NURR1 Involvement in Recombinant Tissue-Type Plasminogen Activator Treatment Complications After Ischemic Stroke
Background and Purpose—Despite the effectiveness of recombinant tissue-type plasminogen activator (r-tPA) during the acute phase of ischemic stroke, the therapy remains limited by a narrow time window and the occurrence of occasional vascular side effects, particularly symptomatic hemorrhages. Our aim was to investigate the mechanisms underlying the endothelial damage resulting from r-tPA treatment in ischemic-like conditions.
Methods—Microarray analyses were performed on cerebral endothelial cells submitted to r-tPA treatment during oxygen and glucose deprivation to identify novel biomarker candidates. Validation was then performed in vivo in a mouse model of thromboembolic stroke and culminated in an analysis in a clinical cohort of patients with ischemic stroke treated with thrombolysis.
Results—The transcription factor NURR1 (NR4A2) was identified as a downstream target induced by r-tPA during oxygen and glucose deprivation. Silencing NURR1 expression reversed the endothelial-toxicity induced by the combined stimuli, a protective effect attributable to reduced levels of proinflammatory mediators, such as nuclear factor-kappa-beta 2 (NF-κ-B2), interleukin 1 alpha (IL1α), intercellular adhesion molecule 1 (ICAM1), SMAD family member 3 (SMAD3), colony stimulating factor 2 (granulocyte-macrophage; CSF2). The detrimental effect of delayed thrombolysis, in conditions in which NURR1 gene expression was enhanced, was confirmed in the preclinical stroke model. Finally, we determined that patients with stroke who had a symptomatic hemorrhagic transformation after r-tPA treatment exhibited higher baseline serum NURR1 levels than did patients with an asymptomatic or absence of cerebral bleedings.
Conclusions—Our results suggest that NURR1 upregulation by r-tPA during ischemic stroke is associated with endothelial dysfunction and inflammation and the enhancement of hemorrhagic complications associated to thrombolysis.
Ischemic stroke is a leading cause of death and permanent disability in adults worldwide. Nevertheless, recombinant tissue-type plasminogen activator (r-tPA; Actilyse) is the only acute treatment currently available. When efficient in dissolving blood clots, r-tPA improves clinical outcomes in patients with ischemic stroke.1 Unfortunately, it is administered to <5% of affected patients, partly because of its narrow therapeutic time-window after symptom onset2 and the incidence of intracranial hemorrhage after thrombolysis.3 This fits with the noxious actions of r-tPA on the vascular/cerebral interface, which increase endothelial permeability, alter blood–brain barrier, and favor brain edema.4 If thrombolytic complications could be predicted, the number of patients who benefit from thrombolysis could be significantly increased.
To study the mechanisms underlying the endothelial damage resulting from r-tPA treatment, we performed microarray analyses using cultured human cerebral endothelial cells submitted to r-tPA treatment during oxygen and glucose deprivation (OGD). The nuclear receptor subfamily 4, group A, member 2 (NURR1; NR4A2) was selected from the validated genes as a candidate molecule potentially altered by r-tPA during ischemia. NURR1 is a nuclear hormone receptor, member of the orphan nuclear receptors, highly expressed in T-cells5 and brain tissue,6 mainly present in glial cells,7 neurons,8 and endothelial cells.9 In the central nervous system, NURR1 plays an important role in the transcriptional activation of tyrosine hydroxylase,10 which is essential for dopaminergic neuronal development.8 However, little is known about NURR1 involvement in ischemic stroke; it has been reported that NURR1 mRNA is upregulated in neurons after global cerebral ischemia in gerbils11 and permanent focal middle cerebral artery occlusion in rats,12 whereas NURR1 protein has been found to be downregulated in the basal ganglia after transient middle cerebral artery occlusion in mice.13
Materials and Methods
The cerebral human microvascular endothelial cell line (hCMEC/D3) was provided by Dr Couraud, Cochin Institute, France.14 Cells were placed in wells precoated with rat collagen I (1:50) (Cultrex; Gaithersburg, MD), grown in Endothelial Basal Medium (Lonza, Barcelona, Spain) supplemented with 1:2 of the provided factors; human fibroblast growth factor basic (hFGF-B), vascular endothelial growth factor (VEGF), human epidermal growth factor (hEGF), recombinant analog of insulin-like growth factor (R3-IGF-1), hydrocortisone, heparin, ascorbic acid, and 2% fetal bovine serum and maintained at 37°C in a humidified incubator containing 5% CO2 and atmospheric oxygen.
OGD and Cell Treatments
Eighty per cent confluent cultures were exposed to OGD by placing them in an anaerobic chamber (Invivo2, Ruskinn, Pencoed, UK) containing 0.5% O2, 4.5% CO2, and 95% N2 at 37°C for 6 hours with deoxygenated glucose-free RPMI (Gibco, Madrid, Spain). Normoxic cultures (control condition) were maintained at 37°C in the incubator containing 5% CO2 and atmospheric oxygen with RPMI with glucose (2 g/L) (Sigma-Aldrich, Madrid, Spain). Cultures were treated with 13 μg/mL (estimated concentration of r-tPA in plasma of patients with ischemic stroke receiving intravenous thrombolytic therapy) or 100 μg/mL of r-tPA (Actilyse, Boehringer, Ingelhem, Germany) or distilled water as vehicle. Cell viability was determined by MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) reduction and citotoxicity by lactate dehydrogenase (LDH) release (methods I in the online-only Data Supplement).
Microarray Gene Expression Analyses
A microarray study was performed to identify differentially expressed genes using an adjusted P<0.05 as a cutoff for the effect of r-tPA during OGD compared with control conditions. Gene expression profiles were obtained using GeneChip Human Genome U133 Arrays (Affymetrix, Santa Clara, CA). Complete information is described in methods II in the online-only Data Supplement. A validation study to confirm the expression of the selected genes was performed by quantitative real-time polymerase chain reaction using new cultures. Complete information is described in methods III in the online-only Data Supplement.
Western blotting was performed as described in methods IV in the online-only Data Supplement. The antibodies used were anti-NURR1 (1:1500, Sigma-Aldrich) or anti-β actin (1:10000; Sigma-Aldrich).
NURR1 gene silencing was performed with passive small interfering RNA (siRNA) delivery via Accell SMART pool siRNA (Thermo Scientific, Rockford, IL) following the manufacturer’s protocol. Cells were plated at 80 000 cells/mL and treated separately with the respective siRNA solutions (1 μmol/L). Nontargeting siRNA was used to measure the side effects of the assay. Cyclophilin B siRNA was used to test the efficiency and specificity of NURR1 silencing. Cells were preincubated for 48 hours in siRNA-containing Accell media, before being replaced with RPMI for the corresponding treatments.
TaqMan Array Human Immune Panel
cDNAs from 4 independent cultures subjected to NURR1 gene silencing or noncoding siRNA were mixed with equal volumes of TaqMan 2XPCR master mix (Applied Biosystems, UK), and 400 ng of cDNA were loaded into each port of the TaqMan Human Immune panel (Applied Biosystems). The plates were run in a 7900HT Fast Real-Time PCR system (Applied Biosystems). Relative gene expression values were obtained as described in methods III in the online-only Data Supplement using the cells treated with noncoding siRNA in control conditions as the calibrator for each experiment.
Thromboembolic Stroke Model
This model of thrombus formation in the middle cerebral artery is based on the activation of the coagulation cascade induced by thrombin injection.15 r-tPA–induced thrombolysis (10 mg/kg; Actilyse; tail vein injection, 10% bolus, 90% perfusion during 40 minutes) was initiated either 20 minutes or 4 hours after the induction of stroke. The lesion volumes were quantified 24 hours after stroke on MRI as previously described.16 Complete information is described in methods V in the online-only Data Supplement.
Analyses of NURR1 Expression in the Mouse Brain
NURR1 gene expression was analyzed from the RNA extracted from each brain hemisphere after 24 hours by using the NucleoSpin RNA II kit (Macherey-Nagel, Hoerdt, France) and reverse-transcribed using the iScript Select cDNA Synthesis Kit (BioRad, Marnes-la-Coquette, France). Relative mRNA transcription was calculated using the 2─ΔΔCt method (methods III in the online-only Data Supplement). Final values are expressed as the ratio of NURR1 expression in the ipsilateral and contralateral hemispheres in each condition. Five animals per group were used for the gene expression analyses. NURR1 protein expression was analyzed by Western blot (methods IV in the online-only Data Supplement) or immunohistochemistry (methods VI in the online-only Data Supplement) using anti-NURR1 (mouse 1:1000; Sigma-Aldrich) in 3 animals per group.
Serum From Patients With Ischemic Stroke
Peripheral blood samples were obtained from patients with ischemic stroke at baseline (before r-tPA administration), 1 hour (by the end of r-tPA infusion), 12 and 24 hours after stroke onset. Patients received r-tPA at the standard 0.9 mg/kg dose. A detailed history of vascular risk factors was obtained from each patient. Our cohort included 28 stroke r-tPA treated patients and 9 sex- and age-balanced stroke-free controls. Stroke severity was assessed with the National Institute of Health Stroke Scale, and neurological improvement was defined as a decrease in National Institute of Health Stroke Scale score of 4 points and deterioration as either death or an increase in National Institute of Health Stroke Scale score by 4 points at 24 or 48 hours after symptom onset. All patients underwent computed tomographic examination within the first 3 hours of stroke onset and repeated between 24 and 48 hours later (or earlier in cases of rapid neurological deterioration) to evaluate the presence of hemorrhagic transformation (HT). Hemorrhagic infarction was defined as a petechial infarction without a space-occupying effect, and parenchymal hemorrhage was defined as hemorrhage with a mass effect.17 Symptomatic intracranial hemorrhage was defined as blood at any site in the brain on the computed tomographic scan and documentation of neurological deterioration. The cohort was selected consecutively from our biobank according to the presence or absence of HT after receiving r-tPA. Fifty per cent of the patients had experienced any type of HT, and among them, 6 had experienced symptomatic HT. Serum samples were processed by centrifugation at 3500 rpm for 15 minutes (4°C) and stored at −80°C until use. The study was approved by the Ethics Committee of Vall d’Hebron Hospital, and all patients or their relatives provided written informed consent.
An ELISA kit from Cusabio Biotech (Wuhan, China) was used for quantitative determination of NURR1 concentration according to the manufacturer’s instructions. Fifty microliter of human serum samples was analyzed in duplicate.
Statistical differences were assessed with Mann–Whitney U and Kruskal–Wallis or one-way ANOVAs with Bonferroni post hoc test, as appropriate, using the SPSS 15.0. Variables from the experimental models are presented as mean±SEM. Data from human samples (normally distributed) are presented as mean±SD. Receiver operating characteristic curves were used to calculate the sensitivity and specificity with which NURR1 levels could predict symptomatic HT. P<0.05 was considered statistically significant at a 95% confidence level.
Toxic Effect of r-tPA on Brain Endothelial Cells Under Ischemia-Like Conditions
A 6-hour treatment with 13 μg/mL r-tPA was not toxic to cultured cerebral endothelial cells, whereas 100 μg/mL r-tPA induced a significant reduction in cell viability and metabolism (16.6±2.13%), as shown by MTT reduction. LDH release showed a nonsignificant change in cytotoxicity by r-tPA. OGD triggered a higher reduction in cell viability (19.4±12%), and the combination of OGD and r-tPA treatment produced an additive decrease in an r-tPA dose-dependent manner (22.6±14 and 35.6±13.3 in r-tPA 13 and 100 μg/mL treatments, respectively). However, results from LDH release indicated a cell death threshold, only exceeded when OGD was combined with 100 μg/mL r-tPA (25.5±7.7% cell death; Figure 1A–1C).
Gene Expression Profiles Induced by r-tPA Under OGD in Brain Endothelial Cells
A microarray study was performed to conduct a genome-wide comparison of brain endothelial cells treated with either a 13 or 100 μg/mL concentration of r-tPA under OGD. A total of 2500 analysed transcripts (Adj-P<0.05) were differentially regulated by 6 hours OGD (Figure 2A and 2B). We focused on the identification of genes exclusively altered by r-tPA during OGD, and the final list of candidates is presented in Figure 2C. When the expression of those genes was assayed by quantitative real-time polymerase chain reaction, the overexpressed patterns identified for BCL10, NURR1, FGF5, and THBS1 were validated, whereas the gene expression differences found for DKK1 and NEDD9 were not corroborated. Regarding those genes downregulated after treatments, results from AVIL and API5 expression were also not validated because OGD induced a downregulation in both r-tPA-treated and nontreated cells (Figure 2C).
NURR1 Expression in Cultured Brain Endothelial Cells
NURR1 was selected as an interesting candidate gene related to the adverse effects of r-tPA treatment. We confirmed its protein overexpression after 13 μg/mL, and especially at 100 μg/mL r-tPA treatment, only under OGD (Figure 3A). Gene silencing experiments were designed to test the effect of NURR1 upregulation on the cytotoxicity induced by high doses of r-tPA in OGD-treated hCMEC/D3 cells. NURR1 siRNA specifically depleted the mRNA levels and downregulated the protein expression (Figure I in the online-only Data Supplement). NURR1 knockdown protected cells from the cytotoxicity observed after 100 μg/mL r-tPA treatment during 6 hours OGD compared with the cells transfected with a nontargeting siRNA control (Figure 3B and 3C). A human immune array was conducted to evaluate the gene expression changes related to inflammation and apoptosis because of NURR1 silencing. NURR1 knockdown resulted in the downregulation of 4 targets exclusively after 100 μg/mL r-tPA treatment during OGD (CSF2, IL1α, NFKB2, and signal transducer and activator of transcription 3 (STAT3); Table I in the online-only Data Supplement). Additionally, NURR1 downregulation prevented the overexpression of CSF2, IL1α, NF-κ-B2, ICAM1, and SMAD3 induced by the combination of r-tPA and OGD (Figure 3D).
NURR1 Expression in Mice Thromboembolic Stroke
We used a mouse model of thromboembolic stroke induced by a local injection of thrombin in the middle cerebral artery. Consistent with clinical observations, ischemic lesion volumes after 24 hours were reduced by 26.26±7.39% when thrombolysis was performed 20 minutes after stroke onset, while these volumes were increased by 34.59±5.77% when r-tPA was administered 4 hours after the initiation of stroke (Figure 4A). NURR1 gene expression in the ipsilateral hemisphere was enhanced after delayed r-tPA administration compared with saline-treated mice (Figure 4B). At the protein level, most likely because of the extensive cell death after 24 hours of ischemia, we observed higher NURR1 expression in the contralateral than in the ipsilateral hemisphere in saline or early r-tPA treated mice. However, NURR1 expression in the ipsilateral was also enhanced after delayed r-tPA administration (Figure 4C), although the differences did not reach statistical significance. NURR1 expression was confirmed in the ipsilateral area of late-thrombolysed mice (Figure 4D1). In addition, brain vessels in that area were immunolabeled with NURR1 as evidenced by the colabeling with von Willebrand factor (Figure 4D2).
Soluble NURR1 Levels in Patients With Ischemic Stroke
Serum NURR1 protein levels were analyzed in controls and patients with ischemic stroke who received thrombolysis, selected according to the presence or absence of HT after r-tPA treatment. The time of r-tPA administration after the symptoms onset was not different among patients who presented or not an HT (159±58 minutes, n=14 versus 188±47 minutes, n=14, P=0.308). No statistical differences were found regarding demographic characteristics and main vascular risk factors between controls and patients with ischemic stroke (Table II in the online-only Data Supplement). However, the associations between baseline NURR1 levels and the demographic data and risk factors of the participants are presented in Table. Diabetes mellitus (P=0.004) and baseline glucose levels (P=0.04) exhibited positive associations with serum baseline NURR1 levels. Regarding clinical complications, the patients who had a parenchymal hemorrhage presented with higher baseline serum NURR1 levels (456.0±136.4 versus 322.1±134.9 pg/mL, P=0.019), and those who experienced a symptomatic HT exhibited the highest baseline NURR1 concentrations compared with patients with asymptomatic or absent HT (499.1±129.6 versus 335.2±134.6 pg/mL, P=0.013; Figure 5A and 5B). Receiver operating characteristic curves identified the optimal cutoff for the association of NURR1 levels with symptomatic HT; this cutoff was 442.48 pg/mL (P=0.007), and the sensitivity and specificity were 83.3% and 81.7%, respectively. Baseline NURR1 was the only independent predictor of symptomatic HT in this cohort. Figure 5A displays the temporal profiles of the serum NURR1 levels in the patients with ischemic stroke segregated by HT type after r-tPA administration. Interestingly, serum NURR1 levels increased after 1 hour of r-tPA treatment, especially in those patients who presented a symptomatic HT after the treatment (597.1±218.9 versus 346.5±178.1 pg/mL, P=0.008, Figure 5C). At 12 and 24 hours after the onset of stroke symptoms, serum NURR1 concentration decreased in all groups and was lower than in the controls (Figure 5A).
Despite the widely demonstrated efficiency of r-tPA in the treatment of patients during acute ischemic stroke,2 it has also been associated with vascular injury, inducing edema formation, and hemorrhagic complications.18 Symptomatic HT impairs the prognosis and increases the incidence of mortality among patients with stroke who receive r-tPA. Understanding the role of the cerebral endothelium may facilitate the identification of predictive biomarkers of the deleterious effects of the thrombolysis treatment after stroke. Thus, we adopted a translational approach, beginning with a massive screening strategy in cultured human brain endothelial cells subjected to r-tPA treatment during OGD. We selected r-tPA concentrations that ensured the toxicity on endothelial cells during OGD to simulate the occasional vascular damage induced by thrombolysis in patients with stroke.
The transcription factor NURR1 was selected as a possibly candidate involved in r-tPA activation pathway because of its unidentified involvement in ischemic stroke. NURR1 expression was altered only when cells were challenged to r-tPA during OGD, suggesting its role in ischemia merely when endothelium was severely affected. Across functional studies, we identified NURR1 as a proinflammatory regulator because the silencing of its expression reversed the cell toxicity and the induction of inflammatory molecules, including NF-κ-B2, IL1α, ICAM1, SMAD3, and CSF2, induced by the combined toxic stimuli.
We confirmed the participation of the endothelium in NURR1 overexpression in mice that had been subjected to thromboembolic stroke, particularly when delayed r-tPA treatment was applied. Based on our observations, it is tempting to speculate that NURR1 inhibition during thrombolysis is a promising approach for avoiding endothelium inflammation during acute stroke. This interpretation may contrast with the role of NURR1 in other systems, which has been reported to act as a crucial molecular check-point for neuronal survival.19 However, on the other hand, it is known that aberrant NURR1 expression is also a proinflammatory mediator in peripheral diseases, such as rheumatoid arthritis, psoriasis, and multiple sclerosis,9,20,21 suggesting its multiple roles in the regulation of specific targets depending on the cell type and the stimulus that induces its activation.22
We tested the possible clinical implications of our findings by determining the association between NURR1 levels in stroke patient’s serum and the posterior complications related to r-tPA therapy. We found that soluble NURR1 concentration before r-tPA administration was associated with the occurrence of symptomatic HT. Furthermore, those levels were higher in patients who had experienced symptomatic HT after receiving r-tPA for 1 hour, which suggests that in vivo overexpression of NURR1 is induced by infused r-tPA. Our results suggest that NURR1 expression levels are associated with the risk of HT. The underlying mechanism could be that as a transcription factor, NURR1 activates the expression of proinflammatory mediators, which would favor leukocyte migration, infiltration, and degranulation, eventually leading to blood–brain barrier disruption and bleeding. This merits future investigations in experimental models.
A limitation of our study using human serum samples is the relatively low number of patients included. Thus, our clinical findings need to be confirmed in a larger cohort; nevertheless, the determination of NURR1 levels in stroke patient serum before r-tPA treatment seems to be a promising strategy for avoiding symptomatic complications and broadening the therapeutic window of thrombolysis.
Our results suggest that the upregulation of NURR1 by r-tPA during cerebral ischemia is associated with endothelial dysfunction and enhances the adverse effects related to the r-tPA treatment. Additionally, the assessment of serum NURR1 levels at the onset of symptoms might be considered in the clinical management of patients with acute stroke who are eligible for thrombolysis.
We are grateful to Dr Gauberti for performing the model of thromboembolic stroke and corresponding MRI analyses.
Sources of Funding
This work was funded by European Stroke Network (EUSTROKE 7FP Health F2-08-202213), ERANET-NEURON program, INSERM, the French ministry of higher education and research, and the University of Caen Basse-Normandie. The Neurovascular Research Laboratory takes part into the INVICTUS network (RD12/0014/0005). C. Merino-Zamorano is supported by the fellowship FI12/00089 and Drs Hernández-Guillamon, Rosell, and Fernández-Cadenas by the Miguel Servet programme (CP12/03259, CP09/00265, and CP12/03298, respectively) from the Carlos III Institute of Health, Spain.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006826/-/DC1.
- Received July 21, 2014.
- Revision received October 17, 2014.
- Accepted November 12, 2014.
- © 2014 American Heart Association, Inc.
- Hacke W,
- Donnan G,
- Fieschi C,
- Kaste M,
- von Kummer R,
- Broderick JP,
- et al
- Zetterström RH,
- Solomin L,
- Jansson L,
- Hoffer BJ,
- Olson L,
- Perlmann T
- Sakurada K,
- Ohshima-Sakurada M,
- Palmer TD,
- Gage FH
- Weksler BB,
- Subileau EA,
- Perrière N,
- Charneau P,
- Holloway K,
- Leveque M,
- et al
- Orset C,
- Macrez R,
- Young AR,
- Panthou D,
- Angles-Cano E,
- Maubert E,
- et al
- Macrez R,
- Obiang P,
- Gauberti M,
- Roussel B,
- Baron A,
- Parcq J,
- et al
- Doi Y,
- Oki S,
- Ozawa T,
- Hohjoh H,
- Miyake S,
- Yamamura T