reproducible. The variation of repeated experiments was generally below 6%. The effect of trypsin on HT29 cells has been studied earlier at a concentration of 2.5 mg/ml [14]. We have now extended the study to a wide range of different proteases and concentrations. As shown in table 1, trypsin at 2.5 mg/ml over a period of 15 min at 37°C induced the formation of junctions in about 50-60% of Exp Cell Res I56 (1985)

106 Cohen et al.

Fig. 1. Freeze-fractured membrane of HT29 cells treated trypsin (2.5 m&ml) for 15 min at 37°C. x23000. Fig. 2. Freeze-fractured membrane of untreated HT 29 cells.

X

16000.

observed membrane surfaces (fig. 1) as compared to less than 1% in untreated cells (fig. 2). The fraction of membranes with TJ decreased to about 30% at trypsin concentrations of 25 &ml although the complexity of the junctions and the degree of anastomosing was not reduced significantly. Even lower concentrations induced the formation of TJ when the time of incubation was increased (table 1). Table 2 summarizes experiments in which TJ were induced by a variety of proteases of different origins. TJ could be induced by each of eight endopeptidases analysed, irrespective of the origin of the enzyme from vertebrates, plants or microorganisms. The structure and complexity of these junctions (number and organization of strands, degree of anastomosing) was similar as in the experiments with trypsin (fig. 3). The induction by the bacterial protease mixture, pronase, showed a marked optimum of about 50 % membranes with TJ at an enzyme concentration of l&100 &ml. Higher concentrations yielded progressively fewer TJ. In additional experiments, cells were treated with pronase at a concentration of 1 mg/ml and samples were analyzed after 1, 6 and 13 min, respectively. In each case the fraction of membranes with TJ was about 34 %. The viabihty of the cells was not

Tight junction formation in epithelial cells. Z 107 Table 2. Induction of TJ in HT29 cells by various proteases Membranes Protease

Cont. Wml)

Total”

With TJb

% TJc

Aminopeptidase C

400

54

1

1.8

Aminopeptidase M

500 250

53 52

1 1

1.8 2.0

Carboxypeptidase

1 250 125

81 45

2 0

2.4 0

a-Chymotrypsin

1 250 125 12.5 1.25

55 53 50 76

34 28 15 1

62 53 30 1.3

Collagenase

1 250 125

201 53

75 0

37 0

Elastase

1 250 125 12.5 1.25

52 63 46 72

28 29 6 3

54 46 13 4.2

1 250 500 125 12.5

71 50 45 73

41 13 5 0

58 26 I1 0

Plasmin

1 250 125 12.5

53 63 49

16 5 0

30 7.9 0

Pronase E

1 250 125 12.5 1.25 0.125

327 246 125 118 71

14 81 65 46 4

4.3 33 52 39 5.6

Thrombin

5000 1 250 500 50 5 0.5

60 62 203 146 90 110

0 2 7 7 5 3

0 3.2 3.4 4.8 5.5 2.7

lkypsin

1 250 125 12.5 1.25

72 54 49 85

40 26 16 10

56 48 33 12

Cells were incubated for 30 min at 37°C with a solution of the respective protease in DMEM. The specific activities of the proteases used were as follows: 0.1 U/mg aminopeptidase C; 7.6 U/mg aminopeptidase M; 24 U/mg carboxypeptidase; 34 U/mg chymotrypsin; 0.22 U/mg collagenase; 40 U/mg elastase; 20 U/mg papain; 4 U/mg plasmin; 8 U/mg pronase; 320 U/mg thrombin; 96 U/mg trypsin. - See table I.

Exp Cell Res 156 ff985)

108 Cohen et al.

Fig. 3. Freeze-fractured membranes of HT29 cells treated at 37°C for 30 mm with (A) elastase (1.2 mg/ml); (E) papain (1.2 mg/ml); (C) pronase (12.5 ug/ml). X30000.

aEected by treatment with high pronase concentration (1.25 mg/ml) as shown by trypan blue exclusion and by subculturing. TJ induced in HT 29 cells by treatment with trypsin were not disrupted by treatment of the cells with high pronase concentrations (table 3). On the other hand, the formation of TJ could not be induced by trypsin in cells which had been pretreated with 1 mg of pronase per ml for 30 min. The induction of TJ by thrombin was maximal at concentrations of 5-50 &ml. However, the fraction of cells with TJ was modest (about 5%) as compared with Table 3. Trypsin and Pronase treatment of HT29 cells Membranes Exp. no.

Pretreatment

Posttreatment

Total”

With TJ’

% TJ’

1 2 3

llypsin Tkypsin Pronase

Pronase ‘Ifypsin

54 72 92

20 24 0

37 35 0

HT29 cells were treated with trypsin (25 ug/ml, 5 min) and pronase (I mg/ml, 30 mitt) at 37’C as indicated. With pronase, cells came off the substrate in small clumps. In expt 3, cells were centrifuged after pretreatment with pronase and subsequently treated with trypsin in suspension. - See table 1. Exp Cell Res 156 (1985)

Tight junction formation in epithelial cells. Z 109 Table 4. Trypsin treatment of HT29 cells at low temperature Membranes llypsin (mdml) 0 0 0.05 2.5 2.5 2.5

Time

No. of expts.

Total’

With TJb

% TJ’

2 min 3h 3h 2 min 15 min 3h

1 2 2 5 3 6

98 102 135 282 142 275

1 0 1 5 2 19

1 0 0.7 1.7 1.4 6.94

Cells were incubated at 0°C in DMEM with or without trypsin. Samples were analysed by freezefracture electron microscopy. Some of these data were published earlier as a preliminary communication [20]. a4 See table 1. d Tight junctions were short and poorly developed.

the other proteases studied. The other endopeptidases tested showed an increase of TJ formation at higher concentrations. Collagenase and plasmin induced the formation of TJ only at relatively high concentrations. In contrast to these findings with endopeptidases, each of three exopeptidases analysed (aminopeptidase M, aminopeptidase C and carboxypeptidase, all of mammalian origin) did not induce the formation of TJ. All experiments described so far were performed at 37°C. Extensive experiments were subsequently carried out with trypsin at low temperature. No junctions were formed by treatment of cells with trypsin (50 &ml) at 0°C during a period of 2 min up to 3 h (table 4). Treatment at 0°C with a high concentration of trypsin (2.5 mg/ml) produced some TJ only after incubation for a long period; in addition, these junctions were very poorly developed, usually consisting of short, single strands (fig. 4). However, we have found that trypsin can cleave the artificial substrate p-toluenesulfonyl-L-argininemethylester [25] at 0°C rather efftciently (16% of the velocity at 37°C). In subsequent experiments, cells were treated at 0°C with different concentrations of trypsin for variable periods. Trypsin was then removed by repeated washing with soybean trypsin inhibitor (STI) (1 mg/ml) in cold DMEM. The cells were then warmed up to 37°C in DMEM in the presence of ST1 and kept for 15 min. Abundant assembly of TJ was observed even in experiments with brief trypsin treatment at low concentration (table 5, fig. 5). Up to 80% of membranes with well-developed TJ were observed. Control experiments were performed using the same experimental conditions without the protease. When cells were kept at 0°C without trypsin for various periods and subsequently incubated at 37°C for 15 min, no TJ were formed. Additional control experiments showed that ST1 did not induce TJ in experiments without trypsin under any experimental conditions (table 5). Exp Cell Res 156 (1985)

110 Cohen et al. Table 5. Separation of induction and assembly of TJ Membranes Agent

Cont. (mdml)

1 0.025 0.0: 2.5 0.0025 0.005 0.025 0.05 2.5 2.5

ST1 Trypsin Trypsin Trypsin Trypsin Trypsin Trypsin Trypsin Tiypsin Trypsin

Time

No. of expts

Total=

2 min 3h 3h 2 min 2 min 2 min 15 min 15 min I5 min 15 min 15 min 3h

2 2 1 1 4 5 2 1 1 4 2 3

144 147 97 84 191 273 214 185 50 159 51 191

With TJb 1 1 1 49 74 201 13 34 27 92 41 123

% TJ’ 0.7 0.6 1 58 39 74 6 18 54 58 80 64

Cells were incubated at 0°C under the conditions indicated. They were then washed twice with ST1 (1 mg/ml) in DMEM at 0°C and subsequently incubated in DMEM with ST1 at 37°C for 15 min. Some of these data were published earlier as a preliminary communication [20]. o--CSee table 1.

Next, we studied whether a brief trypsin treatment in the cold can produce a latent effect which can be retained at low temperature for an extended period (table 6). Cells were treated with trypsin (2.5 mg/ml) for 2 min and washed with ST1 as described above. Following the removal of trypsin, the cells were kept in medium with ST1 at 0°C for periods up to 4 h. The fraction of membranes with TJ after 4 h was 24 %; the junctions were poorly developed and frequently consisted of single strands (fig. 6). However, when the cells were ultimately incubated at 37°C for 15 min in the presence of STI, junctions were formed abundantly (table 5, fig. 7). Catalytic activity of trypsin is absolutely required for the induction of TJ formation. Virtually no junctions were formed by trypsin which had been pretreated with protease inhibitors such as ST1 or phenylmethylsulfonyl fluoride (PMSF) (table 7). Fig. 4. Freeze-fractured membrane of HT29 cells treated with trypsin (2.5 mg/ml) for 3 h at 0°C.

x59000. Fig. 5. Freeze-fractured membranes of HT 29 cells treated with trypsin (0.05 m&l)

for 2 min at 0°C and subsequently incubated with ST1 for 15 min at 37°C. x33000. Fig. 6. Freeze-fractured membrane of HT29 cell treated with 2.5 mg/ml trypsin for 2 min at O”C, and subsequently incubated in DMEM with soybean inhibitor for 4 h at 0°C. ~26ooO. Fig. 7. Freeze-fractured membrane of HT29 cell treated with 2.5 mg/ml trypsin for 2 min at 0°C kept in soybean inhibitor for 4 h at 0°C and then warmed up for 15 min at 37°C. x32ooO. Fig. 8. Freeze-fractured membrane of HT 29 cells incubated for 2 min at 0°C with trypsin (2.5 mg/ml), washed with 4 mM PMSF in DMEM at 3°C and kept at 37°C for 15 min in the presence of 4 mM PMSF in DMEM. x 14300. Exp Cell Res 156 (1985)

S-858331

112 Cohen et al. Table 6. Eflect of incubation of trypsin-treated cells at 0°C Membranes Incubation 0°C

Incubation 37°C

Total”

With TJ*

% TJ’

15 min 15 min 15 mitt

117 105 88 151 135

1 25 71 92 71

0.8 24d 80 61 52

4h lh 4h

HT 29 cells were treated with trypsin (2.5 mg/ml) in DMEM for 2 min at 0°C. Cells were washed with DMEM containing ST1 (1 mg/ml) and subsequently incubated in DMEM containing ST1 at 0°C for the period indicated. They were then incubated at 37°C for the period indicated and analysed by freezefracture electron microscopy. - See table 1. d Tight junctions were single-stranded and short.

However, proteolytic activity is not required during the assembly phase as shown by the following observations. Cells were treated at 0°C with trypsin (2.5 mg/ml, 2 min). They were then washed with medium containing 4 mM PMSF and kept at 37°C for 15 min in the presence of the inhibitor. Under these experimental conditions, junctions were formed abundantly (67%) (fig. 8). It appears trivial that close contacts of neighbouring cell membranes are necessary in order to assemble tight junctions. Cell aggregation experiments were performed to assess the role of membrane contact. A suspension consisting mostly of single cells was produced by trypsin/EDTA treatment of cells. In order to allow for recovery of cell surface proteins, the cells were kept in suspension in bacteriological dishes in medium containing 10% FCS for 4 or 24 h. They were then washed and transferred to Erlenmeyer flasks (25 ml) in 3 ml of medium with FCS (lo6 cells/flask). The flasks were flushed with 5 % CO2 and then incubated on a rotary shaker at 37°C. Samples were taken at intervals. Multicellular aggregates of viable cells were apparent after 2-3 h. After 6, 8 or 24 h, aliquots Table 7. Induction of TJ in HT29 cells by trypsin; injluence of protease inhibitors Membranes Protease inhibitor

Total”

With TJb

% TJ’

None ST1 (1 mg/ml) PMSF (4 mM)

52 64 90

25 0 2

50 0 2.2

A solution of trypsin (50 &ml in DMEM) was preincubated for 10 min with the supplements indicated. Cells were treated with the solutions at 37°C for I5 min. - See table 1. Exp Cell Res 156 (1985)

Tight junction

Table 8. Rotary aggregation

formation

in epithelial

cells. I

113

of HT29 cells

Membranes Aggregation (h)

Total”

With TJbsd

% TJ’

0 6 8 24

99 161 86 184

3 1 0 4

3 0 0 2

Single cells obtained by trypsin/EDTA treatment were kept in suspension for 4 h to allow for recovery of surface proteins. Subsequently, the suspension was agitated on a rotary shaker for the periods indicated at 37°C in an atmosphere of 5 % COr. - See table 1. d Tight junctions were short and single-stranded.

were fixed and processed for electron microscopy. Virtually no TJ were found. However, the aggregates showed well-developed desmosomes (table 8). DISCUSSION It has been known for some time that some proteases can induce the formation of TJ in cultured cells or in tissue [12-141. However, to the best of our knowledge, no systematic study using different proteases in the same experimental system has been performed. The advantage of the present system is that untreated HT29 cells are virtually without TJ; this feature facilitates the quantitation of formation under a wide variety of experimental conditions. We have found that each of the three exopeptidases tested was not able to induce the formation of TJ in HT29 cells. On the other hand, each of the eight endopeptidases studied could induce TJ formation, although the yield was rather low in the case of thrombin. It was also shown that proteolytic action is required, since trypsin pretreated with protease inhibitors was inefficient. The data suggest that endoproteolytic modification of membrane proteins can induce the formation of TJ, irrespective of the specificity of the enzyme. It should be noted that some enzymes studied (collagenase, plasmin, papain) were effective only at relatively high concentrations. It is conceivable that the TJ observed in these experiments were actually induced by small amounts of highly effective proteases of different specificity (such as trypsin) which may have been present as contaminants in the commercial enzyme samples used. This possibility cannot be ruled out at present. In one experiment, we have checked the effect of collagenase (1.25 mg/ml) in the presence of ST1 (0.9 mg/ml). The fraction of membranes with TJ in this experiment was 33 % (19 of 58 membranes) in the presence and 42% (28 of 67 membranes) in the absence of STI. This difference may not be significant. It has been shown earlier that low concentrations of pronase cause the formaExp Cell Res 156 (1985)

114 Cohen et al. tion of TJ in isolated pancreatic islets [ 131.We have found that the induction of TJ in HT29 cells shows a pronounced concentration optimum. Treatment of cells with concentrations of 10-100 l&ml produced tight junctions in about 50% of membranes analysed. At higher concentrations the value dropped to about 4%, although the viability of the cells was not affected as shown by dye exclusion and by subculturing of the treated cells. As shown by independent experiments, pronase cannot disrupt trypsin-induced TJ once assembled. However, it appears possible that high concentrations of pronase can remove or alter som protein necessary for TJ formation which is vulnerable prior to junction assembly. The trypsin-mediated formation of TJ proceeds with very low velocity at 0°C although the enzyme can cleave artificial substrates rather efficiently at low temperature. TJ which were formed at 0°C after incubation for several hours were short and often consisted of single strands. However, it should be noted that the assembly process is not completely suppressed. On the other hand, when cells were treated even briefly with trypsin at 0°C and subsequently incubated at 37°C after removal of the enzyme, junctions formed abundantly. Thus it is possible to separate experimentally the induction and assembly of TJ. Whereas the induction requires the catalytically active enzyme, the assembly of TJ could proceed at 37°C in the presence of protease inhibitors (ST1 or PMSF). It appears that treatment of cells with trypsin in the cold leads to a long-lived modulation of the cell membrane. When cells were subsequently incubated at 0°C in the presence of the protease inhibitor, STI, the assembly of TJ proceeded at low speed, and only small junctions of low complexity formed. However, when the cells were subsequently warmed up to 37”C, large and complex junctions formed abundantly. The data suggest tentatively that the yield of TJ assembled at 37°C is somewhat reduced by long preincubation in the cold. It follows that the protease stimulus can be retained for considerable time at low temperature. It was recently found that the assembly of gap junctions in prostate epithelium can be induced by incubation of the tissue at 0°C followed by incubation at 37°C [31]. It has been suggested that this experimental procedure leads to a perturbation of the cytoskeleton. It is well known that microtubuli depolymerize at low temperature [32-341. Release of the control by microtubuli may enable the confluence of preformed connection precursors. In agreement with this hypothesis, the formation of gap junctions could also be induced by incubation of the tissue with colchicine. Cytochalasin B was also found to induce the formation of gap junctions. As shown in this paper, the induction and assembly of TJ formation in HT29 cells shows quite different characteristics. Cold incubation followed by incubation at 37°C did not induce the formation of TJ. However, as shown earlier, the formation of TJ could be inhibited by cytochalasin B, but not colchicine [19, 35-381. These findings suggest that microtubuli play no important role in the induction and assembly TJ, whereas actin filaments appear to be involved. Exp Cell Res 156 (1985)

Tight junction formation in epithelial cells. Z 115 The mechanism of protease-mediated TJ formation is at present not clear. It has been proposed that membrane fluidity could be the major controlling factor, and that proteases operate by increasing the fluidity [14]. However, it should be noted that TJ can form slowly under the influence of trypsin even at 0°C where the membranes are rigid. Furthermore, the concentration optimum observed with pronase suggests that fluidity is not the only controlling factor, since high concentrations of the enzyme should maximize the fluidity. The aggregation experiments show that close contact is not an important controlling factor of TJ formation, since no TJ were induced during cell aggregation, although desmosomes were formed. As part of our effort to find out whether proteases have similar effects on other epithelial cells, we have performed some preliminary experiments on the human mammary carcinoma cell line T47D. The results obtained so far indicate that at least trypsin has an identical effect on the induction of TJ formation in these cells. The induction and assembly of TJ in experimentally separable phases could enable a detailed study of the factors required to operate in each respective phase, such as membrane fluidity, cytoskeletal involvement etc. Such experiments are now in progress. This work was supported by the Deutsche Forschungsgemeinschaft (grant Ba574/7-1) and by the Israeli Academy for Science (to Y. B. S.). We wish to thank Professor L. Bachmann, Munich, for his help and advice and Professor J. Keidar, Tel-Aviv, for the mammary tumour cell line. We also acknowledge the cooperation of MS Christine Holch, Munich, the technical assistance of Mr Felix Scandrany, Tel-Aviv, and the secretarial assistance of MS Angelika Kohnle.

REFERENCES 1. 2. 3. 4.

Farquhar, M G & Palade, G E, J cell biol 17 (1%3) 375. Friend, D & Gilula, N B, J cell biol 53 (1972) 758. Staehelin, L A, J cell sci 13 (1973) 763. Bullivant, S, 9th International congress on electron microscopy (ed J M Sturgess) vol. 3, p. 675. The Imperial Press Ltd, Toronto (1978). 5. Hirokawa, N, J ultrastruct res 80 (1982) 288. 6. Van Deurs, B & Luft, S H, J ultrastruct res 68 (1979) 160. 7. Kachar, B & Reese, T S, Nature 2% (1982) 464. 8. Pinto da Silva, P & Kachar, B, Cell 28 (1982) 441. 9. Griepp, E B, Dolan, W J, Robbins, E S & Sabatini, D D, J cell bio196 (1983) 693. 10. Stevenson, B R & Goodenough, D A, J cell bio198 (1984) 1209. 11. Martinez-Palomo, A, Meza, I, Beaty, G L Cereijido, M, J cell biol 87 (1980) 736. 12. Shimono, M & Clementi, F, J ultrastruct res 59 (1977) 101. 13. Orci, L, Amherdt, M, Henquin, J C, Lambert, A E, Unger, R H & Renold, A E, Science 180 (1973) 647. 14. Polak-Charcon, S, Shoham, J & Ben-Shaul, Y, Exp cell res 116 (1976) 1. 15. Elias, P M & Friend, D S, J cell bio168 (1976) 173. 16. Azamia, R, Dahl, G & Loewenstein, W R, J membrane biol 63 (1981) 133. 17. Madam, J L, J cell bio197 (1983) 125. 18. Kachar, B & Pinto da Silva, P, Science 213 (1981) 541. 19. Polak-Charcon, S & Ben-Shaul, Y, 9th International congress on electron microscopy (ed J M Sturgess) vol. 2, p. 334. The Imperial Press Ltd, Toronto (1978). 20. Talmon, A, Cohen, E, Bather, A & Ben-Shaul, Y, Biochim biophys acta 769 (1984) 505. Exp Cell Res 156 (1985)

116 Cohen et al. 21. Bergmeyer, H U, Methoden der enzymatischen Analyse, vol. 1, p. 995. Verlag Chemie, Weinheim (1974). 22. - Ibid p. 466. 23. - Ibid p. 1056. 24. Walsh, K A & Wilcox, P E, Methods enzymol 19 (1970) 38. 25. Walsh, K A, Methods enzymol 19 (1970) 41. 26. Wilnsch, E & Heidrich, H G, Hoppe Seyler’s z physiol them 333 (1963) 149. 27. Gertler, A L Hofmann, T, Can j biochem 48 (1970) 384. 28. Amon, R, Methods enzymol 19 (1970) 226. 29. Lin, Y, Means, G E & Feeney, R E, J biol them 244 (1969) 789. 30. Sherry, S & Troll, W, J biol them 208 (1954) 95. 31. Tadvalkar, G & Pinto da Silva, P, J cell bio196 (1983) 1279. 32. Frankel, F R, Proc natl acad sci US 73 (1976) 2798. 33. Olmsted, J B & Borisy, G G, Ann rev biochem 42 (1973) 507. 34. Stephens, R E & Edds, K T, Physiol rev 56 (1976) 709. 35. Meza, I, Ibarra, G, Sabanero, M, Martinez-Palomo, A 8~Cereijido, M, J cell bio187 (1980) 746. 36. Martinez-Palomo, A, Meza, I, Stefani, E & Cereijido, M, J cell biol 83 (1979) 89~ (abstr.). 37. Meza, I, Sabanero, M, Martinez-Palomo, A & Cereijido, M, J cell biol 79 (1978) 243 a (abstr.). 38. Cereijido, M, Meza, I & Martinez-Palomo, A, Am j physiol240 (1981) C96-C 102. Received February 9, 1984 Revised version received June 1, 1984

Exp Cell Res 156 (1985)