Endothelial catabolism of extracellular adenosine during hypoxia: the role of surface adenosine deaminase and CD26

Abstract

Extracellular levels of adenosine increase during hypoxia. While acute increases in adenosine are important to counterbalance excessive inflammation or vascular leakage, chronically elevated adenosine levels may be toxic. Thus, we reasoned that clearance mechanisms might exist to offset deleterious influences of chronically elevated adenosine. Guided by microarray results revealing induction of endothelial adenosine deaminase (ADA) mRNA in hypoxia, we used in vitro and in vivo models of adenosine signaling, confirming induction of ADA protein and activity. Further studies in human endothelia revealed that ADA-complexing protein CD26 is coordinately induced by hypoxia, effectively localizing ADA activity at the endothelial cell surface. Moreover, ADA surface binding was effectively blocked with glycoprotein 120 (gp120) treatment, a protein known to specifically compete for ADA-CD26 binding. Functional studies of murine hypoxia revealed inhibition of ADA with deoxycoformycin (dCF) enhances protective responses mediated by adenosine (vascular leak and neutrophil accumulation). Analysis of plasma ADA activity in pediatric patients with chronic hypoxia undergoing cardiac surgery demonstrated a 4.1 +/- 0.6-fold increase in plasma ADA activity compared with controls. Taken together, these results reveal induction of ADA as innate metabolic adaptation to chronically elevated adenosine levels during hypoxia. In contrast, during acute hypoxia associated with vascular leakage and excessive inflammation, ADA inhibition may serve as therapeutic strategy.

Figure 1.

ADA is rapidly induced by hypoxia. (A) Microarray analysis of individual nucleoside/nucleotide deaminases in response to hypoxia (ADA, CDA, AMPD₂, DCTD, and AMPD₁). Confluent HMEC-1s were exposed to normoxia (pO₂ 147 mm Hg, 18 hours) or hypoxia (24-hour exposure to pO₂ 20 mm Hg) and the relative expression of individual deaminases was quantified from total RNA by microarray analysis. ^*Increased fluorescence, P < .01. #Decreased fluorescence, P < .01. Note the dramatic induction of ADA with hypoxia (47.5-fold). (B) Confluent HMEC-1 or HUVEC monolayers were exposed to normoxia (pO₂ 147 mm Hg, 18 hours) or hypoxia (pO₂ 20 mm Hg) as indicated. Total RNA was isolated, and ADA mRNA levels were determined by real-time RT-PCR. Data were calculated relative to an internal housekeeping gene (β-actin) and are expressed as fold increase over normoxia ± SD at each indicated time. Results are derived from 3 experiments in each condition (^*P < .01). (C) Human saphenous veins were obtained from patients undergoing aorto-coronary bypass surgery and exposed ex vivo to ambient normoxia (pO₂ 147 mm Hg, 24 hours) or hypoxia (pO₂ 20 mm Hg for 2, 8, or 24 hours). After total RNA isolation, real-time PCR was performed to investigate ADA inducibility by hypoxia. Data were calculated relative to an internal control (β-actin) and are expressed as fold increase over normoxia ± SD at each indicated time. Results are derived from 3 experiments in each condition (^*P < .01). (D) Increase in total ADA protein with hypoxic exposure. Confluent HMEC-1 monolayers were exposed to indicated periods of hypoxia (0-72 hours), washed, and lysed. Lysates were resolved by SDS-PAGE, and resultant Western blots were probed with a polyclonal goat antibody directed against human ADA. A representative experiment of 3 is shown. (E) Increase in surface ADA protein with hypoxic exposure. Confluent HMEC-1 monolayers were exposed to indicated periods of hypoxia (0-48 hours), monolayers were washed, surface proteins were biotinylated, and cells were lysed. ADA was immunoprecipitated with a polyclonal goat antibody directed against human ADA. Immunoprecipitates were resolved by SDS-PAGE, and resultant Western blots were probed with avidin-peroxidase. A representative experiment of 3 is shown.

Figure 2.

ADA induced by hypoxia is enzymatically active. (A) To measure ADA enzyme activity on the endothelial surface, HMEC-1s were exposed over indicated time periods to hypoxia (pO₂ 20 mm Hg), 50 μM adenosine was added to the supernatant (HBSS) of intact HMEC-1s, and inosine generation was measured via HPLC. The HPLC tracing obtained from the supernatant is displayed and retention times for inosine (3.2 min) and for adenosine (3.7 min) are indicated (black line indicates HBSS alone; gray line, HBSS after addition of adenosine [50 μM]). (B) Induction of ADA activity in total-cell lysates of endothelia by hypoxia. To obtain an estimate of total ADA activity increase with hypoxia, the ADA activity in lysates of HMEC-1s was measured. (C) Induced, enzymatically active ADA is localized to the cell surface. ADA enzyme activity was measured in intact HMEC-1s by adding 50 μM adenosine to the supernatant (HBSS) and measuring inosine generation. All experiments were performed in the presence of 10 μM dipyridamole to prevent endothelial adenosine uptake. To confirm that the observed increase in inosine generation with hypoxia exposure reflects ADA activity, controls with 100 nM dCF were performed in normoxic (dCF [Norm]) or posthypoxic (dCF [Hyp 48 h]) endothelia. Note the robust hypoxia induction of functional ADA to the cell surface. (D) ADA release into the supernatant is increased with hypoxia. To measure ADA release into the supernatant, HMEC-1s were exposed to normoxia or hypoxia (HMEC-1, pO₂ 20 mm Hg, 48 hours). During the last 4 hours, the media was replaced with HBSS and ADA activity in the cell supernatant was measured. In controls, dCF (dCF [Hyp 48 h], 100 nM) inhibited soluble ADA activity in posthypoxic supernatants. In contrast, the HIV-1 envelope glycoprotein gp120 (gp120 [Hyp 48 h], 100 nM) did not affect ADA activity in the supernatant. Error bars indicate SD.

Figure 3.

ADA-complexing protein CD26 is coordinately induced by hypoxia. (A) Real-time PCR was used to confirm induction of CD26 mRNA by hypoxia in cultured endothelial cells (HMEC-1s and HUVECs). Data were calculated relative to internal housekeeping gene (β-actin) and are expressed as fold increase over normoxia ± SD at each indicated time. Results are derived from 3 experiments in each condition (^*P < .01). (B) Increase in total CD26 protein with hypoxic exposure. Confluent HMEC-1 monolayers were exposed to indicated periods of hypoxia, washed, and lysed. Lysates were resolved by SDS-PAGE, and resultant Western blots were probed with mAb directed against human CD26. A representative experiment of 3 is shown. (C) Increase in surface CD26 protein with hypoxic exposure. Confluent HMEC-1 monolayers were exposed to indicated periods of hypoxia, monolayers were washed, surface proteins were biotinylated, and cells were lysed. CD26 was immunoprecipitated with mAb directed against human CD26. Immunoprecipitates were resolved by SDS-PAGE, and resultant Western blots were probed with avidin-peroxidase. A representative experiment of 3 is shown. (D) ADA surface induction requires interaction with CD26. To confirm that the observed increase in enzymatically active cell-surface ADA is bound to hypoxia induced CD26, HIV-1 gp120, a specific inhibitor of ADA interaction with CD26, was used. Postnormoxic or posthypoxic (pO₂ 20 mm Hg, 48 hours) HMEC-1s were incubated for 10 minutes with 100 nM gp120 in HBSS at 37°C and washed; ADA activity was measured. Note the reduced ADA activity after gp120 treatment in postnormoxic and posthypoxic endothelia (P < .01 by ANOVA). (A, D) Error bars indicate SD.

Figure 4.

Effects of ADA activity on endothelial adenosine signaling in vitro. (A) Modulation of adenosine signaling by hypoxia. Indicated concentrations of adenosine in HBSS were added to the apical surface of confluent normoxic (48-hour exposure to pO₂ 147) or posthypoxic (48-hour exposure to pO₂ 20 mm Hg) HMEC-1s and permeability to FITC-dextran (70 kDa) were quantified. Transendothelial flux was calculated by linear regression (3 samples over 60 minutes) and normalized as a percentage of control (HBSS). Data are derived from 6 monolayers in each condition. ^*Significant differences from baseline (P < .05). #Differences from baseline and from normoxia (P < .05). (B) Effect of extracellular ADA on paracellular permeability. Measurement of adenosine elicited barrier responses in normoxic endothelia (HMEC-1) with or without the addition of 0.1 nM bovine ADA. Note the dramatic decrease in adenosine-induced enhancement of endothelial barrier function in the presence of 0.1 nM ADA. ^*Significant differences from baseline (P < .05). #Differences from baseline and untreated controls (P < .05). (C) Effect of ADA inhibitor dCF on adenosine-elicited barrier responses. Posthypoxic endothelia (HMEC-1, pO₂ 20 mm Hg, 48 hours) were assessed for adenosine-elicited barrier responses in the presence of the highly specific ADA inhibitor dCF (100 nM). Note the dramatically increased adenosine elicited barrier responses with 100 nM dCF. ^*Significant differences from baseline (P < .05). #Differences from baseline and from untreated controls (P < .05). (D) Effect of inhibiting ADA binding to the ADA-complexing protein CD26 with gp120. Posthypoxic endothelia were washed with 100 nM gp120 in HBSS prior to measuring adenosine-elicited barrier function. Note the increased adenosine elicited barrier responses after gp120 treatment. ^*Significant differences from baseline (P < .05). #Differences from baseline and from untreated controls (P < .05). (A-D) Data are expressed as mean ± SD of percent control flux with HBSS only.

Figure 5.

Influence of dCF on pulmonary edema, vascular permeability, and PMN accumulation in vivo. BL/6/129 mice were given injections of dCF (1 mg/kg intraperitoneally and 1 mg/kg subcutaneously) or PBS, and exposed to normoxia (room air) or normobaric hypoxia (8% O₂ and 92% N₂) for 4 hours. (A) Assessment of lung water content in normoxia (▪) and hypoxia () after dCF or PBS treatment. Data are expressed as mean ± SD mg H₂O/mg dry tissue, and are pooled from 6 animals per condition. ^*Significantly different between hypoxia and normoxia (P < .025). #Significantly different between dCF treatment and vehicle control (P < .025). (B) To assess vascular barrier function, animals were administered intravenous Evan blue dye solution (0.2 mL of 0.5% in PBS) prior to normoxia/hypoxia exposure. Animals were killed, and the lung, colon, and liver were harvested. Organ Evan blue concentrations were quantified following formamide extraction (55°C for 2 hours) by measuring absorbances at 610 nm with subtraction of reference absorbance at 450 nm. Data are expressed as mean ± SD Evan blue optical density (OD)/50mg wet tissue, and are pooled from 6 animals per condition. Note that Evan blue retention increases with hypoxia and decreases with dCF treatment. ^*Significant differences between normoxia/hypoxia exposure (P < .01). #Significant differences between dCF/PBS treatment groups (P < .025). (C) Organ assessment of PMN accumulation by myeloperoxidase (MPO) measurements in the indicated organs after 4 hours of normoxia/hypoxia exposure (^*P < .01 compared with hypoxia; #P < .025 compared with vehicle control). Error bars indicate SD.

Figure 6.

Plasma ADA enzyme activity is increased in patients with hypoxia. (A) To develop a technique for measuring ADA activity in plasma, different concentrations of adenosine were added to the plasma and assessed for dCF-inhibited inosine generation. Optimal resolution was found with using 20 μM adenosine (black line indicates plasma alone; gray line, plasma after addition of 20 μM adenosine). (B-C) VEGF and ADA are increased in the plasma of patients with hypoxia. Plasma samples were obtained from 10 pediatric patients with chronic hypoxia (mean oxygen saturation of 82%) undergoing cardiac surgery (bidirectional Glenn procedure) and 10 age- and sex-matched controls (mean oxygen generation of 96%) undergoing noncardiac surgery. Plasma VEGF was measured with standard ELISA technique. ADA activity was measured via HPLC as percent adenosine conversion to inosine. ^*Significant differences with hypoxia (P < .01). #Differences with dCF treatment (100 nM; P < .001). (B-C) Error bars indicate SD. (D) Correlation of ADA plasma activity with VEGF plasma levels displayed for individual patients (▪, hypoxia) and control patients (•, normoxia). Note the strong correlation between ADA activity and circulating VEGF levels (r² = .89, P < .01). The diagonal represents the linear regression line.

Figure 7.

Proposed model of coordinated induction of ADA and CD26. In areas of ongoing inflammation and diminished oxygen supply, hypoxia coordinates the induction of endothelial ADA and CD26. Following induction of ADA mRNA, the enzyme is synthesized and released from the endothelial cell and binds to its cell-surface receptor CD26. Such increases in endothelial cell-surface ADA modulate vascular adenosine signaling during hypoxia. In general, extracellular adenosine can activate endothelial adenosine receptors. Due to increased ADA surface activity with hypoxia, adenosine is metabolized to inosine, thus terminating vascular adenosine signaling and increasing extracellular inosine concentration. Inhibition of ADA with dCF or inhibition of ADA binding to its receptor CD26 (GP120) can contribute to increasing vascular adenosine effects during acute hypoxia.