Augmentation of Ca2+ signaling in astrocytic endfeet in the latent phase of temporal lobe epilepsy

Astrocytic endfeet are specialized cell compartments whose important homeostatic roles depend on their enrichment of water and ion channels anchored by the dystrophin associated protein complex (DAPC). This protein complex is known to disassemble in patients with mesial temporal lobe epilepsy and in the latent phase of experimental epilepsies. The mechanistic underpinning of this disassembly is an obvious target of future therapies, but remains unresolved. Here we show in a kainate model of temporal lobe epilepsy that astrocytic endfeet display an enhanced stimulation-evoked Ca2+ signal that outlast the Ca2+ signal in the cell bodies. While the amplitude of this Ca2+ signal is reduced following group I/II metabotropic receptor (mGluR) blockade, the duration is sustained. Based on previous studies it has been hypothesized that the molecular disassembly in astrocytic endfeet is caused by dystrophin cleavage mediated by Ca2+ dependent proteases. Using a newly developed genetically encoded Ca2+ sensor, the present study bolsters this hypothesis by demonstrating long-lasting, enhanced stimulation-evoked Ca2+ signals in astrocytic endfeet.


INTRODUCTION
Evidence is accruing that perivascular astrocytic endfeet are highly specialized cell compartments in terms of molecular organization and functional roles (Nagelhus and Ottersen, 2013). Many of the unique features of these processes can be explained by their expression of the brain dystrophin DP71 which orchestrates a molecular assembly that includes the water channel aquaporin-4 (AQP4; Frigeri et al., 2001;Neely et al., 2001;Enger et al., 2012;Waite et al., 2012). The endfoot complement of AQP4 determines the rate by which water accumulates in brain in conditions favoring the development of brain edema (Vajda et al., 2002;Haj-Yasein et al., 2011b). The endfeet are also enriched in the inwardly rectifying K + channel Kir4.1 (Nagelhus et al., 1999;Higashi et al., 2001). This channel is thought to mediate K + siphoning in the retina (Kofuji et al., 2000) and contributes to K + spatial buffering in the CNS at large (Chever et al., 2010;Haj-Yasein et al., 2011a). The unique features of the astrocytic endfeet imply that astrocytes are highly polarized cells, biochemically as well as functionally.
It was recently found that loss of astrocyte polarization is common to several neurological conditions. The endfoot pool of AQP4 drops abruptly after an ischemic insult (Frydenlund et al., 2006;Steiner et al., 2012), and is also strongly reduced in models of Alzheimer's disease (Yang et al., 2011) and traumatic brain injury (Ren et al., 2013). Similarly, loss of astrocyte polarization-with reductions in AQP4 as well as Kir4.1-has been described in the hippocampus of patients with temporal lobe epilepsy (Schröder et al., 2000;Eid et al., 2005;Heuser et al., 2012). These changes are reproduced in experimental models of epilepsy, including the kainate model (Lee et al., 2012;Alvestad et al., 2013). The loss of Kir4.1 in particular is likely to be pathophysiologically relevant, as glial-conditional Kir4.1 knockout animals display deficient K + spatial buffering and severe epilepsy (Chever et al., 2010;Haj-Yasein et al., 2011a). Disassembly of endfoot protein complexes emerges as one of several mechanisms whereby astroglia may contribute to hyperexcitability and epileptogenesis Binder and Carson, 2013;Crunelli et al., 2014).
The mechanisms underlying the loss of astrocyte polarization in epilepsy have not been resolved. One possible mechanism is that an early injury causes Ca 2+ accumulation in endfeet, leading to proteolytic cleavage of the dystrophin associated protein complex (DAPC) at these sites. Such a mechanism is plausible, as astrocytes activated by injury contain calpain (Shields et al., 2000)-a protease that shows affinity to dystrophin and that is activated by Ca 2+ (Yoshida et al., 1992). This hypothesis cannot be tested by conventional Ca 2+ imaging, as bulk-loaded synthetic Ca 2+ dyes mainly reveal Ca 2+ signals at the level of the cell bodies (Reeves et al., 2011). Here we use an approach that allows Ca 2+ signals to be monitored in the fine astrocytic processes, including the perivascular endfeet. Specifically, we employed recombinant adeno-associated virus (rAAV) gene delivery of the genetically encoded Ca 2+ indicator GCaMP5E (Akerboom et al., 2012) to hippocampal astrocytes in a mouse model of temporal lobe epilepsy. Two-photon Ca 2+ imaging of acute hippocampal slices obtained in the epilepsy latent phase revealed elevated stimulation-evoked astrocytic Ca 2+ signals, both in endfeet and in astrocytic cell bodies. Indeed, the Ca 2+ signals in endfeet outlasted those in cell bodies. The present data point to endfoot Ca 2+ signaling as a possible mechanism underpinning the loss of astrocyte polarization in epilepsy.

ANIMALS
Male C57BL/6N mice of 2-4 months of age (Charles River) were used for all experiments. All procedures were approved by the animal use and care committee of the Institute of Basic Medical Sciences, University of Oslo, and the Centre for Comparative Medicine, Oslo University Hospital.

PLASMID CONSTRUCTS
The plasmid constructs were generated as described in a separate paper (Tang et al., 2015). In brief, the GCaMP5E DNA sequence was directly taken out from the expression vector pRGCAMP5E (Akerboom et al., 2012) by restriction digest with BamHI and HindIII, and subcloned into the rAAV vector pAAV-6P-SEWB (Shevtsova et al., 2005) with the human SYNAPSIN-1 (SYN) promoter to generate the construct of pAAV-SYN-GCaMP5E. The human GFAP promoter (Hirrlinger et al., 2009) was then inserted with MluI and BamHI into the pAAV-SYN-GCaMP5E vector resulting in the pAAV-GFAP-GCaMP5E construct.

INTRACORTICAL KAINATE INJECTION MODEL FOR MESIAL TLE
We used deep cortical (juxtahippocampal) kainate injection to elicit an initial status epilepticus (SE). Using this approach, more than 90% of injected animals developed recurrent behavioral seizures after a 5-8 day long latent period. For kainate injections, mice were anesthetized with a mixture of medetomidine (0.3 mg/kg, i.p.) and ketamine (40 mg/kg, i.p.) and kept on a heating blanket. A small craniotomy was performed in a stereotactic frame and kainate (50 nl; 20 mM; Tocris) was injected by a Hamilton pipette (Hamilton Company, NV) at a depth of 1.7 mm at the following coordinates relative to Bregma: anteroposterior −2 mm, lateral +1.5 mm (right). Anesthesia was stopped with atipamezol (300 mg/kg, i.p.) and SE was observed either clinically or by telemetric EEG recording and video monitoring. The animal model has been described in detail in a separate paper (Bedner et al., 2015). The non-injected side served as control for the kainate injected side.

IMAGING ANALYSIS
Time-series of fluorescence images were first imported into Fiji ImageJ (Fiji), and regions of interest (ROIs) were manually selected based on morphology. Astrocytic cell bodies were identified by their projecting branches and endfeet by their characteristic circular pattern around transversely cut vessels and elongated, linear appearance along obliquely cut vessels. ROIs over processes were chosen at least 5 µm away from the perimeter of the soma. The relative change in fluorescence (∆F/F) in each ROI, the individual traces and the histograms were all calculated and plotted by MATLAB (R2011b, MathWorks, Inc.) with custom written scripts. Standard deviation (SD) images were generated from time-lapse image recordings by Fiji ImageJ.

STATISTICAL ANALYSIS
Statistical analyses were performed using Prism (Version 6.0b for Mac OSX, GraphPad Software). One-way ANOVA with Tukey multiple comparisons test was used for comparison of GCaMP5E fluorescence changes in astrocytic somata, processes and endfeet following stimulation of Schaffer collaterals/commissural fibers. Paired t-test was used for comparison before and after washin with MCPG and MPEP. P < 0.05 was considered statistically significant.

VIRAL TRANSDUCTION YIELDED EXPRESSION OF THE Ca 2+ INDICATOR GCaMP5E IN ADULT MOUSE HIPPOCAMPAL ASTROCYTES
Injection of the rAAV-GFAP-GCaMP5E construct into the hippocampus yielded robust and selective GCaMP5E expression in hippocampal astrocytes, as revealed by immunolabeling with antibodies against green fluorescent protein (GFP) and glial fibrillary acidic protein (GFAP; Figures 1A,B). Notably, GCaMP5E was expressed within all astrocytic compartments, including the fine astrocytic processes and endfeet adjacent to CD31-immunopositive blood vessels ( Figure 1B).

THE AUGMENTED STIMULATION EVOKED ASTROCYTIC Ca 2+ RESPONSES FOLLOWING KAINATE INJECTION WAS DEPENDENT ON mGLuR5
Administration of the group I/II mGluR antagonist MCPG significantly reduced the amplitude of stimulation evoked Ca 2+ signals in all astrocyte compartments at day 1 post kainate injection (soma, P = 0.04, 32 somata, 9 slices, 6 mice; processes P = 0.0001, 36 processes, 9 slices, 6 mice; endfeet P = 0.04, 14 endfeet, 9 slices, 6 mice). The duration and latency of the Ca 2+ signals were not affected by MCPG, whilst rise rate was significantly reduced only in processes ( Figure 3A). As mGluR5 receptors have been shown to mediate enhanced astrocytic Ca 2+ signaling following pilocarpine induced SE (Ding et al., 2007), we applied the mGluR5 selective antagonist MPEP in our model. Similarly to MCPG, administration of MPEP significantly reduced the amplitude of stimulation evoked Ca 2+ signals at day 1 after kainate injection in astrocytic somata (P = 0.003, 19 somata, 5 slices, 4 mice), processes (P < 0.0001, 19 processes, 5 slices, 4 mice) and endfeet (P = 0.04, 7 endfeet, 5 slices, 4 mice). MPEP reduced the amplitudes of the Ca 2+ transients by 30-40%, i.e., to the level at the noninjected side (Figure 3C). The nonselective antagonist MCPG reduced the Ca 2+ transients to the same extent, suggesting that mGluR5 alone is mediating the enhanced Ca 2+ signal amplitude in the latent phase. Similarly to MCPG, MPEP did not affect the duration and latency of the Ca 2+ signals, and had inconsistent effects on transient rise rate in the three compartments ( Figure 3C).

Frontiers in Cellular
Neither MPEP nor MCPG significantly affected the fEPSP amplitudes (Figures 3B,D).

DISCUSSION
Astrocytes are highly polarized cells, structurally as well as functionally, opening for the possibility of a compartmentation of Ca 2+ signaling analogous to that found in neurons. With the advent of genetically encoded Ca 2+ sensors this possibility can be experimentally explored. A key question is whether Ca 2+ signaling in the astrocytic endfeet could play a role in epileptogenesis, by initiating a sequence of events that lead to disassembly of the DAPC in the endfoot plasma membrane. This complex, known to be critical for K + and water homeostasis in brain, is lost in patients with mesial temporal lobe epilepsy (Eid et al., 2005;Heuser et al., 2012) and in the latent phase of kainate induced epilepsy (Alvestad et al., 2013).
Here we show that intracortical kainate application leads to a stimulation evoked Ca 2+ signal in the endfeet that outlasts the Ca 2+ signal in the astrocytic cell bodies. This underlines the idea that endfeet are distinct subcompartments of astroglia (Nagelhus and Ottersen, 2013) and, more specifically, that endfeet serve as diffusion-limited subcellular compartments (Nuriya and Yasui, 2013).
The Ca 2+ signal in endfeet is attenuated by blockade of group I/II mGluRs and thus dependent on Ca 2+ mobilization from intracellular stores. However, mGluR blockade does not cancel out the difference between endfeet and cell bodies when it comes to the duration of the Ca 2+ signal. This suggests that the increased signal duration primarily reflects reduced clearance of Ca 2+ . An uncoupling of astrocytes could contribute to reduced clearance (Bedner and Steinhäuser, 2013;Bedner et al., 2015).

Frontiers in Cellular Neuroscience
www.frontiersin.org February 2015 | Volume 9 | Article 49 | 5 A disassembly of the DAPC in astrocytic endfeet and the loss of astrocyte polarization that this entails now emerge as a signature event in mesial temporal lobe epilepsy and epilepsy models (Nagelhus and Ottersen, 2013). The disassembly is reflected by a loss of dystrophin DP71, while β-dystroglycan remains (Heuser et al., 2012). Beta-dystroglycan is a member of the DAPC and normally serves to link this complex to extracellular matrix molecules of the pericapillary basal lamina (Neely et al., 2001;Amiry-Moghaddam and Ottersen, 2003).
It has been proposed that the molecular disassembly in astrocytic endfeet is caused by calpain activation (Nagelhus and Ottersen, 2013). Calpain is capable of cleaving DP71, and the expression of this protease is increased in activated astrocytes (Shields et al., 2000). It has not been resolved, however, whether endfeet sustain Ca 2+ signals necessary for activation of calpain or any other Ca 2+ dependent protease with affinity for dystrophin or dystrophin associated molecules. The present study fills this void and shows that endfeet display Ca 2+ signals that even outlast those in the astrocytic cell bodies. The cascade of events underlying the molecular disassembly in endfeet is an obvious target for future therapies.

AUTHOR CONTRIBUTIONS
KS designed experiments, acquired, analyzed and interpreted data, and wrote the paper; KH conceived the study and supervised experiments, acquired, analyzed and interpreted data, and wrote the paper; WT designed experiments, acquired, analyzed and interpreted data, and wrote the paper; VJ designed and supervised experiments, interpreted data, and wrote the paper; RE analyzed and interpreted data, and wrote the paper; PB designed the animal model, interpreted data and commented upon the manuscript; CS designed the animal model, interpreted data and commented upon the manuscript; ET conceived the study, interpreted data and commented upon the manuscript; OPO conceived the study, interpreted data and wrote the manuscript; EAN conceived the study and supervised experiments, interpreted data and wrote the manuscript. All authors revised the work critically and approved the manuscript.