Polar Secretion of Endothelin-1 by Cultured Endothelial Cells*

The aim of this study was to determine the permea- bility of endothelial monolayers for endothelin-1 and a possible directionality of the endothelin-1 secretion process. Human umbilical vein endothelial cells were cultured on acellular amniotic membranes, dividing the tissue culture wells into an apical (luminal) and a basolateral (abluminal) compartment. Whereas in the absence of endothelial monolayers 44.9 f 2.3 and 43.5 f 2.0% of the unilaterally added endothelin-1 per- meated from the apical to the basolateral side and from the basolateral to the apical side, respectively, only 6.5 f 0.6 and 6.6 2 0.4% diffused in the presence of endo- thelial cells. Analyzing endothelin-1 secretion, -80% of the total amount of synthesized endothelin-1 was found in the basolateral compartment; thrombin (10 units/ml) stimulated the production of endothelin- 1 -2-fold, but did not change the relative distribution of endothelin-1 between the apical and basolateral com- partments. In the presence of dexamethasone (lo" M), a decrease in the level of endothelin-1 was found in the apical compartment, whereas the total

a decrease in the level of endothelin-1 was found in the apical compartment, whereas the total amount of endothelin-1 produced was not affected. Dexamethasone did not influence the permeability of human umbilical vein endothelial cell monolayers for endothelin-1. These results strongly support the hypothesis that endothelin-1 is a local paracrine regulator of vasotone.
Endothelin-1 is a potent vasoconstrictor peptide isolated originally from the supernatant of porcine aortic endothelial cells (1). Comparing the sites of endothelin-1 messenger RNA production localized by in situ hybridization and exogenous '251-endothelin-l binding in rat tissues, it was reported that synthesis of messenger RNA occurs in close proximity to the binding sites of endothelin-1 in many organs, e.g. lung, kidney, intestine, and eye (2). This suggests that endothelin-1 acts locally in a paracrine fashion mediated by endothelin-1 receptors expressed closely to the sites of endothelin-1 production in these organs. However, in other organs, e.g. heart and renal cortex, endothelin-1-binding sites are present in the absence of endothelin-1 mRNA, indicating a humoral mode of endothelin-1 action (2). This key question concerning the role of endothelin-1 in cardiovascular control (paracrine uersw endocrine mode of action) nevertheless is still in debate. A predominantly paracrine mode of action of endothelin-1 would imply a polarized secretion pattern; in fact, in 1989, Masaki (3) speculated that endothelial cells may release endothelin-1 in a polar fashion directed toward the underlying * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. intimal smooth muscle. This hypothesis, however, has so far not been tested. The aim of this investigation therefore was to determine whether endothelin-1 produced by cultured endothelial cells is released preferentially toward one side. We found that HUVECs' grown to confluent monolayers on acellular amniotic membranes release endothelin-1 predominantly into the basolateral compartment.

EXPERIMENTAL PROCEDURES
Cell Culture-HUVECs were isolated essentially as described previously (4). The endothelial origin of the cultured cells was confirmed by typical cobblestone morphology (51, immunofluorescence staining with anti-von Willebrand factor antibodies (6), and uptake of acetylated low density lipoprotein (7). HUVECs from passage 2 were seeded at a high density onto acellular amniotic membranes prepared as described previously (8). Pieces of human amniotic tissue were fastened to Teflon rings, and the amniotic epithelium was lysed by incubation in 0.25 M NH,OH for 2 h. Teflon rings holding the membranes were placed into tissue culture wells. HUVECs were plated on the stromal surface of the prepared amniotic tissue and were grown to confluent monolayers. HUVECs reached confluency -2-3 days after seeding. The endothelial monolayer grown on the amniotic membranes thus divided the culture wells into two compartments, an upper one (1 ml) on the apical side and a lower one (2 ml) on the basolateral side, thereby imitating in uivo conditions with a luminal and an abluminal side.
Analysis of Permeability of HUVEC Monolayers for Endothelin-l-For analysis of permeability of endothelial cell monolayers for endothelin-1, confluent monolayers of HUVECs grown to confluency on acellular amniotic membranes were rinsed twice with Hanks' balanced salt solution. Thereafter, the cells were incubated in Medium 199 with Earle's salts to which '251-endothelin-l (1145 Ci/mmol, Peninsula Laboratories, Inc.) was added on the apical or the basolat-era1 side at a final concentration of 0.1 nM. Cells were maintained up to 4 h at 37 or 4 "C. To investigate the possible effect of thrombin or dexamethasone on the permeability of HUVEC monolayers for endothelin-1, cells were incubated in the medium described above containing human thrombin (10 units/ml, Sigma) or dexamethasone (lo" M, CL Pharma) at 37 "C. After 1, 2, 3, and 4 h, aliquots (30 wl) of either compartment were collected, and the activity was counted on a y-counter (Packard Instrument Co.). Permeability of amniotic membranes without endothelial cells for endothelin-1 was analyzed in an analogous fashion.

Analysis of Production and Polarity of Secretion of Endothelin-I by HUVEC Monolayers Grown on Acellular Amniotic Membranes-
HUVECs grown to confluency on amniotic membranes were rinsed twice with Hanks' balanced salt solution. Thereafter, cells were incubated in Medium 199 with Earle's salts and in medium containing human thrombin (10 units/ml) or dexamethasone (10" M). After 1, 2,3, and 4 h, the conditioned media from either side of the membrane were collected separately, put on ice immediately, centrifuged at 300 X g at 4 "C, and stored at -80 "C until further use.
cartridges (Analytichem International, Inc.). The eluted samples were then applied to a radioimmunoassay based on rabbit anti-endothelin-1 serum, a synthetic endothelin-1 standard, and a '251-endothelin-l tracer (Peninsula Laboratories, Inc.). The sensitivity of the assay was 0.2 fmol at 90% binding, and the intra-and interassay coefficients of variation were <8%.
Statistical Eualuation-Dunnet's test was used for statistical evaluation (11). Values are presented as mean t S.D.

RESULTS
Permeability of HUVEC Monolayers for Endothelin-l-Monolayers of HUVECs grown on amniotic tissue for 4-5 days prior to the experiments were used to examine the diffusion of endothelin-1. '251-Endothelin-l was added to the upper or the lower chamber at a concentration of 0.1 nM (1145 Ci/mmol) at 37 "C, and the amount of '251-endothelin-1 was measured in the lower or the upper chamber after 1, 2, 3, and 4 h (Fig. 1, A and B ) . The diffusion of '251-endothelin-1 was linear over the whole period of time, and there was no difference in regard to the diffusion from the upper to the lower and from the lower to the upper chamber. Thus, within the first hour of incubation, 6.5 f 0.6% of the added lz5Iendothelin-1 diffused from the upper to the lower compartment as compared to 6.6 f 0.4% in the reverse direction. In the absence of endothelial cells, 44.9 f 2.3% of the total amount of '251-endothelin-l applied diffused from the apical to the basolateral side as compared to 43.5 f 2.0% from the basolateral to the apical side within the first hour. Thus, almost complete equilibrium of concentrations was achieved within 2 h. The addition of cold endothelin-1 (1 p~) did not influence the diffusion of '251-endothelin-l (6.7 f 1.0% transfer from the apical to the basolateral side and 6.3 f 0.7% from the basolateral to the apical side within the first hour of incubation). Neither thrombin (10 units/ml) nor dexamethasone M ) had any effect (Fig. 1, A and B ) . Recovery of '"I-endothelin-1 in either experiment was >95%.
Polarity of Secretion of Endothelin-1 by HUVEC Monolayers-To investigate whether endothelin-1 secretion proceeds in a polar fashion, HUVECs were grown to confluent monolayers on amniotic membranes. After 1, 2, 3, and 4 h of incubation, the conditioned media from the upper and the lower compartments were collected separately. Under basal conditions, a total amount of -5.2 fmol of endothelin-1 was produced by lo5 cells within 1 h. The increase in the total amount of endothelin-1 over time was linear for the whole observation period of 4 h. Approximately 80% of the overall amount of synthesized endothelin-1 was secreted into the lower (basolateral) compartment. The relative distribution of endothelin-1 between the upper and the lower compartments did not change over the observation time (Fig. 2).
When thrombin (10 units/ml) was added to the culture medium, an exponential increase in the total amount of endothelin-1 produced over time was observed (Fig. 2). After an incubation period with thrombin of 4 h, an -2-fold increase in the level of endothelin-1 was found. The relative distribution of endothelin-1 between the upper and the lower compartments did not change over time and was not different from basal conditions. Again, -80% of the synthesized endothelin-1 was found in the basolateral (abluminal) compartment at each time point (Fig. 2). M), the total amount of endothelin-1 produced was equal to control conditions at every time point (Fig. 2). In the presence of dexamethasone, only 5.6 & 1.3 fmol of endothelin was found in the upper compartment after 4 h of incubation as compared to 10.4 -t 2.1 fmol in the absence of dexamethasone ( p < 0.05). In the basal compartment, 46.8 f 2.8 and 41.7 f 5.4 fmol of endothelin-1 were found in the presence and absence of dexamethasone, respectively. As described above, dexamethasone had no effect on the diffusion of endothelin-1.

DISCUSSION
Since the first description of endothelin-1 (l), considerable attention has been focused on plasma concentrations of endothelin-1 in both healthy and diseased humans as well as on the pharmacological effects of the peptide. It was reported that endothelin-1 circulates in human plasma (9, 12) and is found to be increased in plasma under a variety of pathological circumstances including essential hypertension (13), uremia (14), and cardiogenic shock (15). Endothelin-1 exerts a vasoconstrictor potential in uiuo when applied systemically in several species (1) including humans (10). However, because endothelin-1 may act locally on the underlying smooth muscle rather than as a circulating hormone and plasma endothelin-1 may represent merely the overflow of a local paracrine peptide system, the pathophysiological importance of these Polarity of secretion of endothelin-1 by HUVEC monolayers. HUVECs were grown to confluency on prepared acellular amniotic membranes, thus dividing the tissue culture well into an apical (top) and a basolateral (bottom) compartment. The cells were incubated in Medium 199 with Earle's salts (open bars) and in medium containing human thrombin (10 units/ml) (striped bars) or dexamethasone M) (dotted bars). Conditioned media from either compartment were collected separately, and the content of endothelin-1 was determined. * , p < 0.05 (comparison of endothelin-1 content in control media with endotbelin-1 content in either thrombinor dexamethasone-supplemented media collected and analyzed separately from apical and basal compartments); 0, p < 0.05 (comparison of overall (apical plus basal) endothelin-l content in control media with overall endothelin-1 content in thrombinor dexamethasonesupplemented media). observations remains to be established.
The key question concerning the role of endothelin-1 in cardiovascular control (local paracrine uersus humoral systemic action) is still in debate. In fact, a polar secretion of the peptide by endothelial cells toward the underlying intimal smooth muscle has been suggested (3). Endothelial cells are polarized, exhibiting an apical and a basolateral side. This polarized state is preserved also under cell culture conditions. It has been reported that cultured HUVECs stimulated with tumor necrosis factor secrete urokinase-type plasminogen activator predominantly to the basolateral side (16). The extracellular matrix proteins collagen and fibronectin as well as several matrix-degrading metalloproteases are secreted predominantly in the basal direction by bovine aortic endothelial cells (17). Furthermore, insulin as well as angiotensin I1 are transported unidirectionally through a monolayer of cultured endothelial cells from the luminal to the extraluminal side (18,19). The finding of a differential vascular sensitivity to luminally and adventitially applied endothelin-1 (20) strongly supports the hypothesis of a local paracrine role of endothelin- 1. In these experiments, rabbit femoral veins and arteries in which the luminal and adventitial surfaces could be perfused separately were used. Luminally and adventitially applied endothelin-1 induced similar vasoconstriction in segments in which the endothelium had been removed. In segments where the endothelium was intact, however, vasoconstriction was observed only when endothelin-1 was applied at the adventitial surface, whereas luminally applied endothelin-1 showed no vasomotor response, indicating that the endothelium forms a tight barrier to circulating endothelin-1. Alternatively, an increased endothelin-1-induced release of endothelium-derived relaxing factor or prostacyclin may also explain these results (21).
In this study, we investigated the hypothetical barrier function of an endothelium for exogenous endothelin-1 as well as the suggested polarity of endothelial cells concerning the secretion of endogenous endothelin-1. HUVECs cultured on amniotic membranes exhibited a significant barrier for the diffusion of endothelin-1 in either direction. This finding allowed us to study the polarity of the endothelin-1 secretion. Under basal conditions, -80% of the total amount of endothelin-1 synthesized by the endothelial cells was found on the basal (abluminal) side of the endothelial monolayer. At least half of the amount of endothelin-1 found in the apical (abluminal) compartment may originate from diffusion from the abluminal side. The significant increase in the level of endothelin-1 upon stimulation with thrombin is consistent with the data of others (1,22). Thrombin did not influence the polarity of the secretion. Dexamethasone treatment, however, which did not influence the total amount of secreted endothelin-1, led to an even more pronounced polarity.
In conclusion, an intact endothelium appears to be a major barrier for the diffusion of endothelin-1. The secretion of endothelin-1 is clearly polarized under basal as well as thrombin-stimulated conditions whereby the majority of the endothelin-1 is released into the basolateral compartment, where under in uiuo conditions the smooth muscle cells would be located. These findings are in good agreement with the hypothesis that endothelin-1 acts in a local paracrine way, leading to constriction of the underlying vascular smooth muscle located close to the site where endothelin-1 is produced.