Angiotensin-Receptor-Associated Protein Modulates Ca2+ Signals in Photoreceptor and Mossy Fiber cells

Fast, precise and sustained neurotransmission requires graded Ca2+ signals at the presynaptic terminal. Neurotransmitter release depends on a complex interplay of Ca2+ fluxes and Ca2+ buffering in the presynaptic terminal that is not fully understood. Here, we show that the angiotensin-receptor-associated protein (ATRAP) localizes to synaptic terminals throughout the central nervous system. In the retinal photoreceptor synapse and the cerebellar mossy fiber-granule cell synapse, we find that ATRAP is involved in the generation of depolarization-evoked synaptic Ca2+ transients. Compared to wild type, Ca2+ imaging in acutely isolated preparations of the retina and the cerebellum from ATRAP knockout mice reveals a significant reduction of the sarcoendoplasmic reticulum (SR) Ca2+-ATPase (SERCA) activity. Thus, in addition to its conventional role in angiotensin signaling, ATRAP also modulates presynaptic Ca2+ signaling within the central nervous system.


Results
Expression of ATRAP in the mouse retina. The first sets of experiments served to verify the expression of ATRAP in the retina and to establish its localization in specific cell types. We detect mRNA expression of ATRAP by RT-PCR in whole mouse retina (Fig. 1a). Using laser capture microdissection of the mouse retina, we find that ATRAP mRNA is present in both the outer and inner nuclear layer (Fig. 1b). ATRAP immunoreactivity is strong in the outer plexiform layer of the wild type (Atrap +/+ ) mouse retina (Fig. 1c, middle and right panel) but not in the retina of mice lacking ATRAP (Fig. 1c left panel; Atrap −/− ) 6 . Because ATRAP has been reported to modulate angiotensin receptor signaling, we examined the expression of angiotensin receptor (AT1R) in the retina. As previously reported 30 , weak AT1R immunostaining by an antibody that gives no signal in AT1R knockout mouse kidney is seen in the retinal pigment epithelium but is absent from the outer plexiform layer (OPL) (Fig. 1d), suggesting that in the OPL ATRAP is not involved in conventional angiotensin receptor signaling. Therefore, we use the retina to assess the function of ATRAP without the angiotensin-signaling mediator present.

ATRAP is not expressed in horizontal cells, rod and ON cone bipolar cells. Our results show that
ATRAP is robustly expressed in the OPL, which is composed of a complex network of photoreceptor terminals and horizontal and bipolar cell dendrites. We therefore performed cell type-specific immunohistochemistry to localize ATRAP within the OPL. Using the retina of the Atrap −/− mice we verified that the antibody used to detect ATRAP does not produce false positive signals (Fig. 1c, left panel; Fig. 2 right panel). We find that ATRAP was almost not detectable in horizontal cells, which are identified by calbindin immunostaining 31 (Fig. 3a,d). ATRAP is also barely detectable in rod bipolar cells identified by Goα immunostaining 32 and PKCα immunostaining 33 and in ON cone bipolar cells, identified by Goα immunostaining 32 (Fig. 3b-d).

ATRAP localizes in ribbon synapses of photoreceptor cells. The finding that ATRAP is expressed
in the OPL ( Fig. 1) but not in a variety of second-order neurons (Fig. 3) leaves photoreceptor cells or OFF cone bipolar cells as likely candidates for ATRAP expression. We first use a transgenic mouse line (Rac3-eGFP) in cone photoreceptor cells express enhanced green fluorescence protein (EGFP) to test whether these cells also express ATRAP (Fig. 4a,d). We find robust co-localization of ATRAP with EGFP in the cone photoreceptor cells photoreceptor cells pedicles (Fig. 4a) and some ATRAP labeling beyond the boundaries of the cone photoreceptor terminals, which presumably corresponds to rod photoreceptor terminals (Fig. 4a). Next, we explore the subcellular localization of ATRAP in photoreceptors. ATRAP immunostaining localizes to synaptic terminals identified with the vesicular glutamate transporter 1 (vGluT1) 34,35 ; (Fig. 4b,d). Furthermore, double-labeling sections of mouse retina with antibodies against ATRAP and C-terminal binding protein 2 (CtBP2), a transcriptional corepressor and specific marker for photoreceptor ribbon structures 36 , shows a strong co-localization of both proteins at photoreceptor terminals (Fig. 4c,d). These results demonstrate that ATRAP is likely expressed at the presynaptic side of both rod and cone photoreceptor terminals.
We further analyze the ultrastructural distribution of ATRAP in photoreceptor terminals by means of immunogold labeling in transmission electron microscopy. ATRAP immunoreactivity localizes to membranous structures inside photoreceptor terminals; the postsynaptic dendrites of bipolar cells and horizontal cells show no ATRAP labeling (Fig. 5).  co-localizes with the ribbon marker CtBP2 (Fig. 6a,b). Thus, presynaptic terminals of photoreceptors abundantly express both SERCA2 and ATRAP.
To test whether ATRAP has an effect on Ca 2+ signaling at photoreceptor terminals, we perform Ca 2+ -imaging experiments on vertical slices of acutely isolated retinae from both Atrap +/+ and Atrap −/− mice. Since our experimental setup does not allow to combine imaging with light responses, we simulate the switch from light to dark with a depolarizing stimulus. We repeatedly expose the retinae to extracellular KCl (150 mM, 3 s) depolarize the membrane of photoreceptors to approximately 0 mV, causing the opening of voltage-gated Ca 2+ channels. We tested whether the dihydropyridine Ca 2+ channel blocker isradipine had a differential effect on Ca 2+ influx into photoreceptor terminals of Atrap +/+ and Atrap -/animals. Slices were incubated with the Ca 2+ indicator Fluo4-AM and subjected to extracellular application of 150 mmol/l KCl. KCl-depolarization of Atrap -/retinal sections induces 2+ responses that are significantly smaller than in Atrap +/+ retinal sections (Fig. 6c-e). Application of thapsigargin (1 µM), an inhibitor of SERCA2, significantly reduces the depolarization-induced 2+peaks in retinae of Atrap +/+ mice ( (Fig. 6c,e), indicating that the release from intracellular stores contributes to the Ca 2+ signal at the presynaptic terminal. In retinae of Atrap -/mice, by contrast, thapsigargin treatment has no significant effect on the amplitude of depolarization-induced Ca 2+ peaks (Fig. 6d,e). Thus, compared to Atrap +/+ mice, the depolarization-evoked Ca 2+ signals in the photoreceptor presynaptic terminals of Atrap -/mice are smaller and insensitive to thapsigargin inhibition. Figure 7a shows the response recorded from a single region of interest (ROI) from a wildtype mouse retina. Due to its size, the ROI encompassed several photoreceptor terminals. Cells were depolarized twice in the presence of isradipine (10 µmol/l) to allow for complete block of voltage-gated Ca 2+ channels. The control response was reduced to about 50% by isradipine. Since we observed no differences of Ca 2+ amplitudes between the two KCl applications in the presence of isradipine, the arithmetic mean was used for further data analysis. In photoreceptors of Atrap -/mice, isradipine had a very similar effect on the KCl-induced Ca 2+ response (Fig. 7b). This observation was confirmed when all measurements were averaged and normalized (Fig. 7c,d). The ratios of peak amplitudes in the absence and presence of isradipine, showed no significant difference (Fig. 7e). Mean values were 0.559 ± 0.026 (13 ROIs from 3 animals) in Atrap +/+ and 0.494 ± 0.029 (26 ROIs from 5 animals) in Atrap -/mice.
To get further insight into the functional role of ATRAP, we analyze temporal aspects of individual depolarization-induced Ca 2+ signals in photoreceptor terminals (Fig. 8). Inhibiting SERCA2 with thapsigargin   leads to a faster decay of the Ca 2+ signal in retinal slices from both Atrap +/+ and Atrap −/− mice compared to their respective controls, indicating that SERCA2 activity modulates the time course of Ca 2+ responses (Fig. 8a,b,e). Furthermore, the decay of the Ca 2+ responses in retinal slices from Atrap −/− mice is faster than in retinal slices from Atrap +/+ mice (Fig. 8c,e). Interestingly, the decay of Ca 2+ signals in retinal slices from Atrap −/− mice under control conditions has the same time constant as that of Ca 2+ signals in retinal slices from Atrap +/+ mice in the presence of thapsigargin (Fig. 8d,e). These results indicate that the lack of ATRAP alters the Ca 2+ signaling in the presynaptic terminal in a manner that closely resembles the effects of blocking SERCA2. Recent work showed that Ca 2+ mediates the changes in the morphology of the photoreceptor synaptic ribbons between light and dark adaptation 37,38 . To investigate whether the morphology of photoreceptor synaptic ribbons (SRs) between light and dark conditions is affected in Atrap −/− mice, we compare the shape of SRs between Atrap +/+ and Atrap −/− mice following dark or light adaptation according to Fuchs et al. 39 . We examine several hundred electron micrographs for each genotype and illumination condition and classified SRs according to their shape into three different categories:rod-shaped, club-shaped, and spherical-shaped ( Fig. 9a-c). In dark-adapted Atrap +/+ mice, the majority of SRs was rod-shaped (95.8% of 320 SRs), while the remaining SRs appeared club-shaped (4.2% of 320 SRs). Similar results were obtained for Atrap −/− mice, with the majority of SRs being rod-shaped (97.0% of 346 SRs) and only very few club-shaped (2.4% of 346 SRs) and spherical-shaped SRs (0.6% of 346 SRs) (Fig. 9d). Light adaptation caused in both genotypes a comparable reduction in the number of rod-shaped SRs (Atrap +/+ : 56.2% of 362 SRs; Atrap −/− : 57.4% of 443 SRs) and an increase in the number of ATRAP modulates Ca 2+ signaling at the mossy fiber terminals in the cerebellum. The photoreceptor synapse is specialized for the continuous release of glutamate. As ATRAP mRNA is also found in the brain 4 , it might be possible that ATRAP has a more generalized function at chemical synapses of the brain. We find strong ATRAP immunolabeling in the cerebral cortex and the cerebellum of Atrap +/+ mice. (Fig. 2). The cerebellum is a highly organized and stratified brain area that facilitates the analysis of synaptic structure and function 40 . Thus, we focused the next step of our investigation onto the cerebellum to prove the hypothesis that www.nature.com/scientificreports www.nature.com/scientificreports/ the synaptic ATRAP function does not only apply for photoreceptors. Strong immunoreactivity of ATRAP colocalize with the presynaptic glutamate transporter 1 (vGluT1) in the glomeruli formed by mossy fiber terminals and granule cell dendrites (Fig. 10a) and is absent in Atrap −/− mice (Fig. 10b). In GFAP-ECFP/Thy1-EYFP double-transgenic mice, the mossy fiber terminals can be readily identified in the granule cell layer, while fluorescently tagged Bergmann glial cells outline the Purkinje cell layer below the molecular layer (Fig. 10c,d). Since the specific localization of ATRAP at presynaptic terminals of the cerebellum suggests a similar role as in the retina, we test this hypothesis by Ca 2+ -imaging on acute cerebellar slices isolated from both Atrap +/+ and Atrap −/− mice. We discriminate cell somata from mossy fiber synaptic terminals by the unique morphology of Fluo4-AM fluorescent glomerular structures (yellow circles in Fig. 10e). Control labeling of cell nuclei by DRAQ5 reveals a clear distinction between synaptic and somatic Fluo4-AM loading (Fig. 10e,f). Local application of 150 mM KCl via a patch-pipette is used to activate voltage-gated Ca 2+ channels in cerebellar slices for 5 s. As shown in Fig. 10g,i, KCl-depolarization induces similar Ca 2+ responses in the somata of granule cells in both Atrap +/+ and Atrap −/− mouse cerebellum in the presence or absence of thapsigargin, as one would expect for a neuronal compartment in which plasma membrane-localized voltage-gated Ca 2+ channels predominate. In the glomerular structures containing the presynaptic terminals of mossy fibers, however, depolarization-evoked Ca 2+ responses are also mediated by Ca 2+ release from intracellular stores, since inhibition of SERCA by thapsigargin significantly reduces the response amplitude (Fig. 10h,j). By contrast, in the synaptic terminals of Atrap −/− mice, application of thapsigargin does not change the Ca 2+ response (Fig. 10h,j red vs maroon). Together, these results suggest that, just as in photoreceptors, ATRAP is involved in store-mediated Ca 2+ signaling at the synaptic terminal of mossy fibers in the cerebellum.

Discussion
While studying the functional role of ATRAP in the retina to further understand the physiological role of the RAS in the retina, we found that ATRAP shapes synaptic Ca 2+ transients through mechanisms not involving classical AT1R signaling. Here we present molecular, structural and physiological evidence for a novel function of ATRAP as a synaptic protein within the retina and the cerebellum that participates in the generation of Ca 2+ signals. ATRAP appears to function by modulating the Ca 2+ release from intracellular stores, as we find in  www.nature.com/scientificreports www.nature.com/scientificreports/ selected synapses of brain regions as diverse as cerebellum and retina. Therefore, we hypothesize that ATRAP may contribute more broadly to Ca 2+ signaling in neurons, rather than in its conventional role in the angiotensin receptor signaling.
ATRAP is expressed in a variety of organs including the heart, aorta, lung, adrenal glands, liver, spleen, testis, and brain, and is most abundant in the kidney 4 . ATRAP interacts with the cytosolic side of AT1R 4 . Although the classical function attributed to ATRAP is to modulate angiotensin-receptor signaling, we recently showed that ATRAP is also important in Ca 2+ store-dependent Ca 2+ signaling in the RPE 6 . Moreover, in cardiac myocytes ATRAP physically interacts and stimulates SERCA2 activity 7 . ATRAP further binds to proteins not related to angiotensin II signaling, such as Ca2 + modulating cyclophilin ligand and receptor for activated C kinase-1 8,41 . Along these lines, ATRAP has also been associated with retinal degradation type B protein, a member of the phosphatidyl inositol transfer protein family that functions in retrograde transport from the Golgi to the endoplasmic reticulum 42 . In the retina, we found ATRAP predominantly expressed in the RPE and the OPL. However, as previously shown 30 , we find no expression of the AngII receptor in the OPL (Fig. 1D) when using an antibody against AT1R that shows no false positive staining in the AT1R knockout mouse kidney. Because the OPL lacks expression of AngII-receptor and because ATRAP mainly localizes to intracellular structures distant from the plasma membrane, we postulate that ATRAP is not involved in the conventional AngII-signaling pathway in the mammalian retina. Instead, ATRAP's localization suggests a role in Ca 2+ signaling in photoreceptors presynaptic terminals.
We gain insight into the role of ATRAP in photoreceptor terminals by performing Ca 2+ imaging in the OPL in acute retinal slices from both Atrap +/+ and Atrap −/− mice. We focus on intracellular Ca 2+ stores because: 1) the lack of ATRAP in RPL cells or in cardiac myocytes reduces the store-mediated Ca 2+ response upon extracellular stimulation 6,7 and 2) Ca 2+ release from intracellular stores contributes to the depolarization-induced Ca 2+ signaling at photoreceptor presynaptic terminals that corresponds with a transition from light to darkness 20,21,25 . To mimic this transition, we used pulses of extracellular high K + concentrations that lead to photoreceptor cell depolarization. In the presence of the L-type Ca 2+ channel blocker isradipine, we observed a similar reduction of the Ca 2+ transients (by 85-90%) in Atrap −/− and Atrap +/+ mice. This result indicates that the level of Ca 2+ signals were predominantly measured in photoreceptor terminals and that the level of depolarization together with number of activated L-type channels were equal in both Atrap −/− and Atrap +/+ mice. Photoreceptor terminals in retinae from Atrap −/− mice show a smaller and thus faster recovery of depolarization-evoked Ca 2+ transients. It was shown 37,38 that the shape of the synaptic ribbon changes between light and dark adaptation through a Ca 2+ dependent mechanism. Using electron microscopy techniques, we here showed that the shape of the photoreceptor synaptic ribbons from Atrap +/+ and Atrap −/− mice under light and dark adaptation are similar. Thus, the absence of ATRAP has no influence on the synaptic ribbon, indicating that the function is not influenced as well. In addition, depolarization-evoked Ca 2+ transients in retinae from Atrap −/− mice are insensitive to thapsigargin and have similar decay times to that of Atrap +/+ retinae in the presence of the SERCA blocker thapsigargin (Fig. 6E), suggesting that ATRAP is required to activate the SERCA pump during presynaptic Ca 2+ signaling in photoreceptors. The absence of ATRAP has similar effects than SERCA inhibition. Inhibition of SERCA results in a reduced concentration of Ca 2+ in cytosolic Ca 2+ stores because Ca 2+ constantly leaks out of the stores and is not pumped back. Therefore, SERCA inhibition alone suffices to induce store-operated Ca 2+ entry in many different cell types 43 . The smaller amplitude of depolarization-evoked Ca 2+ transients results from reduced amounts of Ca 2+ released from stores in the Atrap −/− mouse photoreceptors. Thus, ATRAP likely has a similar molecular function in photoreceptor synapses than in cardiac myocytes 7 , however, differing by the fact that in photoreceptors ATRAP does not need AT1R receptor activation.
Our data indicate that ATRAP is involved in Ca 2+ store-dependent signaling in the photoreptor synapse. Therefore we speculate that the physiological role of ATRAP in the photoreceptor synapse is linked to regulate the Ca 2+ store signaling of the light-dependent increase in glutamate release. Transition from light to dark depolarizes photoreceptors, which activates L-type Ca 2+ channels and stimulates Ca 2+ release from endoplasmic Ca 2+ stores via Ca 2+ dependent activation of ryanodine receptors 9 . Ca 2+ -release in turn activates store-operated Ca 2+ entry 44 . This additional store-operated Ca 2+ entry further increases the Ca 2+ transient to strengthen and amplify the signal 20,44 . Importantly, the store-operated Ca 2+ entry boosts vesicular glutamate release at the photoreceptor synapse in general 14,16,21,45 . ATRAP activates SERCA2, which in turn pumps Ca 2+ into cytosolic Ca 2+ stores, that implying that ATRAP supports glutamate release by maintaining high levels of Ca 2+ inside the stores. Finally, because the Ca 2+ signals from Atrap −/− photoreceptors were insensitive to thapsigargin, it is likely that the boost effect does not occure in the absence of ATRAP.
Since the expression of ATRAP is not restricted to the retina and can be found in the cortex and in the cerebellum, we analyzed KCl-evoked Ca 2+ transients in presynaptic structures of cerebellar mossy fibers. Early studies already showed that Ca 2+ release from stores plays a role in generating large miniature inhibitory postsynaptic currents (IPSCs) in cerebellar Purkinje cells [17][18][19] . Here, the generation of large-amplitude miniature IPSCs in cerebellar Purkinje cells depends on activation of ryanodine receptors at the presynapse of interneurons 17,18 . In our experiments, inhibition of SERCA by thapsigargin reduces depolarization-evoked presynaptic Ca 2+ signals in mossy fibers, further implying a contribution of Ca 2+ release from stores. However, in contrast to wildtype mice, the Ca 2+ transients in mossy fibers of Atrap −/− mice are insensitive to thapsigargin. This result equals to that in photoreceptor synapses from Atrap −/− mice and it is tempting to speculate that ATRAP fulfills a related role in Ca 2+ signalling in cerebellar synapses. Moreover, because the cerebellum, unlike the photoreceptors, shows a broad AT1R expression 46 , the function of ATRAP in mossy fibers synapses could also suggest the existence of novel mechanisms of the renin-angiotensin system network in the brain.
In contrast to photoreceptor terminals, the depolarization-evoked Ca 2+ transient amplitude is unchanged in cerebellar mossy fiber cells from Atrap −/− mice. Furthermore, cardiac myocytes from Atrap −/− mice also show unchanged Ca 2+ amplitudes in response to depolarization compared to those from wild-type mice 7  www.nature.com/scientificreports www.nature.com/scientificreports/ lack of ATRAP has very different effects on depolarization-evoked Ca 2+ transients in photoreceptors, mossy fibers and cardiac myocytes. This suggests that, depending on the cell type, ATRAP-dependent modulation of SERCA2 activity affects in different ways the dynamics of Ca 2+ signaling.
To date, ATRAP function was assumed to be limited to modulating the AT1R-mediated angiotensin II signaling cascade. Our study shows that, in addition to its conventional role in angiotensin signaling, ATRAP may be also crucial for the Ca 2+ signaling in synapses of the retina and the cerebellum. Since ATRAP expression localizes to different brain areas it is possible that this synaptic ATRAP function is of more broad relevance.

Methods
Animals and ethical approval. Animals and ethical approval followed the conditions as published previously 6 . Adult male and female ATRAP knockout (Atrap −/− ) mice (C57BL/6 × 129SvEv) from a local colony were used in the study 4 . Atrap −/− and Atrap +/+ littermates were used from heterozygous breeding pairs. All experimental procedures were performed following the guidelines approved by the Institutional Animal Care and Use Committee at University of Regensburg and the Association for Research in Vision and Ophthalmology (ARVO) statement for the use of animals in vision research. All animal experiments were formally approved by the German authorities (Bavarian Administration/Regierung Oberpfalz under number: 54-2532.1-06/10).
GFAP-ECFP/Thy1-EYFP transgenic mice were bred in the animal facility of the University of Saarland in Homburg and used for confocal imaging of the cerebellum after perfusion fixation as previously described 47 .
Tg(Rac3-EGFP)JZ58Gsat/Mmcd (Rac3-EGFP) mice were obtained from the Mutant Mouse Regional Resource Center, a NCRR-NIH funded strain repository, and were donated to the MMRRC by the NINDS funded GENSAT BAC transgenic project.
Laser capture microdissection. For laser capture microdissection, unfixed eyes were prepared, frozen and sectioned at a thickness of 25 µm using a cryostat. Defined regions from the outer and inner plexiform layers were dissected and collected on the same day using the PALM Micro Beam system (Carl Zeiss MicroImaging) equipped with a nitrogen laser (337 nm). After microdissection, the samples were ejected from the object plane with a single laser shot and catapulted directly into a microtube cap (Adhesive-Cap, Carl Zeiss MicroImaging) for subsequent reverse transcriptase-PCR.
The sections were then post-fixed with 2% (w/v) OsO4 in cacodylate buffer for 1 hr, dehydrated in a graded series of ethanol (30-100%), followed by propylene oxide, and flat-embedded in Epon 812 (Serva, Heidelberg, Germany). Ultrathin sections were cut and stained with uranyl acetate. Control vibratome sections lacking primary antibodies were processed as described above. These produced no staining. Ultrathin sections were examined and photographed with a Zeiss EM10 electron microscope.
Conventional electron microscopy. Sample preparation and image categorization was performed as described earlier 39 : For conventional electron microscopy, retinae were fixed in 4% paraformaldehyde and 2.5% glutaraldehyde for 2 hours at room temperature. Tissue contrasting was carried out by incubation in 4% osmium tetroxide in cacodylate buffer (0.1 M, pH 7.4) for 1.5 hours. Retinae were dehydrated using a rising ethanol series and propylene oxide. The tissue was embedded in Epon resin (Fluka, Buchs, Switzerland). Ultrathin sections (60 nm) were cut and counterstained with uranyl acetate and lead citrate in an automated Leica EM AC20 contrasting system (Leica Microsystems, Wetzlar, Germany). Image acquisition was performed using a Zeiss EM10 electron microscope (Zeiss, Oberkochen, Germany) and a Gatan SC1000 Orius TM CCD camera (GATAN, Munich, Germany) in combination with the Digital Micrograph 3.1 software (GATAN, Pleasanton, CA). Images were adjusted for contrast and brightness using Adobe Photoshop CS6. For the quantification of synaptic ribbon shapes, random images of the outer plexiform layer were taken for each genotype and experimental condition. According to their shape, SRs were classified into rod-, club-and spherical-shaped. ca 2+ imaging of retinal slices. Ca 2+ signals from retinal slices were recorded according to the method by Regus-Leidig et al. 54 . For Ca 2+ imaging experiments, sagittal slices of Atrap +/+ and Atrap −/− mouse retinas were cut at 200 µm with a vibratome (Leica, Mannheim, Germany). Subsequently, slices were incubated for 30-60 min at 37 °C in an atmosphere of 5% CO 2 / 55% O 2 with 1 µM Fluo-4 AM (Life Technologies, Grand Island, NY) and 0.5 µl pluronic acid (Life Technologies) in a solution containing (in mM): 117 NaCl, 3 KCl, 2 CaCl 2 , 1 MgCl 2 , 0.4 NaH 2 PO 4 , 25 NaHCO 3 , 15 Glucose (pH 7.4). Following incubation, slices were washed twice and immersed in an extracellular solution containing (in mM): 132 NaCl, 5.4 KCl, 5 CaCl 2 , 1 MgCl 2 , 5 Hepes, 10 glucose (pH 7.4). A Zeiss Examiner D1 microscope (Zeiss, Germany) equipped with a 63x water immersion objective was used to visualize regions of interest (ROIs), and the images were captured by an AxioCam Hsm camera (Zeiss). Small rectangular ROIs (~25 × 5 µm) were placed in the outer plexiform layer. Each region of interest contained several rod and cone terminals, but individual responses of rods and cones could not be distinguished within a single ROI. Therefore, each response measured within a single ROI represents an average of a small sample of both rods and cones. Photoreceptor terminals in the slice preparation were depolarized with a solution containing 150 mM KCl. By application of Nernst's equation, we calculated the resulting membrane potential to be near 0 mV. The KCl solution was applied to the preparation with a focal perfusion system (ALA Scientific Instruments, Farmingdale, NY) controlled by the Patchmaster software (Heka, Lambrecht, Germany). The tip of the perfusion system was located at a distance of ~500 µm from the tissue and it was operated with minimum pressure to rule out motion artefacts. Under these experimental conditions, a concentration of 150 mM KCl proved to reliably evoke depolarization-induced Ca 2+ influx. Retinal slices were incubated with 1 µM thapsigargin for 10 min after loading with Fluo-4 AM. Imaging data were acquired with the Axiovision software (Zeiss, Jena, Germany) at frame rates ranging from 5 to 20 Hz. Since ratiometric imaging using Fura-2AM led to increased acquisition rates beyond the physiological time frame of synaptic responses, imaging experiments were carried out with Fluo-4 AM. Data analysis was performed with custom-made scripts using the software packages Matlab (MathWorks, Natick, MA) and Origin (Microcal, Northampton, MA). Time-dependent decrease of mean ΔF/F0 was fitted with a 1st order exponential function using Microcal Origin (Northampton, MA): where A 1 is the amplitude at t = 0, τ is the time constant of decay, and y 0 represents an offset. Rundown of fluorescence in imaging experiments was estimated by repeated application of 150 mmol/l KCl to photoreceptor terminals before pharmacological treatment with isradipine or thapsigargin. The fluorescence signal usually decreased by 10-15% before a constant amplitude was obtained. The peak amplitude of the first KCl-induced Ca 2+ signal has been used throughout as control. ca 2+ imaging of cerebellar slices. Using the methods as previously published 55 , Ca 2+ signals were recorded from mouse cerebellar slices. The Atrap +/+ and Atrap −/− mice were anesthetized by isofluran before decapitation, and their cerebella were removed from the skull and immersed in an ice-cold, oxygenated (5% CO 2 /95% O 2 , pH 7.4) slice preparation solution containing (in mM) 87 NaCl, 3 KCl, 25 NaHCO 3 , 1.25 NaH 2 PO 4 , 3 MgCl 2 , 0.5 CaCl 2 , 75 sucrose and 25 glucose. Sagittal slices of 300 µm were prepared with a vibratome (Leica VT 1200 S, Leica Instruments, Nussloch, Germany) and transferred to a Nylon basket slice holder for incubation in (2019) 9:19622 | https://doi.org/10.1038/s41598-019-55380-8 www.nature.com/scientificreports www.nature.com/scientificreports/ artificial cerebral spinal fluid (ACSF) containing (in mM) 126 NaCl, 3 KCl, 25 NaHCO 3 , 15 glucose, 1.2 NaH 2 PO 4 , 1 CaCl 2 , and 2 MgCl 2 at 34°C. The slices were allowed to recover in ACSF with continuous oxygenation for at least 0.5 h.
Before imaging, slices were incubated with 1 µM Fluo-4 AM as described for the retina. Subsequently, the slices were washed twice and immersed in extracellular solution containing (in mM) 126 NaCl, 3 KCl, 25 NaHCO 3 , 15glucose, 1.2 NaH 2 PO 4 , 2.5 CaCl 2 , and 1 MgCl 2 . To depolarize, KCl (150 mM) in extracellular solution was applied to the cerebellar slices for 5 s by a focal custom-made application system. To block sarco-endoplasmic Ca 2+ -ATPase (SERCA) activity, slices were incubated with 5 µM thapsigargin for 5 min after loading with Fluo-4 AM; then 10 mM caffeine was applied for 1 min to deplete intracellular calcium. A Zeiss microscope (Axioskop 2 FS mot, Germany) equipped with a 40x water immersion objective was used to visualize the region of interest and the images were captured by a QuantEM 512SC camera (Photometrics, Tucson). Imaging acquisition was controlled by Imaging Workbench software 5.2.20.6 (INDEC BioSystems, USA) at 20 Hz frame rate. Data analysis was performed with custom-made scripts by Matlab (MathWorks, USA) and Graphpad Prism 5.0 (La Jolla, USA). All data were shown as mean ± SEM. Post-hoc Tukey one-way ANOVA was used for multiple group comparison.
Statistical analysis. All data were given as mean and standard error of the mean. Experiments were repeated at least five times. All animal data were obtained from 5-6 animals. If not otherwise stated, test for statistical significance was performed by ANOVA. To test variability the Levene test was used. All statistical analyses were performed by GraphPad, Systat, Sigmaplot or Excel. We used ImageJ to quantify the Pearson's correlation coefficient of immunostaining co-localization.