INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

ADENOSINE PLAYS A ROLE in maintaining the barrier function of endothelial cells and may be important in maintaining vascular integrity, especially during inflammation or reperfusion of ischemic tissues. In addition to inhibiting release of oxidant species from neutrophils (6), adenosine may prevent edema by acting directly on the endothelium. In an in vitro model of the vasculature, adenosine decreased permeability of fetal aortic endothelial monolayers to paracellular tracers in a reversible, concentration-dependent manner (9). Through the use of receptor-specific adenosine analogs, the effect was determined to be mediated via endothelial A₂ purinoreceptors. In another in vitro system, an A₁-specific agonist mimicked the ability of platelet-derived adenosine to reduce basal permeability across bovine pulmonary artery endothelial monolayers, whereas an A₂-specific agonist did not mimick this ability (25). In the heart and lung, adenosine released during short periods of ischemia is thought to protect against ischemia-reperfusion injury by acting at A₁ receptors (16, 18, 19). These studies point to potentially divergent mechanisms for adenosine's action on endothelial cells, since A₂ receptors stimulate adenyl cyclase to produce adenosine 3',5'-cyclic monophosphate (cAMP), whereas A₁ receptors inhibit the enzyme (7). The concept of adenosine exerting permeability-decreasing effects through A₂ receptors is bolstered by substantial evidence that the increases in cAMP enhance the barrier function of endothelial cells (2, 4, 5, 11, 22, 29, 30), whereas other studies point to a cAMP-independent mechanism for control of endothelial permeability (15, 21, 26).

To date most in vitro evidence for the permeability-lowering effect of adenosine comes from studies using uninjured endothelial cells from aorta or pulmonary arteries, vessels that are not usually involved in edema formation; the permeability-decreasing effect of adenosine on endothelial cells of venular origin has not been reported. In addition, the effect of adenosine on endothelial cells directly exposed to oxidant species has not been demonstrated in vitro. We hypothesized that adenosine lowers basal permeability of venular endothelial cells, and, by its direct action on endothelial cells, adenosine prevents the permeability increase associated with oxidant injury to the vasculature. Using human umbilical vein endothelial cell (HUVEC) monolayers exposed to xanthine plus xanthine oxidase (X/XO) as a model of oxidant injury, we tested the ability of adenosine to maintain the barrier function of these venular endothelial cells in the absence or presence of oxygen radicals. In addition, several receptor-specific adenosine analogs were evaluated for their ability to lower basal permeability and prevent oxidant-induced permeability changes. Because cAMP has been implicated as a mediator of endothelial permeability, we sought to determine whether alterations in cAMP are associated with direct oxidant injury of endothelial cells. Because adenosine can either inhibit or stimulate adenyl cyclase through A₁ or A₂ receptors, respectively, we investigated whether cAMP levels of HUVEC were altered by levels of adenosine that protect against X/XO-induced permeability changes.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Protocol. Using an in vitro assay that measures the passage of Evan's blue dye-labeled albumin (EB-albumin) across confluent monolayers, we investigated the ability of adenosine and adenosine receptor agonists to alter basal permeability of HUVEC. The effect of X/XO exposure on permeability was also investigated, as was the ability of adenosine and adenosine receptor agonists to prevent X/XO-induced permeability changes. Intracellular and extracellular levels of cAMP were measured in HUVEC exposed to levels of X/XO that increase permeability. Intracellular cAMP was also measured in uninjured and oxidant-injured HUVEC that had been pretreated with adenosine or receptor-selective agonists at concentrations that lower permeability. In both the permeability and cAMP studies, cells were incubated with adenosine or analog 10 min before the addition of X/XO. In the permeability assay, samples were taken 15, 30, 60, and 90 min after X/XO exposure as described in Permeability assay. In the cAMP study, samples were collected 15 min after the addition of X/XO, processed, and assayed as described cAMP levels in HUVEC.

The following adenosine receptor agonists (Research Biochemical International, Natick, MA) were evaluated for their ability to alter permeability and cAMP levels of uninjured and X/XO-injured HUVEC: N⁶-cyclopentyladenosine (CPA), R()-N⁶-(2-phenylisopropyl)adenosine (R-PIA), and adenosine amine congener (ADAC), selective for A₁ receptors; 5'-(N-ethylcarboxamido)adenosine (NECA), nonselective for A₁ and A₂ receptors; CGS-21680 hydrochloride (CGS-21680) and N⁶-[2-(3,5-dimethoxyphenyl)-2-(2-methylphenyl)ethyl]adenosine (DPMA), both selective for A₂ receptor; and 1-deoxy-1-[6-[[(3-iodophenyl)methyl]amino]-9H-purin-9-yl]-N-methyl--D-ribofuranuronamide (IB-MECA), selective for A₃ receptor. CGS-21680, R-PIA, CPA, DPMA, and IB-MECA were initially dissolved in dimethyl sulfoxide, and ADAC was dissolved in acetic acid (0.1 M). Adenosine and NECA were dissolved in water. Dilutions of stock solutions were made with water to obtain working concentrations of each reagent.

Cell culture. HUVEC were obtained as cryopreserved primary cultures (Clonetics, San Diego, CA) and were subcultured in 25-cm² flasks through passage 10 in endothelial cell growth medium (Clonetics CC-3024). For the permeability assays, HUVEC were seeded at a density of 5 × 10⁴ cells/cm² and grown to confluence on microporous cell culture inserts (Falcon 3096, 3-µm pore size; Becton-Dickinson Labware, Franklin Lakes, NJ). For the cAMP studies, HUVEC were seeded at a density of 8 × 10³ cells/cm² and grown to confluence in 24-well tissue culture plates (Falcon 3504) that had been gelatinized. Experiments were performed 4-5 days after cells reached confluency.

Preparation of EB-albumin. EB-albumin was prepared according to the procedure of Mineau-Hanschke et al. (17) and was reconstituted with minimal essential media (MEM) without phenol red to a concentration of 2.25%.

Permeability assay. The assay for measurement of permeability across endothelial monolayers was modified from the method developed and characterized by Garcia et al. (8), which was later adapted by Patterson et al. (24) to use EB-albumin as a tracer of albumin leak. MEM containing 1% bovine serum albumin (BSA; BSA Bovuminar Reagent Pure Powder, Intergen, Purchase, NY) was used as the assay buffer, and cells were incubated at 37°C for the duration of the assay. Confluent endothelial monolayers on cell culture inserts were washed twice with buffer and transferred to new 24-well tissue culture plates. MEM/BSA (400 and 1,000 µl) was placed in the luminal chamber (insert) and abluminal chamber (well), respectively. Adenosine (Sigma Chemical, St. Louis, MO) or adenosine-receptor agonist was added to the luminal chamber. Ten minutes later, oxidant injury was induced by the addition of xanthine (200 µM final concentration; Sigma) and xanthine oxidase (30 mU/ml final concentration; Sigma) to the luminal chamber. EB-albumin was then added to the luminal chamber to a final concentration of 0.45%, marking the zero time point. Plates were swirled gently upon addition of each reagent. At selected time points, 100 µl of media were removed from the abluminal chamber and placed in a 96-well plate for measurement of absorbance at 600 nm (A₆₀₀). The clearance of EB-albumin at each time point was determined by the formula where V_cl is clearance volume, V_abl is volume of media in abluminal chamber before sampling, A_abl is absorbance of sample taken from the abluminal chamber, and A_lum is absorbance of media in luminal chamber. The absorbance of EB-albumin in the luminal chamber did not change significantly during the time course of the protocol. Therefore, the value of A_lum for each experiment was obtained by mixing 100 µl EB-albumin with 400 µl MEM/BSA to yield the same initial concentration of EB-albumin in the luminal chambers and measuring A₆₀₀ of 100 µl of this solution. This approach eliminated the need to remove aliquots from the luminal chamber at each time point.

cAMP levels in HUVEC. Confluent cells in 24-well plates were washed twice with MEM/BSA, then exposed to adenosine, adenosine-receptor agonist, or vehicle in MEM/BSA (0.5 ml) at 37°C. After 10 min, xanthine and xanithine oxidase were added to each well to final concentrations of 200 µM and 30 mU/ml, respectively, and incubation continued at 37°C. After 15 min, the extracellular media were collected into tubes containing cold ethanol (4.5 ml). Samples were centrifuged (2,000 g, 4°C, 15 min) to remove precipitated proteins; supernatants were decanted into new tubes and evaporated to dryness in a vacuum centrifuge (Savant, Farmington, NJ). Residues from evaporated supernatants were reconstituted in the appropriate assay buffer. Immediately after the extracellular media were removed from cells, 0.5 ml of lysing buffer (0.02 N HCl, 5 mM EDTA, 0.1 mM isobutylmethylxanthine, 25 µg/ml aprotinin, 25 µg/ml leupeptin; Sigma) was added to each well. Cells were mechanically disrupted with the rubber plunger from a tuberculin syringe, and cells and media were transferred into microfuge tubes. After incubation (100°C, 3 min), the tubes were centrifuged (10,000 g, 10 min) to sediment denatured protein. Samples were analyzed for cAMP by radioimmunoassay (RIA, RPA 509 nonacetylated, Amersham, Buckinghamshire, UK). Results are reported as concentration of cAMP (nM) in media or lysing buffer (0.5 ml) removed from each 2.0 cm² well of confluent cells.

Statistics. All data are presented as means ± SE, and n is the number of individual monolayers tested. Differences among treatments were determined by analysis of variance. Dunnett's test identified significant differences between treatments and control (no treatment) group or the X/XO-injured group at each time point. P < 0.05 was considered significant.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Effect of adenosine on basal permeability of HUVEC monolayers. To examine the effect of adenosine on basal permeability of HUVEC, confluent monolayers were exposed to adenosine (0.01-100 µM) in the absence of oxidant injury. At all time points examined, there was a significant difference (P < 0.05) in the permeabilities of monolayers treated with 10 and 100 µM adenosine compared with untreated HUVEC, as indicated by decreased clearance of EB-albumin (Fig. 1). Lower concentrations of adenosine (0.01-1 µM) did not affect permeability significantly, although a concentration-dependent trend was observed (Fig. 2).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Effect of adenosine on basal permeability of human umbilical vein endothelial cells (HUVEC). Confluent endothelial cell monolayers were exposed to adenosine (0.01-100 µM) 10 min before addition of Evan's blue dye-labeled albumin (EB-albumin). Clearance of EB-albumin was measured up to 90 min. Data points are mean values; SE bars are not shown for clarity; n = 14. * P < 0.05 vs. untreated HUVEC.

View larger version (10K): [in this window] [in a new window]   Fig. 2.   Effect of adenosine concentration on basal permeability of HUVEC. Confluent endothelial cell monolayers were exposed to adenosine (0.01-100 µM) 10 min before addition of EB-albumin. Data shown are mean clearance values at 30 min ± SE; n = 14. * P < 0.05 vs. untreated HUVEC.

Effect of X/XO on permeability of HUVEC monolayers. X/XO were used to chemically generate oxygen radicals as a means of injuring endothelial cell monolayers. After 15 min of exposure to X/XO, there was a significant increase in permeability of the monolayers as indicated by increased clearance of EB-albumin compared with untreated HUVEC (Fig. 3). The clearance of EB-albumin in the X/XO-treated group was the same as that of an insert without cells at the 15- and 30-min time points and continued to increase at the 60- and 90-min time points, though not as rapidly as the insert without cells. The addition of either xanthine or xanthine oxidase alone also caused an increase in albumin clearance but not as great as the combination of substrate and enzyme. Viability of endothelial cells was not affected by X/XO treatment. In all subsequent studies, 200 µM xanthine was used with 30 mU/ml xanthine oxidase to induce oxidant injury to HUVEC.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Effect of xanthine plus xanthine oxidase (X/XO) on permeability of HUVEC monolayers. Cells were exposed to xanthine (200 µM) and xanthine oxidase (30 mU/ml) immediately before addition of EB-albumin. Clearance of EB-albumin was measured up to 90 min. Data points are mean values; SE bars are not shown for clarity; n = 20. * P < 0.01 vs. untreated HUVEC.

Effect of adenosine on X/XO-induced injury of endothelial cell monolayers. To examine whether adenosine would protect endothelial cells from injury upon direct exposure to oxygen radicals, cells were exposed to adenosine (100 µM) 10 min before addition of X/XO. Pretreatment of monolayers with adenosine before exposure to X/XO prevented the permeability increase associated with oxidant injury as demonstrated by albumin clearance values that remained below untreated HUVEC at all time points examined (Fig. 4). In additional experiments, pretreatment with adenosine (0.1-100 µM) for 10 min reduced permeability of X/XO-exposed monolayers in a concentration-dependent fashion. Clearance values for cells pretreated with 10 or 100 µM adenosine and then exposed to X/XO were significantly different from those of X/XO-injured cells at all time points examined. For 10 and 100 µM adenosine, clearance of EB-albumin at the 30-min time point was 115 ± 15 and 87 ± 9% of untreated HUVEC, respectively, compared with 206 ± 27% for X/XO-treated cells (see Fig. 8 for additional information).

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Effect of adenosine pretreatment on X/XO-induced injury of HUVEC. Cells were exposed to adenosine (100 µM) for 10 min before addition of xanthine (200 µM) and xanthine oxidase (30 mU/ml) and EB-albumin. Clearance of EB-albumin was measured up to 90 min. Values are means ± SE; n = 14. * P < 0.01 vs. cells treated only with X/XO.

Effect of adenosine receptor agonists on basal permeability of HUVEC monolayers. Adenosine receptor agonists (100 µM) were examined in the permeability assay to test their ability to mimic the permeability-decreasing effect of adenosine on HUVEC monolayers (Fig. 5). To compare the effect of treatments from different experiments, mean clearance values for the 30-min time point as a percentage of control (untreated HUVEC). NECA, an agonist that is nonselective for A₁ and A₂ receptors, lowered basal permeability to 72 ± 32% of untreated HUVEC, but because of large variability, the effect was not significant. CGS-21680, an A₂-selective agonist, increased basal permeability to 133 ± 13% of untreated HUVEC. The A₂-selective agonist DPMA and A₃-selective agonist IB-MECA had no significant effect on basal permeability at any concentration used (100, 10, and 1 µM). Of the three A₁-selective agonists examined, CPA and R-PIA did not significantly lower basal permeability at any concentration studied. ADAC was the only A₁ agonist to significantly lower basal permeability (55 ± 8% of untreated HUVEC) at 100 µM (Fig. 5). Like adenosine, which lowered basal permeability in a concentration-dependent manner, ADAC exhibited a concentration-dependent reduction of permeability (Fig. 6). Furthermore, the extent to which ADAC lowered basal permeability was nearly the same as adenosine, except for 0.1 µM, where ADAC caused an insignificant increase in basal permeability. Comparable dilutions of vehicles (dimethyl sulfoxide or acetic acid) had no significant effect on basal permeability of HUVEC (data not shown).

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Effect of adenosine receptor agonists on basal permeability of HUVEC monolayers. Confluent monolayers were exposed to receptor agonists (100 µM) 10 min before addition of EB-albumin. Data shown are clearance values at 30 min and expressed as percentages of untreated HUVEC. Values are means of duplicates ± SE; n = 6-10. * P < 0.05 vs. untreated HUVEC, ** P < 0.01 vs. untreated HUVEC. See text for definitions of agonist abbreviations.

View larger version (16K): [in this window] [in a new window]   Fig. 6.   Effect of adenosine and adenosine amine congener (ADAC) on basal permeability of HUVEC monolayers. Confluent monolayers were exposed to adenosine or ADAC 10 min before addition of EB-albumin. Data shown are clearance values at 30 min and expressed as percentages of untreated HUVEC. Values are means of duplicates ± SE; n = 6-10. * P < 0.05 vs. untreated HUVEC, ** P < 0.01 vs. untreated HUVEC.

Effect of adenosine receptor agonists on X/XO-injured HUVEC monolayers. Adenosine-receptor agonists (0.1-100 µM) were tested for the ability to reduce or prevent the permeability increase induced by X/XO exposure. In all experiments, treatment with X/XO induced a significant increase in permeability of the monolayers (P < 0.05 vs. untreated HUVEC), even though the extent of injury varied between experiments (Figs. 7 and 8). CGS-21680, DPMA, CPA, R-PIA, and IB-MECA had no significant effect on X/XO-induced injury at any concentration studied (Fig. 7). Like adenosine, NECA and ADAC at 10 and 100 µM prevented X/XO-induced injury of HUVEC monolayers, maintaining permeability of oxidant-exposed monolayers at basal level (Fig. 8). For 10 and 100 µM NECA, clearance of EB-albumin was 99 ± 19 and 118 ± 21% of untreated HUVEC, respectively, compared with 232 ± 31 for X/XO-treated cells. For 10 and 100 µM ADAC, clearance of EB-albumin was 79 ± 20 and 95 ± 17% of untreated HUVEC, respectively, compared with 143 ± 9% for injured cells. For both agonists, pretreatment with 10 µM resulted in lower clearance values than pretreatment with 100 µM after X/XO. Pretreatment with 1 µM NECA also resulted in significant protection against X/XO injury, with a clearance value of 145 ± 21% of untreated HUVEC (Fig. 8). Pretreatment with vehicles for each agonist did not reduce permeability changes in X/XO-injured cells (data not shown).

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Effect of adenosine receptor agonists on xanthine plus xanthine oxidase (X/XO)-induced injury. Cells were exposed to receptor agonists 10 min before addition of xanthine (200 µM) and xanthine oxidase (30 mU/ml) and EB-albumin. Data shown are clearance values at 30 min and expressed as percentages of untreated HUVEC. Open bars represent X/XO injured, and solid bars represent agonist pretreatment plus X/XO treatment. Values are means of duplicates ± SE; n = 6-10. * P < 0.01 vs. X/XO-injured cells. For all experiments, P < 0.05 for X/XO injury vs. untreated HUVEC. See text for complete definitions of abbreviations.

View larger version (35K): [in this window] [in a new window]   Fig. 8.   Effect of adenosine, NECA, and ADAC on X/XO injury. Cells were exposed to receptor agonists 10 min before addition of xanthine (200 µM) and xanthine oxidase (30 mU/ml) and EB-albumin. Data shown are clearance values at 30 min and expressed as percentages of untreated HUVEC. Gray bars represent X/XO injured, and all other bars represent agonist pretreatment plus X/XO treatment. Values are means of duplicates ± SE; n = 3-10. * P < 0.05 vs. X/XO-injured cells, ** P < 0.01 vs. X/XO-injured cells. For all experiments, P < 0.05 for X/XO injury vs. untreated HUVEC.

Effect of adenosine and receptor agonists on intracellular cAMP levels of uninjured HUVEC. To determine whether the permeability-lowering effect of adenosine, NECA, or ADAC is accompanied by an alteration in cAMP level, intracellular cAMP levels were determined after exposure of HUVEC to concentrations of these agonists that lower permeability. Untreated HUVEC yielded 0.93 ± 0.08 pmol cAMP/well of confluent cells, with values ranging from 0.40 to 1.53 pmol/well. To compare values from separate experiments, cAMP levels are expressed as percentages of untreated HUVEC grown on the same tissue culture plate. Adenosine (100 µM) significantly increased cAMP to 139 ± 8% of untreated HUVEC; the increase caused by 10 µM adenosine (114 ± 4% of untreated HUVEC) was not statistically significant (Fig. 9). NECA caused a significant increase in intracellular cAMP at both concentrations studied: 10 and 100 µM resulted in cAMP concentrations that were 174 ± 12 and 175 ± 21% of untreated HUVEC, respectively. ADAC (10 and 100 µM) did not increase intracellular cAMP significantly (128 ± 6 and 125 ± 10% of untreated HUVEC, Fig. 9).

View larger version (21K): [in this window] [in a new window]   Fig. 9.   Effect of adenosine receptor agonists on intracellular cAMP. Cells were exposed to adenosine receptor agonists for 25 min, then lysed; cAMP in intracellular fractions was measured by radioimmunoassay. Data are expressed as percentages of untreated HUVEC. Values are means ± SE; n = 6, except "no treatment" group where n = 12. * P < 0.05 vs. no treatment, ** P < 0.01 vs. no treatment.

Effect of adenosine and receptor agonists on intracellular cAMP levels of X/XO-exposed HUVEC. We investigated whether or not direct oxidant injury with X/XO at concentrations that cause an increase in permeability would alter cAMP levels of HUVEC. Exposure of HUVEC to X/XO did not significantly alter intracellular cAMP levels (104 ± 12% of untreated HUVEC, Fig. 10). Extracellular cAMP levels were unaffected by X/XO exposure as well (data not shown). Because adenosine, NECA, and ADAC prevented X/XO-induced permeability increase in HUVEC monolayers (Fig. 8), the effect of these agonists on cAMP levels in X/XO-exposed HUVEC was examined. Intracellular cAMP was not increased in cells treated with adenosine and then exposed to X/XO. NECA caused a significant increase in cAMP to 200 ± 14 and 228 ± 11% of untreated HUVEC levels for 10 and 100 µM concentrations, respectively, following X/XO-exposure of cells. Likewise, ADAC significantly increased cAMP to 184 ± 10 and 194 ± 12% of untreated HUVEC levels for 10 and 100 µM, respectively (Fig. 10).

View larger version (21K): [in this window] [in a new window]   Fig. 10.   Effect of agonists on intracellular cAMP of X/XO-exposed HUVEC. Cells were pretreated with receptor agonists for 10 min, then exposed to xanthine (200 mM) and xanthine oxidase (30 mU/ml) for 15 min, then lysed; cAMP in intracellular fractions was measured by radioimmunoassay. Data are expressed as percentages of untreated HUVEC. Values are means ± SE; n = 6, except groups that received no treatment or received only X/XO where n = 12. ** P < 0.01 vs. no treatment and vs. X/XO.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Adenosine significantly decreased basal permeability of HUVEC monolayers to albumin in a concentration-dependent manner at 10 and 100 µM. These findings in human venular cells are consistent with previous reports of the ability of adenosine to lower basal permeability of endothelial cells (10, 25) and provide additional evidence that adenosine directly enhances the barrier properties of the endothelium. Two adenosine receptor agonists, NECA and ADAC, mimicked the permeability-lowering effect of adenosine, suggesting that the effect is receptor mediated and not due to metabolic consequences of adenosine uptake by the cell, such as increased intracellular ATP.

Reactive oxygen species released from neutrophils or generated in the bloodstream after reperfusion of ischemic tissue are thought to contribute to vascular injury by increasing endothelial permeability (13, 28, 34). The enzymatic reaction of xanthine oxidase with purine substrate has been used extensively as a means of generating oxygen radicals in vitro, including studies of endothelial barrier dysfunction (3, 27). The permeability-increasing effects of reactive oxygen species on endothelial cells of venular origin were demonstrated using concentrations of X/XO comparable to those used previously by other investigators. In similar studies with bovine aortic endothelial cells, hypoxanthine or xanthine (0.2 mM) added in combination with xanthine oxidase (20 mU/ml) resulted in increased transfer of dye-labeled albumin across monolayers (3). In these studies, addition of either hypoxanthine or xanthine oxidase alone caused a slight increase in albumin transfer that was not statistically significant. In contrast, exposure of HUVEC to either substrate or enzyme alone resulted in significant increases in albumin clearance across monolayers (Fig. 3). However, neither xanthine nor xanthine oxidase alone increased permeability of HUVEC to the extent of an insert without cells, as did the combination of substrate and enzyme (Fig. 3). The increase in permeability resulting from the addition of enzyme by itself may reflect the presence of endogenous purine substrates in the cells or media. The increase in permeability resulting from addition of xanthine alone suggests the presence of endogenous xanthine oxidase. Although xanthine oxidase levels of HUVEC in culture were not determined in our study, endothelial cells previously isolated and cultured from human aorta, coronary artery, or the microvasculature did not exhibit measurable xanthine oxidase activity (23).

Although the ability of adenosine to decrease basal permeability of endothelial monolayers has been demonstrated in vitro, the barrier-maintaining effect of adenosine on endothelial cells injured by direct exposure to oxygen radicals has not been reported until now. Pretreatment with adenosine prevented the permeability increase that results from exposure of HUVEC to X/XO (Fig. 4). Addition of adenosine to media immediately before addition of X/XO did not confer protection against injury (data not shown), indicating that adenosine was not acting simply as an inhibitor of xanthine oxidase or as a sink for free radicals. NECA and ADAC also prevented the permeability increase of X/XO injury, while other agonists examined did not.

These results are consistent with the report by Haselton et al. (10) showing that adenosine (1-100 µM) decreased permeability of fetal bovine aortic endothelial cells to the paracellular tracers polyethylene glycol and cyanocobalamin within 15 min and in a concentration-dependent manner. Using A₁- and A₂-receptor agonists and an A₂ antagonist, Haselton et al. (10) concluded that adenosine decreases permeability of these cells via A₂ receptors. In contrast, we found that the A₂-selective agonists, CGS-21680 and DPMA, did not lower basal permeability of HUVEC, whereas the A₁-selective agonist, ADAC, lowered basal permeability significantly at 10 and 100 µM. These findings are consistent with Paty et al. (25), who found ADAC, but not an A₂ agonist, capable of lowering permeability of cultured bovine pulmonary artery endothelial cells (BPAE). However, the effects of adenosine and ADAC were evident at concentrations much lower than this current study. Basal permeability of BPAE was lowered to ~70% of baseline by adenosine concentrations that ranged from 10⁴ to 10¹ µM. ADAC lowered basal permeability of BPAE to 70% of baseline at 10⁵ and 10³ µM but not significantly at other concentrations. Their results using BPAE are consistent with our HUVEC results in that adenosine at 1 µM and ADAC at 0.1 and 1 µM had no significant effect on basal permeability of either cell type. The two data sets do not necessarily contradict one another, since Paty et al. (25) did not include concentrations of ADAC >1 µM, and we did not examine concentrations <0.1 µM. Cell passage number, seeding density, or other culture conditions may account for the discrepancies.

Adenosine and ADAC were the only agents among those studied to significantly decrease basal permeability of HUVEC (Figs. 5 and 6). However, ADAC and adenosine (10 µM) did not increase intracellular cAMP significantly (Fig. 9). As an A₁-selective agonist, ADAC is not predicted to increase cAMP, yet it lowered basal permeability to the same extent as adenosine at equal concentrations. These data suggest that adenosine and ADAC may lower basal permeability by a mechanism that is independent of cAMP. Despite its ability to increase intracellular cAMP significantly, NECA decreased basal permeability only moderately compared with adenosine and ADAC (Figs. 5 and 9). Taken together, these data demonstrate no correlation between intracellular cAMP levels of HUVEC after agonist exposure and the extent to which these agonists altered basal permeability. In addition, these findings suggest that cAMP concentration is not a sole determinant of basal permeability, as evidenced by the lesser extent to which NECA lowered basal permeability, despite its ability to increase cAMP more than adenosine or ADAC.

Many agents that increase cAMP through receptor-dependent or -independent stimulation of adenyl cyclase or inhibition of phosphodiesterase, as well as cAMP analogs, have been shown to decrease basal endothelial permeability (4, 15, 29, 30, 32). The effect, however, may be specific to cell type. Watanabe et al. (32) reported that NECA decreased permeability of porcine aortic endothelial cells to labeled albumin, as did isoproterenol and forskolin. All three agents stimulated production of cAMP in these cells, which supports the concept that stimulation of adenyl cyclase by binding of adenosine A₂ receptors would result in decreased endothelial permeability. In rat coronary microvascular endothelial cells, however, these same agonists had the opposite effect on macromolecule permeability, even though cAMP was increased (12, 21, 32). These results suggest that increases in cAMP may have opposite effects on barrier properties of different cell types. An alternative explanation is that these stimuli affect permeability by a cAMP-independent mechanism, which our data support.

Whereas cAMP may mediate permeability in response to some agents (2, 4, 5, 11, 22, 29, 30), these current findings implicate a cAMP-independent mechanism for control of endothelial barrier function that may be operative in HUVEC. Other investigators have also provided evidence for control of endothelial permeability that is independent of cAMP. Langeler et al. (15) found a direct correlation between the relative increase in cAMP and the relative decrease in permeability as measured by passage of peroxidase and transendothelial electrical resistance after treatment of human artery endothelial cells with forskolin, isoproterenol, or iloprost. With norepinephrine, however, the decrease in permeability was much greater than what might be predicted from the relatively small increase in cAMP, suggesting that an additional mechanism, independent of cAMP, may be responsible for the effect of norepinephrine (15). The phosphodiesterase inhibitor pentoxifylline lowered basal permeability and reduced endotoxin-induced permeability increase of bovine pulmonary artery endothelial monolayers yet failed to raise cAMP levels in these cells (26). It was concluded that this effect of pentoxifylline in the absence of neutrophils was partly mediated via direct effects on endothelium that may be independent of cAMP.

Adenosine has been found to be protective against reperfusion injury in heart tissue by an A₁-receptor mechanism (16). Pretreatment with adenosine or A₁-receptor agonists mimicked the protective effects of "preconditioning" ischemia, attenuating ischemia-reperfusion injury and limiting infarct size in hearts of several species (16, 36). Adenosine released during a brief period of preconditioning ischemia may protect the heart from prolonged ischemia and reperfusion by uncoupling A₁-adenosine receptors from the signal transduction mechanisms that lead to injury (18). This same protective mechanism may be present in other organ systems, as well, such as the brain and lung, where rapid desensitization of A₁ receptors following activation has been observed (1, 20). Our findings provide evidence for an A₁-protective mechanism against direct oxidant injury in cells of venular origin. The uncoupling of signaling pathways by adenosine and receptor-specific analogs at A₁ receptors may be a plausible explanation for their protective effect in HUVEC against direct oxidant injury with X/XO.

Alternative regulators of endothelial barrier properties may include ATP-sensitive K⁺ channels and guanosine 3',5'-cyclic monophosphate (cGMP). In studying ischemia-reperfusion injury in isolated rat lungs, Khimenko et al. (14) described a protective mechanism that is independent of the cAMP-protein kinase A pathway. Opening K⁺-ATP channels with cromakalim (10 µM) did not induce endothelial injury but prevented and reversed the permeability increase associated with ischemia-reperfusion, a form of oxidant injury. cGMP has also been proposed as a regulator of endothelial barrier properties, since 8-bromo-cGMP mediated a decrease in basal permeability of endothelial monolayers obtained from both rats (12) and humans (33). Like cAMP, cGMP may help maintain cell-to-cell contact of endothelial cells by inhibiting contraction of cytoskeletal elements, analogous to its function in smooth muscle cells.

Exposure of HUVEC to X/XO did not significantly alter intracellular cAMP levels (Fig. 10), suggesting that the mechanism by which X/XO exposure induces a permeability increase in HUVEC is independent of changes in intracellular cAMP. Extracellular cAMP levels also were unaffected by X/XO exposure (data not shown), demonstrating that direct exposure of HUVEC to X/XO at these permeability-increasing doses did not cause sufficient damage to cell membranes to allow leakage of cAMP from the cells. Prevention of X/XO-induced permeability increase by adenosine, likewise, does not seem to be mediated by increased intracellular cAMP. Whereas adenosine prevented permeability increase in X/XO-exposed cells (Fig. 8), cAMP was not increased in X/XO-exposed cells that had been pretreated with effective concentrations of adenosine (Fig. 10). Intracellular cAMP was increased significantly in ADAC- and NECA-treated cells after X/XO exposure, but it is not known whether this increase in cAMP is necessary or responsible for the protective effect of these agonists. It is interesting to note that cAMP was increased in cells treated with the A₁ agonist ADAC and then exposed to X/XO, whereas neither ADAC nor X/XO treatment alone increased cAMP. X/XO also enhanced the increase in cAMP caused by NECA, because cells treated with this agonist and X/XO had higher cAMP levels than cells treated with agonist alone.

By acting directly on endothelial cells, adenosine decreased basal permeability of HUVEC to albumin in a concentration-dependent manner and prevented a permeability increase that resulted from exposure of the cells to X/XO. The permeability-increasing effect of X/XO proceeds by a mechanism that is independent of changes in cAMP, and the protective effect of adenosine against oxidant injury to endothelium may include a cAMP-independent component as well. The speculative A₁ receptor-mediated mechanism may be analogous to receptor desensitization described in the protection of heart tissue against ischemia-reperfusion injury by adenosine.