Neutrophil-derived 5'-adenosine monophosphate promotes endothelial barrier function via CD73-mediated conversion to adenosine and endothelial A2B receptor activation

Abstract

During episodes of inflammation, polymorphonuclear leukocyte (PMN) transendothelial migration has the potential to disturb vascular barrier function and give rise to intravascular fluid extravasation and edema. However, little is known regarding innate mechanisms that dampen fluid loss during PMN-endothelial interactions. Using an in vitro endothelial paracellular permeability model, we observed a PMN-mediated decrease in endothelial paracellular permeability. A similar decrease was elicited by cell-free supernatants from activated PMN (FMLP 10(-6) M), suggesting the presence of a PMN-derived soluble mediator(s). Biophysical and biochemical analysis of PMN supernatants revealed a role for PMN-derived 5'-adenosine monophosphate (AMP) and its metabolite, adenosine, in modulation of endothelial paracellular permeability. Supernatants from activated PMN contained micromolar concentrations of bioactive 5'-AMP and adenosine. Furthermore, exposure of endothelial monolayers to authentic 5'-AMP and adenosine increased endothelial barrier function more than twofold in both human umbilical vein endothelial cells and human microvascular endothelial cells. 5'-AMP bioactivity required endothelial CD73-mediated conversion of 5'-AMP to adenosine via its 5'-ectonucleotidase activity. Decreased endothelial paracellular permeability occurred through adenosine A2B receptor activation and was accompanied by a parallel increase in intracellular cAMP. We conclude that activated PMN release soluble mediators, such as 5'-AMP and adenosine, that promote endothelial barrier function. During inflammation, this pathway may limit potentially deleterious increases in endothelial paracellular permeability and could serve as a basic mechanism of endothelial resealing during PMN transendothelial migration.

Figure 1

Activated PMN decrease endothelial paracellular permeability. PMN with FMLP (10⁻⁶ M) and FITC-dextran 70 kD were added to monolayers. Transendothelial flux was calculated by linear regression (6 samples over 60 min) and normalized as percent of control (HBSS). Data are derived from six monolayers in each condition; results are expressed as mean of percent control flux ± SEM. An asterisk indicates P < 0.05 by ANOVA.

Figure 2

Supernatants from activated PMN decrease endothelial paracellular permeability. Confluent HUVEC on permeable supports were exposed (apical surface only) to cell-free supernatants from activated (FMLP 10⁻⁶ M) PMN. Supernatants from activated PMN were added to monolayers. Supernatants, undiluted or diluted (as indicated) decreased transendothelial flux of FITC-dextran (*P < 0.05 vs. control [HBSS]; ANOVA). Supernatants (undiluted) decreased flux by 50 ± 7%. Data are from 10 monolayers in each condition. Results are expressed as mean reduction in permeability ± SEM.

Figure 3

Chromatographic identification of 5′-AMP and adenosine in activated PMN supernatants. 5′-AMP and adenosine were identified by their characteristic retention times and UV absorption spectra. Representative chromatograms and UV spectra (insets) of authentic 5′-AMP (A) and adenosine (B) standards are shown by dashed lines. Representative chromatograms and UV spectra of activated PMN supernatants are shown by solid lines.

Figure 4

5′-AMP and adenosine concentrations in activated PMN supernatants. PMN (10⁸/ml) were activated in HBSS with FMLP (10⁻⁶ M) in glass tubes. PMN suspensions were sampled at indicated time points after activation. 5′-AMP and adenosine concentrations were measured via HPLC. Data are from eight donors. Results are expressed as mean concentration ± SEM.

Figure 5

5′-AMP and adenosine decrease endothelial paracellular permeability. 5′-AMP or adenosine were added to HUVEC monolayers (left) or HMVEC monolayers (right). Both 5′-AMP and adenosine decreased transendothelial FITC-dextran flux in a dose-dependent manner (*P < 0.05; ANOVA). 5′-AMP and adenosine were equally potent. Data are from 6–8 monolayers in each condition. Results are expressed as mean ± SEM.

Figure 6

5′-Ectonucleotidase inhibition reverses the 5′-AMP–mediated decrease in endothelial paracellular permeability. (A) CD73 is expressed on the cell surface of confluent HUVEC and HMVEC as demonstrated by mAb 1E9 immunoprecipitation of biotinylated protein (arrow). CD73 was also immunoprecipitated from confluent T84 cells, an intestinal epithelia cell line previously known to express this ectoenzyme. (B) Addition of 5′-AMP to HUVEC monolayers decreased transendothelial FITC-dextran flux (*P < 0.05 vs. HBSS; ANOVA). APCP, a 5′-ectonucleotidase inhibitor, reversed this decrease in endothelial paracellular permeability (^‡ P < 0.05; ANOVA). Data are from six monolayers in each condition. (C) Addition of 5′-AMP or adenosine to HUVEC monolayers decreased transendothelial FITC-dextran flux (*P < 0.05 vs. control [HBSS]; ANOVA). 1E9, an anti-CD73 mAb, attenuated this decrease in permeability for 5′-AMP but not for adenosine (^‡ P < 0.05 vs. no 1E9; ANOVA). Data are from 8 monolayers in each condition. (D) HUVEC 5′-ectonucleotidase activity was decreased (*P < 0.05 vs. control) by APCP (3 μM) or 1E9 (10 mcg/ml) pretreatment. Data are from 12 monolayers in each condition. Results are expressed as mean ± SEM.

Figure 7

NECA pretreatment reverses the 5′-AMP– mediated decrease in endothelial paracellular permeability. Addition of 5′-AMP (50 μM) to HUVEC monolayers decreased transendothelial FITC-dextran flux (*P < 0.05 vs. HBSS). Pretreatment of HUVEC monolayers with NECA diminished ([NECA] = 300 nM) or abolished ([NECA] = 3 μM) this decrease in flux (^‡ P < 0.05 vs. 5′-AMP without NECA). Data are from eight monolayers in each condition. Results are expressed as mean ± SEM.

Figure 8

Adenosine A₂ receptor activation decreases endothelial paracellular permeability. ADAC and CPCA, adenosine receptor agonists, were added to HUVEC monolayers. Only CPCA (A₂ agonist) decreased transendothelial FITC-dextran flux (*P < 0.05 vs. control [HBSS]). Data are from 6–8 monolayers in each condition. Results are expressed as mean ± SEM.

Figure 9

Adenosine A_2B receptor antagonist diminishes the 5′-AMP–mediated decrease in endothelial paracellular permeability. Addition of 5′-AMP (50 μM) to HUVEC monolayers resulted in a decrease in transendothelial FITC-dextran flux (*P < 0.05 vs. control [HBSS]). (A) CSC (A_2A receptor antagonist) had no effect on the 5′-AMP–mediated decrease in endothelial paracellular permeability. (B) Alloxazine (A_2B receptor antagonist) inhibited the 5′-AMP–mediated decrease in endothelial paracellular permeability (B; ^‡ P < 0.05 vs. 5′-AMP only). Data are from eight monolayers in each condition; results are expressed as mean ± SEM. (C) Adenosine A_2B receptor is present in confluent HUVEC, HMVEC, and T84 cells as shown by cell lysate Western blot (arrow).

Figure 10

Proposed model of PMN-derived 5′-AMP–mediated decrease in endothelial paracellular permeability. PMN-derived 5′-AMP undergoes conversion to adenosine via endothelial 5′-ectonucleotidase (CD73). Adenosine activates the endothelial A_2B receptor (A_2BR), and via elevated intracellular cAMP levels, decreases endothelial paracellular permeability through actions at the endothelial intercellular junction.