Canalicular Retention as anin VitroAssay of Tight Junctional Permeability in Isolated Hepatocyte Couplets: Effects of Protein Kinase Modulation and Cholestatic Agents

FUNDAMENTAL AND APPLIED TOXICOLOGY ARTICLE NO.

37, 71–81 (1997)

FA972309

Canalicular Retention as an in Vitro Assay of Tight Junctional Permeability in Isolated Hepatocyte Couplets: Effects of Protein Kinase Modulation and Cholestatic Agents Marcelo G. Roma, Dominic J. Orsler, and Roger Coleman School of Biochemistry, The University of Birmingham, Edgbaston, Birmingham B15 2TT, United Kingdom Received August 29, 1996; accepted March 21, 1997

compounds (Watanabe et al., 1987; Kitamura et al., 1990; Fentem et al., 1990; Boyer, 1993; Wilton et al., 1994; Maglova et al., 1995), lipid secretion (Crawford and Gollan, 1988), cytoskeleton function (Oshio and Phillips, 1981; Watanabe and Phillips, 1984), gap junction function (Reverdin and Weingart, 1988; Saez, et al., 1989), and electrophysiological investigations on membrane potential and electrolyte transport (Graf et al., 1984, 1987; Boyer, 1993). In addition, this preparation has proved to be a very useful tool for the study of the disruption of canalicular function induced by hepatotoxic (Fentem et al., 1990; Stone et al., 1994; Coleman et al., 1995) and cholestatic agents (Thibault et al., 1992; Roma´n and Coleman, 1994). Hepatocyte couplets retain the structural and functional polarity of intact hepatocytes. Electron microscopy studies on plasma membrane re-assembling (Boyer et al., 1985; Gautam et al., 1987) and immunolocalization of the tight junction-associated protein ZO-1 (Boyer, 1993) revealed that the canalicular membrane reorganizes within 4–5 hr of culture, reestablishing the biliary pole between the two adjacent cells as a vacuole surrounded by a belt of tight junctional cell membrane contact. This structure seals the canalicular space from the extracellular medium, limiting the access of extracellular markers into the canalicular space. For example, ruthenium red, an electrondense, ionic dye, is excluded from the canalicular space of 90% of couplets (Gautam et al., 1987; Boyer, 1993), and the same holds true for the protein, horseradish peroxidase (HRP)1 (Nathanson et al., 1992; Boyer, 1993). Resealing of the canalicular tight junctional apparatus is indispensable for the secretory unit to retain previously secreted materials in its newly formed canalicular vacuole. Indeed, following recovery of cell polarity,

Canalicular Retention as an in Vitro Assay of Tight Junctional Permeability in Isolated Hepatocyte Couplets: Effects of Protein Kinase Modulation and Cholestatic Agents. Roma, M. G., Orsler, D. J., and Coleman, R. (1997). Fundam. Appl. Toxicol. 37, 71–81. A simple, fast method to evaluate acute changes of tight junctional permeability in isolated hepatocyte couplets is proposed. The method consists of the recording of the number of canalicular vacuoles able to retain the previously accumulated fluorescent bile acid analogue cholyl-lysyl-fluorescein (CLF), as visualized by inverted fluorescent microscopy, following acute exposure to the compounds under study. The method was validated by (i) making a systematic documentation of the effect on CLF retention of a variety of hormonal modulators (vasopressin and phorbol esters), as well as several cholestatic (taurolithocholic acid, cyclosporin A, and estradiol 17b-glucuronide) and hepatotoxic agents (menadione, A23187, and t-butyl hydroperoxide), all known to affect biliary permeability in intact liver, and (ii) carrying out a comparative analysis of the results obtained with those recorded using rapid canalicular access of horseradish peroxidase (HRP) as an alternative procedure. The compounds tested all decreased canalicular vacuolar retention of CLF in a dose-dependent manner. Vasopressin- and phorbol ester-induced decline in CLF retention were prevented by pretreatment with the protein kinase C inhibitors H-7 and staurosporine, thus confirming a role for this enzyme in canalicular permeability regulation. A significant direct correlation (r Å 0.934, p õ 0.001) was obtained when the decrease in canalicular retention of CLF was compared with the increment in the canalicular access of HRP. Image analysis revealed that cellular fluorescence was not increased following exposure to these compounds, suggesting a paracellular rather than transcellular route for CLF egress. These results all support canalicular vacuolar retention of CLF as a suitable method to readily evaluate acute changes in tight junctional permeability in isolated hepatocyte couplets induced by physiological modulators or hepatotoxic agents. q 1997 Society of Toxicology.

1 Abbreviations used: CLF, cholyl-lysyl-fluorescein; cVR, canalicular vacuolar retention; CyA, cyclosporin A; DAB, 3,3*-diaminobenzidine tetrahydrochloride; DMSO, dimethyl sulfoxide; 17b-EG, estradiol 17b-glucuronide; H-7, 1-(5-isoquinolinylsulfonyl)-2-methyl piperazine; HRP, horseradish peroxidase; L-15, Leibovitz-15; PDB, phorbol 12,13-dibutyrate; PKC, protein kinase C; PMA, phorbol 12-myristate 13-acetate; t-BuHP, t-butyl hydroperoxide; TDHC, taurodehydrocholate; TLC, taurolithocholate; VP, vasopressin.

Isolated hepatocyte couplets provide an accessible, functional, primary unit of canalicular bile formation. This model has been employed for the study of a variety of hepatocyte physiological functions comprising transport of cholephilic 71

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

0272-0590/97 $25.00 Copyright q 1997 by the Society of Toxicology. All rights of reproduction in any form reserved.

ftoxa

72

ROMA, ORSLER, AND COLEMAN

hepatocyte couplets are able to retain, in the canalicular vacuole, organic anions such as fluorescein or fluorescein derivatives of bile acids, either after ionophoresis through microelectrodes or following uptake and secretion of these compounds (Boyer, 1993; Wilton et al., 1993b). In turn, resealing of the canalicular space represents the first step for the overall rearrangement of the canalicular and basolateral membranes following loss of polarity induced by cell isolation, including expression of the respective membrane transport systems in these surface domains (Gautam et al., 1987; Boyer, 1993; Roelofsen et al., 1995). Similarities between the permeability properties of the hepatocyte couplet paracellular barrier and those of the intact liver make the former a good potential tool for the study of biogenesis, regulation, and integrity of hepatocellular tight junctional structures, without the interference of other complicating factors such as extrahepatic influences or hepatic hemodynamic changes. Despite these advantages, only a few methodological approaches have been described to evaluate tight junctional permeability in couplets, and no systematic analysis of the employed methods was performed. Penetration of the extracellular marker, ruthenium red, into the canalicular vacuole, as assessed by electron microscopy, has been regarded as indicative of the tightness of tight junctional structures (Gautam et al., 1987), but this procedure only permits qualitative rather than quantitative evaluation of this property. An alternative quantitative approach was introduced by Nathanson et al. (1992), who recorded changes in paracellular permeability induced by hormonal messengers, by assessing the percentage of couplets which were penetrated by exogenously administered HRP. In preliminary studies from our laboratory (Stone et al., 1994; Roma´n and Coleman, 1994) we observed that canalicular retention of the preaccumulated, fluorescent, conjugated bile acid analogue cholyl-lysyl-fluorescein (CLF) was impaired by 2-methyl1,4-naphthoquinone (menadione) and cyclosporin A (CyA), two compounds known to affect biliary permeability in intact liver (Kan and Coleman, 1990; Lora et al., 1991). Since canalicular retention of cholephilic compounds requires tight junctions to be sealed (Wilton et al., 1993b), these observations raise the interesting possibility that changes in canalicular retention of CLF may actually reflect modifications of paracellular permeability in couplets. To investigate this possibility further, the present study provides a systematic documentation of the effect on CLF retention of a variety of hormonal modulators and hepatotoxic compounds known to affect biliary permeability in intact liver. A comparative analysis of the results with those recorded using rapid canalicular access of HRP is also provided. METHODS Materials. CLF, kindly provided by Dr. Charles O. Mills (Birmingham, UK), was synthesized and its purity confirmed according to the methods

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

of Mills et al. (1991); the synthetic procedures gave a high yield of CLF, which appeared as a single spot after HPTLC. Collagenase type A from Clostridium histolyticum was purchased from Gibco (Paisley, UK). Taurolithocholic acid (sodium salt) (TLC) and taurodehydrocholic acid (sodium salt) (TDHC) were from Calbiochem-Novabiochem Ltd. (Nottingham, UK). CyA was from Sandoz A.G. (Basel, Switzerland). Leibovitz-15 (L-15) tissue culture medium, bovine serum albumin (fraction V), 3,3*-diaminobenzidine tetrahydrochloride (DAB), HRP, phorbol 12,13-dibutyrate (PDB), phorbol 12-myristate 13-acetate (PMA), vasopressin (VP), staurosporine, 1-(5-isoquinolinylsulfonyl)-2-methyl piperazine (H-7), 2-methyl-1,4-naphtoquinone (menadione), A23187, t-butyl hydroperoxide (t-BuHP) and estradiol 17b-glucuronide (17b-EG) were obtained from Sigma Chemical Co. (Poole, Dorset, UK). All other chemicals were of reagent grade. Animals. Male Wistar rats bred at in the University of Birmingham (220–240 g) were used throughout. Before the experiments, the animals were maintained on a standard laboratory diet (41B maintenance diet, Pilsbury, Birmingham, UK) and tap water ad libitum. Rats were anaesthetised using Ketalar (ketamine hydrochloride, 6 mg/100g body wt) with Domitor (medetomidine, 25 mg/100g body wt). Surgery was started between 8:00 AM and 9:00 AM to minimize circadian variations. Isolation, enrichment, and culture of hepatocyte couplets. Hepatocyte couplets were obtained from rat liver according to the two-step collagenase perfusion procedure described by Wilton et al. (1993b), adapted from Gautam et al. (1989). Following a perfusion (20 ml/min) with a Ca2/- and Mg2/-free Hanks’ balanced salt solution into the portal vein for 12.5 min, approximately 0.02% (depending on batch) of collagenase in Hanks’ balanced salt solution was infused for 4.5 min. The liver was then removed, chopped, and agitated in ice-cold Krebs–Henseleit solution for 5 min. The final suspension was filtered through 150-mm nylon gauze and the remaining tissue was reincubated in the collagenase solution for approx 7 min (dependant upon collagenase batch activity) at 377C to liberate a second cell preparation with a high overall viability (ú90%), as assessed by trypan blue exclusion using an improved Neubauer hemocytometer. This initial preparation contained 24 { 4% (n Å 20) of couplets with high viability (ú93%). If either cell of a couplet stained positively with trypan blue, then the whole unit was considered nonviable. Couplets were further enriched using centrifugal elutriation as described by Wilton et al. (1991). The resulting preparation, containing 69 { 5% of couplets (n Å 20), was plated in L-15 containing 50 U/ml penicillin and 50 mg/ml streptomycin onto 35mm plastic culture dishes (2 ml/dish) at a density of 0.5 1 105 units/ml, and incubated at 377C for 5 hr. This time was described to be appropiate for couplets to reach maximal capability to transport and accumulate CLF into their canalicular vacuoles (Wilton et al., 1993b). Since this capability then remains almost constant for at least 4 hr (Wilton et al., 1993b), experiments were performed within this time period. Assessment of tight junctional permeability. Paracellular permeability in hepatocyte couplets was assessed by two methods, namely, (i) canalicular vacuole retention (cVR) of CLF and (ii) rapid appearance of HRP within the canalicular space. A schematic representation describing both methods is shown in Fig. 1. cVR of CLF was assessed by determining the percentage of couplets able to retain this fluorescent bile acid analogue (Roma´n and Coleman, 1994; Stone et al., 1994). For this purpose, CLF (2 mM, final concentration in the dishes) was added to each plate and incubated at 377C for 15 min before washing twice at 377C with 2 ml of L-15. This time was chosen because canalicular accumulation of CLF was shown to reach an apparent steady state within this time (Wilton et al., 1993b). Where ‘‘hormones’’ or cholestatic or hepatotoxic compounds were studied (see later section on treatments), they were added at this point into the dishes containing 2 ml of L-15 medium after removing CLF by washing twice with 2 ml of L-15 at 377C. cVR of CLF was monitored by using an inverted fluorescent microscope (Olympus IMT2-RFL; Olympus Optical Ltd., London, UK) equipped with a 100 W mercury light source and an incubator to maintain the cells at

ftoxa

CANALICULAR RETENTION AND PERMEABILITY IN COUPLETS

73

FIG. 1. Schematic representation of the different assays employed to evaluate paracellular permeability in isolated hepatocyte couplets. Under normal conditions, tight junctional apparatus reseals after a short-term culture. This resealing is indispensable for the couplets to retain previously secreted materials in, and to exclude extracellular material from, their newly formed canalicular vacuole. Opening of tight junctions would cause accumulated materials to escape from the canalicular vacuole and extracellular material to penetrate into this space. Opening of tight junctions can therefore be quantified by counting the number of couplets able to retain previously secreted cholyl-lysyl-fluorescein (CLF) (retention assay) or the number of couplets which were penetrated by exogenously administered horseradish peroxidase (HRP) (penetration assay).

377C during observation. Preaccumulated CLF is easily visualized as a brightly fluorescent vacuole between adjacent hepatocytes. Due to progressive bleaching of the fluorophore with continuous exposure to ultraviolet light, such fluorescent microscopy can only be used as a series of ‘‘snapshots’’ rather than continuous recording. Therefore, the estimation of cVR of CLF was determined by randomly choosing a field of vision and all the couplets visible in the field (ú30) were then counted within 30–45 sec. Results were expressed as the percentage of total couplets counted displaying sufficient CLF to be visible in the canalicular vacuole. In some experiments, canalicular and cellular couplet fluorescence were quantified by performing image analysis. For this purpose, fluorescent microphotographs from randomly chosen fields of vision were digitalized to obtain computer images on which cellular and canalicular CLF fluorescence in the couplets could be quantified by densitometry with the aid of an image analysis program. The intensity of cellular fluorescence (as a measure of CLF content in the cell body) was derived by dividing the amount of fluorescence inside the cells by total couplet area. Canalicular fluorescence (as a quantitative estimation of CLF retention in the canalicular vacuole) was expressed as a percentage of total (canalicular plus cellular) fluorescence. All the couplets in a field (ú30) were examined, irrespective of whether they displayed canalicular fluorescence. Assessment of rapid canalicular HRP penetration was based on the method described by Nathanson et al. (1992). The cells were exposed to a 4 mg/ml solution of HRP in L-15 for 1 min. The cells were then washed once with L-15 and fixed by incubating for 30 min with 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.2, at room temperature. The cells were then washed twice with 0.1 M phosphate buffer (pH 7.4) and left overnight at 47C in 0.1 M phosphate/2.5% sucrose buffer (pH 7.4). HRP was developed via the DAB reaction. For this purpose, the cells were washed once with 0.1 M Tris–HCl buffer (pH 7.4) and left for 5 min at room temperature. The buffer solution was then removed and the cells exposed consecutively to a 0.2% DAB solution in 0.1 M Tris–HCl buffer (pH 7.4) for 30 min and to the same solution containing 0.01% H202 for 10 min. Finally, the cells were washed twice with 0.1 M Tris–HCl buffer (pH 7.4) and observed using light microscopy (Olympus IMT2-RFL, Olympus Optical Ltd.). A darkened canaliculus was considered indicative of the presence of HRP/ DAB reaction product. Canalicular HRP was determined in 200 consecutive

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

couplets and results were expressed as the percentage of the total couplets analysed. Treatments. Changes in cVR of CLF or HRP penetration in response to a number of hormones and second messengers acting through protein kinase C (PKC) were studied. The compounds tested comprised (final concentrations in plate, volume of vehicle used to deliver): VP (1007 –10011 05 09 M, 1 ml in acetic acid 0.05%), PDB (10 –10 M, 2 ml in dimethyl sulfoxide, DMSO), and PMA (1005 –1009 M, 2 ml in DMSO), with or without the PKC inhibitors H-7 (1004 M, 10 ml in saline) or staurosporine (1006 M, 10 ml in saline). Stocks of these compounds were prepared in the solvents indicated above and stored in frozen aliquots until use. Cells incubated with VP, PDB, or PMA were exposed to these agents for 30 sec and couplets pretreated with H-7 or staurosporine received the inhibitor for 1 min prior to stimulation. These doses and times of exposure were similar to those employed Nathanson et al. (1992) to assess the effect of hormonal messengers on canalicular permeability in isolated hepatocyte couplets. A variety of hepatotoxic and cholestatic agents were also tested for their ability to affect cVR of CLF. They included (final concentrations in plate) CyA (25–1000 nM), 17b-EG (50–400 mM), TLC (0.5–5 mM), A23187 (0.125–5 mM), menadione (5–100 mM), and t-BuHP (25–350 mM). The compounds were all dissolved in DMSO and added in a volume of 10 ml of stock solutions, which were stored in frozen aliquots until use. The time of exposure of the compounds was 15 min for CyA and menadione and 30 min for 17b-EG, TLC, A23187, and t-BuHP. These times of exposure were chosen on the basis of preliminary experiments showing that a maximal effect had been reached at these respective time periods. For every experiment, a control dish receiving vehicle alone was treated simultaneously, thus taking into account any change in cVR of CLF with time. Viability was systematically assessed before and after each treatment (trypan blue exclusion assay) and showed no significant change. When rapid access of HRP into the canaliculus was determined, only the maximal dose of each compound was tested. Statistical analysis. Results are expressed as means { SE and were considered significant when the p value was õ 0.05 using paired Student’s t test. Regression line analysis was carried out using the least squares method and significance of the difference against zero of the correlation

ftoxa

74

ROMA, ORSLER, AND COLEMAN

induced decrease of cVR of CLF. Complete prevention of this effect was seen when couplets were pretreated with either inhibitor at these concentrations. H-7 and staurosporine alone had no independent effect on cVR of CLF (data not shown). Effect of Hepatotoxic and Cholestatic Agents on cVR of CLF

FIG. 2. Time course of the decline in canalicular vacuolar retention of previously accumulated cholyl-lysyl-fluorescein (CLF) following removal of the fluorescent bile acid from the incubation medium in hepatocyte couplets. Couplets were allowed to accumulate CLF (2 mM, final concentration in the dishes) for 15 min and then the fluorescent bile acid was removed with two washes of L-15 medium (time, 0 min). Canalicular CLF retention was expressed as the percentage of total couplets retaining CLF referred to the initial value. All values are means { SE of four experiments. *Significantly different from the initial value (p õ 0.05).

coefficient r was calculated using appropriate tables (Snedecor and Cochran, 1969).

Figure 5 depicts the dose–response effect on cVR of CLF of different compounds known to affect tight junctional permeability in intact liver. These agents comprise the cholestatic compounds CyA, 17b-EG, and TLC, the calcium ionophore A23187, and the oxidative stress inducers menadione and t-BuHP. None of these compounds affected cell viability over the entire range of doses tested. Instead, these compounds all reduced the percentage of canaliculi retaining CLF in a dose-dependent manner. Although different patterns of dose–response effect were observed, the rate of decrease of this parameter was generally higher at lower doses and then decreased slower as the dose increases to approach an apparent minimal value. This asymptotic value appeared to be different depending on the considered compound. In fact, according to the patterns obtained, two groups could be differentiated. CyA, 17b-EG, and menadione seem to reach a relatively high plateau (cVR of CLF at the highest dose: 42.6, 35.7, and 40.5%, respectively), whereas this asymptotic value appears to be lower for TLC, A23187, and t-BuHP (cVR of CLF at the highest dose: 23.5, 16.5, and 12.7%, respectively).

RESULTS

Stability of cVR of CLF Time-dependent changes in the capability of hepatocyte couplets to retain previously accumulated CLF are shown in Fig. 2. The number of couplets retaining CLF remained around 75% following 15 min of CLF exposure (Table 1). Following CLF removal from the incubation medium, this value (expressed as percentage of original value) remained roughly stable for 30 min, but then declined gradually so that after 2 hr, canaliculi retaining CLF were approximately half of the initial value (Fig. 2).

Comparison of cVR of CLF and HRP Penetration to Evaluate Tight Junctional Permeability in Hepatocyte Couplets Comparative data of the number of canaliculi retaining CLF and the values recorded using HRP canalicular penetration as a reference method following exposure to the different compounds tested in this study are summarized in Table 1. As shown in Fig. 6, a clear linear correlation was obtained between the values recorded using both methods, as suggested by the high correlation coefficient obtained (r Å 0.934, p õ 0.001).

Effect of Vasopressin (VP) and Phorbol Esters on cVR of CLF

Distribution of CLF Fluorescence in Isolated Hepatocyte Couplets

As can be seen in Fig. 3, both VP and the phorbol esters PDB and PMA induced a dose–response decrease of the number of canaliculi retaining CLF after 15 min. This decline was apparent at 10010 and 1007 M for VP and phorbol esters, respectively, although it reached statistical significance at doses of one order of magnitude higher than these. Figure 4 shows the effect of the PKC inhibitors H-7 (1004 06 M) and staurosporine (10 M) on the VP- or phorbol ester-

Fluorescence microphotographs exhibiting retention of CLF in the canalicular lumen of control couplets and in cells treated with VP (1008 M), 17b-EG (400 mM), and A23187 (5 mM) are shown in Fig. 7. CLF accumulation into the canalicular space can be easily visualised as a brightly fluorescent vacuole between adjacent cells. A clear diminution in the number of canaliculi containing CLF can be observed following exposure to all the three tested compounds.

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

ftoxa

75

CANALICULAR RETENTION AND PERMEABILITY IN COUPLETS

TABLE 1 Comparison between CLF Retention and Rapid HRP Access into the Canalicular Vacuole to Evaluate Changes in Tight Junctional Permeability in Hepatocyte Couplets as Modified by Different Treatments Canaliculi retaining CLF

Canaliculi accumulating HRP

% of total

% of change

{ { { {

% of total

% of change

{ { { {

Control (DMSO, 10 ml, 30 sec) VP (1008 M, 30 sec) PDB (1 mM, 30 sec) PMA (1 mM, 30 sec)

78.2 50.2 54.0 53.2

1.2 3.7* 1.9* 5.0*

— 036 { 2 031 { 2 032 { 3

12.4 22.3 19.2 21.5

1.9 3.0* 2.3* 3.2*

— /80 { 10 /55 { 7 /73 { 10

Control (DMSO, 10 ml, 15 min) CyA (1 mM, 15 min) Menadione (100 mM, 15 min)

76.3 { 4.5 35.3 { 3.7* 26.5 { 2.2*

— 054 { 2 065 { 2

13.7 { 1.5 33.6 { 2.9* 30.8 { 3.7*

— /145 { 12 /125 { 13

Control (DMSO, 10 ml, 30 min) 17b-EG (400 mM, 30 min) TLC (5 mM, 30 min) A23187 (5 mM, 30 min) t-BuHP (350 mM, 30 min)

74.7 26.3 20.7 12.5 12.3

— 065 { 073 { 083 { 085 {

14.4 32.3 40.2 50.8 46.9

{ { { { {

6.0 2.8* 2.5* 0.9* 1.7*

2 1 3 2

{ { { { {

1.7 3.2* 3.4* 2.7* 1.9*

— /124 { /179 { /257 { /230 {

15 16 14 10

Note. Hepatocyte couplets were exposed to the compounds for 30 sec (VP, PDB, and PMA), 15 min (CyA and menadione), or 30 min (17b-EG, TLC, A23187, and t-BuHP). Percentage of change for each compound was calculated on the base of its respective control. Data are expressed as means { SE (n Å 4–5). CLF, cholyl-lysyl-fluorescein; CyA, cyclosporin A; DMSO, dimethyl sulfoxide; 17b-EG, estradiol 17b-glucuronide; HRP, horseradish peroxidase; PDB, phorbol, 12,13-dibutyrate; PMA, phorbol 12-myristate 13-acetate; t-BuHP, t-butyl hydroperoxide; TLC, taurolithocholate; VP, vasopressin. * Significantly different from the respective control (p õ 0.05).

The effect of five representative compounds comprising two ‘‘hormonal’’ modulators (VP and PDB) and three hepatotoxicants (17b-EG, TLC, and A23187) on CLF distribution was quantitatively examined in hepatocyte couplets by image analysis. Results are summarized in Table 2. Image analysis revealed that both canalicular fluorescence compared to total couplet fluorescence (as a quantitative estimation of CLF retention in the canalicular space) and cellular fluorescence per unit of total couplet area (as a measure of CLF content in the cellular body) were decreased by all five compounds. Furthermore, the decrease in cellular fluorescence appeared to correlate directly with the decrease in canalicular fluorescence (r Å 0.992, p õ 0.001). In turn, when cVR of CLF was compared with the decrease in cellular fluorescence, a good correlation was also obtained (r Å 0.997, p õ 0.001). DISCUSSION

The need to study hepatocyte function in a rapid and reproducible manner, avoiding the use of a large number of animals, has encouraged the development of alternatives in vitro that mimic hepatocyte function in vivo. Among these, the hepatocyte couplet model has a preferential place since this preparation retains the structural and functional polarity of the hepatic epithelium, making this system particularly useful for studies in which preservation of these features is

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

essential. The investigation of functional modulation of tight junctions and their impairment following toxicological insult is one such application. The present work proposes a simple, fast method to assess tight junctional permeability in hepatocyte couplets. This procedure takes advantage of the modified technique of couplet preparation and further purification by centrifugal elutriation described by our laboratory (Wilton et al., 1991), which enables an examination of the behaviour of a large, representative number of canalicular vacuoles rather than of preselected, individual units. The method involves the recording of the number of canalicular vacuoles present in a field of vision that are able to retain the previously accumulated fluorescent bile acid analogue CLF, which is used here as an indicator of canalicular accumulation. Our approach is based on the widely held view that cholephilic compounds require integrity of the structures that seal the canaliculus, to be retained in this space. Therefore, any perturbation involving the opening of tight junctions would cause preaccumulated materials to escape from the canalicular lumen. Furthermore, taking into account that CLF is negatively charged and recognizing the reported permselectivity of the bile canaliculus (Bradley and Herz, 1978), this method would also offer an opportunity to examine changes in permselectivity not associated with changes in permeability. Any such perturbations would be ultimately visualized as a reduction in the number of canalicular vacuoles retaining the fluorescent

ftoxa

76

ROMA, ORSLER, AND COLEMAN

The validity of the method to evaluate tight junctional permeability was tested (i) by making a systematic documentation of the effect on CLF retention of a variety of hormonal messengers and hepatotoxic agents known to affect biliary permeability in intact liver; and (ii) by carrying out a com-

FIG. 3. Dose–response curves of the effect of VP, PDB, or PMA on canalicular vacuolar retention of previously accumulated CLF in isolated rat hepatocyte couplets. The cells were exposed to the hormonal modulators (or vehicle alone in controls) for 30 sec. Canalicular CLF retention was expressed as the percentage of total couplets retaining CLF referred to the control values. All values are means { SE (n Å 4–5). *Significantly different from control values (p õ 0.05). CLF, cholyl-lysyl-fluorescein; PDB, phorbol 12,13-dibutyrate; PMA, phorbol 12-myristate 13-acetate; VP, vasopressin.

bile acid analogue, compared with an initial high proportion (ú70%) of couplets being able to accumulate the marker in control conditions. This parameter appears to provide similar information to other, more quantitative, but also more laborious, alternatives using image analysis. As indicated by the good correlation seen between the number of canalicular vacuoles retaining CLF and the percentage of canalicular fluorescence expressed as a proportion of total couplet fluorescence, no further effort to quantify intensity of canalicular fluorescence seems necessary to improve our methodology.

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

FIG. 4. Effect of the PKC inhibitors H-7 (1004 M) and staurosporine (1006 M) on VP (1009 M), PDB (1006 M), or PMA (1006 M) induced decrease of canalicular vacuolar retention of previously accumulated CLF in isolated rat hepatocyte couplets. Cells incubated with VP, PDB, or PMA were exposed to these agents (or vehicle alone in controls) for 30 sec and cells pretreated with H-7 or staurosporine received the inhibitor for 1 min prior to stimulation. Canalicular CLF retention was expressed as the percentage of total couplets retaining CLF referred to the control values. All values are means { SE (n Å 4–5). *Significantly different from control value (p õ 0.05). CLF, cholyl-lysyl-fluorescein; PDB, phorbol 12,13-dibutyrate; PMA, phorbol 12-myristate 13-acetate; VP, vasopressin.

ftoxa

CANALICULAR RETENTION AND PERMEABILITY IN COUPLETS

77

FIG. 5. Dose–response curves of the effect of different cholestatic or hepatotoxic agents on canalicular vacuolar retention of previously accumulated CLF in isolated rat hepatocyte couplets. Canalicular CLF retention was assessed following 15 min (CyA, menadione) or 30 min (17b-EG, TLC, A23187, t-BuHP) of exposure to the toxicants (or vehicle in controls) and expressed as the percentage of total couplets retaining CLF referred to control value. All values are means { SE (n Å 4–6). *Significantly different from control value (p õ 0.05). CLF, cholyl-lysyl-fluorescein; CyA, cyclosporin A; 17bEG, estradiol 17b-glucuronide; TLC, taurolithocholate; t-BuHP, t-butyl hydroperoxide.

parative analysis of the results obtained with those recorded using rapid canalicular access of HRP as an alternative approach. Hormonal regulation of hepatic tight junctional permeability is a well-documented phenomenon which has been con-

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

firmed by experiments in isolated perfused rat liver (Lowe et al., 1988) and in hepatocyte couplets (Nathanson et al., 1992). These studies pointed to the involvement of PKC in this pathway, as indicated by the actions of the PKC activators VP and PDB. We took advantage of this observation to

ftoxa

78

ROMA, ORSLER, AND COLEMAN

FIG. 6. Correlation between the percentage of decrease of cholyl-lysylfluorescein (CLF) canalicular retention and the percentage of increase of rapid horseradish peroxidase (HRP) canalicular access in the experimental conditions detailed in Table 1. Correlation coefficient (r Å 0.934) was significantly different from zero (p õ 0.001).

analyse the suitability of cVR of CLF to visualize the changes in canalicular permeability induced by these compounds, adding PMA as a third PKC activator. This experimental test is somewhat of a ‘‘tour de force’’ for the system, since the effect induced by these hormonal modulators is extremely fast (õ 30 sec), and this time period could be regarded as too short to permit the diffusion of canalicular materials from the canalicular vacuole. However, a significant, dose-dependent, reduction of cVR of CLF was observed following this short-term exposure (see Fig. 3), and such an effect was completely prevented by previous exposure of the cells to the PKC inhibitors H-7 and staurosporine (see Fig. 4). These results confirm and extend previous observations pointing to a role of PKC on hepatic tight junctional permeability and, more importantly for our purpose, validate cVR of CLF as a suitable method to detect rapid, modulatory changes of this function. Changes in biliary permeability are a frequent feature in hepatotoxicity and cholestasis. We analyzed the effect on cVR of CLF of a variety of toxicants which were shown to affect biliary permeability in intact liver by different mechanisms. They comprise the cholestatic agents CyA (Lora et al., 1991), 17b-EG (Kan et al., 1989), and TLC (Boyer et al., 1976; Vu et al., 1992), as well as the calcium ionophore A23187 (Kan and Coleman, 1988) and the oxidative stress inducers menadione (Kan and Coleman, 1990) and t-BuHP (Ballatori and Truong, 1989). These compounds all decreased cVR of CLF in a dose-dependent manner and at concentrations at or below those that had been instrumental in impairing biliary permeability in the studies quoted above. Instead, TDHC, a nontoxic bile salt which had been shown not to affect biliary permeability in isolated perfused rat liver

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

(Reichen and Le, 1983), induced no effect at concentrations showing impairment of cVR of CLF when TLC was the bile salt administered (results not shown). These results emphasize the suitability of the method to selectively detect changes of paracellular permeability following hepatotoxic insult. It should be pointed out, however, that this methodology requires hepatocellular function to be unaffected in order to permit normal accumulation of CLF. Therefore, this approach cannot be used to evaluate the effect of in vivo pretreatments resulting in altered hepatocyte transport function. In this case, alternative methodologies assessing entrance of poorly permeable solutes like HRP need to be used. Our methodology, instead, is particularly useful in complementing studies of canalicular function in which acute effects of toxicants on canalicular vacuole accumulation of CLF are evaluated (see Coleman et al., 1995). In fact, impairment of the couplet’s ability to accumulate CLF may involve alteration of a number of processes, namely, uptake, intracellular transport, canalicular excretion, and intracanalicular retention. This last possibility can be investigated by analyzing changes in cVR of CLF. It is important to note that, in this case, control experiments in which couplets are exposed to the vehicle for a similar period should be simultaneously run, since time-dependent changes in this parameter occur spontaneously (see Fig. 2). The pattern of cVR reduction induced by these compounds showed an exponential profile, the rate of decrease being lower as the dose increases, finally approaching an apparent asymptotic value (see Fig. 5). This kind of nonlinear pattern suggests heterogeneity in the susceptibility of the cell population to the hepatotoxicants, some couplets being fully resistant to the effect of some hepatotoxicants, even at high doses. Couplet preparations are mixed populations of periportal, mid-zone, and perivenous hepatocytes, which do not necessarily have similar sensitivities to modulatory or toxicological perturbations. In line with this view, menadione, which, together with CyA and 17b-EG, appears to be able to affect only a limited proportion of couplets (see Fig. 5), has been shown to affect differentially periportal from perivenous couplets, as judged by their ability to impair canalicular CLF accumulation (Wilton et al., 1993a). Extension of this study to the analysis of cVR of CLF for this and the other toxicants in subpopulations of periportal and perivenous couplets obtained by centrifugal elutriation (Wilton et al., 1993a) should contribute to a validation of this observation. A previous study by Nathanson et al. (1992) has evaluated paracellular permeability in couplets by determining the rapid access (approx 1 min) into the canalicular space of the exogenously added protein HRP. This method relies on the fact that in intact liver, HRP was shown to pass mostly through the paracellular route following a short-term (õ5 min) exposure (Lowe et al., 1985). HRP is a protein of 40 kDa and, therefore, severe restrictions to its passage are

ftoxa

CANALICULAR RETENTION AND PERMEABILITY IN COUPLETS

79

FIG. 7. Fluorescent microphotographs showing control isolated rat hepatocyte couplets which were allow to accumulate CLF (2 mM, final concentration in the dishes) for 15 min (A) and couplets exposed to VP (0.01 mM, 30 sec) (B), 17b-EG (400 mM, 30 min) (C), or A23187 (5 mM, 30 min) (D). Preaccumulated CLF is visualised as a brightly fluorescent vacuole between adjacent cells. Note the reduction in the number of canaliculi retaining CLF following exposure to the above mentioned agents. CLF, cholyl-lysyl-fluorescein; 17b-EG, estradiol 17b-glucuronido; VP, vasopressin.

expected to occur during its movement through the tight junctional region. This explains the relatively low (õ15%) percentage of canalicular vacuoles exhibiting HRP reaction product under control conditions (see Nathanson et al., 1992, and Table 1). We took advantage of this alternative approach to validate the results obtained using cVR of CLF. The good correlation between the results obtained using both methods (see Fig. 6) further supports the credibility of our method.

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

Even when all the evidence points to a preferential egress of CLF from the bile canaliculus through the paracellular pathway following tight junction disruption, the possibility also exists that the accumulated bile acid analogue leaves the canalicular lumen through a transcellular route, by reentering one or both of the hepatocytes. However, this is unlikely to contribute significantly to the observed effect because, if this happens, it would have been expected to cause

ftoxa

80

ROMA, ORSLER, AND COLEMAN

TABLE 2 Quantitative Assessment of CLF Distribution in Hepatocyte Couplets as Modified by Different Treatments Canaliculi retaining CLF

Canalicular fluorescence

Cellular fluorescence

% of total

% of control

% of total couplet fluoresce

% of control

Cellular fluorescence per unit of cellular area (arbitrary units)

% of control

Control (DMSO, 10 ml, 30 sec) VP (1008 M, 30 sec) PDB (1 mM, 30 sec)

77 47 55

100 61 71

41.1 { 3.0 27.4 { 3.8* 32.9 { 3.1*

100 67 80

11.6 { 0.7 10.6 { 0.5 10.9 { 0.8

100 91 94

Control (DMSO, 10 ml, 30 min) 17b-EG (400 mM, 30 min) TLC (5 mM, 30 min) A23187 (5 mM, 30 min)

75 28 20 13

100 37 27 17

40.4 17.7 13.3 9.3

{ { { {

100 45 33 23

11.3 9.5 9.1 8.6

{ { { {

100 84 81 76

2.7 4.1* 2.1* 3.7*

0.6 0.8* 1.1* 1.0*

Note. Hepatocyte couplets were exposed to the compounds for 30 sec (VP, PDB), or 30 min (17b-EG, TLC, A23187). Then, microphotographs of randomly selected fields were taken and the number of canaliculi retaining CLF counted. Every fluorescent photomicrograph was then digitalized to obtain a computer image on which canalicular fluorescence (as a percentage of total couplet fluorescence) and cellular fluorescence per unit of couplet area could be quantified with the aid of an image analysis program as indicated under Methods. Number of couplets analyzed was ú30. Data are expressed as means { SE. CLF, cholyl-lysyl-fluorescein; DMSO, dimethyl sulfoxide; 17b-EG, estradiol 17b-glucuronide; PDB, phorbol 12,13-dibutyrate; TLC, taurolithocholate; VP, vasopressin. * Significantly different from the respective control (p õ 0.05).

an increase (or no change, if canalicular ingress is simultaneously counterbalanced by basolateral egress) rather than a decrease in cellular fluorescence following exposure to the tested substances. The reported decrease in cellular fluorescence is compatible rather with increased secretion of CLF into the canalicular space driven by the continuous loss of CLF from the space via the paracellular pathway. In conclusion, assessment of cVR of CLF provides a means to evaluate, rapidly, acute changes in tight junctional permeability induced by hepatotoxic agents or hormonal modulators in a reproducible and sensitive manner, without requiring laborious microscopic techniques involving fixation and staining. Its application to the analysis of tight junction performance may aid in our understanding of the role of this structure in hepatocellular function, its modulation by hormonal agents, and its impairment following toxicological insult. ACKNOWLEDGMENTS M. G. Roma, from the Institute of Physiology, University of RosarioCONICET, Argentina, was a recipient of a postdoctoral fellowship of The Royal Society-CONICET Interchange Program. D. J. Orsler is in receipt of an MRC collaborative Ph.D. studentship with Astra Charnwood. Laboratory expenses were supported by The Wellcome Trust, The Royal Society, and Astra Charnwood. The authors thank Mr. Jose´ Manuel Pellegrino for his assistance in image analysis and graphic design.

REFERENCES Ballatori, N., and Truong, A. T. (1989). Altered hepatic junctional permeability, bile acid excretion and glutathione efflux during oxidative challenge. J. Pharmacol. Exp. Ther. 251, 1069–1075.

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

Boyer, J. L., Layden, T. J., and Hruban, Z. (1976). Mechanism of cholestasis: Taurolithocholate alters canalicular membrane composition, structure and permeability. In Membrane Alterations as Basis of Liver Injury (H. Popper, L. Bianchi, and W. Reutter, Eds.), pp. 353–369. MTP Press, Lancaster. Boyer, J. L., Ng, O.-C., and Gautam, A. (1985). Formation of canalicular spaces in isolated rat hepatocyte couplets. Trans. Assoc. Am. Phys. 98, 21–29. Boyer, J. L. (1993). Isolated rat hepatocyte couplets: A model for the study of bile secretory function. In Hepatic Transport and Bile secretion (N. Tavoloni and P. D. Berk, Eds.), pp. 597–606. Raven Press, New York. Bradley, S. E., and Herz, R. (1978). Permselectivity of biliary canalicular membrane in rats: Clearance probe analysis. Am. J. Physiol. 254, E570– E576. Coleman, R., Wilton, J. C., Stone, V., and Chipman, J. K. (1995). Hepatobiliary function and toxicity in vitro using isolated hepatocyte couplets. Gen. Pharmacol. 26, 1445–1453. Crawford, J. M., Vinter, D. W., and Gollan, J. L. (1988). Taurocholate induces pericanalicular localization of C6-NBD-ceramide in isolated hepatocyte couplets. Am. J. Physiol. 260, G119–G132. Fentem, J. H., Mills, C. O., Coleman, R., and Chipman, J. K. (1990). Biliary excretion of fluorescent cholephiles in hepatocyte couplets: An in vitro model for hepatobiliary and hepatotoxicity studies. Toxicol. in Vitro 4, 452–457. Gautam, A., Ng, O.-C., and Boyer, J. L. (1987). Isolated hepatocyte couplets in short culture: Structural characteristics and plasma membrane reorganization. Hepatology 7, 216–223. Gautam, A., Ng, O.-C., Strazzabosco, M., and Boyer, J. L. (1989). Quantitative assessment of canalicular bile formation in isolated hepatocyte couplets using microscopic optical planimetry. J. Clin. Invest. 83, 565–573. Graf, J., Gautam, A., and Boyer, J. L. (1984). Isolated rat hepatocyte couplets: A primary unit for electrophysiological studies of bile secretory function. Proc. Natl. Acad. Sci. USA 81, 6516–6520. Graf, J., Henderson, R. M., Krumpholz, B., and Boyer, J. L. (1987). Cell

ftoxa

CANALICULAR RETENTION AND PERMEABILITY IN COUPLETS membrane and transepithelial voltages and resistances in isolated rat hepatocyte couplets. J. Membr. Biol. 95, 241–254. Kan, K. S., and Coleman, R. (1988). The calcium ionophore A23187 increases the tight-junctional permeability in rat liver. Biochem. J. 256, 1039–1041. Kan, K. S., Monte, M., Parslow, R. A., and Coleman, R. (1989). Oestradiol 17b-glucuronide increases tight junctional permeability in rat liver. Biochem. J. 261, 297–300. Kan, K. S., and Coleman, R. (1990). Menadione increases hepatic tight junctional permeability. Its effect can be decreased by butylated hydroxytoluene and verapamil. Biochem. J. 270, 241–243. Kitamura, T., Gatmaitan, Z., and Arias, I. M. (1990). Serial quantitative analysis and confocal microscopy of hepatic uptake, intracellular distribution and biliary secretion of a fluorescent bile acid analog in rat hepatocyte doublets. Hepatology 12, 1358–1364. Lora, L., Martines, D., Plebani, M., Floreani, A. R., Salugnini, M., and Naccarato, R. (1991). Cyclosporin A increases tight junctional permeability in rat liver. J. Hepatol. 13, S45. [Abstract] Lowe, P. J., Kan, K. S., Barnwell, S. G., Sharma, R. K., and Coleman, R. (1985). Transcytosis and paracellular movements of horseradish peroxidase across liver parenchymal tissue from blood to bile. Effect of anaphthylisothiocyanate and colchicine. Biochem. J. 229, 529–537. Lowe, P. J., Miyai, K., Steinbach, J. H., and Hardison, W. G. M. (1988). Hormonal regulation of hepatocyte tight junctional permeability. Am. J. Physiol. 255, G454–G461. Maglova, L. M., Jackson, A. M., Meng, X.-J., Carruth, M. W., Schteingart, C. D., Ton-Nu, H.-T., Hoffmann, A. F., and Weinman, S. A. (1995). Transport characteristics of three fluorescent conjugated bile acid analogs in isolated rat hepatocyte couplets. Hepatology 22, 637–647. Nathanson, M. H., Gautam, A., Ng, O. C., Bruck, R., and Boyer, J. L. (1992). Hormonal regulation of paracellular permeability in isolated rat hepatocyte couplets. Am. J. Physiol. 262, G1079–G1086. Oshio, C., and Phillips, M. J. (1981). Contractility of bile canaliculi: implications for liver function. Science 212, 1041–1042. Reichen, J., and Le, M. (1983). Taurocholate, but not taurodehydrocholate, increases biliary permeability to sucrose. Am. J. Physiol. 245, G651– G655. Reverdin, E. C., and Weingart, R. (1988). Electrical properties of the gap junctional membrane studied in rat liver cell pair. Am. J. Physiol. 254, C226–C234. Roelofsen, H., Bakker, C. T. M., Schoemaker, B., Heijn, M., Jansen, P. L. M., and Oude Elferink, R. P. J. (1995). Redistribution of canalicular organic anion transport activity in isolated and cultured rat hepatocytes. Hepatology 21, 1649–1657.

AID

FAAT 2309

/

6k1c$$$101

05-16-97 03:55:37

81

Roma´n, I. D., and Coleman, R. (1994). Disruption of canalicular function in isolated rat hepatocyte couplets caused by cyclosporin A. Biochem. Pharmacol. 48, 2181–2188. Roma´n, I. D., Johnson, G. D., and Coleman, R. (1996). S-Adenosyl-L-methionine prevents disruption of canalicular function and pericanalicular cytoskeleton integrity caused by cyclosporin A in isolated rat hepatocyte couplets. Hepatology 24, 134–140. Saez, J. C., Gregory, W. A., Watanabe, T., Dermietzel, R., Hertzberg, E., Bennett, M. V., and Spray, D. C. (1989). cAMP delays disappearance of gap junctions between pairs of rat hepatocytes in primary culture. Am. J. Physiol. 236, C1–C11. Snedecor, G. W., and Cochran, W. G. (1969). Statistical Methods. The Iowa State Univ. Press, Ames, IA. Stone, V., Johnson, G. D., Wilton, J. C., Coleman, R., and Chipman, J. K. (1994). Effect of oxidative stress and disruption of Ca2/ homeostasis on hepatocyte canalicular function in vitro. Biochem. Pharmacol. 47, 625– 632. Thibault, N., Claude, J. R., and Ballet, F (1992). Actin filament alteration as a potential marker for cholestasis: a study in isolated rat hepatocyte couplets. Toxicology 73, 269–279. Vu, D. D., Tuchweber, B., Raymond, P., and Yousef, I. M. (1992). Tight junction permeability and plasma membrane fluidity in lithocholate-induced cholestasis. Exp. Mol. Pathol. 57, 47–61. Watanabe, S., and Phillips, M. J. (1984). Calcium causes active contraction of bile canaliculi: Direct evidence from microinjection studies. Proc. Natl. Acad. Sci. USA 81, 6164–6168. Watanabe, S., Tomono, M., Hirose, M., Takeuchi, M., Kitamura, T., and Nishima, T. (1987). Microscopic fluorometric analysis of Na-fluorescein transport in the smallest secretory unit (couplet hepatocytes) and the effect of the cytochalasin D. Lab. Invest. 56, 146–150. Wilton, J. C., Williams, D. E., Strain, A. J., Parslow, R. A., Chipman, J. K., and Coleman, R. (1991). Purification of hepatocyte couplets by centrifugal elutriation. Hepatology 14, 180–183. Wilton, J. C., Chipman, J. K., Lawson, C. J., Strain, A. J., and Coleman, R. (1993a). Periportal- and perivenous-enriched hepatocyte couplets: Differences in canalicular activity and in response to oxidative stress. Biochem. J. 292, 773–779. Wilton, J. C., Coleman, R., Lankester, D. J., and Chipman, J. K. (1993b). Stability and optimization of canalicular function in hepatocyte couplets. Cell Biochem. Funct. 11, 179–185. Wilton, J. C., Matthews, G. M., Burgoyne, R. D., Mills, C. O., Chipman, J. K., and Coleman, R. (1994). Fluorescent choleretic and cholestatic bile salts take different paths across the hepatocyte: Transcytosis of glycolithocholate leads to an extensive redistribution of annexin II. J. Cell Biol. 127, 401–410.

ftoxa