Glial Bmal1 role in mammalian retina daily changes

1 Vol.:(0123456789) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports Glial Bmal1 role in mammalian retina daily changes Serena Riccitelli 1,2,7, Fabio Boi 1,7, Davide Lonardoni 1, Lidia Giantomasi 1, Olga Barca‑Mayo 3,6, Davide De Pietri Tonelli 3, Silvia Bisti 1,4,5, Stefano Di Marco 2,5 & Luca Berdondini 1* Visual information processing in the retina requires the rhythmic expression of clock genes. The intrinsic retinal circadian clock is independent of the master clock located in the hypothalamic suprachiasmatic nucleus and emerges from retinal cells, including glia. Less clear is how glial oscillators influence the daily regulation of visual information processing in the mouse retina. Here, we demonstrate that the adult conditional deletion of the gene Bmal1 in GLAST‑positive glial cells alters retinal physiology. Specifically, such deletion was sufficient to lower the amplitude of the electroretinogram b‑wave recorded under light‑adapted conditions. Furthermore, recordings from > 20,000 retinal ganglion cells (RGCs), the retina output, showed a non‑uniform effect on RGCs activity in response to light across different cell types and over a 24‑h period. Overall, our results suggest a new role of a glial circadian gene in adjusting mammalian retinal output throughout the night‑day cycle. All living organisms are exposed to daily environmental changes1 and set the time at which physiological and behavioral events occur through internal self-regulating timekeeping mechanisms with respect to the 24h period. In mammals, the central organizer of the timekeeping system is in the suprachiasmatic nucleus (SCN) located in the hypothalamus2, while independent oscillatory systems are located in many peripheral organs and tissues3. The synchronization between environmental cycles and endogenous oscillators in the SCN is regulated by different factors, e.g., sleep deprivation, social interactions, behavioral activity, aging, and food availability4. Among them, the external day-night cycle is considered the most relevant5,6. A cycle of the internal timekeeping mechanism completes in ~ 24h and, at the molecular level, it arises from self-sustained intracellular transcription-translation feedback loops that involve the essential clock transcription factors CLOCK (Circadian Locomotor Output Cycles Kaput) and BMAL1 (Brain and Muscle Aryl hydrocarbon receptor nuclear translocator-Like protein 1, also known as ARNTL1). They heterodimerize and constitute the positive loop of the molecular clock, enhancing the transcription of target genes, including Per and Cry. Negative feedback is achieved by PER: CRY heterodimers translocated back to the nucleus to inhibit their transcription. Rhythmic expression of clock genes drives the transcription of thousands of clock-controlled genes (CCGs) that are not directly involved in modulating the circadian rhythm itself but generate rhythms in physiological processes in the brain and peripheral tissues, including the retina7,8. The mammalian retina has an autonomous circadian clock7,9,10 that escapes the control of the hypothalamic SCN, and light has been suggested to control retinal circadian rhythmicity directly11. Although fundamental retinal processes—from disc shedding to the release of melatonin and dopamine12—are under intrinsic self-sustained circadian control, the contribution of different retinal cell clocks to the diurnal control of visual information processing remains unclear. Daily modulation of retinal physiology is manifested as circadian variation in the amplitude of some components of the invivo electroretinogram (ERG)13, in the circadian tuning of photopic and mesopic retinal responses14, as well as by an altered retinal function observed upon the constitutive knockdown of Bmal1 gene in the retina11. Recent publications have shown that glial cells are part of the brain circadian circuit regulating daily behaviors15–18. In particular, we have recently provided evidence that the conditional deletion of Bmal1 in Glutamate Aspartate Transporter positive (GLAST+) astrocytes modifies circadian locomotor activity and cognitive function in adult mice17. However, despite the well-known importance of glial-neuron communication in the mammalian retina19,20, glial contribution to the regulation of day-night retinal rhythms remains unknown. OPEN 1NetS3 Lab, Fondazione Istituto Italiano di Tecnologia, Genova, Italy. 2Department of Biotechnology and Applied Clinical Sciences, University of L’Aquila, L’Aquila, Italy. 3Neurobiology of miRNA Lab, Fondazione Istituto Italiano di Tecnologia, Genova, Italy. 4Istituto Nazionale di Biostrutture e Biosistemi (INBB), Roma, Italy. 5Center for Synaptic Neuroscience and Technology, Fondazione Istituto Italiano di Tecnologia, Genova, Italy. 6Circadian and Glial Biology Lab, University of Santiago de Compostela, Santiago de Compostela, Spain. 7These authors contributed equally: Serena Riccitelli and Fabio Boi. *email: [email protected] 2 Vol:.(1234567890) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ In the retina, GLAST-positive glial cells include astrocytes and Müller cells21. Astrocytes are confined to the nerve fiber layer (NFL) at the inner border of the retina, mainly contacting axons of retinal ganglion cells (RGCs)22,23. In contrast, a unique population of glia, the Müller cells, resides in a strategic position spanning the entire retina from the outer to the inner limiting membrane and playing a pivotal role in maintaining retinal function and homeostasis in health and disease24,25. Interestingly, isolated Müller cells exhibit circadian rhythms in clock gene expression, and Bmal1 is necessary for robust circadian rhythms in mouse and human Müller cells26. Here, we report that the conditional deletion of the gene Bmal1 in astrocytes and Müller cells of adult mice retina is sufficient to alter the organization of retinal clock genes expression pattern throughout the day and retinal function, as revealed by electrophysiological recordings obtained with invivo ERG and exvivo highdensity multielectrode arrays (HD-MEA). Results Selective deletion of Bmal1 in retinal astrocytes and Müller cells decreases the b‑wave amplitude under light‑adapted conditions. To evaluate the glial contribution to the day-night retinal changes, we conditionally deleted the core clock gene Bmal1 in astrocytes and Müller cells by crossing a knock-in mouse line bearing a Tamoxifen (TM)-inducible CreERT2 recombinase under the control of Glutamate Aspartate Transporter (Glast) promoter27, with a mouse carrying a floxed basic helix-loop-helix Bmal1 domain allele (corresponding to the exon 8)11 as we previously reported17,28. Two months after oral TM treatment to 6 to 8-weeks-old mice, the deletion of the exon 8 of the Bmal1 gene from GLAST+ cells was ascertained at the DNA level (Fig.1a,b, DNA schematics in the left panel). Analysis of genomic DNA performed on retinae, optic nerves (ON), and cerebellum (CBM) of TM-treated Glast-CreERT2; Bmal1flx/flx (here referred to as Bmal1cKO) and TM-treated Bmal1flx/flx mice (here referred toas Ctrl) showed that GLAST positive tissues29 present an amplified band in correspondence of the 0.27kb only in Bmal1cKO, indicating the excision of the floxed allele, together with a ~ 2kb indicative of GLAST negative cells that did not undergo TM-induced Cre-mediated recombination (Fig.1b, right panel and Supplementary Fig.S2a). To further test the specificity of the deletion, Müller glial cells were isolated from retinal explants of adult Bmal1cKO and Ctrl mice after TM treatment and cultured. PCR products from cultured cells cDNA (obtained after RNA extraction) showed Bmal1 exon 8 excision (Fig.1c, cDNA schematics and in Fig.1c, central panel a size band of 0.14kb and Supplementary Fig.S2b) only in Bmal- 1cKO cDNA. A 0.3kb band, indicating the presence of the exon 8, was detected even in conditional Bmal1cKO samples suggesting either that TM did not elicit Cre-mediated recombination in every GLAST-expressing cell or contamination of GLAST negative cells in the adult primary culture. Furthermore, the specificity of Cre expression in the mouse retina was examined by immunohistochemical analysis studies of retinae from Glast-CreERT2 mice crossed with a Cre-inducible red fluorescent reporter mouse line (Td-Tomato)17. Efficient expression of Cre-recombined cells was observed 2 months after TM treatment, as revealed by the Td-Tomato signal (red) that spans throughout the whole-mount retina (Fig.1d, left upper panel). Analysis of vertical sections across the retinae (Fig.1d, right panel) showed that Td-Tomato-positive cells specifically colocalize with the astrocytic marker GFAP (Glial fibrillary acidic protein, in yellow) confined in the nerve fiber layer (NFL) and a Müller cells marker (GS, Glutamine Synthetase, in green) which somata are located in the inner nuclear layer (INL), while the processes span through the entire retina intimately connecting photoreceptors and other neurons25,30. No other cells, except astrocytes and Müller cells, were Td-Tomato positive. Together with the above-presented evidence, this approach effectively induces the selective deletion of Bmal1 in GLAST+ astrocytes and Müller cells of adult mice retinae upon TM treatment. To examine the involvement of glial Bmal1 in the modulation of the retinal response to light, we carried out invivo ERG recordings in Bmal1cKO and compared them with littermate controls. Recordings were performed in “day-dark” conditions as Barnard etal.13, with mice maintained in darkness overnight and recorded during mid-day hours, after 10min of light adaptation as this provides the maximal photopic b-wave amplitude (photopic ERG)11,13. Results showed a reduced b-wave amplitude by ~ 36% (Fig.1e,f n = 7/7 Ctrl and Bmal1cKO, mean ± SEM 111.1 ± 16.90 vs. 70.91 ± 6.97, respectively; P = 0.0483, Student’s t-test) at 30cd s/m2 in Bmal1cKO compared to Ctrl mice, while no significant differences were found analyzing both the a- and b-wave amplitudes in scotopic condition (0.001 to 30cd s/m2, Supplementary Fig.S1). Altered response to light is not driven by significant changes in retinal morphology or activa‑ tion of inflammatory processes. To examine the cellular basis of the reduced photopic ERGs b-wave amplitude in Bmal1cKO, we assessed retina morphology, astrocytes reactivity28,31 and microglia activation (assessed by cells migration from the inner to the outer layers32,33). Results showed that the laminar structure of the retina and its general architecture are morphologically indistinguishable between control and Bmal1cKO (Fig.2a,b). INL and ONL thicknesses were measured on DAPI-stained retinal slices, and no statistically significant differences between genotypes were found (Fig.2c; n = 4/4 Ctrl and Bmal1cKO; Total retinal thickness:167.20 ± 7.00μm vs. 175.34 ± 2.85μm, INL 28.69 ± 0.99μm vs. 32.12 ± 1.26μm and ONL 67.58 ± 2.97μm vs. 69.25 ± 1.70μm, respectively in Ctrl vs. Bmal1cKO retinae; P > 0.05, Multiple t-tests). Moreover, it is also known that, in the retina, the stress of any origin induces an immediate reactive response of Müller cells that orchestrate downstream events such as activation of self-protective mechanism (fibroblast growth factor 2, FGF-2) and activation of microglial cells. Hence, we analyzed Ctrl and Bmal1cKO retinal sections immuno-labeled for glial fibrillary acidic protein (GFAP)—whose expression in Müller cells is known as a hallmark of early gliosis34—and Ionized calcium-binding adaptor molecule 1 (IBA1), a marker of microglial cells. Results revealed no significant differences in GFAP expression level (Fig.2e) and its localization in the retinal nerve fiber layer (Fig.2b), as well as no differencesin the number of microglia and their localizationwithin retinal layers (Fig.2b,d; n = 8/8 Ctrl 3 Vol.:(0123456789) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ Figure1. Bmal1 exon 8 deletion in GLAST+ retina cells (astrocytes and Müller cells) reduces b-wave amplitude in photopic ERG. (a) Schematic representation of the experimental design. Ctrl and Bmal1cKO mice were orally administered on postnatal weeks 6–8 with tamoxifen (5mg) for 2 consecutive days. A third treatment was repeated one week before the experiments. (b) DNA schematics (left panel): transparent color boxes, introns; solid color boxes, exons; triangles, lox P sites; black bars, sizes of PCR products diagnostic of genotypes. PCR products amplified from the retina, optic nerve (ON), and cerebellum (CBM) genomic DNA (gDNA) demonstrate the deletion of exon 8 in Bmal1cKO mice (0.27kb) after TM treatment. (c) cDNA schematics: grey and red boxes, exons; black bars, sizes of PCR products. Bmal1 exon 8 deletion from adult Müller cells of Bmal1cKO after TM treatment was tested by PCR using primers described in the figure (left). Amplified PCR products show fragments diagnostic of disrupted Bmal1 allele in Bmal1cKO (PCR product 0.14kb). On the right, representative immunocytochemistry indicates cultured cells as Müller cells GS positive (red). Scale bar, 20μm. (d) Glast-CreERT2-Td-Tomato reporter expression throughout the retina, in the whole mount, and the transverse retina (left panel) following TM treatment. Cre recombinase activity is confined explicitly to astrocytes GFAP positive (yellow, a white asterisk in the magnification below) and Müller Cells GS positive (green, white arrow in the magnification below) (right panels). GFAP, Glial fibrillary acidic protein; GS, Glutamine synthetase; OD and ON, optic disc and nerve; GCL, ganglion cell layer; INL, inner nuclear layer, and ONL, outer nuclear layer. Scale bars: left 500μm, right upper 50μm, and right bottom 10μm. (e) Photopic light-adapted ERG averaged waveforms (mean ± SEM) elicited by 1Hz flashes of 30cd s/m2 for Ctrl (n = 7, blue) and Bmal1cKO (n = 7, red) mice (a-and b-waves are labeled on the waveforms). (f) Quantification of fERG responses showing b-wave amplitude reduction by ~ 36% in Bmal1cKO mice under light-adapted conditions (cone pathway ERG). Data are shown as mean ± SEM; animals were recorded at ZT6, *P < 0.05 vs. Ctrl, Student’s t-test. 4 Vol:.(1234567890) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ and Bmal1cKO; GCL + IPL: 6.76 ± 0.29 vs. 7.29 ± 0.26, INL + OPL:6.03 ± 0.41 vs. 6.66 ± 0.35 and ONL:0.52 ± 0.15 vs. 0.35 ± 0.15, respectively in Ctrl vs. Bmal1cKO per 1000μm segment of the retina; P > 0.05, Student’s t-test). Our data indicate that conditional deletion of Bmal1 in glial cells through the Cre-loxP system and TM treatment did not cause any detectable structural change or glial reactivity. Therefore, we decided to address whether clock gene expression-related changes could drive the observed functional effects of Bmal1 deletion on the ERG photopicresponse. Moreover, since photopic light-adapted ERG reflects cone-mediated light responses, our data suggest an altered transmission of information from cones to the bipolar cells in the Bmal1cKO mice retina. Loss of glial Bmal1 globally alters clock gene expression rhythms in the retina in the LD cycle. We determined the impact of the adult glia-selective deletion of Bmal1 on the core clock gene oscillations in the global retina, consistent with a role for glial clock gene networks in setting the intrinsic circadian cycle of the retina. Transcript levels of clock genes were measured over a daily cycle of 12h of light and 12h of dark (LD, light–dark) using quantitative real-time PCR (qPCR) in retinae from adult Ctrl and Bmal1cKO mice two months after TM treatment. Samples were harvested at four equally distributed time points during the day (ZT, Zeitgeber Time). Consistent with previous reports11,35, Bmal1, Cry1, Per1, and the clock-regulated gene Dbp, which expression is under the direct control of CLOCK-BMAL1 activity36, were rhythmically expressed in control mice retinae in the 12:12h LD condition (Fig.3; n = 4–6/4–6 Ctrl and Bmal1cKO). Oscillation amplitudes in retinae from control animals were compatible with those previously reported in this tissue37,38. In particular, as formerly described35–37, Per1 and Cry1 mRNA reached their maximum expression levels at ZT12, while Bmal1 and Dbp peaked at around ZT6 in controls (Fig.3, blue). In contrast, the deletion of Bmal1 in GLAST+ cells advanced the acrophase of all analyzed core clock gene transcripts in the whole retinal tissue, except for the BMAL1 target Dbp, however it did not affect their amplitudes (Fig.3, red). The similar oscillation of Dbp between controls and Bmal1cKO mice retinas (Fig.3, bottom right) indicates that compensatory mechanisms might account for its rhythm or that some retinal cells might have retained their rhythmic expression. Altogether, our results show that the loss of the Bmal1 gene in retinal Figure2. GLAST+ cells specific deletion of Bmal1 does not induce changes in retinal morphological or glial cell reactivity. (a) Representative transverse retinal sections from Ctrl and Bmal1cKO stained with Hematoxylin&Eosin to show the cellular and extracellular morphology. (b) Fluorescence images of retinal sections stained with DAPI (blue) to reveal cell nuclei. Ctrl and Bmal1cKO retinal sections do not show any apparent difference in neural architecture. IBA1+ cells (pink) are localized in the INL, GCL, and plexiform layers. Any noticeable structural changes in the microglial cells are detectable. GFAP immunoreactivity (yellow) is confined in the astrocytic layer in both conditions (yellow box), indicating the innermost retinal layer astrocyte population. Not detected reactivity of Müller cells. (c) Quantitative comparison of the total retina, INL, and ONL thickness. No statistically significant differences were detected between genotypes. n = 4/4, P > 0.05 vs. Ctrl, Student’s t-test. (d) Quantitative analysis of the number of IBA1+ cells in each layer per 1000μm segment of retina. No statistically significant differences were detected between genotypes. n = 8/8, P > 0.05 vs. Ctrl, Student’s t-test. (e) Graphs showing GFAP immunoreactivity in Ctrl and Bmal1cKO per 15e + 3 μm2, including the inner and outer retina. No statistically significant differences were detected between genotypes. n = 6/6, P > 0.05 vs. Ctrl, Student’s t-test. ONL outer nuclear layer, INL inner nuclear layer, GCL ganglion cell layer, OPL outer plexiform layer, IPL inner plexiform layer, GFAP glial fibrillary acidic protein, IBA1 ionized calcium-binding adapter molecule 1. Data are shown as mean ± SEM. Scale bar, 50μm. 5 Vol.:(0123456789) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ astrocytes and Müller cells alters the overall retinal circadian clock gene expression by changing the timing of the expression peaks. This observation expands the evidence of the critical interplay of neuronal and glial cells in regulating global molecular clock expression within the retina, in agreement with results observed in other tissues such as the cortex, hippocampus, and SCN15,17,39. Functional heterogeneity within the RGCs population is not altered in the Bmal1cKO mice ret‑ ina. Retinal ganglion cells (RGCs) represent the output neurons of the retina that provide spike-train encoded visual information to downstream brain circuits. Given the dysfunctional retinal response to light assessed with the invivo ERGs and the altered phasing of circadian clock genes in the retina of Bmal1cKO mice, we wondered whether these changes might impact the retinal output at the level of light-evoked RGCs responses40. To evaluate the effects of glial Bmal1 deletion on the retinal functional output, we first investigated whether RGCs retain their light response and could be classified according to their polarity preference as ON, ON–OFF, and OFF RGCs. We used a 4096 electrodes CMOS high-density multielectrode array (CMOS HD-MEA, 3Brain AG, Wädenswil, Switzerland) to record light-evoked activity in response to full-field flash stimuli presented to Ctrl and Bmal1cKO mice retinae under different Michelson contrast conditions (CT25, 50, 75 and 100, see Material and Methods) and in both mesopic and photopic light regimes. In mesopic conditions, the response of around 27000 single units acquired at 6-h intervals over the day/night cycles (i.e., ZTs 0–6–12–18) from 26 Ctrl and 25 Bmal1cKO mice retinae were considered for further analyses. Every unit was classified as ON, ON–OFF or OFF based on the shape of the cumulative distribution spike trains to alternating flashes (1s white, 1s black) at maximum contrast using a greedy template matching approach (see Material and Methods)41. Accordingly, ON cells preferentially respond to light increments, OFF cells to light decrements, and ON–OFF to both. An average of 241 ± 14 ON, 97 ± 6 OFF, and 194 ± 14 ON–OFF RGCs (mean ± SEM) per experiment were included in the analysis. Since we used a unique set of templates to classify cell polarity in both Ctrl and Bmal1cKO retinae, we could pool data across all recordings and cluster RGCs responses to 2s of light stimulation. In the first step, a k-means unsupervised clustering algorithm based on the cumulative response to the full-field flashing stimuli (Fig.4c,h,m) was used to assess the degree of light response diversity in overall recorded cells (TOT = 27147 from Ctrl and Bmal1cKO mice retinae; ON RGCs n = 12290, ON–OFF RGCs n = 9919, OFF RGCs n = 4938 in mesopic condition). We then used a Silhouette analysis to Figure3. Altered rhythmic clock-genes expression in the retina upon GLAST+ cells-specific Bmal1 deletion. Altered daily rhythms expression of some clock genes, i.e., Bmal1, Cry1, Per1, and Dbp, in the retina of Bmal1cKO (red) compared to Ctrl (blue) in adult mice under 12:12h LD cycles. The yellow background depicts the light phase (ZT 0–12), and the gray represents the dark phase (ZT 12–24). Samples were harvested at four equally distributed time points during the day, and ZT24 time points are the ZT0 ones shown again. Each curve depicts the best fitting (P < 0.05) between a cosine fitting with a period of 24h or linear regression. Mean ± SEM of 4–6 animals per group at each time point, performed in triplicates (n = 4–6 per strain, Multiple t-tests, *P < 0.05, **P < 0.01, and ***P < 0.001 vs. Ctrl) are shown. 6 Vol:.(1234567890) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ Figure4. Functional characterization of ON, ON–OFF and OFF type RGCs detected from responses to fullfield flashes in Ctrl and Bmal1cKO retinae under mesopic conditions. (a–f–k) The mean Silhouette analysis (average silhouette values 0.30, 0.26, and 0.28 for ON, ON–OFF, and OFF RGCs, respectively) suggests an optimal number of 10 clusters for ON and ON–OFF RGCs, and 12 for OFF RGCs. “k” represents the number of clusters that maximize the silhouette width. (b–g–l) Principal component analysis (PCA) of ON, ON–OFF, and OFF RGCs. (c-h-m) Cumulative spike trains distribution to alternating 1s white and 1s black flashes (20 repetitions, CT100) for ON, ON–OFF, and OFF RGCs. Each line represents the averaged response of the cells belonging to the color-coded cluster. (d–i–n) Population response for ON, ON–OFF, and OFF cells (10 ON, 10 ON–OFF, and 12 OFF clusters) per cluster. Each row shows the normalized PSTH of a single cluster. The numbers on the right of the PSTHs represent the numerosity of each cluster. (e–j–o) The histograms denote the relative proportion of recorded RGCs from each retina (Ctrl in blue and Bmal1cKO in red) assigned to the cluster. Cluster numerosity does not exhibit bias toward one of the two strains (Multiple t-tests correct). Data are shown as mean ± SEM. On the bottom right of each panel, the total amount of ON, ON–OFF, and OFF RGCs recorded in Ctrl and Bmal1cKO. Clusters ON 5, 8, 6, 10 and clusters OFF 1, 4, 5, and 9 represent the most abundant and further analyzed clusters. 7 Vol.:(0123456789) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ determine the optimal number of clusters of RGCs functional subtypes in an unsupervised way. Under mesopic conditions at 100% Michelson contrast (CT100), this yielded a total of 32 different RGCs’ qualitative responses to full-field stimuli (Fig.4c,h,m) and the identification (Fig.4a,b,f,g,k,l) of 10 clusters (k) for ON (pink), 10 clusters for ON–OFF (yellow) and 12 clusters for OFF RGCs (green). The cluster-mean normalized peristimulus time histogram (PSTH) obtained as an average response of all recorded units per cluster to 20 flashes repetitions at CT100 (Fig.4d,i,n) revealed the presence of known RGCs in the mouse retina42. Similarly, in the photopic condition, we recorded RGCs light-evoked response in a subset of our initial cohort counting 18 Ctrl and 19 Bmal1cKO mice retinae (6 Ctrl/5 Bmal1cKO ZT0, 4 Ctrl/5 Bmal1cKO ZT6, 3 Ctrl/4 Bmal1cKO ZT12, 5 Ctrl/5 Bmal1cKO ZT18). The contrast preference (ON, ON–OFF, and OFF RGCs) has been measured separately in the photopic condition. We obtained 9 clusters for the ON (TOT = 7755 from Ctrl and Bmal1cKO mice retinae; Ctrl n = 4027 and Bmal1cKO n = 3748), 8 clusters for the OFF (TOT = 7513 from Ctrl and Bmal1cKO mice retinae; Ctrl n = 4010 and Bmal1cKO n = 3503), and 10 clusters for the ON–OFF (TOT = 18213 from Ctrl and Bmal1cKO mice retinae; Ctrl n = 8855 and Bmal1cKO n = 9358) RGC subtypes (see Supplementary Fig.S3 for the functional characterization of ON, ON–OFF and OFF RGCs types in the photopic condition). Next, to evaluate differences between Ctrl and Bmal1cKO retinae, we restricted our analysis to ON and OFF cell types because most OFF RGCs and a large fraction of ON RGCs switch their response preference to ON–OFF type RGCs depending on the stimulus size and ambient luminance43,44. To assess the uniformity of RGCs classification across different experiments, we calculated the relative proportions of cells from Ctrl or Bmal1cKO mice retinae assigned to each cluster (across different timesof the day). Bar plots show that clusters did not exhibit a bias toward one of the genotypes (Fig.4e,j,o; P > 0.05, Multiple t-tests corrected for multiple comparisons using the Holm-Sidak method). This suggests that RGCs retain their characteristic light‐evoked preferential ON and/ or OFF responses to full-field flashes and their ability to transmit visual information to central visual areas upon deletion of the glial Bmal1 clock gene. Glial Bmal1 influences the light response amplitude of RGCs across different times of the day. We then investigated whether stimulus-driven responses of RGC subtypes, which encode distinct aspects of the visual scene, are affected by day-night cycles and by glial knockout of Bmal1. For the four assessed different Zeitgeber times (ZT, i.e., 0, 6, 12, 18; the relative proportions of ON and OFF RGCs sampled at different ZTs in Ctrl and Bmal1cKO retinae are reported in Figs.5b, 6b) and Michelson contrast levels (CT, i.e., 25, 50, 75, 100), we extracted PSTHs of single cells from the four most abundant clusters for both ON and OFF RGCs and we calculated the light-evoked maximum firing rate value (peak of responses). Specifically, we further studied the activity throughout the day of the most basic functional groups, sustained ON RGCs from clusters #8 and #5, and transient ON RGCs from clusters #10 and #6 classified based on the duration of their response (Fig.5a, averaged PSTHs of all cells for each of the indicatedclusters at CT100. See Supplementary Fig.S4a,b for functional ‘fingerprint’ of ON selected cells). Similarly, we chose clusters #1, #4 and #5, #9 for OFF RGCs that show a sustained and transient response to a flashing spot, respectively (Fig.6a, averaged PSTHs of all cells for each of the indicatedclusters at CT100. See Supplementary Fig.S5a,b for functional ‘fingerprint’ of OFF selected cells). Results for ON RGCs responses (Fig.5c) revealed that the responses to the alternating full-field stimulus of ON sustained RGCs (clusters #8 and #5) in Ctrl conditions were stronger at ZT 0 and ZT 12. In contrast, peak amplitudes for ON transient RGCs (clusters #10 and #6) were more stable over the day (One-way ANOVA with Tukey–Kramer post-hoc correction. Asterisks on the plots refer only to differences between groups). Instead, both sustained and transient analyzed OFF RGCs (Fig.6c) in Ctrl exhibited a daily variation in light sensitivity. Those features did not vary with the stimulus contrast (CTs) and were altered only in the Bmal1cKO group. Overall, the comparison of the peak amplitude averaged across cells and normalized to the maximal response over the day in the control condition for both ON and OFF RGCs, clearly revealed a more prominent light response in Bmal1cKO RGCs at ZT18 (Figs.5c and 6c; Figs.5d and 6d, cyan curve (ZT18, Bmal1cKO vs. Ctrl) below the unity line), indicating a higher sensitivity during nighttime in Bmal1cKO retinae. Presenting the stimuli at different contrasts did not affect this outcome. These results indicate that the response amplitude of RGCs sub-types (i.e., cluster of cells with similar spike trains cumulative distribution) is influenced by glial Bmal1 across different timesof the day, suggesting there may be circuit-specific contributions of retinal glia to the circadian regulation of visual processing. Glial Bmal1 effects on luminance dependent RGCs polarity preference. Our ERG data showed that deletion of the core clock gene Bmal1 in astrocytes and Müller cells alters cone ERG in the light-adapted condition but not in the dark-adapted one. This result aligns with previous reports that showed a tendency for cone-isolating conditions to intensify the effects of the intrinsic retinal circadian clocks on retinal function11,14. However, single-unit recordings from a large population of retinal ganglion cells revealed that in mesopic conditions, the glial Bmal1 deficiency is sufficient to alter the daily modulation of the maximal response to a flash of light in some RGCs clusters, but not the heterogeneity of RGCs physiological subtypes. Given that most OFF- type RGCs and a large fraction of ON-type RGCs switch their response preference depending on stimulus size and ambient luminance43,44, we investigated whether this shift was altered as a function of ambient light levels in Bmal1cKO compared to Ctrl retinae. Hence, we analyzed the polarity preference of 16515 single-unit RGCs identified from 37 datasets collected from CMOS HD-MEA recordings under both mesopic and photopic regimes (Fig.7). A similar amount of RGCs in Ctrl and Bmal1cKO retinae kept their polarity preference as ON RGCs with increasing luminance (Fig.7a, 36.9%, and 35.4%, respectively; Fisher’s exact test n.s.). Interestingly, the fraction of OFF RGCs identified in the mesopic condition that switched their polarity preference in the photopic regime was higher in Bmal1cKO than 8 Vol:.(1234567890) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ Ctrl retinae (Fig.7a, 49.9% Bmal1cKO vs. 64.9% Ctrl OFF RGCs kept their OFF polarity preference; Fisher’s exact test, P < 0.05). We then investigated whether these luminance-depended shifts in contrast preference are under day-night cycle regulation and to which extent they might be affected by the glial Bmal1 conditional deletion. Since more than 50% of ON–OFF RGCs showed no transitions in polarity with increasing luminance in both Ctrl and Bmal1cKO retinae, we restricted our analysis to ON and OFF RGCs. This analysis revealed that OFF RGCs not only varied their contrast switching behavior at different times, but the “polarity switching profile” is different between Ctrl and Bmal1cKO groups (Fig.7b). Indeed, while the polarity switching profile of OFF RGCs in controls oscillated over the 24-h day (overall 64.9% kept their preferential response), the Bmal1cKO OFF RGCs were shifting more their contrast preference at ZT0 being more stable at ZT18 (Fig.7b). These findings suggested the involvement of glial Bmal1 in tuning day/night RGCs contrast preference when responding to full-field flash stimuli in different light conditions. Figure5. Bmal1 deletion affects ON RGCs maximum firing rate in response to light in mesopic conditions. (a) Averaged normalized PSTH (mean ± SEM) of all RGCs belonging to the most abundant ON clusters (both genotypes, all ZTs at CT100) #8, #5, #10, and #6 in response to full-field white-black alternating flashes (stimulus indicated above). (b) The proportion of recorded RGCs per ZTs (Top, Ctrland bottom, Bmal1cKO). All cells from different experiments pulled together. (c) Analysis of the maximum firing rate in response to alternating bright and dark flashes performed on selected ON RGCs clusters (different rows) at four different ZTs (0, 6, 12, 18) and different contrasts (CTs, different columns). Data are shown as mean ± SEM between cells belonging to a specific cluster, separately for Ctrl (blue) and Bmal1cKO (red). Asterisks mark a significant difference between groups. (Kolmogorov–Smirnov test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). (d) Scatter plots of RGCs peak amplitudes normalized to the maximal response of the control throughout the day in Ctrl (y-axes) vs. Bmal1cKO (x-axes) conditions. Different colors indicate results obtained at diverse ZTs, and the symbols denote contrast conditions. Dashed line: unity line. 9 Vol.:(0123456789) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ Discussion This study shows that the glial clock controls the intrinsic retinal circadian machinery and contributes to the daily variation of retinal physiology. In agreement with previous studies that show the presence of a circadian tuning of photopic and mesopic responses45, our results indicate, for the first time, the involvement of glial cells (astrocytes and Müller cells) in the regulation of retinal function throughout the day. The entire cellular population of neurons and glia in the retina express all canonical clock genes except rod photoreceptors46. Accordingly, the original paper from Storch and colleagues11 showed the critical role of the core clock gene Bmal1 in the retinal global circadian clock and function by deleting this gene from all retinal cells since development. Surprisingly, our results revealed that selective adult Bmal1 deletion in glial cells is sufficient to modify the clock genes expression temporal profile (not their expression level) in the whole retina and to reduce the b-wave amplitude of photopic ERGs, as we observed through light-adapted fERG45. Notably, the deletion of Bmal1 in all retinal cells11 and the selective deletion of other clock genes such as Cry1/Cry2 do not lead to detectable retinal morphological alterations14,47,48. Consistently, in our work, the selective deletion of Figure6. Bmal1 deletion affects OFF RGCs maximum firing rate in response to light in mesopic conditions. (a) Averaged normalized PSTH (mean ± SEM) of all RGCs belonging to the most abundant OFF clusters (both genotypes, all ZTs at CT100) #1, #4, #5, and #9 in response to full-field white-black alternating flashes (stimulus indicated above). (b) The proportion of all recorded RGCs per ZTs (Top, Ctrl, and bottom, Bmal1cKO). (c) Analysis of the maximum firing rate in response to alternating bright and dark flashes performed on selected OFF RGCs clusters (different rows) at four different ZTs (0, 6, 12, 18) and different contrasts (CTs, different columns). Data are shown as mean ± SEM between cells belonging to a specific cluster, separately for Ctrl (blue) and Bmal1cKO (red). Asterisks mark a significant difference between groups. (Kolmogorov–Smirnov test; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). (d) Scatter plots of RGCs peak amplitudes normalized to the maximal response of the control throughout the day in Ctrl (y-axes) vs. Bmal1cKO (x-axes) conditions. Different colors indicate results obtained at diverse ZTs, and the symbols denote contrast conditions. Dashed line: unity line. 16 Vol:.(1234567890) Scientific Reports | (2022) 12:21561 | https://doi.org/10.1038/s41598-022-25783-1 www.nature.com/scientificreports/ 68. Zeck, G. M. & Masland, R. H. Spike train signatures of retinal ganglion cell types. Eur. J. Neurosci. 26, 367–380 (2007). 69. Statistics and Machine Learning Toolbox—MATLAB. https:// www. mathw orks. com/ produ cts/ stati stics. html. Acknowledgements We thank M. Götz (Physiological Genomics, Biomedical Center, Ludwig-Maximilians-University Munich, Germany) for kindly providing the Glast-CreERT2 mouse line. We thank M. Nanni and R. Pelizzoli, and IIT-NBT technical staff for their excellent support. We also thank M. Morini and the IIT Animal Facility Central Research lab staff for animal work assistance. This study was supported by Fondazione Istituto Italiano di Tecnologia. DDPT and OBM were partly funded by the European Research Executive Agency (REA) through the FP7- PEOPLE-2014-IEF “ASTROCLOCK” (629867) and Fondazione CARIPLO research grant (2015-0590). Author contributions Conceptualization, S.R., F.B., S.B., S.D.M.and L.B.; methodology, S.R., F.B., L.G., O.B.M., D.D.P.T. and S.D.M.; formal analysis, F.B., S.R., and D.L.; data curation, F.B., S.R., and D.L.; writing—original draft preparation, S.R., F.B, S.B., and L.B.; writing—review and editing, all authors; supervision, D.D.P.T., L.B., and S.B.; project administration, L.B.; funding acquisition, L.B., O.B.M., and D.D.P.T. All authors have read and agreed to the published version of the manuscript. Competing interests The authors declare no competing interests. Additional information Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598- 022- 25783-1. Correspondence and requests for materials should be addressed to L.B. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2022