Adenosine Opposes Thrombin-Induced Inhibition of Intercellular Calcium Wave in Corneal Endothelial Cells

Abstract

purpose. In corneal endothelial cells, intercellular Ca²⁺ waves elicited by a mechanical stimulus involve paracrine intercellular communication, mediated by ATP release via connexin hemichannels, as well as gap junctional intercellular communication. Both mechanisms are inhibited by thrombin, which activates RhoA and hence results in myosin light chain phosphorylation. This study was conducted to examine the effects of adenosine, which is known to oppose thrombin-induced RhoA activation, thereby leading to myosin light chain dephosphorylation, on gap junctional intercellular communication and paracrine intercellular communication in cultured bovine corneal endothelial cells.

methods. An intercellular Ca²⁺ wave was elicited by applying a mechanical stimulus to a single cell in a confluent monolayer. The area of Ca²⁺ wave propagation was measured by [Ca²⁺]_i imaging using the fluorescent dye Fluo-4. Gap junctional intercellular communication was assessed by fluorescence recovery after photobleaching. Activity of hemichannels was determined by uptake of the hydrophilic dye Lucifer yellow in a Ca²⁺-free medium containing 2 mM EGTA. Adenosine triphosphate (ATP) release in response to mechanical stimulation was measured using the luciferin-luciferase technique. Gap26, a connexin mimetic peptide, was used to block hemichannels.

results. Exposure to thrombin or TRAP-6 (a selective PAR-1 agonist) inhibited the Ca²⁺ wave propagation by 70%. Pretreatment with adenosine prevented this inhibitory effect of thrombin. NECA (a potent A2B agonist) and forskolin, agents known to elevate cAMP in bovine corneal endothelial cells, also suppressed the effect of thrombin. The A1 receptor agonist CPA failed to inhibit the effect of thrombin. Similar to the effects on Ca²⁺ wave propagation, adenosine prevented the thrombin-induced reduction in the fluorescence recovery during photobleaching experiments. Furthermore, pretreatment with adenosine prevented both thrombin and TRAP-6 from blocking the uptake of Lucifer yellow in a Ca²⁺-free medium. However, adenosine was ineffective in overcoming the Gap26-mediated block of Lucifer yellow uptake. In consistence with Lucifer yellow uptake through hemichannels, the thrombin-induced inhibition of ATP release was overcome by pretreatment with adenosine.

conclusions. Adenosine prevents thrombin-induced inhibition of hemichannel-mediated paracrine intercellular communication and of gap junctional intercellular communication. The mechanism involves an increase in cAMP, which results in inhibition of RhoA and a subsequent decrease in myosin light chain phosphorylation. Since myosin light chain dephosphorylation causes a decrease in contractility of the actin cytoskeleton, the results suggest possible effects of the actin cytoskeleton on gap junctions and connexin hemichannels.

Gap junctions are proteinaceous channels that interconnect the cytoplasm of adjacent cells, enabling a direct transfer of signaling molecules (Ca²⁺, cAMP, inositol-1,4,5-trisphosphate) and metabolites (<1.2 kDa). ^{1 2} This form of intercellular communication is essential for tissue homeostasis, control of cell proliferation, and synchronization of response to extracellular stresses. ^{1 2 3} A large number of gap junction channels, organized into symmetrical arrays, form plaques close to adherens junctions. Adherens junctions facilitate this accretion by providing the tethering forces necessary for propitious interlocking of two hemichannels (also called connexons) contributed by two adjacent cells to form a gap junction channel. ²  

Connexons, which consist of six integral transmembrane proteins called connexins, are assembled along the trans-Golgi network and delivered to the plasma membrane in a closed state as hemichannels. ^{1 2 4} Such hemichannels then undergo lateral diffusion and may eventually dock with another hemichannel on an adjacent cell to form gap junctions. ^{1 2 4} Although hemichannels can assemble into gap junctions, a number of recent reports ^{1 2 4 5 6 7} also indicate that hemichannels in the plasma membrane can open and permit leakage or uptake of a variety of molecules into and from the extracellular space. ^{4 5 6 7 8} Recent studies have implicated hemichannels in the release of ATP in several cell types. ^{5 6 7 9 10} Diffusional spread of the released ATP and its metabolite adenosine diphosphate (ADP), ¹¹ and subsequent interaction of the nucleotides with the P2 family of purinergic receptors in the neighboring cells has been shown to contribute to the Ca²⁺ wave propagation. ^{7 9 10}  

It has been demonstrated that certain connexins interact with a number of linker proteins associated with adherens and tight junctions. ^{12 13 14 15 16 17 18 19 20 21} Connexin Cx43, a major subtype, binds to the junctional adhesion molecule-associated proteins zonula occludens-1 (ZO-1) and β-catenin, and thereby gap junction channels are structurally linked to the actin cytoskeleton. ²¹ In addition to microfilaments, microtubules are also known to play a role in trafficking and/or activity of certain connexons. ^{22 23 24} Whereas Cx43-dependent connexons bind to microtubules, those formed from Cx26 do not possess such an affinity. ^{2 25 26} The binding of connexons to the microtubules may influence the functional properties of connexons in gap junctions and/or hemichannels, possibly due to altered trafficking and/or mobility in the plasma membrane. ^{2 16 26}  

Intercellular Ca²⁺ wave propagation evoked by a point mechanical stimulus is a distinct paradigm for functional evaluation of gap junctions as well as connexin hemichannels. ^{7 10 27 28 29 30 31 32 33 34 35} While examining such a Ca²⁺ wave propagation in bovine corneal endothelial cells (BCECs), we observed contributions of both gap junctions (leading to gap junctional intercellular communication, GJIC) and hemichannels (leading to paracrine intercellular communication, PIC). As in other cell types, ^{7 9} ATP released via hemichannels was found to be the paracrine factor. ¹⁰ Furthermore, while investigating the factors that could influence PIC and GJIC, we observed that the serine protease thrombin, which is often implicated in inflammation and wound healing, ^{36 37 38} significantly inhibited the propagation of the Ca²⁺ wave. ³⁹ While thrombin inhibited GJIC to a modest degree, its inhibitory effect on the ATP-dependent PIC was profound. ³⁹ Moreover, the effect of thrombin could be prevented by pretreatment with inhibitors of MLCK (myosin light chain kinase), Rho kinase (downstream effector of the small GTPase RhoA), and PKC. Because these kinases converge at the point of phosphorylation of the regulatory light chain of myosin II (myosin light chain or MLC), ^{40 41 42} our findings suggest an involvement of contractility of the actin cytoskeleton and/or of the cortical actin organization in the regulation of the activity of hemichannels involved in PIC. 

The thrombin-induced loss of the intercellular communication in the corneal endothelium is reminiscent of a loss of barrier integrity produced by the protease through contraction and disruption of the cortical actin cytoskeleton. ^{41 43 44} Centripetal forces generated by contraction of the peripheral actomyosin ring (PAMR) found in the corneal endothelial cells, ^{41 43 44} oppose the intercellular tethering forces. This is implicated in the breakdown of the tight junctions. ^{45 46 47 48} Since the thrombin-induced MLC phosphorylation is mediated through activation of the RhoA-Rho kinase axis, the loss of the barrier integrity could be prevented by inhibition of RhoA through elevated cAMP. ^{41 43 44} This protective effect of elevated cAMP led us to investigate whether the second messenger also had a similar effect on the thrombin-induced inhibition of intercellular Ca²⁺ wave propagation. Moreover, we wanted to investigate the differences, if any, in the effect of the second messenger on PIC and GJIC. Our results show that adenosine, which increases cAMP through activation of G_αs-coupled A2B receptors in BCECs, ⁴⁹ prevents the effect of thrombin on both GJIC and PIC. Therefore, our findings appear to implicate the contractility and organization of the cortical actin cytoskeleton in the regulation of the activity of hemichannels as well as that of gap junctions. 

Materials and Methods

Cell Culture

Primary cultures of BCECs were established as previously described. ^{10 27 28 39 44 50} The growth medium contained Dulbecco’s modified Eagle’s medium (DMEM; 11960-044; Invitrogen-Gibco, Karlsruhe, Germany) and 10% fetal bovine serum (F 7524; Sigma-Aldrich, Deisenhofen, Germany), 6.6% l-glutamine (Glutamax; 35050-038; Invitrogen-Gibco), and 1% antibiotic-antimycotic mixture (15240-096; Invitrogen-Gibco). Cells were grown at 37°C in a humidified atmosphere containing 5% CO₂. Cells of the first, second, and third passages were harvested and seeded into two chambered glass slides (155380, Laboratory-Tek; Nunc, Roskilde, Denmark) at a density of 165,000 cells per chamber (4.2 cm²). Cells were allowed to grow to confluence for 3 to 4 days before use. 

Fluorescence Recovery after Photobleaching

We have described the FRAP (fluorescence recovery after photobleaching) protocol in BCECs. ¹⁰ Cells were loaded with the Ca²⁺-insensitive dye 6-carboxyfluorescein diacetate (10 μM) in Dulbecco’s PBS (Invitrogen-Gibco) for 5 minutes at room temperature, and fluorescence recovery after photobleaching (FRAP) was measured with a confocal laser scanning fluorescence microscope (LSM510; Carl Zeiss Meditec, Inc., Jena, Germany). A single cell was bleached by exposure to laser light at 488 nm for 50 scans at 95% intensity. The recovery of fluorescence in the bleached cells was measured every 10 seconds over a period of 5 minutes by excitation at 488 nm with emission recorded at 570 nm. The decrease of fluorescence in a region widely distant from the bleached cells was also measured as a reference for correction for bleaching due to the excitation light used for fluorescence detection during fluorescence recovery. Fluorescence recovery in the bleached cell at 3 minutes was compared with that of the prebleach scan, and the percentage recovery was calculated. 

Mechanical Stimulation for Inducing Ca²⁺ Wave

Point mechanical stimulation of a single cell was performed as described previously. ^{10 27 28 39} The stimulation consisted of an acute deformation of the cell by briefly touching <1% of the cell membrane with a glass micropipette (tip diameter <1 μm) coupled to a piezoelectric crystal (Piezo Flexure NanoPositioner P-280, Amplifier-E463; PI Polytech, Karlsruhe, Germany) mounted on a micromanipulator. 

Measurement of [Ca²⁺]_i

The Ca²⁺ wave propagation was assayed by imaging [Ca²⁺]_i as described previously. ^{10 27 28 39} Briefly, the cells were loaded with the Ca²⁺-sensitive dye Fluo-4 AM (10 μM) for 30 minutes at 37°C. The dye was excited at 488 nm, and its fluorescence emission was collected at 530 nm. Spatial changes in [Ca²⁺]_i after point mechanical stimulation were measured with the confocal microscope (LSM510; Carl Zeiss Meditec, Inc.) with a 40× objective. Polygonal regions of interest (ROIs) were drawn to define the borders of each cell. The neighboring cells (NB cells) immediately surrounding the mechanically stimulated (MS) cell are defined as neighboring cell layer 1 (NB1), and the ones immediately surrounding the NB1 cells are defined as neighboring cell layer 2 (NB2), and so on. Fluorescence was averaged over the area of each ROI. Normalized fluorescence (NF) was then obtained by dividing the fluorescence by the average fluorescence before mechanical stimulation. Intercellular propagation of the Ca²⁺ wave was characterized by maximum normalized fluorescence (NF), and percentage of responsive cells (%RC), as well as the total surface area of responsive cells (active area [AA]) with NF ≥ 1.1. 

Measurement of ATP Release

ATP release, after mechanical stimulation on the confocal microscope, was followed using the luciferin–luciferase bioluminescence protocol. ^{10 28 39} One hundred microliters was sampled from the 500 μL bathing solution covering the cells and taken to a custom-built photon-counting setup to measure the luminescence as described previously. ^{10 27 28 39 51} Briefly, photons emitted as a result of the oxidation of luciferin in the presence of ATP and O₂, and catalyzed by luciferase were detected by a photon-counting photomultiplier tube (H7360–01; Hamamatsu Photonics, Hamamatsu City, Japan). Voltage pulses from the photomultiplier module were counted with a high-speed counter (PCI-6602; National Instruments, Austin, TX). Dark count of the photomultiplier tube was <80 counts/s. 

Lucifer Yellow Uptake Assay

The protocol for Lucifer yellow (LY) uptake has been described in our previous studies. ^{10 27 28 39} Cells grown to confluence in chambered slides were incubated in Ca²⁺-rich PBS containing the drug of interest for 30 minutes. Cells were then exposed to PBS containing the 2 mM EGTA and 2.5% LY for 5 minutes in the continued presence of the drug. After a wash with Ca²⁺-containing PBS, cellular fluorescence was recorded with the laser scanning confocal microscope (LSM510; Carl Zeiss Meditec, Inc.) by excitation at 488 nm with emission recorded at 530 nm. 

Chemicals

Fluo-4 AM (F14217) and 6 carboxyfluorescein diacetate (C1362) were obtained from Invitrogen-Molecular Probes (Eugene, OR). ATP, forskolin (F-6886), adenosine (A-9251), thrombin (T-4648), CPA (N6-cyclopentyladenosine; C-8031), NECA (5′-(N-Ethyl-carboxamido)adenosine); E-2387), rolipram (R-6520), and LY (L-0259) were obtained from Sigma-Aldrich. TRAP-6 was obtained from Bachem (Torrance, CA). 

Gap26 (VCYDKSFPISHVR) peptide was synthesized at the Laboratory of Biochemistry, KULeuven. The peptide was analyzed by reversed-phase HPLC (Waters Corp., Milford, MA), on a C18-column (Luna 5u, 250 × 4.60 mm; Phenomenex, Torrance, CA), with a linear gradient of acetonitrile–water, containing 0.06% trifluoracetic acid (TFA). The exact sequence of the peptide was confirmed by ESI-triple quadrupole mass spectrometry (API-3000 mass spectrometer PE-SCIEX; Applied Biosystems, Nieuwerkerk aan den Ijssel, The Netherlands). The purity of the peptide was greater than 95%. 

Data Analysis

One-way ANOVA was used to compare mean values for different treatments, whereas unpaired tests were used to compare results of experiments with a single treatment and a single control (Prism 4.0 for Windows; GraphPad Software Inc., San Diego, CA). P < 0.05 was considered statistically significant. Histograms are expressed as mean ± SEM. N indicates the number of independent experiments (the number of MS cells), and n represents the total number of responsive cells. 

Results

Effect of Adenosine on Thrombin-Induced Inhibition of the Ca²⁺ Wave

Figure 1shows a series of typical [Ca²⁺]_i-sensitive fluorescence images representing Ca²⁺ wave propagation in response to a point mechanical stimulus in a control condition, in the presence of thrombin, and in the presence of thrombin after preincubation with adenosine. The line graphs (to the right of the fluorescence images) show the time course of the Ca²⁺ transients (represented as normalized fluorescence [NF]) in the mechanically stimulated (MS) cell and in the neighboring (NB) cell layers. In control conditions, as we have reported previously, ^{10 27 39} the MS cell shows a transient [Ca²⁺]_i increase, which originates at the point of stimulation and spreads out to the neighboring cell layers (Fig. 1A) . Ca²⁺ transients are observed four to six cell layers away from the MS cell. The corresponding line graphs of the Ca²⁺ transients show that the peak normalized fluorescence decreased with increasing distance from the MS cell. As demonstrated in our previous study, ³⁹ cells subjected to mechanical stimulation after incubation with thrombin (2 U/mL for 5 minutes) also show a Ca²⁺ wave, but the spread of the wave was limited to only about two to four neighboring cell layers (Fig. 1B) . The inhibition of the Ca²⁺ wave propagation by thrombin is overcome by pretreatment with adenosine (200 μM for 30 minutes), as the wave is shown to spread to four to six neighboring cell layers (Fig. 1C) , similar to control conditions (Fig. 1A) . 

Table 1provides a quantitative summary of experiments similar to those shown in Figure 1 . Compared with control conditions, maximum normalized fluorescence after mechanical stimulation in cells treated with thrombin is significantly smaller in all cell layers. However, pretreatment with adenosine prevents the reduction in the value of the normalized fluorescence by thrombin. The maximum normalized fluorescence in the different neighboring cell layers in the presence of adenosine plus thrombin is not significantly different from the value in the presence of adenosine alone. The percentage of responsive cells in neighboring layers is smaller in thrombin-treated cells than in corresponding cell layers in untreated cells. Pretreatment with adenosine also prevented this thrombin-induced reduction of percentage responsive neighboring cells. 

The effect of thrombin and its prevention by adenosine can also be visualized in terms of the area of Ca²⁺ wave propagation (AA). Figure 2Aand Table 1show the summary of experiments in terms of the AA. The AA is reduced by ∼70% from 45,500 ± 1,700 μm² in untreated cells to 13,800 ± 1,000 μm² (P < 0.001; N = 145) in cells exposed to thrombin (Fig. 2A) . Pretreatment with adenosine overcomes the thrombin-induced reduction, resulting in an AA of 38,700 ± 2,000 μm² (P < 0.001 for adenosine plus thrombin versus thrombin alone; N = 145). The AA in the presence of adenosine plus thrombin is not significantly different from the value in the presence of adenosine alone or from the control value. Similar effects of adenosine were obtained when the PAR-1 selective agonist TRAP-6 was used to inhibit the Ca²⁺ wave (N = 25; Fig. 2B ). Thus, results shown in Figures 1 and 2and Table 1demonstrate that adenosine overcomes the inhibitory effect of thrombin on the intercellular propagation of Ca²⁺ waves in BCECs that is mediated by activation of PAR-1 receptors. 

Mechanism of Adenosine Response

A previous study on BCECs has shown expression of A1- and A2B-receptors by RT-PCR and immunocytochemistry. ⁴⁹ To investigate the identity of the receptors involved in the adenosine-mediated suppression of the effect of thrombin on the Ca²⁺ wave, we used adenosine receptor agonists. As shown in Figure 3 , in cells pretreated with CPA (selective A1 receptor agonist; 100 μM for 30 minutes), thrombin reduced the AA from 35,000 ± 3,700 (N = 25) under control conditions to 17,000 ± 1,900 (P < 0.05; N = 25; Fig. 3 ). It can also be noted that CPA itself had no significant effect on the AA compared with control conditions (N = 25; Fig. 3 ). These experiments suggest that the effect of adenosine on the suppression of thrombin-mediated inhibition of the Ca²⁺ wave is not through A1 receptors. In contrast to CPA, NECA is known to enhance cAMP significantly in corneal endothelial cells, ⁴⁹ consistent with its potent activation of A2B receptors, known to be coupled to adenylate cyclase via G_αS. Pretreatment with NECA (100 μM for 30 minutes) overcame the reduction of the AA of the Ca²⁺ wave by thrombin (P < 0.05; N = 25). Specifically, the AA in the presence of the combination of NECA and thrombin did not differ significantly from the AA in the presence of NECA alone (31,100 ± 5,100 in NECA plus thrombin vs. 33,900 ± 4,700 in NECA; N = 25; Fig. 3 ). 

To further establish that the inhibition of the effect of thrombin by adenosine is mediated through cAMP, we used forskolin to activate adenylate cyclase directly and also used the cAMP-dependent phosphodiesterase inhibitor, rolipram (selective for the PDE4 subtype). Pretreatment with forskolin (10 μM for 2 minutes) or the combination of forskolin (10 μM for 2 minutes) and rolipram (200 μM for 30 minutes) significantly inhibited the effect of thrombin when compared with thrombin alone (AA 23,100 ± 4,200 in forskolin plus thrombin and 38,000 ± 6,500 in forskolin plus rolipram plus thrombin versus 9,700 ± 1,800 in thrombin; N = 25; P < 0.005 and P < 0.001, respectively; Fig. 4 ). 

Effect of Adenosine on PIC

As shown in our previous studies, corneal endothelial cells take up LY in Ca²⁺-free medium containing 2 mM EGTA. ¹⁰ The uptake of LY was shown to be inhibited in the presence of thrombin. ³⁹ Figure 5shows the effect of thrombin and TRAP-6 on LY uptake after pretreatment with adenosine. As a control, Figure 5Ashows the uptake of LY in control conditions, and the absence of dye uptake in the presence of thrombin (2 U/mL for 5 minutes) or TRAP-6 (10 μM for 30 minutes). As shown in Figure 5B , pretreatment with adenosine (200 μM for 30 minutes) prevents such an influence of thrombin or TRAP-6 on LY uptake. Since it has also been shown that the LY uptake is attributed to hemichannels, ¹⁰ and that the effect of thrombin is mainly due to inhibition of ATP release via hemichannels, we investigated whether adenosine is able to enhance the Ca²⁺ wave propagation in the presence of a PAR-1 agonist in conditions where hemichannels are blocked by Gap26. Figure 5Cshows that LY uptake is blocked when the cells are pretreated with Gap26, which is known to block hemichannels, and that application of thrombin or TRAP-6 did not alter the LY uptake. Figure 5Ddemonstrates that adenosine fails to prevent the inhibition of dye uptake by PAR-1 agonists when Gap26 is present. These experiments demonstrate that connexin hemichannels are involved in the effect of adenosine. Therefore these experiments provide further evidence that the inhibition of PIC by thrombin and TRAP-6 is due to the inhibition of hemichannels. 

Intercellular Ca²⁺ wave propagation induced by point mechanical stimulation in BCECs is through ATP release via hemichannels, ¹⁰ and we demonstrated in our previous study that such an ATP release is inhibited by thrombin. ³⁹ We investigated the effect of adenosine on the effect of thrombin in terms of ATP release. The experiments showed that pretreatment of the cells with adenosine (200 μM for 30 minutes) prevented the inhibition of ATP release by thrombin (2 U/mL for 5 minutes), confirming that adenosine restores PIC. The median reduction of ATP release by thrombin was 58% versus control, whereas it was only 23% after pretreatment with adenosine (N = 6). Similar results were obtained with NECA (N = 3) and forskolin (N = 6), whereas CPA did not affect the inhibition caused by thrombin (N = 3). 

Effect of Adenosine on GJIC

Experiments with the FRAP protocol indicate functional GJIC, and we had shown that GJIC is partly inhibited by thrombin. ³⁹ Figure 6shows FRAP experiments to investigate the effect of adenosine on the thrombin-induced inhibition of GJIC. In control conditions, fluorescence recovery in the bleached cells reached ∼67% within 3 minutes after photobleaching. In the presence of thrombin, the fluorescence recovery after 3 minutes was reduced to ∼59%, whereas the recovery at 3 minutes was ∼73% after pretreatment with adenosine (P < 0.001; N = 70). Adenosine by itself did not significantly affect the fluorescence recovery after photobleaching (N = 70). These experiments demonstrate that the partial inhibition of GJIC by thrombin is prevented by pretreatment with adenosine. 

Discussion

Exposure of BCECs to thrombin is known to induce MLC phosphorylation through activation of the MLCK, PKC, and RhoA-Rho kinase axis. This results in a disruption of the cortical actin cytoskeleton and appearance of interendothelial gaps. ^{40 42} The enhanced contractility of the cortical actin causes a breakdown of the barrier integrity that is dependent on the tight junctions. ^{40 42} Since activated PKA phosphorylates RhoA at Ser-188, resulting in inhibition of RhoA activation, ^{52 53} elevated cAMP causes MLC dephosphorylation. ^{54 55} Accordingly, it has been demonstrated that the thrombin-induced loss of the endothelial barrier’s integrity can be prevented by pretreatment with adenosine, ^{41 43 44} which is known to activate A2B receptors and elevate cAMP in BCECs. ⁴⁹ In a recent study, ³⁹ we demonstrated that thrombin inhibits the intercellular Ca²⁺ wave propagation through mechanisms also dependent on MLC phosphorylation. In the present study, we investigated whether adenosine could suppress the inhibitory effect of thrombin on intercellular Ca²⁺ wave propagation. The major findings are that (1) pretreatment with adenosine prevents the inhibition of Ca²⁺ wave propagation by thrombin and the purinergic agonist also has a significant effect by itself, (2) adenosine decreases the inhibitory effect of thrombin on dye uptake and ATP release, and (3) the masking of the effect of thrombin by adenosine is mediated via elevated cAMP. These findings, therefore, provide additional evidence for our conclusion from previous studies ³⁹ that connexin hemichannels are involved in PIC and that their opening is inhibited by MLC phosphorylation. 

Interference of Adenosine with PAR-1 Signaling

The findings shown in Figures 1 and 2 , taken together, clearly demonstrate that pretreatment with adenosine prevents the reduction of Ca²⁺ wave propagation by thrombin, whereas adenosine by itself has no significant effect on the AA (Fig. 2) . In our previous study, ³⁹ we showed that thrombin exerts its effect on the Ca²⁺ wave propagation via PAR-1 receptors. Therefore, the present experiments demonstrate that the masking of the effect of thrombin on the spread of the Ca²⁺ wave by adenosine is due to interference with the thrombin-mediated signaling via PAR-1 receptors. In accordance with this argument, adenosine is also effective in masking the effect of TRAP-6, a selective PAR-1 agonist, on the Ca²⁺ wave propagation (Figs. 2 5) . Since we have also shown that the effect of PAR-1 activation on Ca²⁺ wave propagation is RhoA dependent ³⁹ and since adenosine overcomes the effect of PAR-1 activation without having a significant effect by itself, we suggest that adenosine interferes with RhoA-dependent PAR-1 signaling. 

Our data also indicate that the masking of the effect of thrombin on the spread of the Ca²⁺ wave by adenosine is specifically due to elevated cAMP. First of all, we demonstrated that the effect of adenosine is through A2B receptors and not through A1 receptors, since CPA did not have any effect on the wave propagation (Fig. 3) . Second, we demonstrated that forskolin also masks the effect of thrombin on the wave propagation and more so in the presence of rolipram, a PDE4-selective inhibitor (Fig. 4) . These findings are also consistent with the response obtained with NECA (Fig. 3) , a potent activator of A2B receptors, which is also known to elevate cAMP in the corneal endothelial cells. ⁴⁹  

Locus of the Effect of Thrombin

Effect of Adenosine on Gap Junctions.

Experiments with FRAP demonstrated that pretreatment with adenosine opposes the inhibition of GJIC by thrombin (Fig. 6) . The disruption of barrier integrity is attributed to a loss of the tethering forces, ^{45 46 47 48} which are essential to stabilize the interactions of the transmembrane proteins of tight junctions. In a similar vein, we can suggest that GJIC could be affected by a loss of the tethering forces, which may be a prerequisite for formation and maintenance of gap junctions, ² by enabling docking of hemichannels from adjacent cells. ^{56 57}  

Effect of Adenosine on the Activity of Hemichannels.

Despite the similarity between the loss of GJIC and barrier integrity, the profound reduction in the Ca²⁺ wave propagation by thrombin is largely due to the inhibition of PIC, which is mediated by ATP release via connexin hemichannels. ^{10 28} We showed that adenosine is able to suppress the inhibition of ATP release by thrombin. Pretreatment with adenosine also opposed the inhibitory effect of thrombin on LY dye uptake in Ca²⁺-free solution containing 2 mM EGTA (Fig. 5) , which was shown to be mediated via hemichannels. ¹⁰  

Inhibition of LY uptake by the hemichannel blocker Gap26 was not masked by adenosine. This excludes that adenosine opens additional pathways for LY uptake. Moreover, adenosine also failed to reverse the inhibition of LY uptake when cells were treated with both thrombin and Gap26. These findings, taken together, suggest that the effect of adenosine on the thrombin-induced loss of LY uptake is solely on mechanisms involving the activity of hemichannels. Therefore, our experiments demonstrate that adenosine overcomes the effect of thrombin on PIC, which is mediated by ATP release via connexin hemichannels. ^{10 28}  

In contrast to gap junctions, hemichannels are unlikely to be affected by tethering forces, but could be influenced by altered organization of the cortical cytoskeleton or the disposition of scaffolding proteins, contraction, and/or protein–protein interactions. In support of this argument, a recent study demonstrated that increased MLC phosphorylation led to a significant redistribution of certain proteins (e.g., ZO-1) of the junctional protein complex leading to a breakdown of the barrier integrity. ⁵⁸ Since ZO-1 is also known to interact with Cx43, ^{17 20} a redistribution of this component of the tight junction complex or other connexin-interacting proteins could significantly affect trafficking, gating, and internalization of connexin hemichannels, either directly or indirectly through effects on the linker proteins which couple actin cytoskeleton and connexins. 

Interference of Thrombin Signaling by Adenosine

Based on the data just discussed, we suggest that the masking of the thrombin-induced inhibition of the Ca²⁺ wave by adenosine is due to interference of cell signaling at RhoA, leading to inhibition of its activation. Figure 7extends the transduction cascade proposed in our previous studies ^{39 41 43 44} by including our findings on the inhibition of the thrombin effect by adenosine. Activation of RhoA has two known effects: inhibition of myosin light chain phosphatase (MLCP), ^{40 42} and phosphorylation of LIM1 kinase (LIMK1). ⁵⁹ Phosphorylated LIMK1 phosphorylates the actin depolymerizing factor cofilin, causing its inactivation. ^{59 60} Therefore, by inactivating RhoA, adenosine could induce MLC dephosphorylation and/or actin depolymerization. ^{41 43 44} In BCECs, as we reported in an earlier study, MLC dephosphorylation occurs in response to adenosine with and without thrombin. ^{43 44} However, adenosine in the presence of thrombin is not known to induce actin depolymerization. ^{43 44} Thus, if we can preclude the effects through LIMK1, the effect of thrombin could be attributed to the contraction induced by MLC phosphorylation. Since the inhibitory effect of thrombin on PIC is too rapid to involve actin remodeling, we suggest that the effect of the actin cytoskeleton on hemichannels may be due to an effect on their gating. This is supported by the finding that certain connexins form hemichannels that are mechanosensitive. ^{61 62}  

Physiological Relevance to the Corneal Endothelium

Our observations on the effect of adenosine add yet another mechanism to the growing repertoire of defense mechanisms by which adenosine may protect and enhance the physiological functions of corneal endothelium. Thus, adenosine is known to enhance fluid transport, ⁶³ activate cAMP-activated Cl⁻ channels, ^{49 64} and rescue the loss of barrier integrity secondary to MLC phosphorylation. ^{43 44} Despite the focus of this study on the thrombin-induced inhibition of the intercellular communication, our observation is general, in that the target molecule of adenosine in rescuing the intercellular communication was found to be MLC. Since proinflammatory molecules can cause MLC phosphorylation, it is plausible that by inducing MLC dephosphorylation adenosine could be beneficial in overcoming the potential threat to intercellular communication concomitant with its ability in rescuing the barrier integrity. Furthermore, our finding that GJIC is enhanced in response to cAMP is supported by a previous finding that bicarbonate, which activates soluble adenylate cyclase ⁶⁵ enhances dye coupling in the corneal endothelium. ⁶⁶ In contrast to the effects on GJIC, our finding that adenosine also overcomes the thrombin-induced inhibition of hemichannels is paradoxical, but PIC could be helpful if cells are not apposed to each other. 

In summary, our results show that adenosine, which increases cAMP through activation of G_αs-coupled A2B receptors in BCECs, ⁴⁹ prevents the effect of thrombin on intercellular Ca²⁺ wave propagation involving GJIC and PIC. In addition, our findings suggest possible involvement of the contractility and organization of the cortical actin cytoskeleton in the regulation of the activity of hemichannels as well as that of gap junctions. 

 Supported by National Eye Institute Grant EY14415 (SPS) and by a Faculty Research Support Program award from the Office of the Vice Provost for Research at Indiana University (SPS); FWO-Vlaanderen Grant G.0218.03, GOA/2004/07; and IAP Program 5/05 (BH, JV).

 Submitted for publication September 7, 2006; revised November 8, 2006; accepted January 17, 2007.

 Disclosure: C. D’hondt, None; S.P. Srinivas, None; J. Vereecke, None; B. Himpens, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Effects of thrombin on Ca²⁺ wave propagation. Left: Representative pseudocolored fluorescence images showing Ca²⁺ transients at different times after mechanical stimulation of BCECs. Right: time course of normalized fluorescence (NF) in the mechanically stimulated (MS) cells and an average NF in neighboring (NB) cell layers NB1 to NB5. The first image shows the fluorescence intensities before stimulation. Arrow: MS cell. (A) Control conditions: A Ca²⁺ wave propagated to five neighboring cell layers. The total area of cells reached by the wave (AA) was 59,000 μm². (B) Effects of thrombin: Thrombin (2 U/mL for 5 minutes) reduces the spread of the wave to three layers and the AA to 16,500 μm². (C) Effect of thrombin after pretreatment with adenosine (200 μM for 30 minutes): The Ca²⁺ wave propagates to four neighboring cell layers. The AA is 55,300 μm².

Table 1.

 

Quantitation of Fluorescence in Micrographs in Each of the Treatment Groups

	MS	NB1	NB2	NB3	NB4	NB5	AA (μm2)
Control
NF ± SEM	2.58 ± 0.07	3.05 ± 0.04	2.53 ± 0.03	2.07 ± 0.02	1.73 ± 0.02	1.49 ± 0.02	45,490 ± 1,674
% RC	100	99	93	75	48	34
n	75	507	914	1129	884	593	145
Thrombin
NF ± SEM	2.23 ± 0.07*	2.29 ± 0.05*	1.88 ± 0.04*	1.53 ± 0.03*	1.37 ± 0.03*	0.00 ± 0.00*	13,773 ± 993*
% RC	100	81	57	29	17	0
n	41	215	311	243	148	0	145
Adenosine
NF ± SEM	2.57 ± 0.07	2.54 ± 0.04*	2.24 ± 0.03*	1.89 ± 0.02*	1.69 ± 0.02	1.56 ± 0.03*	40,844 ± 1,879
% RC	100	98	82	59	41	31
n	75	457	808	841	712	495	145
Adenosine and Thrombin
NF ± SEM	2.68 ± 0.08,	2.62 ± 0.04* ,	2.20 ± 0.03* ,	1.89 ± 0.02* ,	1.65 ± 0.02* ,	1.60 ± 0.02* ,	38,738 ± 2,026* ,
% RC	100	99	85	64	45	37
n	73	478	831	908	769	540	145

 Average maximum normalized fluorescence (NF), percentage responsive cells (% RC) and active area (AA) in the MS and NB layers during mechanical stimulation in control conditions and after treatment with thrombin, adenosine, and adenosine plus thrombin.

*   P < 0.05 versus control.

  P < 0.05 versus thrombin.

Figure 2.

 

Effect of PAR-1 receptor agonists on the AA of the Ca²⁺ wave in cells pretreated with adenosine. (A) Effect of pretreatment with adenosine (200 μM for 30 minutes) on the reduction in AA by thrombin (2 U/mL for 5 minutes; N = 145). (B) Similar results were obtained with TRAP-6 (10 μM for 30 minutes; N = 25). *P < 0.001 for comparison between AA in the presence versus absence of the PAR-1 agonist (i.e., comparison of ▪ or with □ in each condition). P < 0.001 for comparison between AA in the presence versus absence of adenosine (i.e., comparison of filled (▪ respective to ) bar with corresponding filled bar in control condition.

Figure 3.

 

Effect of thrombin on the AA of the Ca²⁺ wave in cells pretreated with adenosine receptor agonists. Effect of thrombin (2 U/mL for 5 minutes) on AA in cells pretreated with adenosine (200 μM for 30 minutes), the A1 agonist CPA (100 μM for 30 minutes), or the potent A2B agonist NECA (100 μM for 30 minutes; N = 25). *P < 0.001 for comparison between AA in the presence versus absence of thrombin (i.e., comparison of ▪ with □ in each condition). P < 0.001 for comparison between AA in the presence versus absence of the adenosine receptor agonists (i.e., comparison of ▪ in each condition with ▪ in the control condition).

Figure 4.

 

Effect of thrombin on the AA of the Ca²⁺ wave in cells pretreated with forskolin and rolipram. Effect of thrombin (2 U/mL for 5 minutes) on AA in cells pretreated with forskolin (10 μM for 30 minutes) and the phosphodiesterase inhibitor rolipram (selective for PDE4 subtype; 200 μM for 30 minutes; N = 25). *P < 0.001 for comparison between AA in the presence versus absence of thrombin (i.e., comparison of ▪ with □ in each condition). P < 0.001 for comparison between AA in the presence versus absence of forskolin or forskolin plus rolipram (i.e., comparison of ▪ in each condition with ▪ in control condition).

Figure 5.

 

Effect of thrombin, TRAP-6 and adenosine on LY uptake. Cells were exposed to the fluorescent dye LY (2.5% for 5 minutes) in Ca²⁺-free solution containing 2 mM EGTA. (A) Uptake of LY in the control condition and in the presence of thrombin (2 U/mL for 5 minutes) or TRAP-6 (10 μM for 30 minutes; N = 10). (B) Adenosine inhibits the effect of thrombin and TRAP-6 on dye uptake (N = 7). (C) Gap26 inhibits LY uptake in the presence and absence of thrombin or TRAP-6 (N = 4). (D) Adenosine is not able to prevent the effect of thrombin or TRAP-6 on LY uptake after pretreatment with Gap26 (N = 4).

Figure 6.

 

Gap junctional communication analysis by FRAP. Cells were loaded with carboxyfluorescein. The dye of a single cell was bleached by an intense laser beam, and images were taken at intervals of 10 seconds to study the recovery of the fluorescence. The recovery of the fluorescence after photobleaching (corrected for background bleaching) of a single cell is plotted as a function of time after bleaching. Effects of thrombin (2 U/mL for 5 minutes) in the presence or absence of adenosine (200 μM for 30 minutes). Three minutes after bleaching, a recovery of 67% ± 1.36% is noticed in control conditions, whereas in the presence of thrombin, 59% ± 1.66% of the fluorescence is recovered (P < 0.001, N = 70). The presence of adenosine caused a significant enhancement of the percentage of recovery after 3 minutes (P < 0.001).

Figure 7.

 

Postulated mechanisms underlying adenosine-mediated suppression of the effects of thrombin on GJIC and PIC. In corneal endothelial cells, thrombin activates PAR-1 receptors leading to activation of G_αq/11 and G_α12/13 proteins. The former activates MLCK, which drives MLC phosphorylation. Activation of G_α12/13 results in RhoA-Rho kinase signaling. Rho kinase inhibits MLC phosphatase (MLCP). The thrombin-induced signaling thus enhances MLC phosphorylation and inhibits GJIC and PIC, as shown in our previous study, ³⁹ presumably as a consequence of increased contractility of the actin cytoskeleton and reorganization of the cortical actin. ^{41 43 44} Adenosine activates A2B receptors, resulting in elevated cAMP. Since elevated cAMP blocks the activation of RhoA, adenosine interferes with thrombin signaling and thereby prevents the effects on GJIC and PIC by the protease.

The authors thank Raf Ponsaerts for critical discussions and Wendy Janssens for technical assistance and help with the experiments.