Abstract
The endoplasmic reticulum (ER) is the largest intracellular Ca2+ store, serving as the source and sink of intracellular Ca2+. The ER Ca2+ store is continuous yet organized into distinct subcompartments with spatial and functional heterogeneity. In cerebellar Purkinje cells (PCs), glutamatergic inputs trigger Ca2+ release from specific ER domains via inositol 1,4,5-trisphosphate receptors (IP3Rs) or ryanodine receptors (RyRs). Upon ER store depletion, refilling occurs through store-operated Ca2+ entry mediated by stromal interaction molecule-1 (STIM1). Although the significance of STIM1-mediated Ca2+ regulation within PCs is established, STIM1 localization in ER subcompartments in PCs for Ca2+ release and refilling remains elusive. Using validated antibodies, we demonstrated that STIM1 was predominantly localized as intense puncta along dendritic shafts in male and female mice, colocalizing with IP3R1 but not with RyR1. Immunoelectron microscopy revealed that STIM1 was accumulated in the subsurface ER in the dendritic shaft but excluded from those in the dendritic spine, the primary site of metabotropic glutamate receptor 1 (mGluR1)–IP3R-mediated Ca2+ signaling. Ca2+ imaging from control and STIM1-knockdown (STIM1-KD) PCs demonstrated that mGluR1-mediated Ca2+ release is more critically dependent on STIM1 than RyR-mediated Ca2+ release. These findings reveal a spatially organized ER network in PCs, where specialized ER subcompartments differentially regulate Ca2+ release and refilling. These findings suggest that STIM1 preferentially regulates Ca2+ dynamics associated with mGluR1–IP3R signaling, supporting specialized ER subcompartments for Ca2+ release and refilling. These findings highlight the intricate molecular–anatomical organization of dendritic ER Ca2+ signaling in PCs, crucial for synaptic plasticity and motor learning.
- cerebellum
- endoplasmic reticulum
- inositol 1,4,5-trisphosphate
- metabotropic glutamate receptor
- Purkinje cell
Significance Statement
Intracellular calcium (Ca2+) signaling is essential for neuronal function, yet the organization of endoplasmic reticulum (ER) subcompartments that coordinate Ca2+ release and refilling remains unclear. This study demonstrates that stromal interaction molecule-1 (STIM1), a key regulator of store-operated Ca2+ entry, is predominantly localized to the subsurface ER in Purkinje cell (PC) dendrites, which had not been previously identified. STIM1 colocalizes with inositol 1,4,5-trisphosphate receptor type 1 (IP3R1) and sarcoplasmic/ER Ca2+-ATPase 2 but is segregated from ryanodine receptor 1, highlighting specialized ER subdomains for Ca2+ release and refilling. These findings provide new insights into the molecular–anatomical organization of Ca2+ signaling in PCs, which plays key roles in synaptic plasticity, motor learning, and the pathophysiology of neurodegenerative diseases.
Introduction
Cytosolic calcium (Ca2+) is a fundamental secondary messenger that regulates various cellular functions. While Ca2+ is present at submicromolar concentrations in the cytosol at rest, it is at millimolar concentrations in the extracellular fluid. Intracellularly, Ca2+ is predominantly stored in the endoplasmic reticulum (ER), where its concentration reaches hundreds of micromoles (Verkhratsky, 2005; Clapham, 2007). The steep Ca2+ gradient across the plasma membrane (PM) and ER, which is crucial for cellular signaling, is maintained by Ca2+-handling molecules, such as endogenous buffers and calcium pumps, including sarcoplasmic/ER Ca2+-ATPase (SERCA) and PM Ca2+-ATPase (Clapham, 2007).
The ER is important not only for buffering Ca2+ but also for local Ca2+ signaling (Berridge, 1998; Emptage et al., 2001). In central neurons, Ca2+ stored in the ER is released upon stimulation via inositol 1,4,5-trisphosphate (IP3) receptors (IP3Rs) and ryanodine receptors (RyRs; Berridge, 1998). The amplitude of the Ca2+ signal depends significantly on the ER Ca2+ content (Friel and Tsien, 1992; Usachev and Thayer, 1997), making ER store replenishment crucial for proper signaling. Decreased ER Ca2+ concentration triggers Ca2+ influx through the PM via store-operated Ca2+ entry (SOCE; Lewis, 2011). The core components of most SOCE pathways are the ER Ca2+ sensor, stromal interaction molecule-1 (STIM1), and the Ca2+ channel in the PM, Orai (Soboloff et al., 2012). When STIM1 detects a decrease in luminal Ca2+ concentration via its luminal N-terminus (Stathopulos et al., 2006; Wu et al., 2006; Xu et al., 2006; Stathopulos et al., 2008), it clusters and activates Orai channels in the PM (Feske et al., 2006; Wu et al., 2006). Ca2+ entering the cytosol through these store-operated Ca2+–permeable channels is transported into the ER via SERCA (Prakriya et al., 2006; Yeromin et al., 2006). This intricate refilling primes the cell for another signaling round, supporting essential cellular functions.
The ER Ca2+ store forms a continuous network (Terasaki et al., 1994) but is organized into distinct spatial and functional subcompartments (Volpe et al., 1993; Meldolesi and Pozzan, 1998; Blaustein and Golovina, 2001). These ER subcompartments are typically enriched with IP3R, RyR, and SERCA, which can individually unload and load Ca2+ (Blaustein and Golovina, 2001). Cerebellar Purkinje cells (PCs) develop an intricate ER network, where abundant IP3R and RyR show partially overlapping but distinct distributions. IP3Rs are expressed from the soma to the dendritic spine, while RyR is excluded from the spine and is mostly limited to the proximal somatodendritic compartment (Mignery et al., 1989; Ellisman et al., 1990; Satoh et al., 1990; Walton et al., 1991; Takei et al., 1992). PCs receive two glutamatergic inputs at the dendritic spine: >100,000 parallel fibers (PFs) at the distal dendrite (spiny branchlet) and a single climbing fiber (CF) at the proximal dendrite (Palay and Chan-Palay, 1974; Ichikawa et al., 2016). Each input triggers distinct Ca2+ signaling within specific ER domains. Repetitive PF inputs activate metabotropic glutamate receptor 1 (mGluR1)–IP3R-mediated Ca2+ release, increasing cytoplasmic Ca2+ concentration in small dendritic domains or single spines (Finch and Augustine, 1998; Takechi et al., 1998). In contrast, a single CF input induces widespread Ca2+ influx across dendritic arbors, resulting in a cumulative Ca2+ increase in the ER (Okubo et al., 2015) and RyR-mediated Ca2+ release (Kakizawa et al., 2007). Although the STIM1-mediated Ca2+ store regulation in PCs is essential for gene expression (Dhanya and Hasan, 2021), motor coordination (Hartmann et al., 2014), and cerebellar memory consolidation (Ryu et al., 2017), the localization of STIM1 in PCs remains unexplored.
This study investigated the subcellular localization and role of STIM1 in Ca2+ signaling in PCs. STIM1 is preferentially localized to the subsurface ER in the dendritic shaft but is excluded from the dendritic spines. Dendritic STIM1 colocalizes noticeably with IP3R1 and SERCA2 but only faintly with RyR1. Ca2+ imaging in control and STIM1-knockdown (STIM1-KD) PCs demonstrated that mGluR1-mediated Ca2+ release depends more critically on STIM1 than RyR-mediated Ca2+ release. These arrangements suggest that distinct ER subcompartments have specialized roles in Ca2+ regulation.
Materials and Methods
Animals
Animal experiments were performed according to the guidelines laid down by the Hokkaido University Animal Experiment Committee (Protocol Number 19-0111, 23-0033). Both male and female C57BL/6 mice, including neonates [postnatal day 1–5 (P1–P5)], adolescents (P35–P40), and adults (8–12 weeks old), obtained from Sankyo-Lab Service, were used as WT and control mice. Three female Japanese white rabbits (12 weeks old) and Hartley guinea pigs (6 weeks old) were also obtained from Sankyo-Lab Service and used for antibody production.
Knockdown experiments
To suppress STIM1 or SERCA2 expression in PCs, a designed microRNA (miRNA) was introduced into PCs with an adeno-associated virus (AAV; Miyazaki et al., 2021). The following engineered miRNAs designed according to the BLOCK-iT Pol II miR RNAi Expression Vector kit guidelines (Thermo Fisher Scientific) were used to target the 22 bp sequence of STIM1 (GAGGATGAGAAGCTCAGCTTT):
5″-TGCTGAAAGCTGAGCTTCTCATCCTCGTTTTGG CCACTGACTGACGAGG ATGAAGCTCAGCTTT
-3″ and 5″-CCTGAAAGCTGAGCTTCATCCTCGTCAGTCAGT GGCCAAAACGAGGATGAGAAGCTCAGCTTTC-3″ and target 19 bp sequence of SERCA2(GAAACGATCAGCCTTTGTA):
5″-TGCTGTACAAAGGCTGTAATCGTTTCGTTTTGGCCACTG ACTGACGAAACGATCAGCCTTTGTA
-3″ and 5″-CCTGTACAAAGGCTGATCGTTTCGTCAGTCAGTGGCCAAAACGAAACGATTACAGCCTTTGTAC-3″
The fragments were subcloned into pcDNA6.2-GW/EmGFP vector and finally ligated into AgeI/NotI site of pAAV/L7-6-GFP-WPRE vector (Addgene plasmid #126462; Prof. Hirokazu Hirai). For Ca2+ imaging experiments, the EGFP sequence at the DraI site was removed, and the mCherry sequence, which was amplified by PCR from pAAV-mCherry-flex-dtA vector (Addgene plasmid #58536; Prof. Naoshige Uchida), was then ligated into the pcDNA6.2-GW/Stim1-miR vector. The vectors were purified using an Endofree Plasmid Midi Kit (Qiagen) and transfected with AAV-DJ, pHelper (Cell Biolabs) into 293 AAV cells (Cell Biolabs) using the calcium phosphate methods. After 48–60 h post-transfection, cells were solubilized, and AAVs were purified using a HiTrap Heparin column (GE Healthcare).
AAV injections were performed under sterile conditions. P1–P5 pups were anesthetized with ∼5% isoflurane (Pfizer) and secured in a stereotaxic frame. For injection, holes were drilled into the occipital bone for the needles. The coordinates were (0, 0, 0.6 mm), corresponding to positions caudal to the occipital external protuberance, right of the midline, and ventral to the pial surface, respectively, with the manipulator tilted at 40°. Each region received 200 nl of AAV. Mice were analyzed at least 4 weeks postinjection.
Antibodies
Primary antibodies against the following molecules were used: calbindin, EGFP, IP3R1, mGluR1α, RyR1, SERCA2, and STIM1. In the present study, we produced a rabbit anti-SERCA2 antibody and a guinea pig anti-STIM1 antibody. To express glutathione S-transferase fusion proteins, we subcloned cDNA fragments encoding mouse SERCA2 (322–360 amino acid residues, NCBI GenBank, NM_001163336.1) or STIM1 (23–62, NM_009287) into the pGEX4T-2 plasmid (GE Healthcare). Immunization and affinity purification were performed as described previously (Watanabe, 2021). In brief, each fusion protein, emulsified with Freund's complete adjuvant (Difco) in the first immunization and incomplete adjuvant in the following immunization, was injected subcutaneously into a female Japanese white rabbit and a Hartley guinea pig at intervals of 2–4 weeks. Two weeks after the sixth injection, serum was collected, and immunoglobulins specific to antigen peptides were affinity-purified using the antigen-coupled to CNBr-activated Sepharose 4B (Cytiva). The final antibody concentration was determined using OD280 measurement. Information on the molecule, antigen sequence, host species, specificity, reference, NCBI GenBank accession number, and RRID are summarized in Table 1. The dilution of antibodies in each experiment is described in the following subsections.
Details of primary antibodies
Immunofluorescence
Prior to transcardial perfusion, mice were deeply anesthetized with an overdose of pentobarbital (100 mg/kg, i.p.). Perfusion was performed using either glyoxal fixative, 2% paraformaldehyde (PFA), or 4% PFA. For glyoxal fixation, mice were first perfused with 5 ml of saline, followed by 60 ml of glyoxal fixative solution [9% (v/v) glyoxal (Merck), 8% (v/v) acetic acid], pH 4.0. In contrast, PFA perfusion was performed directly without saline preperfusion. Postfixation was carried out in the same fixative overnight. Subsequently, brains were immersed in a 30% sucrose solution/0.1 M PB, pH 7.2. Parasagittal sections (50 μm thick) were prepared from the whole brain or cerebellar vermis using a cryostat (CM1860, Leica Microsystems) and immunostained using the free-floating method in glass test tubes. Phosphate-buffered saline (PBS), pH 7.4, containing 0.1% Triton X-100 (PBST) was used as the dilution and washing buffer. Sections were blocked with 10% normal donkey serum (Jackson ImmunoResearch Laboratories) for 20 min and then incubated overnight at room temperature with a mixture of primary antibodies (1 µg/ml each). Following incubation, sections were washed and then incubated for 2 h at room temperature with Alexa Fluor 488, Cy3, and Alexa Fluor 647-labeled species–specific secondary antibodies (1:200; Jackson ImmunoResearch Laboratories; Thermo Fisher Scientific). After washing, sections were mounted on APS-coated glass slides (Matsunami Glass), air-dried, and coverslipped using ProLong Glass (Thermo Fisher Scientific).
Immunofluorescence using ultrathin sections
Under deep pentobarbital anesthesia (100 mg/kg body weight, i.p.), mice were transcardially perfused with 5 ml of saline, followed by 60 ml of glyoxal fixative solution. The brains were then dissected and postfixed in glyoxal fixative at 4°C for 2 h. Coronal brain sections (400 µm thick) were cut using a VT1000S vibratome (Leica Microsystems) and stained with 2% uranyl acetate at 4°C overnight. After dehydration in graded alcohol and acetone, the sections were embedded in Durcupan (Sigma-Aldrich) and polymerized at 60°C for 48 h. Ultrathin sections (100 nm thick) were prepared using an Ultracut ultramicrotome (UC7, Leica Microsystems) and mounted on silane-coated glass slides (New Silane II, Muto Pure Chemicals). After a brief etching with saturated sodium ethanolate (1–5 s), the sections were processed with ImmunoSaver (Nisshin EM) at 95°C for 30 min in a Decloaking Chamber (NxGen, Biocare Medical). PBST was used as the dilution and washing buffer. Sections were blocked with 10% normal donkey serum (Jackson ImmunoResearch Laboratories) for 20 min and then incubated overnight at room temperature with a mixture of primary antibodies (1 μg/ml each). After three washes, sections were incubated for 2 h at room temperature with Alexa Fluor 488-, Cy3-, and Alexa Fluor 647-labeled species–specific secondary antibodies (1:200; Jackson ImmunoResearch Laboratories; Thermo Fisher Scientific). Following washes, sections were air-dried and mounted with ProLong Glass (Thermo Fisher Scientific).
Immunofluorescence image acquisition and analysis
Whole-brain images were captured using a fluorescence microscope (BZ-X710, Keyence) equipped with a CFI PlanApo λ (10×/0.45) objective lens (Nikon). Double or triple immunofluorescence images were acquired with a confocal laser microscope (FV1200, Evident Scientific) equipped with 473, 559, and 647 nm diode lasers and a UPlanXApo (60×/1.42 oil) objective lens. For quantitative analysis, images were separated into individual channels and converted to 8 bit grayscale. Measurements, including a line-scan analysis, were performed using the MetaMorph software (Molecular Devices). In the present study, dendrites with a diameter greater than ∼2.5 µm were defined as the proximal dendrite, while those with a diameter less than ∼1.5 µm were defined as the spiny branchlet. For Figures 1, 2D, 4H, 5E, and 6D, regions of interest (ROIs) were defined manually on specific brain regions and compartments, and the average signal intensity (in arbitrary units, A.U.) was measured. For Figure 2N–P, immunopositive puncta were semiautomatically detected using the “Inclusive Threshold” and “Create Regions Around Objects” functions.
Pre-embedding immunoelectron microscopy
Under deep anesthesia with pentobarbital (100 mg/kg, i.p.), mice were transcardially perfused with 2% PFA/0.1 M PB, pH 7.2, for 10 min, and the brains were removed. Brains were then postfixed for 2 h in the same fixative, and parasagittal sections (50 μm thick) were prepared using a vibratome (VT1000S, Leica Microsystems). PBS, pH 7.4, containing 0.1% Tween 20 was used as the dilution and washing buffer. Sections were blocked with 10% normal goat serum (Nichirei Biosciences) for 20 min and incubated with guinea pig anti-STIM1 antibody (1 μg/ml) overnight at room temperature. After washing, sections were sequentially incubated with a biotin-SP–conjugated guinea pig IgG antibody (1:200, Jackson ImmunoResearch Laboratories) and streptavidin-conjugated Alexa Fluor 488-FluoroNanogold (1:200, Nanoprobes) for 2 h at room temperature. Sections were washed with HEPES buffer (50 mM HEPES, 200 mM sucrose, 5 N sodium hydroxide), pH 8.0, and then incubated with silver enhancement reagent (AURION R-Gent SE-EM; AURION) for 1 h. Sections were fixed with 1% osmium tetroxide solution on ice for 15 min and then stained with 2% uranyl acetate for 15 min, dehydrated in a graded ethanol series and n-butyl glycidyl ether, embedded in epoxy resin, and polymerized at 60°C for 24 h. Ultrathin sections (∼80 nm thick) in the plane parallel to the section surface were prepared with an ultramicrotome (UC7, Leica Microsystems). Serial sections were mounted on indium tin oxide-coated glass slides (IT5-111-50, Nanocs) and successively stained with 2% uranyl acetate and lead citrate. After washing, colloidal graphite (Ted Pella) was pasted on the glass slides to surround the ribbons. Images were acquired using a SEM with a backscattered electron beam detector at an accelerating voltage of 1.0 kV and a magnification of 15,000× (SU8240, Hitachi High Technologies). The densities of metal particles per unit area of the ER and cytoplasm were calculated using the MetaMorph software (Molecular Devices). Metal particles within 35 nm of the PM were considered as the PM labeling. The criteria for defining the proximal dendrite and spiny branchlet were consistent with those used in the immunofluorescence analysis (see above, Immunofluorescence image acquisition and analysis). For 3D reconstruction, SEM images were loaded into Image Pro 10 (Media Cybernetics), aligned, and the structures of interest were segmented by delineating their boundary contours to create 3D surface renderings.
Postembedding immunoelectron microscopy
Double-labeling postembedding immunogold EM was performed using LR gold-embedded brain samples (Tsuzuki et al., 2024). Under deep anesthesia with pentobarbital (100 mg/kg, i.p.), mice were transcardially perfused with 4% PFA/0.1 M PB, pH 7.2, for 10 min. The brains were then removed and postfixed for 2 h in the same fixative, and coronal sections (300 μm thick) were prepared using a vibratome (VT1000S, Leica Microsystems). The sections were cryoprotected with 30% glycerol in PB and frozen rapidly with liquid propane in the EM CPC unit (Leica Microsystems). Frozen sections were transferred to the AFS freeze-substitution unit (Leica Microsystems), where freeze substitution proceeded as follows: 0.5% uranyl acetate in methanol at −90°C for 24 h; the temperature was increased at 4°C/h to −45°C; and 100% methanol at −45°C for 30 min (three times). The following steps, including the infiltration with LR gold (Electron Microscopy Sciences), were conducted at −25°C. Samples were immersed successively with a mixture of methanol and LR Gold (1:1) for 1 h; 1:2 for 1 h; pure LR Gold for 1 h; and a mixture of pure LR Gold supplemented with 0.1% BENZIL (Electron Microscopy Sciences) overnight. After transferring to fresh LR Gold with 0.1% BENZIL, samples were polymerized under UV light for 48 h.
Ultrathin sections (90 nm thick) were mounted on nickel grids and etched with saturated sodium ethanolate for 1–5 s. Tris-buffered saline, pH 7.4, containing 0.03% Triton X-100 was used as the dilution and washing buffer. The sections were blocked with 2% normal goat serum (Nichirei Biosciences) for 20 min. Next, the sections were incubated with a STIM1 antibody (20 μg/ml) diluted in 1% normal goat serum overnight, followed by washes. The sections were then incubated with colloidal gold-conjugated (5 nm) anti-guinea pig IgG (1:100, British BioCell International) for 2 h. After extensive washes, the grids were blocked with 2% normal guinea pig serum for 20 min and subsequently incubated with either a SERCA2, IP3R1, or RyR1 antibody (20 μg/ml) diluted in 1% normal guinea pig serum overnight. The sections were then incubated with colloidal gold-conjugated (10 nm) anti-rabbit IgG (1:100, British BioCell International) for 2 h. Following washes, grids were rinsed with distilled water and sequentially stained with 1% OsO₄ for 15 min, 2% uranyl acetate for 5 min, and Reynolds’ lead citrate for 1 min. Finally, sections were imaged using a transmission electron microscope (JEM1400, JEOL) at 30,000× magnification.
Electrophysiology
Adolescent (P35–P40) control and STIM1-KD mice that had received AAV infection at P5 were used for analysis. Parasagittal cerebellar slices (250 μm thick) were prepared as described previously (Yamasaki et al., 2006; Miyazaki et al., 2017). The animals were decapitated following anesthesia with ∼5% isoflurane (Pfizer), and the cerebella were rapidly removed and placed in ice-cold artificial cerebrospinal fluid (ACSF) containing the following (in mM): 125 NaCl, 4.5 KCl, 2 CaCl2, 1 MgSO4, 1.25 NaH2PO4, 26 NaHCO3, and 20 glucose, bubbled with 95% O2 and 5% CO2. After cutting, slices were recovered for 30 min at 32°C and kept for up to 6 h at 25°C in ACSF. Whole-cell recordings were made from visually identified PCs using an upright microscope (BX51WI, Evident). Patch pipettes had 4–6 MΩ resistances when filled with an internal solution. The internal solution contained the following (in mM): 148 potassium gluconate, 10 HEPES, 10 NaCl, 0.5 MgCl2, 4 Mg-ATP, 0.4 Na3-GTP, and 0.1 Oregon Green BAPTA 488-1 hexapotassium salt (Thermo Fisher Scientific), pH 7.3. During the recordings, the slices were continuously perfused with ACSF that contained 10 μM bicuculline (Tocris Bioscience). Ca2+-free ACSF contained (in mM) 0.1 mM ethylene glycol-bis (β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (Sigma-Aldrich), 20 glucose, 4.5 KCl, 3 MgSO4, 125 NaCl, 26 NaHCO3, and 1.25 NaH2PO4. Local drug application was performed by a glass pipette with a resistance of ∼6 MΩ, which was connected to a Pneumatic Picopump (PV800, World Precision Instruments). The mGluR1/5 agonist dihydroxyphenylglycine (DHPG; 400 μM; Selleck Chemicals) was dissolved in ACSF, and RyR agonist caffeine (80 mM, Tocris Bioscience) was diluted in caffeine ringer (Chen-Engerer et al., 2019), which contained the following (in mM): 2 CaCl2, 10 HEPES, 2.5 KCl, 1 MgCl2, 120 NaCl, and 1.25 NaH2PO4. The pipette tip was placed at a distance ∼20 μm from the slice surface. Ionic currents were recorded with a patch-clamp amplifier (EPC-10/2; HEKA Elektronik). The signals were filtered at 2 kHz and digitized at 20 kHz. The holding potential was −70 mV without liquid junction potential adjustment. Series resistance and leak currents were monitored continuously, and recordings were terminated if these parameters changed significantly. All experiments were performed at 31°C.
Ca2+ imaging
Imaging of transient Ca2+ changes in PC dendrites was started 25–30 min after establishing the whole-cell configuration. An upright microscope (BX51WI) equipped with a 40× objective (LUMPlanFl 40×/0.80 w; Evident Scientific) was used to acquire fluorescence images from dendritic fields in parallel to the patch-clamp recordings. Full-frame 1,280 × 480 pixel images were recorded at 38 Hz with a confocal unit (MEMS C15890-488S, MAICO, Hamamatsu) controlled by the HCI Image software (Hamamatsu). The recorded frames were further analyzed using ImageJ (http://www.macbiophotonics.ca/imagej/index.htm). The fluorescence intensity in each dendritic ROI was corrected by background subtraction. An ROI outside of the PC arbor was taken as the background. Temporal fluorescence intensity changes in ROIs were expressed as relative percentage changes in fluorescence intensity: ΔF/F (%) = [(F − F0)/F0]. F0 is defined as baseline fluorescence, which means the fluorescence intensity before a given stimulus, and F is the fluorescence change over time. ΔF/F (%) values were calculated and plotted using Igor Pro 5.05 (WaveMetrics).
Statistical analysis
Unless otherwise noted, all values are expressed as the mean ± SEM (where n is the number of analyzed neuronal profiles, cells, or ROIs). Graphing and statistical tests were performed using GraphPad Prism 10 (GraphPad Software). Normality was determined with D’Agostino and Pearson's, Anderson–Darling, Shapiro–Wilk, or one-sample Kolmogorov–Smirnov tests. In all cases, the assumption of normality was not met for all groups; statistics were performed using nonparametric tests. Two groups were compared using a two-tailed unpaired Mann–Whitney U test for independent samples or a two-tailed paired Wilcoxon matched-pair signed–rank test for paired data. For three or more group comparisons, we conducted the Kruskal–Wallis test. If significant differences were detected, Dunn's test assessed post hoc multiple comparisons. In all figures, statistical significance is presented as *p < 0.05; **p < 0.01; ***p < 0.001; and ****p < 0.0001. Details on the number of observations and statistical tests are provided in each figure or in Table 2.
Statistical tests and p values, not provided in the figure, figure legend, or Results section
Availability of data and antibodies
The datasets for this study are available from the corresponding author upon reasonable request. Antibodies used in this study that are presented with RRID can be purchased from Nittobo Medical (https://nittobo-nmd.co.jp/english/product/product_list.php) after publication of this study. Other antibodies are also available from the corresponding author upon reasonable request.
Results
Specificity and distribution of STIM1 immunolabeling
To investigate STIM1 localization, we raised an antibody against the 23–62 amino acid residues in the N-terminus of mouse STIM1, which is located in the ER lumen and shows low sequence homology with STIM2 (Fig. 1A). We compared staining patterns using 4% PFA and two fixatives suitable for detecting transmembrane proteins: 2% PFA and 9% glyoxal (Richter et al., 2018; Konno et al., 2023; Tsuzuki et al., 2024). In all fixatives, immunoreactivities for STIM1 were widely distributed throughout the brain, with marked expression in the cerebellum (Fig. 1B–D). To compare the expression pattern across brain regions, we quantified the mean relative intensity of the immunosignal in each area and normalized it to that in Layers II/III of the somatosensory cortex. We found that the relative intensity in the cerebellum was above 2, indicating the highest expression in this region, while the intensities in the other areas were ∼1 (Fig. 1E–G, left). At higher magnification, intense expression was observed in the soma and dendrites of cerebellar PCs (Fig. 1E–G, right). We also noticed that dendritic shafts were decorated with bright punctate staining (Fig. 1E–G, right, insets). To validate the specificity of STIM1 immunolabeling, we downregulated STIM1 expression in PCs using an AAV-mediated knockdown system with miRNA (Fig. 1H–J). We injected AAV vectors expressing EGFP and miRNA for STIM1-KD, driven by the PC-specific L7 promoter, into the mouse cerebellum at P1. Four weeks postinfection, we performed double immunofluorescence for STIM1 and EGFP under the three fixation conditions (Fig. 1H–M). We set the ROI ∼PC somata and proximal dendrites, measured the fluorescent intensity of STIM1, and compared the gray levels between EGFP-negative control PCs and EGFP-positive STIM1-KD PCs. We confirmed a significant reduction in STIM1 signals in STIM1-KD PCs in all fixatives (Fig. 1K–M; *p < 0.05; **p < 0.01, Wilcoxon matched-pair signed–rank test), validating the specificity of STIM1 labeling.
Specificity of STIM1 immunostaining in cerebellar PCs. A, Antigen information for the STIM1 antibody produced in this study. The underlined amino acid sequence of STIM1 was used for polyclonal antibody production. This region exhibits low homology to STIM2. B–G, Comparing immunoreactivities for STIM1 in different fixatives. 4% PFA- (B, E), 2% PFA- (C, F), and glyoxal- (D, G) fixed brain sections. E–G, Left, Bar graphs showing the relative intensities of signals in various brain regions. After converting color images to grayscale, the relative intensities were calculated from the measured gray levels (in A.U.) in the designated brain regions. These values were then normalized to the intensity in Layer II/III of the primary somatosensory cortex (S1), indicated by a dashed line. n = 6 sections (from 3 mice). Data are shown in mean ± SEM. E–G, Right, Immunoreactivities for STIM1 in cerebellar PCs. Higher-magnification images show punctate signals along proximal dendrites (insets). H–J, Double immunofluorescence for EGFP (green) and STIM1 (red) in the control and STIM1-KD PCs. Asterisks and arrows indicate the soma and dendrites, respectively, in knockdown PCs. K–M, Summary plots comparing the averaged fluorescent intensity in KD PCs (green dots) and adjacent control PCs (white dots) in the same image. The numbers of images examined, which were used for statistical testing, are as follows: 4% PFA (n = 6 images/4 sections/2 mice); 2% PFA (n = 10 images/3 sections/2 mice); glyoxal (n = 10 images/3 sections/2 mice). *p < 0.05; **p < 0.01; Wilcoxon pairwise comparisons. Other abbreviations, Cb, cerebellum; Cx, cerebral cortex; Hi, Hippocampus; Mb, midbrain; ML, molecular layer; MO, medulla oblongata; OB, olfactory bulb; PCL, Purkinje cell layer; Str, striatum; Th, thalamus.
Preferential localization of STIM1 in dendritic shafts over spines
We next examined the subcellular localization of STIM1 by immunofluorescence using glyoxal-fixed samples, which enhances antibody penetration and thus enables more precise and efficient analysis (Konno et al., 2023). To this end, we utilized two markers: calbindin to label the entire compartments of PCs and mGluR1 to preferentially label the dendritic spines of PCs (Mansouri et al., 2015; Yamasaki et al., 2021). Triple immunofluorescence for STIM1, calbindin, and mGluR1 revealed that STIM1 immunoreactivity was detected in the PC soma and dendrite (Fig. 2A). At higher magnification, STIM1 immunoreactivity was observed as discrete puncta along the dendritic shaft of proximal thick dendrite (Fig. 2B) and distal thin dendrite bearing numerous spines (i.e., spiny branchlet; Fig. 2C). In contrast, STIM1 immunoreactivity was quite low in the dendritic spine labeled for mGluR1 (Fig. 2C, arrows). To compare STIM1 expression levels in subcellular compartments, we delineated the soma, proximal dendrite, spiny branchlet, and dendritic spine based on calbindin immunoreactivity and measured the average signal intensity in each compartment (Fig. 2D). We found that the signal intensity in the proximal dendrite and spiny branchlet was significantly higher than in the soma and dendritic spine (Fig. 2D; *p < 0.05; ****p < 0.0001, Kruskal–Wallis test and post hoc Dunn's multiple comparisons). By contrast, there was no significant difference between the soma and dendritic spine or between the proximal dendrite and spiny branchlet [Fig. 2D; p = 0.89 (soma vs spine), p = 0.99 (proximal dendrite vs spiny branchlet), Kruskal–Wallis test and post hoc Dunn's multiple comparisons)]. To examine the subcellular localization of STIM1 more in detail, we performed a line-scan analysis of STIM1 signal intensity along both cross (Fig. 2E,F–I) and tangential (Fig. 2E,J–M) lines through each compartment. The intensity plot along the cross line showed that STIM1 was widely distributed in the soma without prominent peaks (Fig. 2F). In contrast, several prominent peaks, likely corresponding to the STIM1-positive puncta, were observed near the surface in the proximal dendrite (Fig. 2G) and spiny branchlet (Fig. 2H), but not in the dendritic spine (Fig. 2I). Similarly, the intensity plot along the tangential line on the surface showed no discernible peaks in the soma (Fig. 2J). In comparison, there were repetitive peaks along the proximal dendrite (Fig. 2K) and spiny branchlet (Fig. 2L), but not in the dendritic spine (Fig. 2M). When we counted the number of STIM1-positive puncta, it was more densely distributed in the proximal dendrite and the spiny branchlet than in the soma (Fig. 2N; *p < 0.05; **p < 0.01; Kruskal–Wallis test and post hoc Dunn's multiple comparisons). There was no significant difference in the density of STIM1-positive puncta between the proximal dendrite and the spiny branchlet (Fig. 2N; p = 0.99). Moreover, the size (Fig. 2O) and signal intensity (Fig. 2P) of STIM1-positive puncta in the proximal dendrite and spiny branchlet were similar (p = 0.30 for size and p = 0.10 for signal intensity, Mann–Whitney U test). These findings demonstrate that STIM1 exhibits a distinct localization pattern, forming clusters of similar density and intensity in the proximal dendrite and spiny branchlets, while no such clusters are observed in the soma or dendritic spines, where only diffuse or minimal staining is present.
Subcellular distribution of STIM1 in cerebellar PCs. A–C, Single optical section images showing triple immunofluorescence for STIM1 (red), mGluR1α (green), and calbindin (blue). B, Higher-magnification images of the boxed region in A, showing the punctate localization of STIM1 along the proximal dendrite and spiny branchlet. C, Enlarged images of the boxed region in B show that the dendritic spines labeled with mGluR1α (arrows) lack intense STIM1 expression. (D) Summary bar graphs comparing the fluorescent intensity of STIM1 immunoreactivity in each PC compartment. The numbers of profiles examined, which were used for statistical testing, are as follows: n = 9 somata, 9 proximal dendrites (PrDn), 9 spiny branchlets (SpBr), and 38 dendritic spines (Sp) from three images in two mice. *p < 0.05; ****p < 0.0001; Kruskal–Wallis test and post hoc Dunn's multiple comparisons. E, Representative images of the soma (left), dendrites (middle), and spines (right), excerpted from A for a line-scan analysis, are shown in grayscale. F–M, STIM1 signal intensities along the dotted cross lines (green) and tangential lines (magenta) in E were plotted. N, Summary bar graphs comparing the density of STIM1 puncta in each compartment (n = 4 soma, 4 PrDn, and 13 SpBr from 4 images in 2 mice). *p < 0.05; **p < 0.01; Kruskal–Wallis test and post hoc Dunn's multiple comparisons. O, P, Summary bar graphs comparing the size (O) and signal intensity (P) of STIM1 puncta in the PrDn and SpBr (n = 62 puncta in 4 PrDn and 53 puncta in 13 SpBr from 4 images in 2 mice). Mann–Whitney U test. Data are shown in mean ± SEM. Additional information on p values for this and all the following figures are provided in Table 2.
Intracellular localization of STIM1 by pre-embedding immunoelectron microscopy
To determine the subcellular localization of STIM1, we performed pre-embedding silver–enhanced immunoelectron microscopy using serial ultrathin sections (Fig. 3). In the soma, metal particles for STIM1 were predominantly found in the cytoplasm (97 out of 105 particles, 92.3%), where they exhibited a diffuse distribution (Fig. 3A). A smaller fraction (8 out of 105 particles, 7.6%) was associated with the subsurface ER, which typically forms parallel stacks beneath the PM (Rosenbluth, 1962). These particles were detected in 5 out of 34 subsurface ER structures (14.7%). Only a minor proportion of STIM1 particles (5 out of 105, 4.8%) were attached to the PM or the ER membrane.
Pre-embedding immunoelectron microscopy for STIM1 and a partial reconstruction showing the distribution of STIM1-associated ER. A–C, Pre-embedding immunoelectron microscopy for STIM1 acquired with a scanning electron microscope. Three consecutive images in the soma (A), proximal dendrite (PrDn, B), and spiny branchlet (SpBr, C) are shown. Metal particles for STIM1 are encircled with magenta lines, and arrows indicate ER cisternae. PC (soma, PrDn, SpBr), CF, and PF are pseudocolored in pale green, blue, and yellow, respectively. D, Summary bar graphs showing the density of STIM1 particles in the subsurface ER (Sub ER) and cytoplasm (Cyto) in each PC compartment. The numbers of profiles examined, which were used for statistical testing, are as follows: 5 soma, 11 PrDn, and 17 SpBr from two mice. ****p < 0.0001, Kruskal–Wallis test and Dunn's multiple-comparison test. Data are shown in mean ± SEM. E–H, Partial reconstruction of the PrDn (E, F) and SpBr (G, H) using consecutive 37 ultrathin section images, including B and C. Reconstructed PC, CF, ER, and STIM1 are shown in pale green, blue, green, and magenta, respectively. The boxed region in G is enlarged in H.
In contrast, in the proximal dendrite, most STIM1 particles were detected in the subsurface ER (312 out of 360 particles, 86.7%), while they were rarely observed on the PM (3 out of 360 particles, 0.83%; Fig. 3B). Furthermore, subsurface ER structures were consistently labeled with STIM1 particles (61 out of 77, 79.2%; Fig. 3B). STIM1 particles were enriched around the subsurface ER, with 18.5 ± 1.8 metal particles/µm2 on the subsurface ER compared with only 0.47 ± 0.01 metal particles/µm2 in the cytoplasm [p < 0.0001; Mann–Whitney U test; n = 11 dendrites; 312 particles (subsurface ER) and 48 particles (cytoplasm)], highlighting the preferential association of STIM1 with the subsurface ER. In contrast, STIM1 particles were not observed in the dendritic spine (n = 23 spines; Fig. 3B). A similar distribution pattern was observed in the spiny branchlet (Fig. 3C). Most STIM1 particles were detected in the subsurface ER (90 out of 115 particles, 78.3%), while only a small fraction was detected on the PM (5 out of 115 particles, 4.3%). The subsurface ER was frequently labeled with STIM1 particles (25 out of 40, 62.5%; Fig. 3C). STIM1 particles were enriched around the subsurface ER, with 30.3 ± 5.5 metal particles/µm2 on the subsurface ER compared with 5.1 ± 2.2 metal particles/µm2 in the cytoplasm [p < 0.001; Mann–Whitney U test; n = 17 dendrites; 90 particles (subsurface ER) and 25 particles (cytoplasm)], further supporting the preferential association of STIM1 with the subsurface ER. In contrast, STIM1 was rarely detected in the dendritic spine (2 out of 70 spines; 2.9%; Fig. 3C). The labeling density in the subsurface ER of the proximal dendrite and spiny branchlet was significantly higher than in the soma (****p < 0.0001; Kruskal–Wallis test and Dunn's multiple comparisons test), whereas the labeling densities in the deep cytoplasm did not vary among compartments (p = 0.32; Kruskal–Wallis test; Fig. 3D). Three-dimensional (3D) reconstruction of serial ultrathin sections confirmed that STIM1-associated ER is distributed adjacent to the PM in the proximal dendrite (Fig. 3E,F) and spiny branchlet (Fig. 3G,H). These observations collectively suggest that STIM1 preferentially accumulates on and around the subsurface ER in the dendritic shaft of the proximal dendrite and spiny branchlet.
STIM1 expression in the SERCA2-positive structures
Given that SERCA is required for the efficient Ca2+ loading to the ER during SOCE (Sampieri et al., 2009; Manjarres et al., 2010; Soboloff et al., 2012), we next compared the distribution of STIM1 and SERCA2, the most dominant subtype in PCs (Baba-Aissa et al., 1998). We raised an antibody against SERCA2 and confirmed that the immunoreactivity was intense in the soma and dendrites of PCs (Fig. 4A) and diminished in the KD PCs (Fig. 4A–C). Triple immunofluorescence for STIM1, SERCA2, and calbindin showed that intense immunoreactivity for SERCA2 was detected in the soma, proximal dendrite, and spiny branchlet, but only weakly in the spines (Fig. 4D–G). This finding is consistent with a previous report (Takei et al., 1992) and is further supported by quantitative analysis of fluorescent signals across subcellular compartments (Fig. 4H). The signal intensities in the soma, proximal dendrite, and spiny branchlet were much higher than those in the dendritic spine [*p < 0.05 (spiny branchlet vs spine); ***p < 0.0001 (soma or proximal dendrite vs spine); Kruskal–Wallis test and post hoc Dunn's multiple comparisons; Fig. 4H]. To compare the distribution pattern of STIM1 and SERCA2, we performed a line-scan analysis of the signal intensity along the cross line in each compartment (Fig. 4I–M). The intensity plot across the soma showed that SERCA2 signal was intense, widely distributed, and wholly overlapped with STIM1 (Fig. 4J). In the proximal dendrite and spiny branchlet, the intensity plot of SERCA2 yielded similar peaks near the surface and mostly overlapped, but not perfectly aligned, with STIM1 (Fig. 4K,L). Both SERCA2 and STIM1 signals were below the detection threshold in the dendritic spine (Fig. 4M). To enhance resolution, we further performed immunofluorescence on ultrathin sections (100 nm thick; Hirai et al., 2024; Fig. 4O–T). SERCA2 labeling appeared as a punctate and reticular pattern, extending from the soma to the dendrites (Fig. 4O,P). Line-scan analysis across the soma (Fig. 4Q), proximal dendrite (Fig. 4R), and spiny branchlet (Fig. 4T) revealed recurrent intensity peaks, indicating that SERCA2 is localized to distinct interspersed structures. Although not perfectly aligned with STIM1, SERCA2 exhibited similar peak distributions near the surface in the proximal dendrite (Fig. 4P,R) and spiny branchlet (Fig. 4S,T). These findings indicate a close spatial relationship between SERCA2 and STIM1 in the soma and dendritic shaft, particularly near the dendritic surface. However, a key distinction was that STIM1 exhibited a more restricted distribution than SERCA2 in the somatodendritic compartment, notably forming large bright puncta in dendrites—a feature unique to STIM1.
Comparing the subcellular localization of STIM1 with SERCA2. A–C, Specificity of SERCA2 immunolabeling. Double immunofluorescence for EGFP (green) and SERCA2 (red) in the control and SERCA2-KD PCs (A, B). Asterisks and arrows indicate SERCA2-KD PCs somata and dendrites, respectively. A boxed region in A is enlarged in B. A dotted line delineates the contour of KD PC dendrites. SERCA2-labeling intensity significantly decreases in SERCA2-KD PCs (C; n = 18 images/3 sections/2 mice). ***p < 0.001, Wilcoxon pairwise comparisons. D–G, Single optical section images showing triple immunofluorescence for STIM1 (red), SERCA2 (green), and calbindin (blue). E, F, Higher-magnification images of the boxed region in D show the soma (E), proximal dendrite (PrDn), and spiny branchlet (SpBr). G, Enlarged images of the boxed regions in F show that dendritic spines (Sp) labeled with calbindin (arrows) lack intense SERCA2 expression. Contours of dendritic shafts and spines are drawn based on calbindin signals. Note that STIM1 puncta (arrowheads) are associated with SERCA2 labeling. H, Summary bar graphs comparing the fluorescent intensity of STIM1 immunoreactivity in each PC compartment (n = 6 soma, 6 PrDn, 9 SpBr, and 20 Sp from 4 images in 2 mice). *p < 0.05; ***p < 0.001; Kruskal–Wallis test and post hoc Dunn's multiple comparisons. Data are shown in mean ± SEM. I, Representative images of the soma (left), PrDn and SpBr (middle), and Sp (right), excerpted from (D) and (F) for a line-scan analysis. J–M, Signal intensities of STIM1 (red) and SERCA2 (green) along the dotted cross lines in (I) are plotted. O–T, Immunofluorescence on ultrathin (100 nm thick) sections for improved resolution and line-scan analysis. (Q–T) Signal intensities of STIM1 (red) and SERCA2 (green) along the dotted cross lines in (P, S) are plotted.
STIM1 expression in the IP3R1 or RyR1-positive structures
We next examined whether STIM1 was localized to distinct domains of intracellular Ca2+ stores in PCs. To this end, we analyzed STIM1 expression using the two major markers for the intracellular Ca2+ stores in PCs: IP3R1 and RyR1 (Figs. 5, 6). Triple immunofluorescence for STIM1, calbindin, and IP3R1 showed that IP3R1 was intense in PCs and detectable in the entire compartments of the soma, dendritic shaft, and dendritic spine (Fig. 5A–D). Higher-magnification images revealed that the expression levels of IP3R1 varied across the subcellular compartments (Fig. 5B–D). When we compared the signal intensity of IP3R1 in each compartment, they were comparable in the soma, proximal dendrite, and spiny branchlet but significantly lower in the dendritic spine [Fig. 5E; **p < 0.01 (soma vs spine); ***p < 0.001 (spiny branchlet vs spine); ****p < 0.0001 (proximal dendrite vs spine); Kruskal–Wallis test and post hoc Dunn's multiple comparisons]. To compare the distribution pattern of IP3R1 and STIM1, we performed a line-scan analysis of the signal intensity along the cross line in each compartment (Fig. 5F–J). The intensity plot along the cross line showed that IP3R1 was widely distributed in the soma and almost completely overlapped with STIM1 (Fig. 5G). In the proximal dendrite and spiny branchlet, the intensity plot of IP3R1 showed a broad peak across the entire cytoplasm, flanked by two STIM1 peaks just beneath the surface (Fig. 5H,I). Accordingly, the signals for IP3R1 and STIM1 were juxtaposed and partially overlapped in the periphery of the dendritic shaft (Fig. 5H,I), whereas STIM1 signals were below the detection threshold and did not coincide with IP3R1 in the inner domain of the dendritic spine (Fig. 5J). To achieve higher resolution, we performed immunofluorescence on ultrathin sections (Fig. 5K–Q). IP3R1 labeling displayed a punctate and reticular pattern extending from the soma to the dendrites (Fig. 5F). Line-scan analysis across the soma (Fig. 5L,M), proximal dendrite (Fig. 5N,O), and spiny branchlet (Fig. 5P,Q) revealed recurrent intensity peaks, suggesting that IP3R1 is clustered in distinct interspersed structures. In the proximal dendrite and spiny branchlet, IP3R1 and STIM1 signals were juxtaposed and partially overlapped near the dendritic surface (Fig. 5O,Q). Collectively, these findings highlight a notable spatial association between IP3R1 and STIM1, particularly near the surface of the proximal dendrite and spiny branchlet.
Comparing the subcellular localization of STIM1 with IP3R1. A, Single optical section images showing triple immunofluorescence for STIM1 (red), IP3R1 (green), and calbindin (blue). B–D, Higher-magnification images of the boxed region in A, showing the soma (B), proximal dendrite (C), and spiny branchlet (D). D, Enlarged images of the boxed region in C show that the dendritic spines labeled with calbindin (arrows) had notable IP3R1 expression. E, Summary bar graphs comparing the fluorescent intensity of IP3R1 immunoreactivity in each PC compartment (n = 10 soma, 11 proximal dendrites, 14 spiny branchlets, and 20 dendritic spines from 4 images in 2 mice). **p < 0.01; ***p < 0.001; ****p < 0.0001; Kruskal–Wallis test and post hoc Dunn's multiple comparisons. Data are shown in mean ± SEM. F, Representative images of the soma (left), dendrites (middle), and spines (right), excerpted from B and C for a line-scan analysis. A white line delineates the spine contours, drawn based on calbindin signals. G–J, Signal intensities of STIM1 (red) and IP3R1 (green) along the dotted cross lines in (F) are plotted. K–Q, Immunofluorescence on ultrathin (100 nm thick) sections for improved resolution and line-scan analysis. M, O, Q, Signal intensities of STIM1 (red) and IP3R1 (green) along the dotted cross lines in L, N, and P are plotted.
Comparing the subcellular localization of STIM1 with RyR1. A–C, Single optical section images showing triple immunofluorescence for STIM1 (red), RyR1 (green), and calbindin (blue). B, C, Higher-magnification images of the boxed region in A, showing the soma (B), proximal dendrite, and spiny branchlet (C). D, Summary bar graphs comparing the fluorescent intensity of RyR1 immunoreactivity in each PC compartment (n = 7 soma, 6 proximal dendrites, 7 spiny branchlets, and 44 dendritic spines from 4 images in 2 mice). **p < 0.01; ***p < 0.001; Kruskal–Wallis test and post hoc Dunn's multiple comparisons. Data are shown in mean ± SEM. E, Representative images of the soma (left), dendrites (middle), and spines (right), excerpted from B and C for a line-scan analysis. Contours of spines are drawn based on calbindin signals. F–I, Signal intensities of STIM1 (red) and RyR1 (green) along the dotted cross lines in E are plotted. J–P, Immunofluorescence on ultrathin (100 nm thick) sections for improved resolution and line-scan analysis. L, N, P, Signal intensities of STIM1 (red) and RyR1 (green) along the dotted cross lines in K, M, and O are plotted.
Triple immunofluorescence for STIM1, calbindin, and RyR1 showed that RyR1 was abundantly expressed in the soma and proximal dendrite of PCs (Fig. 6). Higher-magnification images revealed distinct distribution patterns for RyR1 and IP3R1, with RyR1 showing prominent expression in the perikarya but only weak signals in other compartments (Fig. 6B,C). Quantification of the signal intensity in the subcellular compartments confirmed this observation (Fig. 6D). RyR1 expression in the dendritic spine was significantly lower than in the soma and proximal dendrite [Fig. 6D; **p < 0.01 (proximal dendrite vs spine); ***p < 0.001 (soma vs spine); Kruskal–Wallis test and post hoc Dunn's multiple comparisons]. To compare the distribution pattern of RyR1 and STIM1, we performed a line-scan analysis of the signal intensity along the cross line in each compartment (Fig. 6E–I). The intensity plot along the cross line showed that RyR1 was widely distributed in the soma with several peaks and showed a notable overlap with STIM1 (Fig. 6E,F, left). In contrast, in the proximal dendrite, RyR1 showed a nonoverlapping pattern with, but close apposition to, STIM1 in both immunofluorescence image and intensity plot (Fig. 6E,G, middle). In the spiny branchlet, the signal intensity of RyR1 was low and did not overlap with STIM1 (Fig. 6E,H, middle). In the dendritic spine, signals for RyR1 or STIM1 were not detected (Fig. 6E,I, right). To achieve finer spatial detail, we performed immunofluorescence on ultrathin sections (Fig. 6J–P). Unlike SERCA2 (Fig. 4O) and IP3R1 (Fig. 5K), RyR1 exhibited a more scattered and discontinuous distribution (Fig. 6J). Line-scan analysis across the soma (Fig. 6K,L) and proximal dendrite (Fig. 6M,N) showed alternating intensity peaks for RyR1 and STIM1 (Fig. 6L,N). No clear intensity peaks were observed in the spiny branchlet (Fig. 6O,P). These findings indicate that the notable colocalization of STIM1 with RyR1 is restricted to the perikarya and deep cytoplasm, away from the PM.
STIM1 preferentially colocalizes with IP3R and SERCA2 in the subsurface ER
We next examined the expression of SERCA2, IP3R1, and RyR1 in the STIM1-labeled subsurface ER structure at the ultrastructural level (Fig. 7). Double-labeling postembedding immunogold EM for STIM1 and SERCA2 (Fig. 7A,B) revealed that immunogold labeling for SERCA2 was associated with the STIM1-labeled subsurface ER (Fig. 7B). Similarly, double-labeling for STIM1 and IP3R1 (Fig. 7C,D) showed that IP3R1 was distributed along the ER extending from the dendritic shaft into the dendritic spine (Fig. 7D). While IP3R1 was present around the ER in dendritic spines, STIM1 remained restricted to the dendritic shaft, where it was particularly enriched on the subsurface ER membrane (Fig. 7D). In contrast, double-labeling for STIM1 and RyR1 (Fig. 7E–H) demonstrated that STIM1- and RyR1-positive ER structures were largely segregated, with minimal colocalization (Fig. 7H). These findings indicate that STIM1 preferentially localizes to ER domains associated with IP3R1 and SERCA2, while its spatial segregation from RyR1 underscores the compartmentalization of ER Ca2+ signaling domains.
Double-labeling postembedding immunogold EM for STIM1 and SERCA2/IP3R1/RyR1 in the subsurface ER. A, B, Double-labeling immunogold EM for STIM1 (5 nm, magenta arrows) and SERCA2 (10 nm, green arrows) in the proximal dendrite. The boxed region in A highlights a subsurface ER and is enlarged in B. C, D, Double-labeling immunogold EM for STIM1 (5 nm, magenta arrows) and IP3R1 (10 nm, green arrows) in the spiny branchlet. The boxed region in D highlights a subsurface ER and is enlarged in B. Note that the ER cisterna extends into the dendritic spine. E–H, Double-labeling immunogold EM for STIM1 (5 nm, magenta arrows) and RyR1 (10 nm, green arrows) in the proximal dendrite. The boxed regions in E highlight subsurface ER and are enlarged in F–H. Note that STIM1 labeling in the RyR1-labeled ER cisterna is weak. Arrowheads flank the edges of PSDs. The contours of dendrites are delineated by thick yellow lines, and ER lumens are pseudocolored in pale blue.
mGluR1-mediated Ca2+ release from internal Ca2+ store depends on STIM1 and refilling of Ca2+ stores at resting potential
Finally, we investigated the role of STIM1 in Ca2+ signaling in PCs by combining whole-cell recordings with confocal Ca2+ imaging to examine mGluR1- and RyR-mediated responses in PC dendrites. Since STIM1 facilitates ER Ca2+ store replenishment by mediating extracellular Ca2+ influx at resting potential over several minutes (Hartmann et al., 2014), we first examined how extracellular Ca2+ depletion affects mGluR1- and RyR1-mediated Ca2+ release in control PCs to assess the contribution of STIM1-driven refilling to these receptor-mediated signals. In normal ACSF containing 2 mM Ca2+, dendritic puff application of the mGluR1 agonist DHPG (400 μM, 8 psi, 300 ms) elicited large dendritic Ca2+ transients at −70 mV (Fig. 8A). However, these DHPG-evoked Ca2+ transients were markedly reduced after switching to Ca2+-free ACSF (Fig. 8A, right), decreasing to 14.5 ± 3.4% of the control level within 20 min (n = 7 PCs). The transients recovered upon returning to Ca2+-containing ACSF (Fig. 8A, bottom), reaching 97.0 ± 8.6% of the control level within 20 min (n = 7 PCs), suggesting that mGluR1-mediated Ca2+ release requires continuous Ca2+ refilling from extracellular sources. In contrast, local application of the RyR agonist caffeine (80 mM, 8 psi, 2 s) also induced large dendritic Ca2+ transients in normal ACSF (Fig. 8C, left), but these responses remained largely unchanged after switching to Ca2+-free ACSF (Fig. 8C, right), maintaining 80.6 ± 7.5% of the control level after 16 min. This indicates that RyR-mediated release is less dependent on acute extracellular Ca2+ availability.
Ca2+ imaging showing Ca2+ transients in control and STIM1-KD PC dendrites. A, B, Ca2+ imaging in control PCs. A, Top, Color-coded confocal images of a dendritic field in an Oregon Green BAPTA 488-1 (OGB1)-filled, voltage-clamped PC. Bottom, Relative changes in OGB1 fluorescence (ΔF/F) showing Ca2+ transients evoked by local dendritic puff application of DHPG (400 μM, 300 ms, 8 psi; applied at 3 min intervals) at the site indicated by the schematic pipette. Dashed lines delineate ROI. Baseline responses in normal ACSF containing 2 mM Ca2+ (left) and after switching to Ca2+-free ACSF at the indicated time points. B, Normalized amplitude of DHPG-evoked Ca2+ transients plotted over time. Amplitudes were normalized to the average of three responses before switching to Ca2+-free ACSF. Data were obtained from seven PCs in seven mice. C, D, Similar experiments using local dendritic puff application of caffeine (80 mM, 2 s, 8 psi; applied at 3 min intervals). Data were obtained from six PCs in six mice. E, STIM1-KD PCs identified by mCherry fluorescence. F–H, Ca2+ transients in control (black) and STIM1-KD (red) PCs. F, Ca2+ transients induced by a depolarizing pulse (1 s, from −70 to 0 mV) and summary comparing depolarization-induced ΔF/F. G, DHPG-evoked Ca2+ transients (400 μM, 300 ms, 8 psi) and summary comparing ΔF/F (right). H, Caffeine-evoked Ca2+ transients (80 mM, 2 s, 8 psi) and summary comparing ΔF/F (right). **p < 0.01, Mann–Whitney U test. The number of PCs tested is indicated in parentheses. Values are shown as mean ± SEM.
Next, we tested whether STIM1 reduction affects mGluR1- and RyR-mediated Ca2+ release in PC dendrites. In STIM1-KD PCs (Fig. 8E), depolarizing pulses produced Ca2+ transients comparable with those observed in control PCs (Fig. 8F; p = 0.11; Mann–Whitney U test), indicating that Ca2+ entry through voltage-gated Ca2+ channels is intact in the absence of STIM1. However, DHPG-evoked Ca2+ transients were nearly abolished in STIM1-KD PCs (Fig. 8G; **p < 0.01; Mann–Whitney U test), demonstrating that mGluR1-mediated Ca2+ release is highly dependent on STIM1-driven ER Ca2+ refilling. In contrast, caffeine application elicited Ca2+ transients in STIM1-KD PCs, but the magnitude was significantly smaller than that in control PCs (Fig. 8H; **p < 0.01; Mann–Whitney U test), suggesting that while RyR-mediated release is partially dependent on STIM1, it is sustained by additional mechanisms. These results collectively indicate that mGluR1- and RyR-mediated Ca2+ release differentially depend on STIM1-mediated ER Ca2+ store refilling. mGluR1-mediated Ca2+ release relies heavily on continuous Ca2+ replenishment via STIM1, whereas RyR-mediated release exhibits partial dependence, likely maintained by additional Ca2+ handling mechanisms.
Discussion
The ER forms an extensive intracellular network, acting as both a source and a sink of Ca2+ crucial for regulating intracellular Ca2+ signaling and homeostasis. To sustain Ca2+ signaling, the ER employs various mechanisms to maintain Ca2+ levels, including SOCE via STIM1. Despite STIM1's well-established role in Ca2+ store regulation and cerebellar function, its expression in PCs has been primarily confirmed at the mRNA level (Klejman et al., 2009; Hartmann et al., 2014; Allen Brain Atlas, available at https://portal.brain-map.org/). Limited protein-level confirmation through immunohistochemical analysis was performed merely to verify its presence in PCs, without detailing its subcellular localization (Klejman et al., 2009; Skibinska-Kijek et al., 2009; Hartmann et al., 2014; Ryu et al., 2017; Dhanya and Hasan, 2021), leaving its subcellular distribution yet to be fully characterized. By combining specific antibody production with optimized fixation protocols (Konno et al., 2023; Tsuzuki et al., 2024), we demonstrated here that STIM1 is preferentially localized to the subsurface ER structures in the dendritic shaft of PCs but excluded from the dendritic spine, the primary site of IP3-induced Ca2+ release (Finch and Augustine, 1998; Takechi et al., 1998; Miyata et al., 2000). Furthermore, dendritic STIM1 coincides with IP3R1, but not with RyR1, in the subsurface ER stack. The main points of our observations are schematically illustrated in Figure 9.
Schematic diagram showing STIM1 distribution in ER subcompartments in cerebellar PCs. STIM1 is distributed in IP3R-positive ER compartments in the proximal dendrite and the spiny branchlet but excluded from spines, the primary site for IP3-induced Ca2+ release. In these compartments, STIM1 is closely associated with SERCA2. By contrast, its colocalization with RyR1 is restricted to the soma. This differential localization suggests that PCs possess specialized ER subcompartments for Ca2+ release and refilling.
The ER Ca2+ stores consist of spatially and functionally distinct compartments (Blaustein and Golovina, 2001). Multiple lines of evidence indicate that RyR and IP3R are localized to distinct subcompartments of ER Ca2+ stores (Mignery et al., 1989; Ellisman et al., 1990; Satoh et al., 1990; Walton et al., 1991) and regulate separate Ca2+ pools (Chen-Engerer et al., 2019), though some studies suggest potential overlap (Khodakhah and Armstrong, 1997). Further supporting this compartmentalization, coordinated units termed “plasmerosomes,” assembled from subsurface ER structures and overlaying small PM microdomains, have been postulated to facilitate distinct regulatory functions (Blaustein and Golovina, 2001). In PCs, calsequestrin and SERCA colocalize in IP3R-enriched ER subdomains (Villa et al., 1991; Takei et al., 1992) and are closely associated with Ca2+-activated potassium channels on the PM (Kaufmann et al., 2009; Indriati et al., 2013). In contrast, RyR-enriched ER subdomains are physically linked to voltage-gated Ca2+ channels on the PM via junctophilin (Kakizawa et al., 2007). Our findings indicate that STIM1 predominantly forms an IP3R-enriched ER subdomain within the dendritic shaft while exhibiting a diffuse distribution in the soma and being largely absent from spines (Fig. 5). In contrast, involvement of STIM1 in the RyR-enriched ER subdomain appears to be primarily restricted to the soma (Fig. 6). This spatial organization aligns with the observation that STIM1 is not essential for store refilling via voltage-gated Ca2+ channels during depolarization (Hartmann et al., 2014), a process that likely involves RyR-sensitive stores, as demonstrated in hippocampal pyramidal cells (Chen-Engerer et al., 2019).
Consistent with this anatomical and molecular distinction, our Ca2+ imaging experiments further reveal that mGluR1-mediated Ca2+ release in PC dendrites is highly dependent on STIM1-driven ER store refilling, as DHPG-evoked Ca2+ transients were drastically reduced in STIM1-KD PCs (Fig. 8G). In contrast, RyR-mediated Ca2+ release exhibited only partial dependence on STIM1, as caffeine-evoked transients were diminished but not abolished (Fig. 8H). These functional findings support the segregation of STIM1-enriched and RyR-enriched ER subdomains, as demonstrated by our histological observations, further highlighting the differential reliance of IP₃R- and RyR-mediated Ca2+ stores on STIM1-dependent refilling. Notably, the requirement for continuous ER Ca2+ refilling from extracellular sources appears to differ between these two pathways. While RyR-mediated release can persist even in Ca2+-free external solutions (Kano et al., 1995; Fig. 8D), mGluR1–IP₃R-mediated release is strictly dependent on ongoing store replenishment (Hartmann et al., 2014; Fig. 8B). Although the precise mechanisms underlying the refilling of these distinct Ca2+ stores remain to be fully elucidated, our results provide converging evidence that IP₃R- and RyR-sensitive stores operate with a degree of functional independence in PCs.
ImmunoEM analysis further revealed that punctate localization of STIM1 along the dendritic surface reflects accumulation in the subsurface ER structures (Fig. 3). The subsurface ER comprises a single cisterna or junctional ER, which is typically located within 15–30 nm from the PM, and multiple cisternae stacked in parallel with the subsurface cistern (Henkart et al., 1976; Wu et al., 2017). As the subsurface cistern harbors ER–PM contact sites, it has been implicated in SOCE (Orci et al., 2009). We detected a few STIM1 particles at the ER–PM contact sites (Fig. 3), likely representing STIM1 oligomers formed upon ER store depletion and recruiting Orai channels. STIM1 particles were not restricted to ER–PM contact sites or ER membranes facing the PM; they were also observed around the ER cisternae facing the cytoplasm and between ER cisternae (Fig. 3). This widespread distribution may imply that a substantial pool of STIM1 is available in the ER, potentially awaiting recruitment to ER–PM junctions to mediate store refilling. This observation aligns with previous findings that STIM1 also facilitates cytosolic Ca2+ clearance during PC firing (Ryu et al., 2017). Together with double-labeling immunofluorescence (Fig. 4) and postembedding immunogold EM (Fig. 7) showing STIM1 accumulation near the dendritic surface alongside SERCA2, this arrangement defines a specialized site for Ca2+ refilling. These molecular and spatial arrangements align with the Ca2+ tunneling hypothesis that Ca2+ entering via SOCE is transported to distant release sites through ER-mediated Ca2+ flow (Petersen et al., 2017). Indeed, PF-induced Ca2+ depletion in the dendritic spine is quickly refilled through diffusion from the parent spiny branchlet (Okubo et al., 2015). The revealed STIM1 localization within ER subcompartments of the dendritic shaft would ensure efficient Ca2+ store refilling, which is essential for PF- and mGluR1-dependent Ca2+ signaling (Hartmann et al., 2014).
The stacks of parallel ER cisternae were initially considered fixation artifacts or indicators of neuronal damage (Herndon, 1964; Karlsson and Schultz, 1965; Bestetti and Rossi, 1980; Hansson, 1981). However, they have been consistently observed in samples that were optimally fixed (Rosenbluth, 1962; Harris and Stevens, 1988; Rusakov et al., 1993; Takei et al., 1994) as well as in brain homogenates (Takei et al., 1994). Furthermore, in hippocampal pyramidal cells, the subsurface ER stack is a functional assembly supporting the protein kinase A signaling and intracellular Ca2+ signaling driven by membrane depolarization (Vierra et al., 2023). Furthermore, the subsurface ER stack is considered dynamic and may reflect the functional state of PCs (Takei et al., 1994; Banno and Kohno, 1996; Ikemoto et al., 2003). Notably, it responds to changes in the microenvironment, increasing under anoxic conditions (Banno and Kohno, 1996; Ikemoto et al., 2003) and decreasing upon reoxygenation (Ikemoto et al., 2003). In PCs, IP3Rs aggregate between cisternae of the ER stacks (Otsu et al., 1990; Satoh et al., 1990; Takei et al., 1992, 1994), likely through head-to-head interactions involving their N-terminal domains (Takei et al., 1994; Chavda et al., 2013). This aggregation may reduce cytosolic Ca2+ levels by limiting IP3R access to the cytosol (Takei et al., 1994). A notable fraction of STIM1 particles was observed between the ER cisternae within the subsurface ER stacks (Fig. 3). STIM1 has been shown to mediate ER remodeling and regulate spatiotemporal calcium signals in filopodia during axon guidance (Pavez et al., 2019). Although not directly tested in this study, the accumulation of STIM1 in the dendritic ER cisterna may also contribute to structural changes, potentially limiting excessive Ca2+ release and regulating Ca2+ signaling during firing and anoxic conditions. As such, the consistent presence and dynamic changes of subsurface ER stacks across various conditions suggest their active role in both physiological and pathophysiological processes.
In conclusion, this study elucidated the previously unreported subcellular localization of STIM1 in PCs through detailed immunohistochemical analysis, and Ca2+ imaging data supported its distinct functional role. While similar observations have been reported separately, our findings integrate these aspects to provide new insights into STIM1 distribution and function. STIM1 was abundantly localized in the subsurface ER structures of the dendritic shaft, where it colocalized with IP3R1 and SERCA2. These findings suggest that intracellular Ca2+ stores in PCs comprise distinct compartments with diverse molecular characteristics, allowing for compartmentalized Ca2+ regulation that is crucial for synaptic plasticity and intracellular calcium homeostasis.
Footnotes
This work was supported by Ministry of Education, Culture, Sports, Science and Technology 20H03410 and 21K06746, 22K06784, 20H0562814, and 21H02589.
↵*S.N. and M.Y. contributed equally to this work.
The authors declare no competing financial interests.
- Correspondence should be addressed to Miwako Yamasaki at k-minobe{at}med.hokudai.ac.jp.
