Amelioration of Inflammation and Cytotoxicity by Dipyridamole in Brain Endothelial Cells

Dipyridamole, as an antiplatelet agent, is known to inhibit adenosine transporter and phosphodiesterase-5; thus, increasing extracellular levels of adenosine and intracellular levels of cAMP and cGMP. These well-documented actions of dipyridamole provide a major clinical strategy for preventing recurrent vascular events of noncardioembolic etiology in patients with stroke or transient ischemic attack [1,2,3].

More recently, however, it has been suggested that dipyridamole may also posses an anti-inflammatory function [1,3]. In monocytes incubated with thrombin-stimulated platelets, dipyridamole but not aspirin attenuated the activation of NFĸB and blocked the expression of monocyte chemotactic protein-1 and matrix metalloproteinase-9 (MMP-9) [4]. In lipopolysaccharide-challenged macrophages, dipyridamole inhibited p38 MAP kinase, prevented NFĸB activation and downregulated cyclooxygenase-2 [5]. In neutrophils, dipyridamole blocked its adhesion to activated endothelial cells, in part by downregulation of macrophage-1 antigen (Mac-1) [6]. In human umbilical vein endothelial cells, dipyridamole attenuated TNFα- or thrombin-induced NFĸB activation and IĸBα phosphorylation [7]. Taken together, it is conceivable that these pleiotropic effects of dipyridamole may also be relevant in the cerebrovascular system, especially within the context of inflammation during and after stroke.

In the present study, we assessed the effects of dipyridamole using in vitro cultures of human brain endothelial cells subjected to various types of stress. Inflammatory stress was initiated with TNFα. Metabolic stress was triggered with oxygen-glucose deprivation (OGD). Our findings showed that dipyridamole partly ameliorated the elevations in adhesion molecules and matrix metalloproteinases triggered by inflammatory stress, and moderately reduced cytotoxicity after metabolic stress.

A previously characterized human brain microvascular endothelial cell line was used [8]. Cells were seeded in 6-well plates at 60–70% confluence and kept at 37°C in 5% CO₂. After 6 h serum starvation, cells were exposed to TNFα (R&D Systems) with or without dipyridamole (Boehringer Ingelheim) at the same time for 20 h, then cell media were collected for gelatin zymography and cells were used for protein assay or viability assay. Tartaric acid (2 mM) was used as vehicle control for dipyridamole.

To induce OGD, cells were changed to glucose-free DMEM medium, then subjected to hypoxia with a modular chamber (Billups-Rothenberg) and perfused (90% N₂, 5% H₂, 5% CO₂) for 30 min [9,10]. The chamber was then sealed and incubated at 37°C. After OGD for 6 h, the cells were returned to a regular incubator with serum-free RPMI-1640 medium for reperfusion over 18 h with or without dipyridamole. Cell media were collected for gelatin zymography and the cells were used for the viability assay.

Cell lysate was collected with cell lysis buffer (Cell Signaling) after the cells were rinsed twice with ice-cold PBS. Each sample was loaded onto 10% Tris-glycine gels, 5 µg protein for ICAM-1 and 20 µg protein for VCAM-1 and PECAM-1. After electrophoresis and transference to PVDF membranes, the membranes were blocked with 5% non-fat dry milk in Tris-buffered saline for 1 h at room temperature, then incubated overnight at 4°C with primary antibody anti-ICAM-1 (Santa Cruz), antibody anti-VCAM-1 (Chemicon) or antibody anti-PECAM-1 (Abcam). Peroxidase-conjugated secondary antibodies were used for visualization with enhanced chemiluminescence (Pierce). The amount of β-actin was used as the loading control. The X-ray films were scanned and the optical density of each band was analyzed with Image J. Note that we only focused on cell-membrane-associated adhesion molecules. Hence, our adhesion molecule blots were performed on cell lysates.

Conditioned media were collected and centrifuged at 10,000 g for 2 min to discard cell debris; the supernatants were concentrated by Vivaspin 500 with a 10-KDa cutoff (Sartorius Stedim Biotech). Gelatin zymography was performed as previously described [10,11]. Briefly, media samples were mixed with SDS (sodium dodecyl sulfate) sample buffer (Invitrogen) and electrophoresed on 10% SDS-polyacrylamide gel containing 1 mg/ml gelatin as the protease substrate (Invitrogen). After electrophoresis, the gels were renatured in Triton X-100 buffer (Novex) for 1 h and then incubated for 48 h at 37°C in developing buffer (Novex). After developing, the gels were stained with Coomassie Brilliant Blue and photographed. The gelatinase zymogram standards for human MMP-2 and MMP-9 (Chemicon) were loaded as positive controls. MMPs were quantified as fold increase compared with normal controls, as measured by optical density. Note that unlike the cellular responses in adhesion molecules, MMP responses to TNF or OGD are manifested as a release of proteases into extracellular space. Hence, MMP responses are measured in conditioned media.

Cell viability was determined by standard MTT assay that measures mitochondrial metabolic activity [12]. Endothelial cytotoxicity was calculated as a percentage of untreated controls, i.e. 100 minus percent cell viability. Percent cell death was also confirmed with a standard LDH release assay.

All of experiments were performed in duplicate or triplicate, and repeated 4–6 times independently. Quantitative data were expressed as mean ± SD, and analyzed with ANOVA followed by Dunnett’s t test. Differences of p < 0.05 were considered significant.

The first set of experiments tested the effects of dipyridamole against an inflammatory stressor. In our brain endothelial cells, the potent cytokine TNFα (12.5–50 ng/ml) clearly increased the protein levels of ICAM-1 and VCAM-1 in a concentration-dependent manner (fig. 1a). There were no detectable responses in PECAM-1 (fig. 1a). Co-treatment with dipyridamole markedly reduced TNFα-induced ICAM-1 expression; 12.5 ng/ml TNFα upregulated ICAM-1 levels by up to 5-fold and 5 µM dipyridamole decreased this response by almost 50% (fig. 1b, c). Dipyridamole also seemed to slightly affect TNFα-induced VCAM-1 responses, but these differences did not reach statistical significance (fig. 1d, e).

TNFα also significantly increased MMP-9 levels in the conditioned media of the brain endothelial cells (fig. 2a). There were no clear effects on MMP-2. At the highest concentration of TNFα tested (50 ng/ml), secreted MMP-9 levels almost reached 20-fold above baseline (fig. 2b, c). The effects of dipyridamole were tested using a 12.5 ng/ml concentration of TNFα. Co-treatment of dipyridamole (1–5 µM) greatly reduced TNFα-induced MMP-9 levels by up to 50% (fig. 2b, c). Again, there were no statistically significant changes on MMP-2 levels (fig. 2d). In all experiments, there were no detectable effects on endothelial cell viability (fig. 2e), suggesting that no direct cytotoxicity was involved in these MMP responses to TNFα stimulation.

In the second set of experiments, OGD was used as a trigger of metabolic stress. Brain endothelial cells were subjected to 6 h OGD followed by 18 h reperfusion/recovery. Injured cells were compared with cells treated with dipyridamole at the onset of reperfusion/recovery. This form of metabolic injury also induced a release of MMP-9 into the conditioned media (about 2-fold, fig. 3a, b). And this response was significantly attenuated by treatment with dipyridamole (fig. 3a, b). No statistically significant changes were noted in MMP-2 (fig. 3c). As expected, this metabolic injury with transient OGD resulted in moderate levels of endothelial cytotoxicity (fig. 3d, e). Treatment with dipyridamole significantly decreased cell death induced by OGD-reperfusion as measured by both MTT (fig. 3d) and LDH assays (fig. 3e).

Dipyridamole is widely used as an antiplatelet agent in patients at risk of cardiovascular disease and stroke [1,2,3]. However, emerging data now suggest that dipyridamole may also have important anti-inflammatory actions. In our proof-of-principle study, we showed that dipyridamole significantly ameliorated ICAM-1 and VCAM-1 expression, MMP-9 upregulation and cytotoxicity in brain endothelial cells subjected to inflammatory and metabolic stress. In the context of stroke and cerebrovascular disease, these cell culture data raise the interesting hypothesis that dipyridamole may have neurovascular effects that influence clinical outcomes in these patients.

Ischemia and reperfusion induce inflammatory responses in endothelial cells, in part via TNFα [13,14]. Downstream of TNFα, many proinflammatory genes can be activated, including those encoding adhesion molecules and neurovascular proteases from MMP family [15,16]. In patients with ischemic heart disease, administration of dipyridamole results in significantly increased plasma levels of ICAM-1 and decreased plasma levels of VCAM-1 [17]. Our findings suggested that in brain endothelial cells, dipyridamole may downregulate ICAM-1 and VCAM-1, but how these cell surface responses may relate to shedding of antigens into the circulation is unclear. Our findings also demonstrate that dipyridamole can ameliorate TNFα-induced elevations in MMP-9. It is reported that dipyridamole can block the activation of NF-ĸB [7]. Since NFĸB is a major transcription factor for adhesion molecules and MMP-9 gene regulation, it is conceivable that this may represent one common mechanism for dipyridamole to reduce ICAM-1, VCAM-1 and MMP-9 expressions in endothelial cells.

In the context of stroke, inflammatory responses in brain endothelial cells are known to contribute to neurovascular pathophysiology [13,18,19,20,21,22]. ICAM-1 can mediate endothelial activation and transmigration of inflammatory cells [23]. MMP-9 can damage the blood-brain barrier, exacerbate edema and hemorrhage, and worsen neurovascular injury and cell death [14,24,25]. Gene knockout or pharmacologic inhibition of ICAM-1 and MMP-9 has been shown to decrease injury in various animal models of stroke [11,20,26,27,28,29,30]. In stroke patients, it has been proposed that MMP-9 levels in blood and brain may correlate with clinical outcomes. Hence, the ability of dipyridamole to attenuate inflammatory ICAM-1 and MMP-9 responses may hint at potential beneficial actions of this drug at the neurovascular level.

Beyond inflammation per se, stroke is obviously a potent metabolic stressor. In cell culture models, this is often mimicked by subjecting cells to OGD. MMP-9 mRNA or activity can be induced by OGD in brain endothelial cells [31,32]. In brain endothelial cells, MMP inhibition reduced hypoxic cell death [10]. Here, we found that dipyridamole reduced MMP-9 levels and attenuated cell death in brain endothelial cells. Is it possible that some form of endothelial protection may contribute to potential beneficial properties of dipyridamole in cerebral ischemia? A recent study demonstrated that subtherapeutic doses of dipyridamole combined with simvastatin were able to exert additive NO-dependent vascular effects on a mouse stroke model with increased cerebral blood flow, leading to decreased stroke volume and improved neurologic outcomes [33]. Further investigations appear warranted to truly dissect these neurovascular pathways of dipyridamole in the context of stroke and cerebrovascular disease.

Taken together, our data here provide initial proof-of-principle that – independent of its antiplatelet actions – dipyridamole can directly ameliorate brain endothelial cell reactions to inflammatory and metabolic stress. However, there are many caveats. First and foremost, our study only provides an initial description of these effects. Further studies are needed to investigate the molecular mechanisms underlying these endothelial responses. Second, our endpoints are limited. Will dipyridamole affect tight junction proteins or protect the blood-brain barrier? Will dipyridamole attenuate endothelial interactions with inflammatory cells? All these potential effects of dipyridamole warrant further investigation. Thirdly, our findings occur out of context insofar as in vivo brains are concerned. How will dipyridamole affect endothelial function when other cell types are present, including astrocytes, pericytes, neurons etc? The effects of dipyridamole on cell-cell homeostasis in the neurovascular unit will have to be carefully dissected before any of these initial single-cell observations can be properly interpreted.

Finally, a major caveat may arise in terms of how the present data relate to ongoing clinical trials. As early as 1996, the European Stroke Prevention Study 2 (ESPS-2) suggested that aspirin plus extended-release dipyridamole may have benefits over aspirin alone in preventing subsequent stroke [34]. In 2006, the European/Australian Stroke Prevention in Reversible Ischemia Trial (ESPRIT) concluded that combination of dipyridamole with aspirin showed additional risk reduction of stroke and transient ischemic attacks in patients compared with aspirin alone without increased bleeding [35]. Taken together these trials seemed to suggest that dipyridamole might have additional pleiotropic effects on vascular cells beyond antiplatelet actions per se [36]. However, outcomes from the recent PROFESS trial comparing dipyridamole to clopidogrel seemed inconsistent with these earlier trials [37]. How cell biology may be able to help sort out these differences remains to be determined. Correlating in vitro cell culture models with in vivo physiology is extremely difficult. We tried to use concentrations of dipyridamole (1–5 µM) that are close to plasma levels in patients, i.e. approximately 4 µM in patients taking Aggrenox (aspirin plus extended-release dipyridamole). However, in our experimental paradigms, single short-term treatments are the norm. How cell signaling responds to short-term versus long-term exposure to dipyridamole is not fully understood.

In conclusion, our cell culture study provided initial proof-of-principle that dipyridamole can directly ameliorate brain endothelial cell responses to inflammatory and metabolic stress. Our findings suggest that beyond its traditional antiplatelet actions, dipyridamole might possess interesting neurovascular effects in brain. Whether and how these cellular observations may relate to stroke and cardiovascular patients remains to be determined.

This study was supported in part by grants from the NIH, a Bugher award from the American Heart Association, and a grant from Boehringer-Ingelheim to S.G.

These experiments were funded in part by an academic research grant from Boehringer-Ingelheim to S.G. Boehringer-Ingelheim had no role in study design, data collection, data analysis, data interpretation, writing of the report or the decision to submit for publication.