Vasoactive intestinal peptide controls the suprachiasmatic circadian clock network via ERK1/2 and DUSP4 signalling

Once established, the effects of VIP, particularly on amplitude, could not be reversed by media change: they represent a permanent re-programming of the SCN circuit (Supplementary Fig. 1e). The emergence of these effects was, however, progressive, with reduced amplitude established by only 2 h of VIP treatment, whereas sustained period lengthening required more than 6 h (Supplementary Fig. 1f–j). All effects of VIP were dose-dependent, and comparable dose-dependent effects were observed following treatment with the specific VPAC2 agonist Bay 55-9837²⁹ at 50 nM or 5 µM (Supplementary Fig. 2a–f).

To compare the phase-shifting action of VIP with that of glutamate, the primary mediator of retinal input to the SCN core, we applied VIP or glutamate via droplet directly on to SCN slices at either CT10 or CT14. Whereas VIP caused significant phase shifts at both times, glutamate caused a phase shift only at CT14 (Supplementary Fig. 2g–i), consistent with previous reports^30,31. Equally, VIP acutely induced PER2 at both time points but glutamate did so only at CT14 (Supplementary Fig. 2j). Furthermore, glutamate did not result in a subsequently reduced amplitude at either phase, in contrast to VIP (Supplementary Fig. 2k). Thus, the effects of VIP on the SCN clock network are distinct from those of glutamate (and by extension light), in terms of their phase-dependence and molecular consequences^23,31.

Induction of PER2::LUC by VIP was not dependent on a functional TTFL, insofar as bioluminescence in arrhythmic Cry1^−/−Cry2^−/− (CryDKO) SCN, which lack circadian organisation^8,32,33, exhibited an immediate induction following addition of VIP (Supplementary Fig. 3a–c). Further, VIP application damped and smoothed the bioluminescent trace and significantly decreased the root mean square of PER2::LUC (Supplementary Fig. 3a, b, d), a measure of noise within the bioluminescent signal in CryDKO slices and thus analogous to amplitude in WT slices. Therefore, the molecular cascades whereby VIP acts within VPAC2-positive target cells to control the TTFL can function independently of the TTFL.

VIP could affect SCN rhythmicity by acting at the cellular and/or circuit levels. A significant feature of circuit-level time-keeping is the spatiotemporal wave of bioluminescence that reflects phase variations in clock gene expression within different regions of the SCN. To characterise the network-level effects of VIP, SCN slices were imaged on CCD camera (Fig. 1c, Supplementary Movie 1) and the spatiotemporal dynamics of PER2::LUC expression were analysed using centre of mass (CoM), which provides an integrated descriptor of the wave³⁴, and thus the phase relationship between SCN sub-regions. All slices showed a clear and consistent disruption of the spatiotemporal wave immediately after VIP application (Fig. 1d, e), mirroring the effects of VIP cell activation with Gq DREADDS³⁴. Not only was the range of the CoM reduced (Fig. 1f), but the directionality of the CoM was consistently altered from the stereotypical dorsomedial-ventrolateral to a more dorsolateral-ventromedial trajectory after VIP (Fig. 1d, e). This may be in part due to the dorsal tip of the slice displaying a high baseline of bioluminescence but very little oscillation (Fig. 1c). Thus, VIP affects the phase relationships between cellular oscillators within the SCN in a consistent, non-random manner.

To investigate the contribution of cell-autonomous actions of VIP, individual SCN cells were defined as regions of interest (ROIs, identified using Semi-Automated Routines for Functional Image Analysis (SARFIA)³⁵ in Igor Pro (Fig. 1g)) and circadian oscillations analysed. VIP had strong effects on the rhythmicity and amplitude of most oscillators (Fig. 1h), abrogating the previously tight phase coherence between cells (Fig. 1i, j). Consistent with the ensemble measures, the majority of ROIs displayed a lengthened period (Fig. 1k, l), and reduced amplitude (Fig. 1m). Thus, exogenous VIP affects cellular TTFLs across the SCN. The reduction in amplitude observed at the network level arises from cell-autonomous effects of VIP as well as network-level phase dispersal, whilst lengthening of ensemble TTFL period is likely cell-autonomous.

To identify potential molecular mediators of the effects of VIP on SCN circadian properties, we used microarrays to interrogate the transcriptomes of slices 2 h (CT12) or 6 h (CT16) after treatment with VIP or vehicle (Veh) at CT10. In all, 1165 significantly regulated transcripts were identified, with CT12 VIP vs. CT12 Veh being the comparison with the greatest number (738 genes with P < 0.05). The majority (65%) of VIP-regulated transcripts showed elevated expression after 2 h, and they also tended to show a greater fold-change in abundance than did downregulated genes. Of potential interest were clock genes (e.g. Per1, Per2, Rorα, Dec1), genes known to be induced in the SCN by light (e.g. Dusp1, Dusp4^38,39), and genes encoding proteins involved in intercellular communication (e.g. Cartpt, Gjb2). Downregulated genes included D-site albumin promoter-binding protein (Dbp), a protein regulated by the core circadian TTFL⁴⁰.

To validate the results of the microarray, 11 genes were selected for qPCR analysis based on criteria of rhythmic expression and SCN enrichment (Allen Mouse Brain Atlas⁴¹). The SCN samples were independent of those used for the microarray and an additional CT16 vehicle group was included to aid interpretation. The qPCR data closely reflected the changes observed in the microarrays (Supplementary Fig. 4), and could again be separated into clusters based on response kinetics: the majority of tested genes increased acutely before subsequently declining, with some returning to baseline levels, whereas Dbp showed downregulation following VIP, and Cartpt and Vgf continued to increase at CT16, as did Per2, consistent with bioluminescence recordings.

The cellular processes of genes significantly regulated by VIP were interrogated through gene ontology (GO) term analysis with subsequent clustering for visualisation (Fig. 3b). The largest group of transcripts responding after 2 h contained transcriptional regulators. This group was closely connected to genes involved in cellular transduction cascades relating to kinases, particularly MAPK pathways. The relevance of some unexpected clusters, including steroid synthesis, heparin binding and cytokine signalling awaits clarification. Notable significant clusters after 6 h included the down-regulation of previously upregulated transcription factors and MAPK-related transcripts (Fig. 3c).

To probe for common mechanisms in the transcriptional responses to VIP we focused on CREs, enhancer elements in genes that are bound by transcription factors, such as CREB, and which have a role in the SCN core in light-mediated phase shifting^15,16,42. Moreover, VPAC2 acts through cAMP- and phospholipase C-signalling^23,26, which could subsequently activate CREs. To identify the extent of CRE-mediated transcription, the VIP-regulated transcripts in the current study were compared to the dataset of Zhang et al.⁴³. Chi-squared analysis revealed that for genes upregulated after 2 h and down-regulated after 6 h there was a significant enrichment for CREs within 300 bp of the TATA box (CRE-TATA; Fig. 3d). GO term analysis showed that they were predominantly transcriptional activators and MAPK-related transcripts (Supplementary Fig. 5a). This was complemented by the observation that genes downregulated at 2 h and upregulated at 6 h showed no enrichment for CREs (Fig. 3d). Furthermore, genes that were regulated by VIP but did not contain CREs separated out into distinct clusters, for example cytokine-related transcripts lacked CREs (Supplementary Fig. 5a, b). Thus, although not universal, CREs were a significantly enriched marker of responses to VIP.

We tested whether the re-aligned rhythms of CRE-transcription and [Ca²⁺]_i led to differential effects on core genes of the TTFL, given the known differences in the number and responsiveness of CREs within their respective promoters^17,43. Vehicle or VIP was applied at CT10 to SCN slices carrying either a Cry1-Luc⁴⁴, Per1-Luc⁴⁵ or PER2::LUC reporter (Fig. 4j). Their responses were markedly different. PER2::LUC displayed acute induction, delayed phase, increased baseline and reduced amplitude. In contrast, the induction of Per1-Luc presented as an increased baseline, but without an associated suppression of peak bioluminescence. These different induction profiles for Per1 and Per2 were confirmed by qPCR (Supplementary Fig. 4). Cry1-Luc did not show an acute induction, consistent with the absence of CREs in the reporter, although the peak was broadened following VIP. The amplitude was reduced, but the baseline did not increase. Interestingly, PER2::LUC and Cry1-Luc stayed in phase following their VIP-induced phase shifts, whereas Per1-Luc phase shifted significantly less (Fig. 4j, k). Finally, the period of all reporters lengthened comparably (Fig. 4l). Thus, VIP signalling re-programmed the normally tight phase relationship of components of genetic (CRE, TTFL) and cytosolic (Ca²⁺) oscillations, revealing significant VIP-directed plasticity in the cell-autonomous TTFL.

Brain-derived neurotrophic factor (BDNF) is a known activator of the ERK1/2 pathway and a component of circadian phase-resetting in response to light⁴⁶. Treatment of PER2::LUC slices with BDNF, however, elicited a response that contrasted with that of VIP (Supplementary Fig. 7a), displaying a phase advance, no period change, and an amplitude increase (Supplementary Fig. 7b-d), with the only commonality between the responses being an increased baseline of PER2::LUC. To confirm further the specific contribution of the ERK1/2 pathway to the VIP response, the involvement of two other kinases frequently implicated in circadian phase-resetting, protein kinase A (PKA)²⁵ and protein kinase C (PKC)⁴⁷, was tested pharmacologically (PKA: 1 µM PKI 14-22 amide myristoylated or 50 µM Rp-8-Br-cAMPS; PKC: 300 nM sotrastaurin). These inhibitors had no significant effect on the response to VIP when applied individually (Supplementary Fig. 7e-h), but when PKI 14-22 and sotrastaurin were applied in combination, they potentiated the phase-shifting (Supplementary Fig. 7e) and fold-induction (Supplementary Fig. 7f), a result not reproduced using Rp-8-Br-cAMPS in place of PKI 14-22. Finally, the observed enrichment of CREs in the promoters of VIP-regulated transcripts suggested that CREB could be involved in the VIP response, potentially downstream of ERK1/2 activation. We therefore applied VIP in the presence of CREB inhibitor 666-15⁴⁸. 666-15 increased period when applied alone (Supplementary Fig. 7g), indicative of a role for CREB in the ongoing TTFL, but, surprisingly, had no impact on any parameter of the response to VIP (Supplementary Fig. 7e-h). In conclusion, ERK1/2 is critical for VIP signalling, but by specific pathways independent of PKA, PKC and CREB.

Confirmation of ERK1/2 involvement directed attention to VIP-responsive transcripts related to the MAPK pathway. These included negative regulators belonging to the dual specificity phosphatase (DUSP) family, including Dusp1, Dusp4, Dusp10 and Dusp14. Due to its strong upregulation after VIP, pronounced SCN localisation (Allen Mouse Brain Atlas⁴¹), rhythmic mRNA expression (CircaDB⁴⁹), close clustering with Per1 responses to VIP (Fig. 4a) and induction by light in the SCN core and shell^27,39, we focussed on Dusp4, which can dephosphorylate MAP kinase proteins such as ERK1/2 and JNK1/2/3⁵⁰, as a potential regulator of the effects of VIP.

The mice were then exposed to experimental ‘jet-lag’ conditions (Fig. 6c). The 12:12 LD lighting was phase-advanced by 8 h, resulting in a gradual resetting of behavioural onset to realign with ‘lights off’. Dusp4^−/− behavioural onsets advanced faster in the first two days following the switch (Fig. 6d), although there was no significant difference in the total time taken to readjust, as evidenced by the 50% phase shift (PS50) value (Fig. 6d inset). In contrast, Dusp4^−/− offsets delayed significantly more slowly to an 8 h delay (Fig. 6e), taking 2 days more to reach PS50 (Fig. 6e inset). Consistent with these directionally specific effects in jet-lag, phase delays in response to a single light pulse at ZT14 were almost completely abolished in Dusp4^−/− mice (Fig. 6f, g). In contrast, no significant difference between genotypes was observed for an advancing light pulse (30 min, 400 lux) delivered at ZT22 (Fig. 6h), although advances for all genotypes at this phase were not robust. To obtain a molecular correlate of this behavioural effect, pERK levels were determined in the SCN of control and Dusp4^−/− mice immediately following a delaying 30-minute light pulse at ZT14. This revealed an attenuated induction of pERK signal (by ~20%) in the mutant mice (Supplementary Fig. 8d, e), consistent with their reduced behavioural sensitivity to light.

All animal work was licensed under the UK Animals (Scientific Procedures) Act 1986, with Local Ethical Review by the MRC. All relevant ethical regulations were complied with. Mouse breeding and genetics were performed by members of the transgenic mouse facility, Ares. All mice older than P14 were culled by a Schedule 1 method: dislocation of the neck followed by exsanguination. All pups P14 and younger were culled by decapitation. DUSP4^−/− mice were obtained from the European Mouse Mutant Archive and were subsequently crossed with the PER2::LUC, generated by Dr. Joseph Takahashi²⁸ (University of Texas Southwestern Medical Center, Dallas). Dusp4^flx/flx were generated by crossing the DUSP4^−/−-PER2::LUC mouse with a Flp recombinase mouse, resulting in the removal of lacZ and neomycin cassettes, leaving loxP sites located either side of exon 3 of the Dusp4 gene. Per1-Luc transgenic mice were obtained from Dr. Hitoshi Okamura⁴⁵ (Kyoto University). Cry1-Luc transgenic mice were generated in-house and validated by our laboratory⁴⁴. CRY double knockout mice (CryDKO) were bred in-house using Cry1^−/− and Cry2^−/− mice generated by Dr. Gijsbertus van der Horst (Erasmus University, Rotterdam)⁷³. The Vpac2-Cre line (obtained from JAX Laboratories, Maine) carries a BAC transgene containing Cre recombinase under the control of the Vipr2 gene promoter.

Mice were individually housed in a light-controllable cabinet for the duration of behavioural monitoring. Their activity patterns were assessed using running wheels (Actimetrics) and passive infrared movement detectors. Mice were typically entrained to a 12:12 LD cycle (which mimicked that of their holding room) for at least 5 days before being exposed to different lighting schedules, including constant dim red light (DD), 8 h phase advances or delays, and Aschoff Type II light pulses (30 min, 400 lux). Food and water were provided ad libitum. Wheel-running data were acquired and stored as wheel revolutions per six-minute bin. Data were analysed using ClockLab (ActiMetrix Inc.) to calculate behavioural period by Chi-squared periodogram, activity duration (alpha), and activity onsets and offsets. Phase angles of entrainment, defined as the difference in time between lights off and activity onset, were calculated based on these onsets in Microsoft Excel. 50% phase shift values (PS50s) were calculated by fitting a sigmoidal curve to onset (phase advance) or offset (phase delay) values using the interpolate function in Graphpad Prism.

SCN organotypic slices were prepared as described⁷⁴. Slices were cultured for at least seven days before any bioluminescence recording. For long-term slice maintenance, slices were kept at 37 °C, 5% CO₂ with a medium change every 10 days. For bioluminescent recording, slices were transferred to 35 mm culture dishes containing 1.2 ml HEPES-buffered medium with 100 µM luciferin⁷⁴ (Biosynth), and were then air-sealed using glass coverslips secured with silicon grease. Bioluminescence produced by firefly luciferase (in PER2::LUC, Per1-Luc or Cry1-Luc slices) was measured by photomultiplier tubes (Hamamatsu) housed above culture dish stages within a retractable light-tight sleeve in a light-tight incubator kept at 37 °C. Photons were registered every second (counts per second, cps) and were integrated over 6 min bins. Time-lapse imaging of bioluminescence was performed using a charge-coupled device (CCD) camera (Hamamatsu Orca II) and inverted microscope with heated stage contained within custom-built light-tight housing. Images were acquired at 1 frame per hour, and compiled in FIJI (ImageJ) post-acquisition. An Olympus LV200 Luminoview was used for combined bioluminescence and fluorescence imaging. All images were acquired at a rate of 2 frames per hour, with exposure times of 1 ms for brightfield, 200 ms for fluorescence and 1.75 × 10⁶ ms (~30 min) for bioluminescence using a C9100-13 EM-CCD camera (Hamamatsu).

For adeno-associated viruses (AAVs), SCN slices were given a medium change immediately prior to AAV addition. A volume of 1 µl of AAV was placed directly on to the SCN slice and left for one week to allow expression before subsequent recording or further treatment. For lentiviruses (provided by Dr. Marco Brancaccio), SCN slices were treated two days after dissection by dropping 5 µl lentivirus directly onto the SCN slice. Superfluous residuum was washed off after two days.

VIP (Tocris) and Bay 55-9837 (Tocris) stocks were prepared by dissolving in HEPES-buffered medium. Glutamate (1 mM; Tocris) was dissolved in 1eq. NaOH, Tetrodotoxin (TTX, 1 µM; Sigma) and BDNF (200 ng/ml; R&D Systems) were dissolved in water. BDNF was applied at CT16, while TTX was applied 24 h prior to VIP treatment. Sotrastaurin (300 nM; Cayman Chemical), SP600125 (3 µM; Tocris), SCH772984 (100 nM; Cayman Chemical) and 666-15 (1 µM; Tocris) were all dissolved in DMSO, Rp-8-Br-cAMPS (50 µM; Biolog) in water, while PKI 14-22 amide myristoylated (1 µM; Tocris) was dissolved in 30% acetonitrile; these 6 drugs were used 30 min prior to VIP treatment. All pharmacological agents were bath-applied to SCN slices unless otherwise stated, and washed off only if stated explicitly. Phase of treatments was extrapolated from the preceding rhythm. Due to their arrhythmic nature, Cry1^−/−Cry2^−/− slices were treated based on Cry1^−/−Cry2^+/- littermate phases (data not shown).

Bioluminescence recordings from adult and pup slice cultures were analysed to calculate circadian period length, amplitude and relative amplitude error (RAE; a measure of the robustness of a rhythm) using the Fast Fourier Transform–Non-Linear Least Squares (FFT–NLLS) function in the BioDare software⁷⁵. Changes in these values following drug treatment, rather than the absolute values themselves, were often used to account for intrinsic variability between SCN slices. Unless otherwise stated, values were calculated based on rhythms four days before and four days after treatment excepting the first 36 h immediately after treatment application. Phase shifts were calculated as follows: the pre-treatment recording was used to predict when subsequent bioluminescence peaks would occur by extrapolating multiples of the measured period to the last identified peak before the treatment. These were compared with the actual observed peaks of bioluminescence following treatment and any differences were calculated in hours. PER2::LUC induction was calculated by identifying the bioluminescence values for three peaks prior to drug treatment, and extrapolating what the next peak would have been using the ‘growth’ function in Microsoft Excel. The actual bioluminescence value of the peak following the drug treatment was compared to this predicted value, and a fold-change could then be calculated. If this approach was not possible (for example for CryDKO slices), induction of bioluminescence was calculated instead by subtracting the normalised bioluminescence value immediately before treatment from the highest value within the 6 h recording window after treatment for each slice. Also for CryDKO slices, root mean square (RMS; a measure of how much a signal deviates from zero) was calculated as a replacement for amplitude as there is no reliable PER2::LUC oscillation in CryDKO slices. Raw Cry1^−/−Cry2^−/− bioluminescence traces were first normalised to the highest value within each dataset (to account for absolute gain differences between PMTs), before performing a 24 h rolling baseline subtraction so that the signal oscillated around zero. Values 48 h before treatment and 48 h after treatment (allowing a 24 h interval which was not included immediately post-treatment) were squared, and the mean for each interval was calculated. The square root of these means was taken and a percentage change between the two intervals was calculated for each slice.

The Semi-automated route for image analysis (SARFIA)³⁵ package for Igor Pro software (v. 6.3; Wavemetrics) was used for time-lapse region of interest (ROI) and centre of mass (CoM) analysis. ROIs within the slice were identified following despeckling and background subtraction in FIJI by thresholding the slice images and optimising pixel limits using SARFIA. The positions of these ROIs, their individual bioluminescence intensity profiles and raster plots generated from these profiles were then produced in Igor Pro. Subsequent circadian analysis of ROIs using BioDare was performed as described. Rayleigh plots displaying phase coherence could be produced from these data by converting phase information of each ROI into circular data using Microsoft Excel, and plotted using the R ‘circular’ package. For CoM analysis³⁴, time-lapse series were processed (despeckled, background subtracted and normalised) and thresholded in FIJI before using an in-house plugin for Igor Pro from which co-ordinates were generated.

Organotypic SCN slices were divided into four treatment groups for subsequent RNA extraction and microarray analysis to determine the acute effects of VIP treatment. The CT10 group was untreated and would serve as a baseline comparison; CT12 Vehicle slices were treated with vehicle (HEPES-buffered medium) at CT10 and harvested at CT12 (the peak of PER2::LUC bioluminescence); CT12 VIP slices were treated with 10 µM VIP at CT10 and harvested at CT12; CT16 VIP slices were treated with 10 µM VIP at CT10 and harvested at CT16 (the approximate new peak of PER2::LUC following VIP treatment based on previous experiments). Following total RNA extraction, samples were sent to Cambridge Genomic Services (Department of Pathology, University of Cambridge) for further processing and microarray analysis. Here the total RNA was checked on an Agilent Bioanalyzer 2100 for integrity, with all samples demonstrating an RNA Integrity Number (RIN) of > 9 apart from two (RIN 7.1 and 8.0). Total RNA was amplified using the Ovation Pico WTA v2 kit (NuGEN Technologies), and the resulting cDNA was fragmented and biotinylated using the BiotinIL kit (NuGEN Technologies). Concentration, purity and integrity of this labelled cDNA were determined using the Nanodrop ND-1000 (Thermo Scientific) and by Bioanalyzer before the cDNA was hybridised to a MouseWG-6 v2 BeadChip overnight. After washing, hybridisation was visualised by staining with streptavidin-Cy3 and scanned using a Bead Array Reader (Illumina).

Raw data were loaded into R using the Bioconductor lumi package⁷⁶ and segregated into pairs (such as CT12 Vehicle vs. CT10) for later comparison. These subgroups were then filtered using the detection p-value from Illumina to exclude any non-expressed probes, defined as not being significantly different (p > 0.01) from negative controls. Data were then transformed using the Variance Stabilisation Transformation (VST)⁷⁷ and then normalized using quantile normalisation to remove technical variation between arrays. A global normalisation to allow visualisation of all groups was also performed. The limma package⁷⁸ was used for comparisons between sample groups, and results were corrected for multiple testing using False Discovery Rate (FDR) correction. Correlations and clustering of the datasets from each sample were then performed to identify potential outliers and assess the quality of the data. A heatmap was generated using Morpheus software (Broad Institute, software.broadinstitute.org/morpheus/↗), and clustering was carried out using the in-built Hierarchical Clustering function in Morpheus. Transcripts found to be significantly regulated by VIP were examined using the DAVID ontology software^79,80 to identify significantly regulated pathways (using gene ontology (GO) terms) through the functional annotation chart function. These significant GO terms were then clustered and visualised using the enrichment map plugin⁸¹ for Cytoscape (v 3.5)⁸². Only terms significant at P< 0.01 and q< 0.05 were visualised. Significantly regulated genes were analysed for the presence of cAMP/Ca²⁺ regulatory elements (CREs) in promoters by comparison with the dataset from Zhang et al.⁴³. Transcripts that were not found in that dataset were excluded from this analysis. Once the circadian or CRE regulation of transcripts had been established, significant enrichment (e.g. for CRE regulation amongst VIP-upregulated transcripts) was tested using the chi-squared test. The observed number of genes regulated in a given way was compared with expected values calculated based on the ratios seen when looking at all transcripts in the CT12 VIP vs. CT12 Vehicle dataset.

The CRISPR-Cas9 system⁸³ was used to knockout Dusp4 in SCN slices. Guide RNA (gRNA) sequences were cloned into pAAV-U6sgRNA(SapI)_hSyn-GFP-KASH-bGH (Addgene #60958)⁵², which contains a gRNA and gRNA scaffold sequence under the control of the human U6 promoter (an RNA polymerase III promoter), as well as a GFP reporter localised to the cytoplasm by a Klarsicht, ANC-1, Syne Homology (KASH) domain under the human synapsin promoter. gRNA sequences were designed using either the gRNA Design Tool by Atum (https://www.atum.bio/products/crispr#4↗) or the MIT CRISPR design tool (http://crispr.mit.edu/↗) and the 20 nucleotide targets were as follows: g1, GCTGCAATACCATCGTGCGG; g2, GCGCCCTCTTAGCCCTCGCC; g3, ACTGCGTTTTGCCGGCGACA; g4, CTCCATCGTCACCATGTCGC; g5, GGCGGCGGCGGCAGCGCGGG; g6, AACAAGAAGGAACCAGGCGA. These oligos, along with complementary 20 nt oligomers, were synthesised (by Sigma) and annealed together. Annealing was carried out by mixing 2 µg of each oligomer (dissolved in water at 100 µM) together with its counterpart, then making the final solution up to 50 µl with annealing buffer (10 mM EDTA, 10 mM Tris (pH 8) and 50 mM NaCl). This was heated at 95 °C for 5 min and then allowed to cool to room temperature. The oligomers were designed such that there would be a 3 nt overhang at each end (ACC and AAC at the 5′ and 3′ ends, respectively) so that, following annealing, the double stranded product could be ligated into pAAV-U6sgRNA(SapI)_hSyn-GFP-KASH-bGH digested with SapI. Final constructs were validated in cells and successful plasmids were packaged into AAVs. Co-transduction was carried out with pAAV.Mecp2.SpCas9 (Addgene #60957).

Conditional DUSP4 overexpression was achieved by cloning Dusp4 cDNA, alongside an mCherry tag, into pAAV-EF1a-double floxed-hChR2(H134R)-mCherry-WPRE-HGHpA (Addgene #20297). mCherry-P2A was amplified from an in-house plasmid using the forward primer 5′-TATGGTACCTTCATAGGGCCGGGATTCTCC-3′ and the reverse primer 5′-TATGTCGACCATGGTGAGCAAGGGCGAG-3′, containing KpnI and SalI restriction sites respectively. The product was digested with KpnI and SalI and ligated into a similarly digested CMV.Dusp4-myc vector (Origene MR222119). mCherry-P2A-Dusp4 was amplified from the resulting vector using the forward primer 5′-TATGCTAGCCGTCGACCATGGTGAGCAA-3′ and reverse primer 5′-TATGGCGCGCCTACAGCTGGGGGAGGTGG-3′, which incorporated NheI and AscI restriction sites, respectively. After digestion with NheI and AscI, the product was ligated into EF1a-double floxed-hChR2(H134R)-mCherry-WPRE-HGHpA (Addgene #20297), which had been similarly digested, to produce Ef1a.DiO.mCherry-P2A-Dusp4. This was validated in cells and subsequently packaged into AAV particles.

SCN slices were washed in sterile PBS before detaching them from the membrane filter and storing them in RNALater (Qiagen) for 1–30 days at 4 °C to preserve the RNA. RNA extraction was carried out with an RNeasy Micro Kit (Qiagen) using an adapted protocol from the manufacturer’s recommendation. RNALater surrounding the slice was removed by pipette and replaced with 350 µl Buffer RLT to disrupt and homogenise the tissue, assisted by 30 s of vortexing. One volume of 70% ethanol was added to the lysate and mixed, then transferred to an RNeasy MinElute spin column and centrifuged for 30 s at 12,000 × g. a volume of 350 μl Buffer RW1 was added to the column and spun through for 30 s at 12,000 × g, before carrying out a DNase I incubation to remove genomic DNA. A volume of 350 μl Buffer RW1 and 2 × 500 μl Buffer RPE washes were then performed, again for 30 s at 12,000 × g. A final wash with 80% ethanol was done, centrifuging for 2 min at 12,000× g, with any remaining residue being removed by a subsequent 5 min spin at full speed (~21,000 × g). RNA was eluted in 14 μl RNase-free water, and its purity and concentration were assessed via Nanodrop. cDNA was subsequently synthesised from RNA using an iScript cDNA synthesis kit (Bio-Rad) according to manufacturers instructions, with the result diluted 1:2 in nuclease-free water before subsequent processing.

Following careful dissection, brains were immediately post-fixed in 10 ml 4% paraformaldehyde (PFA; Alfa Aesar) made up in phosphate buffer (108 mM Na₂HPO₄.2H₂O, 25.3 mM NaH₂PO₄.2H₂O) and shaken for 2 h at 4 °C before being cryopreserved in 20% sucrose (Fisher Chemical, UK) in PBS at 4 °C overnight. The brains were then mounted on a freezing microtome using OCT embedding medium (Thermo Scientific) and 40 µm coronal sections were taken, rostral to caudal. Sections were placed in wells of WHO dimple trays containing 0.01 M PBS. Solution A (0.43 mg ml⁻¹ X-gal, 2% N,N-dimethylformamide, 1 mM MgCl₂; in PBS) was mixed with Solution B (21 mg ml⁻¹ potassium hexacyanoferrate (II) trihydrate, 16.5 mg ml⁻¹ potassium hexacyanoferrate (III); in H₂O) at a ratio of 9:1 to make X-gal staining solution. Following washes in PBS, the sections were incubated in X-gal staining solution at 37 °C in a humidified chamber for 12–48 h. The sections were washed in PBS and then mounted onto slides before being allowed to air dry. The slides were briefly rinsed in water to remove residual salts and then dehydrated in 95% ethanol and 100% ethanol for 2 min each. Slides were placed in Clear-Rite 3 (Thermo Scientific) for 2 min and finally coverslipped using Pertex Mounting Medium (HistoLab). Images were acquired on an Olympus BX41 microscope with a Nikon DS2mv camera attachment.

Neuroblastoma2A cells (N2As; PHE culture collection 88112303) were cultured in DMEM with Glutamax (Invitrogen) supplemented with 10% FCS in a 37 °C humidified incubator at 5% CO₂. For transient transfection, cells were seeded in 12-well plates at a density of 1 × 10⁵ one day prior to transfection. A volume of 1 µg total plasmid DNA was used, incubated with 3 µl FuGene 6 Transfection Reagent (Promega) in 250 µl OptiMEM (GIBCO) per well for 20 min. This transfection solution was then gently added to each well. Cells were typically harvested 72 h after transfection for western blots. For downstream fixation and imaging, cells were seeded in 24-well plates on poly-L-lysine-coated 13 mm coverslips at a density of 5 × 10⁴ per well one day prior to transfection and were transfected with FuGene 6, OptiMEM and 0.5 µg total DNA using the same ratios and protocol as stated above. Ef1a.DiO.mCherry-P2A-Dusp4 was co-transfected with pAAV.CAG.iCRE-2A-H2BGFP.WPRE (kind gift of Ernesto Ciabatti, MRC, LMB). Cells were fixed after 72 h.

N2A cells were washed in warm PBS and harvested in 100 µl RIPA buffer supplemented with one cOmplete, EDTA-free Protease Inhibitor Cocktail (Roche) tablet per 10 ml. Samples were subsequently sonicated to ensure cell lysis and to shear DNA before centrifugation at 21,000 × g at 4 °C for 10 min to obtain the soluble protein fraction. Equal amounts of protein was denatured for 5 min at 90 °C in the presence of 4x NuPAGE LDS Sample Buffer (Invitrogen) and 25 mM DTT. Samples were loaded onto 4–12% Bis-Tris NuPage gels (Invitrogen) alongside a Novex Sharp Pre-Stained Protein Standard (Invitrogen) and run using the MOPS buffer system at 200 V for 45 min. Proteins were transferred using a wet tank system at 30 V for 1 h at 4 °C onto polyvinylidene difluoride (PVDF) membranes (Immobilon-P Membrane, 0.45 µm, Millipore) activated in methanol. Membranes were subsequently blocked for 1 h at room temperature in 5% (w/v) non-fat dried milk (Marvel) in TBS-T and incubated with anti-myc tag (1:2000, Abcam ab9106) overnight at 4 °C. Membranes were then washed three times in TBS-T for 15 min each and incubated in anti-rabbit IgG coeroxinjugated to horseradish pdase (CST 7074). After three more 20 min washes in TBS-T, the ECL Prime (GE Healthcare) chemiluminescence system was used for detection, imaged using a GelDoc system (Bio-Rad) and analysed in Image Lab software (Bio-Rad) Rabbit anti-β-actin (1:5000, CST 4967) was used as a loading control.

Freshly dissected brains were fixed in 4% PFA for 3 h at room temperature, then cryopreserved in 20% sucrose/PBS overnight at 4 °C. The brains were then mounted on a freezing microtome using OCT embedding medium and 40 µm coronal sections containing SCN were taken, rostral to caudal. Sections were permeabilised in 100% methanol for 10 min at −20 °C, washed in PBS and then blocked in blocking buffer (2% normal goat serum, 1% BSA, 0.3% triton X-100 in PBS) for 1 h shaking at room temperature. Primary antisera incubation was carried out overnight at 4 °C using rabbit anti-pERK (CST, 197G2) at 1:1000 in antibody buffer (1% BSA, 0.3% triton X-100 in PBS). Cells were then washed in a 1:3 dilution of antibody buffer in PBS before secondary antibody incubation with goat anti-rabbit Alexa fluor 488 (Invitrogen) at 1:500 in the same buffer. Sections were washed and mounted onto Superfrost Plus slides (Thermo Fisher), rinsed in water and coverslipped using Vectashield Hardset mounting medium with DAPI (Vector labs). Confocal imaging was performed using a Zeiss 780 microscope and a 20x objective, with fluorescence measurements performed on FIJI.

Immunohistochemistry of SCN slices was performed as above with some modifications. Slices were fixed for 1 h at 4 °C 30 min after drug treatment. Permeabilisation, blocking and primary antisera incubation were performed as above, but washes were carried out in undiluted antibody buffer over several days after secondary antibody incubation. Confocal imaging was performed using a Zeiss 780 microscope and ×20 or ×63 objectives.

Immunocytochemistry of cells was performed using the above procedure with some modifications. Cells were fixed for 20 min at 4 °C, there was no methanol permeabilisation step, primary antisera incubation was carried out using rabbit anti-myc tag (Abcam, ab9106) at 1:1000, and the secondary antibody used was goat anti-rabbit Alexa fluor 647 (Invitrogen) at 1:500. Cells were washed, rinsed in water and the coverslips transferred to slides. Images were acquired using a Nikon HCA inverted fluorescence microscope.

HEK293T cells were seeded in 15 cm plates at a density of 5 × 10⁶ one day prior to transfection. Cells were co-transfected with 7 µg AAV 2/1 (Rep/Cap proteins, for serotype 1), 20 µg pHGTI-Adeno1 (adenoviral helper) and 7 µg of the vector plasmid itself per plate, using polyethylenimine (PEI) in a 1:4 ratio DNA:PEI, mixed with 5 ml of DMEM. About 6 h after transfection, the medium was changed, with the harvest beginning 72 h post-transfection. The supernatant from each plate was harvested (and pooled if applicable) before centrifugation in a tabletop centrifuge at 2000× g for 30 min. The resultant supernatant was then filtered through a 0.45 µm Steriflip (Millipore) and transferred to polyallomer 30 ml ultracentrifuge tubes (Beckman Coulter). A volume of 2 ml of sterile 20% sucrose/PBS was gently pipetted to the bottom of these tubes before ultracentrifugation at 40,000 × g at 4°C for 2 h (Optima XPN-90 Ultracentrifuge, Beckman Coulter). The supernatant was subsequently removed and discarded, leaving the AAV pellet. The polyallomer tubes were allowed to air dry before the pellet was resuspended in 25 µl sterile, ice-cold PBS. This was left on ice for 1 h before combining resuspensions (if necessary), aliquotting and storing at −80 °C. Other AAVs were purchased commercially: AAV8.hSyn.GFP::Cre (UNC Vector Core); AAV1.CAG.DiO-tdTomato and AAV1.Syn.NES.jRCaMP1a.WPRE.SV40 (Penn Vector Core)⁸⁴.

Statistical tests and graphical representation of data (typically mean ± SEM) were performed using Prism 6 or 7 software (Graphpad). Two-tailed, unpaired Student’s T-tests were performed in experiments containing two groups separated by one independent variable. More than two groups separated by one independent variable were tested using an Ordinary one-way ANOVA, with Tukey correction for multiple comparisons where every group is compared to every other group, or Dunnett correction where groups are compared to a control group. Two-way ANOVAs with Tukey correction for multiple comparisons between all groups were performed on experiments involving multiple independent factors (e.g. time and drug treatment). If no significant differences are found between individual groups, then any significant differences between factors are reported. Pearson’s correlation coefficient was calculated to determine if two variables were correlated and, if applicable, linear regression was used to see if one variable could be predicted from another.

All unique materials (plasmids, gRNA sequences, qPCR primers, viral vectors) are available from the authors, excluding: two pre-validated qPCR primers for Mat2a and Cxcl10, which are commercially available from BioRad, and two commercially available AAVs: AAV8.hSyn.GFP::Cre (UNC Vector Core); AAV1.Syn.NES.jRCaMP1a.WPRE.SV40 (Penn Vector Core).

Further information on experimental design is available in thelinked to this article. Nature Research Reporting Summary

Supplementary Information Peer Review File Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Reporting Summary