Comparative Ca2+ channel contributions to intracellular Ca2+ levels in the circadian clock

Circadian rhythms in mammals are coordinated by the central clock in the brain, located in the suprachiasmatic nucleus (SCN). Multiple molecular and cellular signals display a circadian variation within SCN neurons, including intracellular Ca2+, but the mechanisms are not definitively established. SCN cytosolic Ca2+ levels exhibit a peak during the day, when both action potential firing and Ca2+ channel activity are increased, and are decreased at night, correlating with a reduction in firing rate. In this study, we employ a single-color fluorescence anisotropy reporter (FLARE), Venus FLARE-Cameleon, and polarization inverted selective-plane illumination microscopy to measure rhythmic changes in cytosolic Ca2+ in SCN neurons. Using this technique, the Ca2+ channel subtypes contributing to intracellular Ca2+ at the peak and trough of the circadian cycle were assessed using a pharmacological approach with Ca2+ channel inhibitors. Peak (218 ± 16 nM) and trough (172 ± 13 nM) Ca2+ levels were quantified, indicating a 1.3-fold circadian variance in Ca2+ concentration. Inhibition of ryanodine-receptor-mediated Ca2+ release produced a larger relative decrease in cytosolic Ca2+ at both time points compared to voltage-gated Ca2+channels. These results support the hypothesis that circadian Ca2+ rhythms in SCN neurons are predominantly driven by intracellular Ca2+ channels, although not exclusively so. The study provides a foundation for future experiments to probe Ca2+ signaling in a dynamic biological context using FLAREs.

Previous studies have implicated multiple Ca 2þ channel subtypes that contribute to Ca 2þ signaling in SCN neurons (27). Membrane depolarization stimulates Ca 2þ influx by activating voltage-gated Ca 2þ channels (VGCCs) including L-, N-, P/Q-, R-, and T-type channels (2)(3)(4)(5). Ca 2þ release from intracellular stores in the endoplasmic reticulum (ER) is mediated by ryanodine receptors (RyR2 and RyR3) (24,(36)(37)(38). IP 3 signaling stimulates Ca 2þ release from the ER by activating inositol 1,4,5-trisphosphate receptors (IP 3 Rs) (2,39). Prior studies have shown that cytosolic Ca 2þ levels may be mediated in part by Ca 2þ release from intracellular ER stores, as pharmacological inhibition of RyRs produces a large decrease in cytoplasmic Ca 2þ (24,40,41). However, Ca 2þ influx through the plasma membrane from voltage-gated Ca 2þ channels may also contribute. Inhibition of L-type voltage-gated Ca 2þ channels with nimodipine or action potential firing with the Na þ channel blocker tetrodotoxin partially reduce cytoplasmic Ca 2þ (40). To date, no single study has directly compared the contributions of the main plasma membrane and intracellular Ca 2þ channels at both the peak and trough of the circadian cycle from intact SCN slices.
This study utilizes a newly developed fluorescent biosensor to provide a quantifiable and direct comparison for the contributions of voltage-gated and intracellular Ca 2þ channels to daytime (peak) and nighttime (trough) Ca 2þ levels in SCN neurons from intact brain slice cultures. Polarization inverted selective-plane illumination microscopy (piSPIM) was used to measure Ca 2þ concentration within SCN using a ratiometric, neuronally expressed Ca 2þ sensor, Venus FLARE-Cameleon (Venus-cp172Venus FLARE-Cameleon) (42). The Venus FLARE-Cameleon sensor is a fluorescence resonance energy transfer (FRET)-based fluorescence anisotropy reporter (FLARE) (42). Ca 2þ concentrations were estimated from in situ calibration of Venus FLARE-Cameleon fluorescence anisotropy signals in SCN slices. Pharmacological inhibitors targeting the major Ca 2þ channel subtypes were applied during the peak and trough phases of the diurnal cycle to evaluate the impact of different Ca 2þ sources on Ca 2þ levels. These data revealed a peak-to-trough difference in cytosolic Ca 2þ concentration that was higher during the day, with ryanodine receptors providing the largest contribution at both times of the diurnal cycle.

Animals and ethical approval
Wild-type C57BL/6J mice were bred in a standard 12:12-h light-dark cycle. Male and female mice were killed for experiments via decapitation at postnatal day 4. All procedures involving mice were conducted in accordance with the University of Maryland School of Medicine Animal Care and Use Guidelines and approved by the Institutional Animal Care and Use Committee.
A subset of slices transduced with Venus FLARE-Cameleon AAVs were plated onto multielectrode arrays on culture day 10 as described previously (43,45). SCN slices cultured on filters were excised from the surrounding filter and flipped (SCN side down) onto multielectrode arrays pretreated overnight with 500 mL of 0.1 mg/mL collagen (C8919; Sigma-Aldrich, St. Louis, MO) and maintained in culture medium as described in Supporting materials and methods.
piSPIM imaging of fluorescence anisotropy Imaging experiments were conducted in 6-h time windows using the peak and trough of firing rhythms as the reference point, in which images were obtained between 5 h before to 1 h after the time of the peak or the trough in action potential firing. Filter sections with SCN slices were excised, rinsed in phosphate-buffered saline, transferred to the microscope chamber, and equilibrated for 20-30 min in 6 mL of prewarmed imaging solution containing 125 mM NaCl, 8 mM NaOH, 5 mM KCl, 1 mM MgCl 2 , 20 mM HEPES, 5 mM D-glucose, and 2.5 mM CaCl 2 (pH 7.20 5 0.01 at 35 C). Fluorescence anisotropy imaging was performed on a polarization inverted selective-plane illumination microscopy (piSPIM) microscope with stage-scanning capability assembled and aligned as described previously (46)(47)(48). The collection arms of the microscope were fitted with filter wheels containing emission filters and an image splitting device, OptoSplit II (Cairn, Faversham, UK), to separate parallel (P) and perpendicular (S) polarizations. The microscope was housed in an environmentally controlled incubator (Okolab, Ambridge, PA) maintained at 37 C. Automated stage and piezo focus control hardware elements were purchased from Applied Scientific Instruments (Eugene, OR). Camera and piezo electronics were controlled using Micromanager software (available at https://micro-manager.org/) (49) on a Z840 workstation (Hewlett Packard, Palo Alta, CA). Volumetric images (16-bit grayscale) were collected on a Nikon Eclipse TE2000-U microscope with water-dipping objectives (MRD07420, 40Â, 0.8 NA; Nikon, Tokyo, Japan) and a digital camera (Flashv4 Orca, C13440; Hamamatsu, Hamamatsu, Japan) as stack files with 20 image slices per volume (1mm spacing, 512 Â 1024 pixels per image slice, 332-nm pixel width and height, 2 Â 2 binning). Samples were excited in 10-s (KCl experiments) or 30-s (Ca 2þ inhibitor experiments) intervals with a 488-nm laser. Images were collected from a $170 Â 340 Â 20 mm area within the center of the SCN, which was visually identified under brightfield illumination at 4Â magnification using the optic chiasm and third ventricle as reference landmarks. After baseline control images were acquired, imaging solution (100-200 mL) was removed from the bath, mixed with the appropriate amount of drug stocks or dimethyl sulfoxide (DMSO) (<0.01%, D2650; Sigma-Aldrich), and reapplied to the bath chamber. The temperature of the bath solution was 35 5 0.1 C.
piSPIM image processing and data analysis Images were processed and analyzed according to Ross et al. (42,50) with some modifications using ImageJ (FIJI) macros and script executed in Python (v3.7). Volumetric image stacks were split to separate P and S channels. Corresponding P and S image stacks (512 Â 1024 pixels) were aligned using a Python script and separated into individual images. The median grayscale value of the background intensities for each image was calculated and subtracted. An adaptive local thresholding method was used to obtain a binary clipping mask to separate cell signals from image background. The local threshold value for each pixel was calculated using the Gaussian-weighted sum of the neighborhood pixel intensities (51,52). Anisotropies (r) were calculated using pixel intensities above the threshold value from the corresponding background-subtracted P and S images using the equation (53) r ¼ P À gS P þ 2gS : The g-factor constant (g) was measured using an isotropic fluorescein solution and calculated to account for the difference between P and S channel transmission efficiencies as previously described (50). The r-values for each image were summed across all images in each stack and plotted as a histogram distribution. A single mean r-value for each image stack was calculated with a Gaussian fit of the r histogram distribution in Prism v8.4 (GraphPad Software, San Diego, CA). Scripts for automated image alignment, background subtraction, pixel thresholding, and r-value calculations were executed in Python.

In situ calcium calibration
Ca 2þ buffering solutions were prepared using the method described in McGuigan et al. (54). To ensure EGTA concentrations in Ca 2þ -EGTA and EGTA solutions were identical, a 2Â EGTA stock solution containing all ingredients (except for NaOH and CaCl 2 ) was prepared and split into two volumes. CaCl 2 and NaOH were added to one volume and diluted to obtain a 1Â Ca 2þ -EGTA solution containing 125 mM NaCl, 44 mM NaOH, 5 mM KCl, 2 mM KOH, 1 mM MgCl 2 , 20 mM HEPES, 1.8 mM 2-deoxy-D-glucose, 5 mM EGTA, 5 mM CaCl 2 , 0.01 mM rotenone, 0.01 mM ionomycin, and 0.01 mM cyclopiazonic acid (CPA) (pH 7.20 5 0.01 at 35 C). NaOH and HCl were added to the second volume to produce a final 1Â EGTA (zero free Ca 2þ ) solution containing 125 mM NaCl, 44 mM NaOH, 5 mM KCl, 2 mM KOH, 1 mM MgCl 2 , 20 mM HEPES, 1.8 mM 2-deoxy-D-glucose, 5 mM EGTA, 0.01 mM rotenone, 0.01 mM ionomycin, and 0.01 mM CPA (pH 7.20 5 0.01 at 35 C). The appropriate quantities of Ca 2þ -EGTA and EGTA solutions were mixed to obtain solutions with known free Ca 2þ concentrations calculated with WebMaxC standard (available online at https://somapp.ucdmc.ucdavis.edu/pharmacology/ bers/maxchelator/webmaxc/webmaxcS.htm). SCN slices were equilibrated in Ca 2þ buffer solutions at least 20 min before imaging. All imaging solutions were prepared with Ca 2þ -free liquid-chromatography mass-spectrometry (LC-MS)-grade water (WX0001-6; Sigma-Aldrich). The dissociation constant (K d ) and Hill coefficient (n) were determined by fitting a plot of the r vs. Ca 2þ concentration data in Prism (GraphPad Software) with the equation n :

Statistics
Statistical tests were performed in Prism v8.4 (Graphpad Software). Changes in anisotropy values across baseline time points were tested with a two-way analysis of variance (ANOVA) with repeated measures. Student's t-tests (two tailed) were used to determine significant differences in anisotropy values and Ca 2þ concentrations between peak and trough time points. One-way ANOVA with Bonferroni's post hoc tests were used to determine significant differences in DCa 2þ between conditions at each time point. Paired t-tests (two tailed) were used to test for changes in anisotropy and Ca 2þ concentration between baseline and drug conditions for individual SCN slices at each condition. Significant differences in GCaMP6f fluorescence across multiple peak and trough time points were tested with a two-way, repeated-measures ANOVA and Bonferroni's post hoc test using the F/Fmax-values from individual cells for all slices across time points.

RESULTS
In the ex vivo organotypic slice preparation, the isolated SCN exhibits intrinsic circadian rhythmicity. First, rhythms in long-term spontaneous action potential activity were recorded by multielectrode array (Supporting materials and methods) to verify robust intrinsic circadian rhythms in cultured SCN slices before imaging. After establishing the diurnal phase using action potential firing, standard confocal imaging was used to verify intracellular Ca 2þ was also rhythmic under these experimental conditions using the Ca 2þ sensor GCaMP6f (Fig. S1). These data were then used to determine the time windows for quantitative Ca 2þ imaging using piSPIM.
To measure intracellular Ca 2þ using Venus FLARE-Cameleon (42), SCN slices were cultured on filter membranes ( Fig. 1 A i) and transduced with AAVs containing Venus FLARE-Cameleon cDNAs (42) expressed under the neuron-specific hSyn1 promoter (44). Ca 2þ binding to the Venus FLARE-Cameleon protein induces FRET between the two Venus fluorophores (55-59), which is detected as a decrease in the polarization (anisotropy) of emitted light from the sensor (42,53,(59)(60)(61). FRETbased measurements from Venus FLARE-Cameleon provide a ratiometric quantification of Ca 2þ concentration that is insensitive to variation in expression levels, cell morphology, illumination, or experimental preparations. Thus, this biosensor circumvents the variability in measurements that are based on fluorescence intensity (42,53,55,56,59,62,63,64), enabling quantitative measurements of Ca 2þ that are comparable across experimental time points and different SCN slices.
Fluorescent signals from neurons expressing the Venus FLARE-Cameleon biosensor were clearly detectable within the SCN ( Fig. 1 A ii). Volumetric images of the polarized fluorescence signals were collected from a cubic area of the SCN (Fig. S2). A local threshold, calculated based on the sum from a Gaussian window, was applied to each image to delineate cell signals (above threshold) from background (below threshold). Anisotropies were calculated from integrating the signal from all pixels above the threshold in each image, which were summed across all images (20 images per stack) to provide a single anisotropy distribution per image stack ( Fig. 1 A iii). Anisotropy histograms were fitted with a Gaussian distribution to calculate a single mean anisotropy value encompassing signals from all cells within the imaging region.
Because the relationship between anisotropy values and Ca 2þ concentration can be sensitive to variations in temperature and pH (65), the Ca 2þ concentration and fluorescence anisotropy for Venus FLARE-Cameleon was calibrated in situ from SCN slices incubated in buffered standards of known Ca 2þ concentration at 35 C (Fig. 1 B). For in situ calibration experiments, ionomycin, which permeabilizes the cell membrane to Ca 2þ ; rotenone, an ATP inhibitor; and CPA, a SERCA-ATPase inhibitor, were added to the bath solution to limit homeostatic compensation and promote the clamping of intracellular Ca 2þ concentration (66).
We found that the dissociation constant (K d ¼ 230 nM) and hillslope (n ¼ À1.0) values obtained from this in situ Ca 2þ calibration curve were similar to those previously reported for this sensor in vitro (42) (Fig. 1 B). The maximal (Rmax) and minimal (Rmin) anisotropy values were 0.259 and 0.184 in 0 and 1 mM Ca 2þ , respectively. These results indicated that the Venus FLARE-Cameleon reporter was functionally expressed and responsive to changes in clamped Ca 2þ concentration.
The in situ calibration values were next used to calculate Ca 2þ concentrations from images obtained during peak and trough of the circadian cycle. Anisotropy values were calculated from all pixels, in all cells within the imaging region. Each image typically contained between two and five neurons expressing Venus FLARE-Cameleon after image thresholding. Average anisotropies obtained during the peak (0.223 5 0.002) were significantly lower compared to anisotropies obtained during the trough (0.228 5 0.002), indicating that more Ca 2þ was bound to the sensor during the peak of the circadian cycle (Fig. 1 C). This corresponded to an estimated Ca 2þ concentration that was 1.3-fold higher at the peak (218 5 16 nM; range 53-408 nM) compared to the trough (172 5 13 nM; range 21-365 nM) (Fig. 1 D). These data show that a circadian rhythm in intracellular Ca 2þ can be detected from SCN neurons with the Venus FLARE-Cameleon sensor, and to our knowledge, provides a new method to track circadian changes in cytosolic Ca 2þ . . Compared to the trough, peak anisotropy was significantly decreased (p ¼ 0.04) and peak Ca 2þ concentration was significantly increased (p ¼ 0.02). *p < 0.05, unpaired Student's t-test. Data points represent measurements from individual SCN slices (one imaging region per slice). N ¼ 39 slices for peak, N ¼ 43 slices for trough.
To define the response of the biosensor to acute changes in Ca 2þ signaling under these in situ conditions, 50 mM KCl was applied to a subset of SCN slices at the peak and trough. Volumetric images were acquired in 10-s intervals for a 2-min baseline period before KCl was applied to the bath chamber and up to 10 min after KCl application. Consistent with prior studies (67), SCN slices responded to KCl treatment with a transient decrease in anisotropy values (peak: À0.024 5 0.005; trough: À0.018 5 0.003), corresponding to a transient increase in neuronal Ca 2þ (peak: þ271 5 77 nM; trough: þ524 5 212 nM). The maximal KCl-evoked responses were compared with the average baseline anisotropies and Ca 2þ concentrations at the circadian peak and trough. KCl produced a transient increase in Ca 2þ levels of 3.66-fold during the peak (baseline: 102 5 25 nM, KCl: 373 5 78 nM; n ¼ 5) and 3.52-fold during the trough (baseline: 208 5 20 nM, KCl: 732 5 228 nM; n ¼ 8). Thus, Venus FLARE-Cameleon detects changes in Ca 2þ evoked during maximal Ca 2þ signaling at both the peak and trough.
Next, to test the contributions of the different voltage-gated and intracellular Ca 2þ channel subtypes to neuronal Ca 2þ in the SCN, we measured the effects of Ca 2þ channel inhibitors on anisotropy (Fig. S3, A-D) and estimated Ca 2þ concentration (Fig. 2, A-D) during the peak and trough of the circadian cycle. For Ca 2þ channel pharmacology experiments, volumetric images were captured in 30-s intervals for 2 min to obtain baseline anisotropy values before drugs or vehicle con-trols were applied. The effects of each drug were analyzed in slices imaged in 30-s intervals for 2 min of baseline and for 10 min after the application of a drug or vehicle control (Veh, <0.1% DMSO), which were added to the bath solution just before time 0. The average change in anisotropy values (Fig. S3, A  and B) and corresponding change in Ca 2þ concentrations (Fig. 2, A and B) relative to the baseline average of each slice were plotted as a function of time. Anisotropy values were stable, with no significant change for the duration of the 2-min baseline recordings in each condition (Fig. S3, A and B) (p ¼ 0.3, two-way repeated-measures ANOVA). As a control for neuronal health after drug treatments, 50 mM KCl was applied to some slices after drug effects were obtained. KCl produced decreases in anisotropy values corresponding to increases in Ca 2þ concentration that were 2-20 times higher than baseline Ca 2þ levels (data not shown). The duration and magnitudes of these transient KCl-evoked responses were similar to those observed for slices in control conditions. These KCl responses obtained at the end of the experimental protocol verify that SCN slices are able respond to stimuli after Ca 2þ channel inhibitors were applied, demonstrating that drug exposure did not affect slice viability.
First, to probe the contributions of voltage-gated channels, we used 10 mM Nim to target L-type Ca 2þ channels and a cocktail containing voltage-gated channel inhibitors (VGCs) targeting N-type (3 mM ConoG-VIA), P/Q-type (200 nM AgaIVA), R-type (30 mM A C B D FIGURE 2 Effects of Ca 2þ channel inhibitors on peak and trough Ca 2þ concentration. (A and B) Time course of the change in Ca 2þ concentration (DCa 2þ ¼ Ca 2þ À average Ca 2þ from 2min baseline) before and after the application of vehicle control (Veh) or Ca 2þ channel inhibitors at the peak (A) and trough (B). Data are mean 5 SEM. (C and D) Plots of median, 25th and 75th percentile (boxes), and minimal and maximal (whiskers) changes in Ca 2þ concentration for individual slices quantified from 9 to 10 min after drugs were applied at the peak (C) and trough (D). Inhibition of L-type Ca 2þ channels with nimodipine (Nim, 10 mM) or inhibition of N/P/Q/R/T-type Ca 2þ channels with VGC (a mixture of 3 mM ConoGVIA, 200 nM AgaIVA, 30 mM nickel, and 1 mM TTA-P2) did not significantly affect peak or trough Ca 2þ levels compared to Veh. Inhibition of ryanodine receptors with dantrolene (Dan, 10 mM), inhibition of SERCA-ATPase with cyclopiazonic acid (CPA, 10 mM) and combined inhibition of voltage-gated Ca 2þ channels and ryanodine receptors with a cocktail containing Dan, Nim, and VGC (cocktail X) significantly decreased Ca 2þ at peak and trough. *p < 0.05, one-way ANOVA and Bonferroni post hoc test between drug and vehicle control conditions at peak (Nim, p ¼ 0. nickel), and T-type (1 mM TTA-P2) Ca 2þ channels. For each slice, paired comparisons were made between baseline, and the Ca 2þ concentration averaged from 9 to 10 min after drugs were applied (Fig. S3, C-F). Application of vehicle control (Veh, <0.1% DMSO) during the peak or the trough did not significantly affect anisotropy values (Fig. S3, C and D) or Ca 2þ concentrations (Fig. S3, E and F) compared to baseline. In contrast, voltage-gated Ca 2þ channel inhibitors produced an increase in anisotropy values (Fig. S3, C and D), which corresponded to a reduction in the Ca 2þ levels in paired comparison to baseline values (Fig. S3, E and F). These data implicate voltage-gated Ca 2þ channels as contributors to the cytosolic Ca 2þ levels during the peak and the trough.
We then tested the contributions of intracellular Ca 2þ channels. Previous studies have shown that inhibiting RyR-mediated Ca 2þ release from the ER produced a decrease in cytosolic Ca 2þ levels (24,41), but the effect of inhibiting intracellular Ca 2þ channels at both the peak and trough has not been systematically tested. 10 mM Dan was used to inhibit RyRs, and 10 mM CPA was used to target the SERCA-ATPase, which inhibits refilling of ER Ca 2þ stores. CPA produces ER store depletion and subsequent inhibition of both RyR-and IP 3 R-mediated Ca 2þ release (68). In a third condition, a cocktail (X) containing a combination of the VGCs, plus nimodipine and dantrolene to collectively inhibit RyRs along with the voltage-gated channels, was applied. Application of cocktail X thus inhibits other channels without blocking IP 3 Rs. We found that each inhibitor of intracellular Ca 2þ channels produced a significant decrease in Ca 2þ levels at the peak and the trough in a paired comparison to the baseline values for each slice (Fig. S3, E and F). These results suggest that intracellular channels also contribute to cytosolic Ca 2þ levels at both the peak and the trough.
Determination of the relative contributions for VGCCs and intracellular Ca 2þ channels across experiments requires accounting for the variation in baseline Ca 2þ levels in each SCN slice. To make this comparison, the average Ca 2þ concentration from 2 min of baseline was subtracted from the average Ca 2þ levels 9-10 min after drug application to obtain the change in Ca 2þ (DCa 2þ ) within each slice. First, we focused on the DCa 2þ produced by each drug during the peak (Fig. 2, A  and C). Starting with the vehicle controls, the DCa 2þ was negligible at À2 5 7 nM (À2 5 5% change) (Fig. 2  C). The voltage-gated channel inhibitors produced a DCa 2þ of À35 5 10 nM (À20 5 4% change) (Nim) and À40 5 13 nM (À19 5 4%) (VGC); however, these decreases were not statistically different than the vehicle control. This may be partly explained by the variability in the DCa 2þ responses of individual slices, which ranged from À6 to À78 nM (Nim) and À5 to À79 nM (VGC) (Fig. 2 C). In contrast, intracellular Ca 2þ channel inhibitors produced decreases in peak Ca 2þ that were significantly larger than the vehicle controls. Dantrolene produced the largest decrease, with a DCa 2þ of À50 5 6 nM (À36 5 6% change). Similarly, the SERCA inhibitor CPA produced a DCa 2þ of À59 5 3 nM (À24 5 5%), and the mixture of VGC inhibitors along with nimodipine and dantrolene added together (X cocktail) produced a DCa 2þ of À67 5 21 nM (À22 5 6%) (Fig. 2 C). The responses of individual slices to intracellular Ca 2þ channel inhibitors were less variable, with the range of À38 to À73 nM for dantrolene, À52 to À71 nM for CPA, and À27 to À114 nM for cocktail X. Taken together, these results suggest that peak intracellular Ca 2þ is predominantly set by RyR channel contribution.
These inhibitors had similar effects on Ca 2þ during the trough of the circadian cycle (Fig. 2 D). Vehicle control had little effect, À5 5 4 nM (À7 5 5% change). Nimodipine decreased Ca 2þ by À28 5 8 nM (À17 5 5%) overall with a range of responses between À2 and À56 nM. Similarly, VGC decreased Ca 2þ by À20 5 6 nM (À16 5 3%), with changes in Ca 2þ ranging from À3.5 to À37 nM. The decreases in trough Ca 2þ were significantly larger for dantrolene, À53 5 12 nM (À28 5 6%); CPA, À43 5 8 nM (À32 5 7%); and cocktail X, À51 5 11 nM (À23 5 5%). These overall decreases were accompanied by a larger range in DCa 2þ responses, which were À14 to À106 nM for dantrolene, À16 to À65 nM for CPA, and À15 to À91 nM for cocktail X (Fig. 2 D). Thus, at both peak and trough time points, the largest decrease in Ca 2þ was produced by dantrolene. In contrast, the contribution of voltagegated channels appears less significant and in some cases has more variability.

DISCUSSION
A primary advance of this study, to our knowledge, is the novel application of Venus-cp172Venus FLARE-Cameleon sensor (42) with piSPIM (59) to measure biologically controlled changes in Ca 2þ in live organotypic brain tissue. The Venus FLARE-Cameleon Ca 2þ sensor captured the collective signal in basal cytosolic Ca 2þ averaged from multiple neurons within a region of the SCN without integrating shorter-timescale Ca 2þ signals, such as action-potential-evoked Ca 2þ transients. Using these techniques, Ca 2þ concentrations (218 5 16 and 172 5 13 nM, respectively) were refined over prior absolute and relativistic estimates (27). These SCN concentrations are consistent with basal Ca 2þ levels typically measured in other neuronal cell types, $40-190 nM (69), and the values reported using other ratiometric Ca 2þ sensors in the SCN. In prior studies, peak values ranged between 50 and 440 nM, and trough values ranged between 50 and 150 nM (27). However, prior measurements using Fura-2 (13,29,41,70) or genetically encoded sensors such as Yellow Cameleons (24,25,33,71) did not utilize in situ calibration of the Ca 2þ sensor. These measurements relied on cell-free in vitro calibrations, which does not account for factors in the intracellular environment that could affect the Ca 2þ estimates (65). In this study, Ca 2þ measurements were obtained using a Ca 2þ sensor that was calibrated in SCN slices under the same experimental conditions at baseline and across different Ca 2þ inhibitor experiments. Venus FLARE-Cameleon also has the added advantage of only occupying a single-color channel, which will allow imaging of multiple biosensors expressed in the same neuron. Thus, this study provides a foundation for future experiments to investigate the cross talk between Ca 2þ and other cellular signaling components toward piecing together how the ensemble circadian clock mechanism functions at a cellular level.
A central feature of SCN neurons is that they express different properties depending on the time of the circadian cycle. SCN neurons exhibit a state of increased excitability and increased activation of voltage-gated Ca 2þ channels during the day (peak of the cycle) and a state of decreased excitability at night (trough of the cycle), during which voltage-gated Ca 2þ channel activity is reduced (3,4,38,72). Yet, it has remained unclear whether this daily increase in voltage-gated Ca 2þ channel activity is involved in maintaining the circadian pattern in cytosolic Ca 2þ , which is also highest during the circadian peak. Furthermore, no single study has directly compared the contributions for these different Ca 2þ channel types to cytosolic Ca 2þ levels at both peak and trough of the circadian cycle in intact SCN slices. Prior studies measured the effects of inhibitors on Ca 2þ levels only at a single time point or employed only a single Ca 2þ channel inhibitor (24,28,30,41,67). As a result of these methodological discrepancies, the relative contributions of the Ca 2þ channel subtypes at both times of the circadian cycle have been less than fully conclusive. For example, in studies measuring the Ca 2þ rhythms from the whole SCN, inhibition of Ltype voltage-gated Ca 2þ channels with nimodipine reduced the magnitude of the day-night difference in Ca 2þ levels (28,30). However, other studies found no effect of nimodipine on Ca 2þ levels (24,73).
With direct comparison of the relative contributions for each Ca 2þ source under equivalent experimental conditions, the results in this study support the current view that intracellular RyR Ca 2þ channels are major contributors to the Ca 2þ levels during both the peak and trough of the circadian cycle. Because the combined inhibition of voltage-gated channels and RyRs did not significantly decrease Ca 2þ levels further compared to inhibiting RyRs alone, it suggests that RyR inhibition produced the majority of the decrease in Ca 2þ . Inhibiting the SERCA-ATPase, which prevents Ca 2þ reuptake into the ER and leads to a depletion of ER stores (68), also did not produce a decrease in Ca 2þ that was larger in magnitude than the decrease observed when inhibiting RyRs alone. This further suggests that IP 3 Rs, which also can mediate ER Ca 2þ release, may have a lesser contribution to cytosolic Ca 2þ compared to RyRs, although this was not tested directly because of a lack of selective IP 3 R inhibitors (74). This study corroborates prior reports of decreased peak Ca 2þ with RyR inhibition (6,24,41). RyR2 messenger RNA (mRNA) and protein also exhibit a daytime peak in expression (38,75). However, it is unlikely that an expression-based mechanism would fully account for the circadian oscillation in Ca 2þ levels, as RyR activity is regulated by increases in intracellular Ca 2þ (27). Although it remains to be determined whether other VGCCs contribute to Ca 2þ -induced Ca 2þ release in SCN neurons, the lesser effect of inhibiting these channels suggests they do not serve as the primary sensors for RyR-mediated Ca 2þ release. Other sources of calcium in the SCN not tested here include ionotropic glutamate receptors, including N-methyl-D-aspartate receptors and a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (20). Ca 2þ homeostasis is also maintained by the activity of Na þ -Ca 2þ exchanger types 1 and 2 (NCX1 and NCX2) (38,75) and endoplasmic reticulum Ca 2þ -ATPases (SERCA) (70), which mediate Ca 2þ efflux or uptake into ER stores, whereas uptake of Ca 2þ via mitochondrial NCX (76) and Ca 2þ -binding proteins buffer cytosolic Ca 2þ (77)(78)(79). There is also potential involvement of store-operated Ca 2þ entry channels (80). Another mechanism that could be involved in regulating basal Ca 2þ over the circadian cycle, in conjunction with the activity of ion channels, is Ca 2þ buffering by Ca 2þ -binding proteins (69), which can alter basal Ca 2þ as well as influence the amplitude and decay of stimulus-evoked Ca 2þ transients (81,82), leading to changes in firing properties of neurons (69). SCN neurons express calbindin D 28K and calretinin (77,83). In SCN, levels of cytosolic calbindin protein have been observed to change over the course of the circadian cycle (77). This evidence suggests that these Ca 2þ binding proteins can be regulated based on the daily requirements of SCN neurons and could play an important role in circadian Ca 2þ rhythms. Further studies will be required to investigate their contributions.
This study did not focus on defining the Ca 2þ regulatory mechanisms by subregion, but the phase of the Ca 2þ rhythm exhibits regional differences. It has not yet been addressed whether the Ca 2þ channels themselves differ by subregion as the basis. The shell region of the SCN, defined by expression of the neuropeptide arginine vasopressin, exhibits a rhythmic Ca 2þ peak 3-5 h before the core, defined by vasoactive intestinal polypeptide (VIP) expression (28,30,33). Shell Ca 2þ rhythms also had higher amplitudes. These regional phase and amplitude differences could contribute to the wide variation in Ca 2þ values observed in the baseline measurements of this study, as the ROIs were located in the center of the SCN and undefined with respect to the core and shell boundaries. In addition, sequential application of inhibitors could reveal whether the relative contribution of Ca 2þ channel types differs between subregions. However, in previous studies, at least one inhibitor (nimodipine) failed to show a regional difference in its effects on Ca 2þ levels (30), leaving open the question of which Ca 2þ channels produce these phase differences. Regional differences are also present in Drosophila clock neurons (84,85), in which basal Ca 2þ levels were sensitive to RNA interference (RNAi) knockdown of intracellular Ca 2þ release via IP 3 Rs or SERCA, but only one set of neurons was sensitive to VGCC knockdown (86).

AUTHOR CONTRIBUTIONS
A.E.P. and A.L.M. designed the experiments and wrote the manuscript. A.E.P. performed the experiments and analyzed the data, V.P.R. wrote the computer code and assisted with image analysis, and M.A.R. designed and built the piSPIM microscope.