All animal procedures were performed in compliance with European Community law 86/609/ECC and were approved by the Ethics Committees of the Instituto de Salud Carlos III and the Universidad Complutense de Madrid.

For procedures in rat, adult male Fischer rats (275 to 300 g) were anesthetized with 1.5% isoflurane in a mixture of 70% nitrogen/30% oxygen. Rats in which the middle cerebral artery (MCA) and both common carotid arteries (CCA) were exposed but not occluded served as sham-operated controls. The femoral artery was cannulated for continuous monitoring of arterial pressure and blood sampling for analysis of pH, gases and glucose. Body and brain temperature was maintained at 36.5 ± 0.5°C throughout the procedure. Monitored physiological variables did not differ significantly between groups of animals before, during or after middle cerebral artery occlusion (MCAO) (data not shown). The surgical procedure was a variant of that previously described 24 25 .

For procedures in mouse, 2-month-old adult male wild-type and Rcan1 knockout (KO) C57BL6 mice weighing 28 to 30 g were used. The KO strain has been described elsewhere 26 , and was backcrossed into the C57BL6 background for more than 20 generations. Oligonucleotide primers used for genotyping analysis were 5'-GGTGGTCCACGTGTGTGAGA-3' and 5'-ACGTGAACAAAGGCTGGTCCT-3'. Mice were subjected to transient focal cerebral ischemia through a combination of both MCAO and ipsilateral common carotid artery occlusion (CCAO) for 90 minutes, followed by disocclusion of the two vessels and reperfusion for 48 h.

For protein extracts and RNA preparation, tissue samples were dissected from the ischemic brain region around the occluded MCA and from the corresponding non-infarcted region in the contralateral hemisphere and rapidly frozen. For immunohistochemistry, rats were deeply anesthetized (see above) 24 h after blood reperfusion and then perfused intracardially with 4% paraformaldehyde in phosphate buffer (PB, pH 7.4). The brains were removed immediately and post fixed overnight at 4°C in the same fixative. Brains were then cryoprotected by incubation in 30% sucrose in PB for 48 to 72 h at 4°C. Frozen coronal sections (50 μm) were cut with a freezing microtome (Leica, SM2000R) and processed for immunohistochemistry.

The infarct volume was estimated on Nissl-stained coronal sections using Cavalieri's principle 27 . Volume measurements were calculated using the 'contour' and 'Cavalieri estimator probe' of the Stereo Investigator software package. The spared tissue in the damaged hemisphere and the contralateral hemisphere were outlined. The software then randomly placed a quadratic grid of points over these areas. Each grid point thus represents an area. The program was used to estimate the area of the contours from the number of selected grid points. The associated volume was calculated by multiplying the area by the mean section thickness. The precision of the volume estimate for each brain was determined by computing the coefficient of error for the estimates. The ratio between the volume of the spared cortex in the damaged hemisphere (RN) and that in the whole tissue of the contralateral hemisphere (L) was calculated, and used to detect differences in the amount of cortex that was damaged by the infarct in each animal. When expressed as a percentage, this ratio indicates the percentage fraction of tissue spared from the ischemia; therefore the percent of neocortex that was infarcted (%I) is readily obtained by the formula %I = (1 - (RN/L)) ×100.

Infarcted tissue in the neocortex was identified by staining slide-mounted coronal sections with Nissl (0.2% (w/v) cresyl violet). Adjacent sections were then processed for immunohistochemistry. Sections were permeabilized and blocked for 3 h at room temperature by incubation with 0.4% (v/v) Triton X-100 in Tris-buffered saline (TBS) containing 10% (v/v) normal serum from the species in which the secondary antibody was raised. Primary antibody was diluted in 0.2% (v/v) Triton X-100 in TBS containing 4% (v/v) of the same serum used for blocking. Goat polyclonal anti-GFAP antibody (Santa Cruz Biotechnology, Inc Santa Cruz, CA, USA, sc-6170) was diluted 1:500, and rabbit polyclonal Rcan1 antibody 26 was diluted 1:100. Sections were incubated with primary antibody overnight at 4°C. After several washes in TBS the sections were incubated for 2 h at room temperature with secondary antibodies: Cy3-conjugated anti-goat IgG or Alexa Fluor 488-conjugated anti-rabbit IgG, (Invitrogen, Grand Island, NY, USA, 1:200). After washing, sections were counterstained for 20 minutes at room temperature with TO-PRO 3 iodide (Molecular Probes) before mounting in Fluor-mount-G solution (Southern Biotech Birmingham, AL, USA). Parallel controls performed without primary antibodies showed very low levels of non-specific staining. Confocal images were acquired with a TCSSP5 confocal microscope (Leica Microsystems GmbH, Wetzlar, Germany).

Cortical primary cultures enriched in astrocytes were prepared as described 28 . Cortical astrocyte cultures were plated at 4 × 10⁵ cells per well in six-well plates or 1 × 10⁶ cells per 10 cm² plate and grown to confluence, with medium changes every 3 days. After 17 to 20 days in culture, immunocytochemical analysis indicated that > 98% of the cells in the culture were GFAP positive.

The Ca²⁺ ionophore A23187 was from Merck Biosciences Darmstadt, Germany. Phorbol 12myristate 13-acetate (PMA) was from Sigma-Aldrich St. Louis, MO, USA. Cyclosporin A (CsA) was from LC Laboratories Woburn, MA, USA..

Confluent primary astrocyte cultures were quiesced in Dulbecco's modified Eagle medium (DMEM) containing 0.5% fetal calf serum (FCS) overnight. Cultures were then subjected to hypoxia and deprived of glucose by incubation for 1 h in glucose-free DMEM (Life Technologies Corporation Carlsbad, CA, USA) plus 0.5% FCS in an anaerobic chamber (COY LaboratoryGrass Lake, MI, USA) in an atmosphere of 3% oxygen and 5% CO₂ balanced with nitrogen. At the end of the required incubation period, cells were lysed and frozen inside the chamber.

For whole-cell extracts, cells grown and stimulated in six-well plates or as indicated were washed twice with cold PBS and lysed for 30 minutes on ice in 100 μl hypertonic buffer with occasional mild agitation, as described 13 . Total tissue extracts were obtained by homogenizing the tissue in hypertonic buffer as before. Lysates were centrifuged for 10 minutes at 13,000 g, and the protein content of supernatants, containing cytosolic and most nuclear proteins, was quantified by Bio-Rad detergent-compatible protein reagent (Hercules, CA, USA). Total extracts were then boiled in 1× Laemmli buffer and resolved by 10% or 6% SDS-PAGE. Proteins were transferred to nitrocellulose membranes that were then immunoblotted as described 13 . The following antibodies were used: anti-NFATc3 polyclonal M75 (Santa Cruz Biotechnology, Inc Santa Cruz, CA, USA, sc-8321), monoclonal anti-α-tubulin (Sigma-Aldrich St. Louis, MO, USA, T8203), monoclonal Rcan1-4, raised and described by Minami et al. 15 , and polyclonal Rcan1 antibody, raised and described by Porta et al. 26 .

Extraction of total RNA from rat or mouse cortical astrocyte cultures and analysis of differential gene expression by quantitative real-time Reverse-Transcription Polymerase Chain Reaction (real time qRT-PCR) were as described previously 13 . 18S rRNA and TATA-binding protein (TBP) transcripts were used as internal control genes and were amplified in the same tube to normalize for variation in input RNA. The amounts of target mRNA in samples was estimated by the 2^-ΔΔCT relative quantification method 29 . Ratios were calculated between the amounts of mRNA from stimulated and non-stimulated control cells.

The adenovirus bicistronically encoding green fluorescent protein (GFP) and human Rcan1-4 (Ad Rcan1-4) and the control adenovirus encoding GFP alone (Ad GFP) were as described 15 . Adenoviruses were generated and purified following standard protocols. Adenoviral infection was carried out on subconfluent primary astrocytes, and expression of the encoded protein was monitored by immunoblot.

CN enzyme activity in brain tissue was analyzed with the Biomol Green Calcineurin Assay kit (Biomol, Plymouth, PA, USA) according to the manufacturer's instructions. Soluble protein extracts from three independent animals of each genetic background (Rcan1 wild-type (WT) and Rcan1 knockout (KO)) were prepared by homogenizing the tissue samples on ice in lysis buffer containing protein inhibitors. Soluble cytoplasmic proteins were prepared by ultracentrifugation (100,000 g for 45 minutes at 4°C). Excess sample phosphate was removed by passing samples through freshly prepared columns containing desalting resin (Bio-Rad, Hercules, CA, USA). To measure calcineurin phosphatase activity, protein extracts were quantified and equal protein amounts incubated for 30 minutes according to the assay kit instructions; liberated phosphate was measured colorimetrically at 620 nm. Calcineurin activity was calculated by comparison with a phosphate (PO₄) standard curve.

Results are expressed as the mean ± SEM or mean ± SD of the indicated number of experiments. Statistical analysis was by one-way analysis of variance (ANOVA) followed by individual comparisons of means (Student's t, Newman-Keuls or Bonferroni tests). Values were considered statistically significant at P < 0.05 (*) and P < 0.01 (**).

The possible accumulation of Rcan1-4 protein in response to ischemia/reperfusion (I/R) brain injury was examined in a rat model. Ischemia was induced by MCAO for 1 h followed by reperfusion for varying periods up to 24 h. In this model, ischemia-induced necrosis is limited to the parietal and sensory-motor cortex, and large infarcts are reproducibly generated in the MCA territory. For each condition, the contralateral (C) hemisphere was analyzed in parallel to control for physiological events not specific to the infarcted (I) area. Tissue extracts from infarcted and contralateral areas were obtained, and the protein levels of the two main Rcan1 isoforms, Rcan1-1 and Rcan1-4, were measured by immunoblot. The level of Rcan1-4 in the infarcted area of the brain, where the occlusion took place, was increased relative to the contralateral area of sham-operated animals (Figure 1). Increased Rcan1-4 expression was noticeable after 5 h reperfusion and was very pronounced after 24 h (Figure 1A, panel i lane 9). Parallel analysis showed that in most cases the expression of the Rcan1-1 isoform varied only slightly throughout the I/R injury period (Figure 1A, panel i), although in some experiments a slight decrease in the levels of Rcan1-1 protein was seen in the infarcted area after the longer reperfusion periods (5 and 24 h). α-Tubulin protein levels generally remained constant, although a slight decline was observed in some experiments after 24 h reperfusion, possibly as a result of damage to the ischemic tissue (Figure 1B, panel ii).

Rcan1-4 mRNA expression in the same brain ischemia model was determined by real time qRT-PCR, together with the mRNA expression levels of the proinflammatory cytokines IL-1β, TNFα and interleukin 6 (IL-6) and the proinflammatory gene cyclo-oxygenase 2 (Cox-2). Rcan1-4 mRNA in the infarcted area was induced 2 h after I/R injury (Figure 1B), preceding the increase in Rcan1-4 protein; Rcan1-4 mRNA expression levels in the contralateral area were unchanged. Expression levels of the inflammatory markers in the infarcted area were also specifically increased after 2 h reperfusion and this increase was maintained at 5 h (Figure 1B, i to iv). A basal elevation in inflammatory marker expression in the infarcted area in the absence of I/R was probably due to meningeal membrane rupture during the experimental procedure. This increased inflammatory marker expression above the levels obtained in sham-operated animals agrees with published reports on other models of stroke (reviewed in 30 ).

We next investigated the cellular location of the increased Rcan1 protein expression in response to brain I/R injury, using an anti-Rcan1 antibody that recognizes Rcan1-1 and Rcan1-4 isoforms 26 . After 24 h, Nissl staining of coronal sections detected a hypochromatic area in the infarcted neocortex, indicating neuronal injury (Figure 2A). The location of Rcan1 expression in relation to the injury site was analyzed by parallel immunohistochemical staining on adjacent sections. In addition, we stained for GFAP. In response to injury, astrocytes at the wound boundary migrate to repair the area. It is well established that astrocytes undergo a reactive process in response to different kinds of brain injury, such as ischemia and neurodegenerative diseases 6 31 32 33 . The reactive astrocyte phenotype is characterized by cellular hypertrophy, hyperplasia and increased expression of GFAP. At 24 h after early reperfusion, immunostaining with different anti-GFAP antibodies detected increased numbers of GFAP-positive cells in the neocortex surrounding the injured tissue (Figure 2B). Colocalization of Rcan1 in the majority of GFAP-positive cells around the infarct was confirmed by double immunostaining with goat anti-GFAP and rabbit polyclonal anti-Rcan1 antibodies (Figure 2C). These results agree broadly with a recent study describing Rcan1 protein accumulation in neuronal and astroglial cells, in which, as here, the antibody used for immunohistochemistry recognizes an Rcan1 C-terminal peptide common to both isoforms, 23 .

To examine the effect of ischemia on astrocyte Rcan1-4 expression, highly pure cultures (> 98%) of primary murine astrocytes were subjected to periods of hypoxia (3% O₂) combined with glucose deprivation (HGD). For comparison, cells were treated with phorbol ester (phorbol 12-myristate 13-acetate) plus calcium ionophore A23187 (PIo); Rcan1-4 expression in PIo-treated astrocytes is maximal after 4 h 28 , although at this time NFATc3 protein appears as a slower migrating form (Figure 3A, lane 2). Compared with PIo, HGD provoked a more rapid and transient accumulation of Rcan1-4 protein, with levels increasing significantly within 1 h of HGD in rat and mouse astrocytes and beginning to diminish after 4 h (Figure 3A, B). A similar profile was observed for Rcan1-4 mRNA, with expression maximal after 60 minutes HGD, and diminishing to levels similar to those in non-stimulated cells after 4 h (Figure 3C). HGD treatment also induced dephosphorylation of NFATc3 protein, observed as a faster migrating band on immunoblot (Figure 3A).

Western blot analysis of NFAT proteins is difficult to interpret, since most NFAT members exist as multiple isoforms. In the case of NFATc3, there are at least four mouse isoforms (according to the Ensemble database), with at least three of them in the range of 1060 to 1076 amino acids. These NFATc3 isoforms all contain the epitope used to raise the antibody used here; therefore to differentiate them we carried out immunoblot analysis on 6% SDS-PAGE gels, which were run until the 97 kDa marker approached the bottom of the gel in order to better resolve high molecular weight proteins (Figure 4A). This analysis shows that the antibody recognizes a number of bands that might correspond to the different mouse NFATc3 isoforms. To confirm that the bands we are studying correspond to NFATc3, we pretreated mouse astrocyte primary cultures for 1 h before HGD treatment with the calcineurin phosphatase inhibitor cyclosporin A (CsA). CsA prevented the appearance of the faster migrating bands after HGD treatment (Figure 4A). Furthermore, CsA changed the mobility pattern in non-stimulated cells toward slower migrating forms.

To further analyze the potential role of Rcan1-4 protein in astroglial cell behavior, we infected three independent primary astrocyte cultures with an adenoviral vector bicistronically encoding Rcan1-4 plus green fluorescent protein (Ad Rcan1-4) or a control vector encoding GFP alone (Ad GFP) 15 . Total cell proteins were separated on 6% SDS-PAGE gels. Rcan1-4 overexpression in primary astrocytes caused the appearance of slower migrating forms of NFATc3 (Figure 4B), which match the NFATc3 forms observed in cells pretreated with CsA (Figure 4A). These results thus indicate that Rcan1-4 accumulation inhibits CN signaling by preventing NFAT dephosphorylation. Our previous work demonstrated that primary astrocytes respond to PIo treatment with a marked induction of Cox-2 protein expression 28 . Adenoviral overexpression of Rcan1-4 in primary astrocytes reduced the amount of Cox-2 induced by PIo by 30% (Figure 5B, C). Consistently, primary astrocytes deficient for Rcan1 expressed increased amounts of Cox-2 protein in response to this treatment (data not shown). These data are in line with the induced Cox-2 transcript expression in brain after I/R injury (Figures 1 and 5).

To investigate the precise role of Rcan1 expression after brain I/R injury we compared injured brain areas in WT and Rcan1 KO mice. For this, we first characterized the procedure for MCAO, previously described for rat, in the mouse. Transient focal cerebral ischemia (90 minutes) was produced in WT and KO mice in the C57/BL6 genetic background. In the brains of Rcan1 KO mice, I/R injury was accompanied by a more pronounced increase in the mRNA expression of inflammatory markers. After 5 h reperfusion, the increases in the expression of TNFα, IL-6, IL-1β and Cox-2 in WT mice were comparable to those encountered in the rat model (Figure 1 and 6). Infarcted cortical tissue extracts from Rcan1 WT and KO mice showed similar levels of CN activity toward the specific substrate RII phosphopeptide (Figure 6B), suggesting that Rcan1 deficiency does not significantly affect soluble CN phosphatase activity in brain cortices.

We next analyzed the functional consequence of Rcan1 deletion in this stroke model. The extent of I/R injury after 48 h reperfusion was determined from the infarct volume, measured from serial Nissl-stained sections. Our results show that infarct volumes were significantly larger in Rcan1 KO mice compared with WT (Figure 6). The MCAO procedure was not associated with animal death in either animal genotype.