Histamine decreases the permeability of an endothelial cell monolayer by stimulating cyclic AMP production through the H₂-receptor

Vascular permeability is regulated by several humoral factors such as histamine, bradykinin, catecholamines and interleukins. In injured patients, histamine is believed to increase vascular permeability (Matoltsy and Matoltsy, 1951; Hayashi et al. 1964). In fact, many investigators have described injury to endothelial cells when histamine is administered in vivo (Matoltsy and Matoltsy, 1951; Majno and Palade, 1961; Udaka et al. 1970; Gabbianti et al. 1970; Pietra et al. 1971; Northover, 1975; McNamee and Grodins, 1975; Kaliner et al. 1982). Histamine concentration is increased in the interstitial fluid local to an injury and in the serum in the early stages after injury (Horakova and Beaven, 1974; Sampson and Archer, 1967). However, the direct effects of histamine on the vascular wall are not established. According to studies using cultured endothelial cells, histamine causes changes in cell shape (Antonov et al. 1986), mediates cell mobility (Bottaro et al. 1985), and accelerates cell growth (D’amore and Shepro, 1977) and prostacyclin production (Baenziger et al. 1980; Baenziger et al. 1981; Alhenc-Gelas et al. 1982). However, the permeability of the endothelial cell layer has not been examined in the in vitro system. We developed an in vitro model for assay of endothelial cell monolayer permeability and examined the direct effect of histamine on the endothelial cell monolayer. Our results show that histamine decreases endothelial cell monolayer permeability to albumin while it increases intracellular cyclic AMP through the H₂-receptor. Accompanying these changes was an increase in cytoplasmic Ca²⁺ via the H ₂-receptor. However, Ca²⁺ seems not to play a major role in the permeability change induced by histamine, since the H ₁-blocker did not inhibit the decrease of permeability caused by histamine.

MCDB151 was obtained from Sigma Chemical Co. (MO, USA). Fetal bovine serum (FBS) was from Cell Culture Laboratories Inc. (OH, USA), and was heat-inactivated at 56°C for 30 min. Endothelial cell growth substance (ECGS) was purified from bovine brain and checked for its growth inducing activity (Imamura and Mitsui, 1987). Fibronectin was a generous gift from Ito Ham (Japan). Most of the other reagents were of special grades and manufactured by Wako Pure Chemical Industries, Ltd. (Japan). The co-culture chamber (Intercell) was purchased from Kurabou (Japan).

Human umbilical vein endothelial cells (HUE147) were isolated as described previously (Imamura and Mitsui, 1987), and an endothelial cell line from a calf pulmonary artery was designated as PACE2. Cells were maintained in MCDB151 supplemented with 15% FBS containing 2.5 ng/ml ECGS and 5 μg/ml heparin to maintain growth, and subcultured after treatment with 0.25% trypsin. Cells were studied between passages 10 and 20.

HUE147 and PACE2 cell lines in the growth phase were seeded at a density of l.6×10 ⁵ /cm² and 8×10 ⁵ /cm², respectively, into co-culture chambers containing 200 μl of culture medium. The Teflon membrane of the chamber was coated with bovine fibronectin at 10 μg/cm². One chamber was placed in each well of a 24-well plate containing 600 of the same culture medium. To prepare a confluent cell monolayer, cells were allowed to proliferate on the Teflon membrane for 2 days at 37°C in a humidified incubator under a 95%:5% (air: CO₂) atmosphere.

Histamine, diphenhydramine hydrochloride and famotidine were dissolved in MCDB151 + 15% FBS at various concentrations. A 600μl; sample of each assay medium containing 2 μM FITC-conjugated bovine albumin was added gently to each chamber and then incubated. After 2 hours the medium in the lower chamber was removed and the fluorescence intensity was measured with a fluorescence spectrophotometer (Shimazu, Japan) (excitation, 490 nm; emission, 525 nm).

Human umbilical vein endothelial cells (HUE147) were seeded into fibronectin-coated 6-well plates and cultured for 2 days. The cells were then exposed to 1 ml of fresh assay medium containing 10^{− 5} M histamine and/or 10^{− 6} M diphenhydramine or 10^{− 8} M famotidine. After incubation for 10 min, the medium was removed to assay for cyclic AMP. Cyclic AMP was measured in fmol/cm² per well using a cyclic AMP [¹²⁵I]RIA kit (Amersham, UK) (non-acetylation protocol).

HUE147 endothelial cells were incubated with culture medium containing 2 μM Indo-1 AM (a calcium-sensitive fluorescent probe) for 20 min. The cells were then washed twice gently with MCDB151 and exposed to assay medium containing histamine and/or diphenhydramine or famotidine. Ca²⁺-bound and free Indo-1 in each cell were excited with a 351-363 nm argon ion laser and the immunofluorescence was measured at 405 nm and 485 nm, respectively, using an ACAS470 fluorimeter (Meridian Instruments, MI, USA).

All values are expressed as the mean ± s.d. Comparisons of data in the permeability study and cyclic AMP assays were made by one-way ANOVA and Bonferoni’s method. Comparisons of data from the Ca²⁺ assay were assessed by the chi-squared test after Yates’ correction. Variations were significant at the P<0.05 level.

Both human umbilical vein and calf pulmonary artery endothelial cells formed a confluent monolayer on the 10 μg/cm² fibronectin-coated Teflon membrane of a coculture chamber (as verified by phase-contrast and scanning electron microscopy). In the confluent human umbilical vein endothelial cell monolayer the permeability to albumin was decreased to 15-25% of the control containing membrane only. The permeability of a confluent monolayer of pulmonary artery endothelial cells was decreased to 60-80% of the control. Fibronectin coating alone did not affect permeability. The permeability to albumin was measured after 2 h when FITC-albumin reaches a detectable concentration under our assay system.

Histamine at a concentration of 10⁻⁶ to 10⁻⁴ M decreased the permeability of the membrane to albumin in a culture of HUE147 cells (Fig. 1 A). The permeability of PACE2 cells to albumin was also decreased by histamine in a dose-dependent manner (Fig. 1 B). In both cases, the permeability decreased but reached a lower plateau level, and in HUE147 cells it even began to increase, at higher doses of histamine. The decrease in permeability of HUE147 cells to albumin in the presence of 10^{− 5} M histamine was blocked by famotidine (Fig. 2 A). Diphenhydramine, however, did not have any statistically significant effect even at the high concentration of 10^{− 4} M (Fig. 2 B).

Histamine also increased cyclic AMP production in a dose-dependent manner and reached a maximum at 10 min after its addition to cells (data not shown). Fig. 3 shows that 10⁻⁸ M famotidine reduced histamine-stimulated cyclic AMP production but treatment with 10^{− 4} M diphenhydramine did not affect the cyclic AMP level.

The cytoplasmic Ca²⁺ concentration of isolated endothelial cells increased rapidly with 10^{− 5} M histamine (Fig. 4 A), regardless of the presence of 10⁻⁸ M famotidine (Fig. 4 C: for experimental details see legend). In contrast, 10⁻⁴ M diphenhydramine, the concentration used in the permeability studies and assay of cyclic AMP, inhibited the oscillations in Ca²⁺ concentration after the initial rapid increase (Fig. 4 B). Neither famotidine nor diphenhydramine themselves had any effect on the cytoplasmic Ca²⁺ concentration. Pretreatment with diphenhydramine 20 min prior to adding histamine inhibited both the initial rapid increase and the latter fluctuations in cytoplasmic Ca²⁺ concentration. Inhibition of the initial rise was observed only after preincubation, although diphenhydramine is a receptor blocker. It has not been determined why diphenhydramine was not immediately effective when applied to cells in culture. Famotidine pretreatment did not alter this effect. This is illustrated in Table 1, which shows the effects of histamine and each antagonist on the frequency of the initial rapid rise and subsequent oscillations of cytoplasmic Ca²⁴ that were observed in a large number of experiments. Pretreatment with famotidine, in contrast to diphenhydramine, did not affect the histamine-induced large rise in cytoplasmic Ca²⁺ but inhibited, somewhat, subsequent Ca²⁺ oscillations.

The effects of histamine on endothelial cells are thought to be mediated by the histamine receptors on these cells (Ash and Schild, 1966; Buonassi and Venter, 1976; Berti et al. 1979; Baenziger et al. 1980; Simionescu et al. 1982; Van de Voorde and Leusen, 1983). The receptors can be subdivided into two classes, Hi and H₂, based on their physiology and pharmacology, and on the second messengers specific to each. It is reported that the second messengers of histamine via the H₁-receptor are phosphatidylinositol, diacylglycerol and calcium, and via the H₂-receptor it is cyclic AMP (Grigorian et al. 1989). In this study, we show that the histamine-induced decrease in the permeability of an endothelial cell monolayer to albumin is accompanied by an increase in the cyclic AMP concentration of endothelial cells. These effects of histamine were inhibited by an H₂-blocker (famotidine) and not by an Hi-blocker (diphenhydramine). Histamine also increases the concentration of Ca²⁺ in endothelial cells. This effect, however, was inhibited by the H₁-blocker but not by the H₂-blocker. On the basis of these observations we suggest that histamine alters the permeability of the endothelial cell monolayer via H₂-receptors and cyclic AMP and that the cytoplasmic Ca²⁺ concentration does not affect cell permeability.

Histamine is reported to stimulate endothelial cell proliferation (D’Amore and Shepro, 1977). Because the population doubling time of HUE 147 is 22 hours and our results were obtained over a two-hour period, the histamine-induced decrease in permeability cannot be due to an increase in the number of cells. Mizuno-Yagyu et al. (1987) reported that PGI₂ (prostocyclin) inhibited dextran transport through the endothelial cell monolayer and that this effect was mediated by increased cyclic AMP production. On the other hand, Baenziger et al. (1980) showed that histamine stimulates the production of PGI₂ via an H₁-receptor-mediated mechanism. Our data, however, support the belief that histamine itself decreases the permeability of the endothelial cell monolayer primarily through the H₂-receptor. This suggests that the decrease in permeability due to histamine does not require PGI₂ synthesis.

Regarding the role of cyclic AMP, it has been reported that membrane-permeable analogues of cyclic AMP, 8-bromo cyclic AMP and dibutyryl cyclic AMP decrease the permeability of the endothelial cell monolayer (Casnocha et al. 1989). Duffey et al. (1981) reported that a correlation exists between electrical resistance and the cyclic AMP content of epithelial cells. Since albumin passes through the endothelial cell layer intercellularly (junctional way) and/or transcellularly (vesicular transport) (Navab et al. 1986), cyclic AMP may increase the number of junctions between endothelial cells and narrow the intercellular space.

Synchronized cytoplasmic Ca²⁺ oscillations have been reported in confluent monolayers of human endothelial cells (Sage et al. 1989; Neylon and Irvine, 1990). Neylon and Irvine reported that the synchronized repetitive spikes in cytoplasmic calcium occur in response to histamine in confluent human umbilical vein endothelial cell monolayers, and that spiking behavior is not seen in non-confluent cell monolayers. In this study, we observed the oscillations in Ca²⁺ concentration in single cells using an ACAS470 fluorimeter. The cells were spread on a coverglass sparsely and had no contact between each other. However, the repetitive spikes in a single cell are still evoked by histamine, and these spikes are inhibited by diphenhydramine (H₁blocker). At present it is not clear whether these oscillations are generated by the same mechanism as those seen in monolayer cells.

Several studies have shown that histamine increases vascular permeability in vivo (Matoltsy and Matoltsy, 1951; Hayashi et al. 1964). We now report that histamine decreases endothelial cell monolayer permeability in vitro. Our in vitro system of an endothelial monolayer was designed to represent a simple model of capillary vessels. There always remains the question of its relevance to the situation in vivo. Albelda et al. (1988) reported that the calculated permeability of albumin across an in vitro monolayer is 10-100 times higher than that found in vivo, although they used fetal bovine aortic endothelial cells. The reasons for the differences are not known. The authors observed occasional gaps between adjacent cells (5-10%) in their in vitro model, and also found a lack of charge selectivity. The influence of the basement membrane and/or interactions between endothelial cells and other cells or humoral factors in vivo would also result in differences between the intact endothelium and in vitro models. However, we believe that the use of an in vitro model for permeability has a number of advantages, such as offering direct access to luminal and abluminal fluid for analysis and being highly simplified and limited to a single cell type. Our results suggest that the direct effect of histamine on the endothelial cell monolayer is to decrease permeability and that when histamine increases the permeability, it does so indirectly by stimulating the surroundings in vivo.

The authors thank Dr. A. Iwashima for advice on the Ca²⁺ assay.

This work was supported by a project grant for Basic Technology for Future Industry from the Ministry of International Trade and Industry of Japan.

Data in Fig. 1 A were presented at the 18th Annual Meeting of the Japanese Association for Acute Medicine, Kurashiki city, Japan, November 9, 1990.