Cryptochrome proteins regulate the circadian intracellular behavior and localization of PER2 in mouse suprachiasmatic nucleus neurons

Significance The suprachiasmatic nucleus (SCN), the master circadian clock of the mammalian brain, coordinates cellular clocks across the organism to regulate daily rhythms of physiology and behavior. SCN timekeeping pivots around transcriptional/translational feedback loops whereby PERIOD (PER) and CRYPTOCHROME (CRY) proteins associate and enter the nucleus to inhibit their own expression. The individual and interactive behaviors of PER and CRY and the mechanisms that regulate them are poorly understood. We combined fluorescence imaging of endogenous PER2 and viral vector–expressed CRY in SCN slices and show how CRYs, acting via their C terminus, control nuclear localization and mobility of PER2 to dose-dependently initiate SCN timekeeping and control its period. Our results reveal PER and CRY interactions central to the SCN clockwork.

Vector Builder (USA) The above AAV constructs were made as described below: C1R was modified from AAV.pCry1.(min).mCry1.EGFP (6), where NcoI and BsrGI restriction sites were used to replace EGFP with mRuby3. mRuby3 was amplified from the mRuby-C1 plasmid (Addgene: plasmid #127808) by PCR using primers the following primers: were again washed twice in 1:3 PBS-BT, followed by 2 washes in PBS, before mounting as described above.

Organotypic SCN slice preparation
SCN slices were prepared as previously described (9) but also outlined briefly below: Both male and female mouse pups aged between P9-P12 were culled by a Schedule 1 method (dislocation and exsanguination) and brain removed, which was then quickly transferred to icecold GBSS (Sigma, USA) dissection medium ( μM Luciferin (Microsynth, Switzerland)). Dishes were sealed with a cover-glass secured using silicone grease to prevent evaporation of media during the recording. Luciferase bioluminescence was detected by a photon multiplier tube (PMT; Hamamatsu, Japan), maintained at 37 °C in a light-tight incubator (CO2 not required as slices were sealed and buffered with HEPES and NaHCO3). Photon counts were recorded every second and counts were combined in 6-minute bins.

Viral transduction of SCN slices
SCN slices were transduced with AAVs after a medium change (culture medium or recording medium, depending on the experiment) immediately prior to transduction. One μL of AAV with titre of at least 1x10 13 GC/mL was dispensed directly on top of the SCN slice and incubated for 7 days before exchanging for fresh medium.

Initiation of Cry expression in Cry1/2 deficient slices
CryDKO SCN slices were assessed for arrhythmic phenotype using PER2::Luc or Cry1-Luc PMT recordings for at least 4 days, or longer in cases where residual short period and unstable oscillations persisted (a known phenotype for Cry deficient tissues). Slices where arrhythmicity was confirmed were transduced with the CRY AAVs. PMT recordings incorporated the baseline phenotyping phase, 7 days transduction phase and at least 7 days after washout of the AAV, the last phase being when expression is stable. "Pre" and "AAV" comparisons were made using the phenotyping phase and the washout phases of the recordings.
For assessing the effect of Cry initiation on PER2::Venus localisation, Cry1-Luc PMT recordings were made as above, but the SCN slices were fixed (as described previously) between 2-4 days after the final medium change, at the CT12 as defined by Cry1-Luc rhythms (peak at CT13 (10)), and imaged by confocal (described later).
Single Cry KO SCN slices (either Cry1 -/or Cry2 -/-) were put through a similar experimental procedure, the only difference being in the baseline phenotyping phase. Here, the baseline recording was for at least 5-7 days before transduction, to allow for "peak-to-peak" period determination. For dual initiation of both Cry1 and Cry2 expression, the slices were transduced first with CRY2-T2A-mCherry AAV for 7 days, followed by a medium change and then subsequently transduced with CRY1::mRuby3, followed by a final medium change, where again, 7 days of recording allowed for a final period determination.

Translational switching
As with non-ts dependent initiation of Cry expression, CryDKO SCN slices were assessed for arrhythmic phenotype using either PER2::Luc or Cry1-Luc reporters. SCN slices were then transduced simultaneously by application of both AAVs: PylRS and tsC1R (1 μL of each AAV). T1/2 values were calculated from the curve fit using the following equation: (where is either 1 or 2 for PER2fast or PER2slow respectively) The t1/2 values were then converted to diffusion coefficients using the Axelrod equation (11), which accounts for different sizes of bleach area: For each experiment, % mobile PER2 molecules is given by the magnitude of fluorescence recovery ( − 0 ) as a percentage of the initial pre-bleach fluorescence ( − 0 ).
Percentage of each fraction is given by the following equations: Mobile fractions: Immobile fraction: FRAP ROIs were either "spot bleaches" within either cytoplasm or nucleus, where each ROI circle was 2 μm in diameter, or whole "compartment bleaches" where the entire cytoplasm or nucleus was targeted.

Confocal snapshot imaging
Fixed SCN tissue (sections and slices) were imaged using either Zeiss LSM710, 780 or 880 systems using a 63x oil immersion apochromatic objective. To image the whole coronal view of the SCN, a tile-scan protocol was used within the Zen acquisition software. Particularly in the DAPI channel, this produced some tile artifacts at the joins between tiles, which can be seen in images. Nonetheless, the tile-scan allowed for higher resolution images of the whole SCN than could be acquired with the highest numerical aperture (NA) 10x objective, an objective that would not have required tiling. Further processing was carried out within FIJI (12).

Circadian rhythm analysis
SCN circadian rhythmicity was assessed through whole-slice emission of luciferase bioluminescence, as measured by PMTs (as described above) or by mean average slice fluorescence (average fluorescence intensity of the slice across each frame). In SCN slices with multiple fluorescence reporters, the fluorescence rhythms could be compared with one another.

BioDare
The Fast Fourier Transform -Linear Non-Least Squares (FFT-NLLS) within the BioDare (www.biodare2.ed.ac.uk) circadian rhythms analysis software package (13) was used to analyse rhythmicity of both luciferase and fluorescence-based recordings, where there was stable rhythmicity of at least 5 days. The first 24 hours after any medium change or treatment were excluded from the analysis. The output from software included best-fit period within a circadian-relevant window of 18-32 h periods, relative peak phase, amplitude, goodness of fit (GOF) and relative amplitude error (RAE), the latter are an assessment of the robustness of the rhythms.

Peak-to-Peak period and delta changes
Peak-to-Peak period was calculated to assess changes of period on a cycle-by-cycle basis, and was carried out within Graphpad Prism (v8 and v9, Graphpad Software). Raw data were first de-trended using a centred fifth exponential fit (unbiased detrending). The "area under the curve" analysis was then used to define peaks that were more than 10% above the baseline (baseline = 0 after detrending). The time between peaks was calculated and plotted as peak- to-peak period. The peak-to-peak Δ was calculated as the difference between consecutive peak-to-peak period intervals. To account for the inter-slice variability in absolute period change across a recording, the peak-to-peak Δ was expressed as a percentage of the maximum period change across the experiment.

Circadian rhythm amplitude and normalisation
In CryDKO SCN slices transduced with Cry AAVs, circadian rhythm amplitude was used as a measure of the robustness of the newly initiated rhythms. Amplitudes of PER2::Luc oscillations were determined on a peak-to-peak basis using the "area under the curve" feature in Graphpad Prism, where peaks were defined as described above. Circadian rhythm amplitude is known to be variable between SCN slices, so peak-to-peak amplitude was normalised to the maximum amplitude during the stable "washout" phase (after the final medium change) of the recording.

Average circadian profile generation and phase mapping of C1R
The average peak phase of C1R was mapped to circadian time by comparing to the peak P2V, which was defined as CT12 (4). Peak-to-Peak intervals were defined for each reporter in Graphpad Prism (Graphpad), as described above, for at least 4 cycles. The individual peak times, expressed in circadian time, were then used to calculate the mean average CT for the peak of C1R for each slice (N =6). Once the mean peak time for C1R was calculated, this was then used to generate average 24-hour circadian profiles for both P2V and C1R. These profiles were generated from a single circadian day (recordings scaled to 24 hours), where the circadian peaks of fluorescence were normalised to 100% for each slice. The peak times were then aligned to CT12 and CT18 for PER2 and CRY1, respectively, across all slices. After peak alignment, each circadian time point (0.5 circadian hours) was averaged across all slices (N =6) to generate the average profile.

Initiation dynamics of AAV-expressed Cry proteins
Confocal time-lapse recordings (as described above) were made at the point of AAV transduction to directly track protein initiation dynamics. A simple linear regression was fit over the first 72 hours of each recording to calculate the slope. This slope was used as an estimate of initial rate of protein production.

ROI and spatio-temporal analysis of time-lapse recordings
The Semi-automated route for image analysis (SARFIA) (14) package within Igor Pro (Wavemetrics, USA) was used for spatio-temporal analyses of time-lapse recordings. Prior to analyses, the raw recordings were processed by removing imaging artifacts (despeckling) and background subtraction from an ROI (outside of the slice) using FIJI (NIH, USA).
For ROI-based analyses, SARFIA was used to determine cell-like ROIs using thresholding and Laplacian edge detection. From this, time-lapse series were generated for each ROI and subsequently analysed using BioDare to assess circadian properties, as described above.
Rhythmic synchrony/coherence of the slices were assessed using the relative peak phase of each ROI (output from BioDare) and applying a Rayleigh analysis using bespoke software (https://github.com/tomoinn/web-statistics). Rayleigh analysis outputs included vector length and a circular plot with peak times of the individual ROIs.
For spatio-temporal analyses, an in-house script was run in the SARFIA package of Igor Pro, which generated a "Centre-of-Fluorescence" (CoF) xy coordinate time-series for each slice.
The xy series for one circadian day for each slice was plotted in Prism (Graphpad, USA) to generate a daily trajectory across the slice. Larger, directional deviations correspond to greater spatio-temporal organisation, whereas small or noisy trajectories correspond to less spatiotemporal organisation.

Intracellular fluorescence measurements and Nuclear:Cytoplasmic ratio calculation
Fluorescence intensity measures were determined manually using FIJI (NIH, USA) in both fixed SCN brain sections from adult mice and fixed SCN brain slices from neonatal pups (as previously described). All measures were background-corrected using fluorescence intensity located outside the slice area. ROIs for were selected for whole-cell, nucleus or cytoplasm.      in non-switched C1R and tsC1R (1mM AlkK) in SCN slices, fixed at CT12 (unpaired t-test). (G) Measure of the functional threshold of CRY1 relative to endogenous levels. As in (F), but instead comparing tsC1R +0.3 mM AlkK (TS) with endogenous C1R knock-in (KI) measured at CT12, the same timepoint as for tsC1R, as well as at the nadir (CT6) and peak (CT18) of CRY1 expression  Video S1.
Whole nucleus FRAP of WT SCN slice at CT12.

Video S2.
Whole cytoplasm FRAP of WT SCN slice at CT12.

Video S4.
Whole cytoplasm FRAP of CryDKO SCN slice at (CT not applicable).