GABA_A receptor subunit composition regulates circadian rhythms in rest–wake and synchrony among cells in the suprachiasmatic nucleus

Significance

The hypothalamic suprachiasmatic nucleus (SCN) synthesize, release, and respond to γ-aminobutyric acid (GABA), but the role of their GABA receptor subtypes is unknown. We recorded clock gene expression or firing rate from SCN neurons after either knocking down or mutating the γ2 or δ GABA_A receptor subunits because they classically mediate tonic or phasic inhibitory neurotransmission, respectively. We found that either of these GABA_A receptor subunit types was necessary to maintain synchrony among circadian neurons and daily rhythms in rest–wake behavior. Overexpression of either of these receptor subunits compensated for the loss of the other. We conclude that the density of GABA_A receptors sets the phase relationships among circadian SCN neurons to coordinate daily behaviors.

Abstract

The mammalian circadian clock located in the suprachiasmatic nucleus (SCN) produces robust daily rhythms including rest–wake. SCN neurons synthesize and respond to γ-aminobutyric acid (GABA), but its role remains unresolved. We tested the hypothesis that γ2- and δ-subunits of the GABA_A receptor in the SCN differ in their regulation of synchrony among circadian cells. We used two approaches: 1) shRNA to knock-down (KD) the expression of either γ2 or δ subunits in the SCN or 2) knock-in mice harboring a point mutation in the M2 domains of the endogenous GABA_A γ2 or δ subunits. KD of either γ2 or δ subunits in the SCN increased daytime running and reduced nocturnal running by reducing their circadian amplitude by a third. Similarly, δ subunit knock-in mice showed decreased circadian amplitude, increased duration of daily activity, and decreased total daily activity. Reduction, or mutation of either γ2 or δ subunits halved the synchrony among, and amplitude of, circadian SCN cells as measured by firing rate or expression of the PERIOD2 protein, in vitro. Surprisingly, overexpression of the γ2 subunit rescued these phenotypes following KD or mutation of the δ subunit, and overexpression of the δ subunit rescued deficiencies due to γ2 subunit KD or mutation. We conclude that γ2 and δ GABA_A receptor subunits play similar roles in maintaining circadian synchrony in the SCN and amplitude of daily rest–wake rhythms, but that modulation of their relative densities can change the duration and amplitude of daily activities.

Daily cycles in physiology and behaviors such as sleep–wake depend on a central circadian clock located in the suprachiasmatic nucleus (SCN) of the hypothalamus (1, 2). In mammals, nearly all cells, including SCN neurons, contain a transcription–translation negative feedback loop of core clock genes and proteins (e.g., Period1 and 2 genes and their products, PER1 and PER2) that generates circadian oscillations and synchronizes to local time and to other cells (3). Unlike most mammalian neurons, SCN neurons can be depolarized or hyperpolarized directly by γ-aminobutyric acid (GABA) depending on cell type, time of day, and photoperiod (4–7). Although GABA is the primary neurotransmitter in the SCN mediating many functional connections that can vary in strength with time of day, its roles appear diverse depending on experimental conditions and timescales and methods of measurement (8, 9).

Both ionotropic GABA_A- and metabotropic GABA_B-receptor types mediate responses to GABA within the SCN (8, 10). GABA_A receptors pass chloride and are significant drug targets for anesthesia and disorders including insomnia, anxiety, and epilepsy (11). Notably, the SCN is rarely considered in the context of GABA_A receptor therapy or the genetic basis of diseases. Most GABA_ARs are composed of two α subunits, two β subunits, and one γ subunit, although many isoforms are possible with the 19 known subunits: six α, three β, three γ, three ρ, and one each of the δ, ε, π, and θ. For example, the γ2 subunit is essential for postsynaptic clustering of GABA_A receptors and phasic responses to GABA (12–15) and δ subunits can localize to extrasynaptic sites where they mediate tonic inhibition to low GABA concentrations (16). The γ2 and δ subunits, among the most highly expressed GABA receptor subtypes in the central nervous system, can be found in the same neurons and variable in their expression across cell types (11). In the SCN, pharmacological, immunohistochemical, and molecular biology studies have reported the presence of subunits α1–α6, β1–β3, γ1–γ3, δ, ε, π, and θ in postnatal and adult SCN neurons, but it is still not clear whether all subunits are present at early postnatal ages (CircaDB, http://circadb.hogeneschlab.org/ (10, 17–24). The relative roles of synaptic and extrasynaptic GABA_A receptors have not been studied in the SCN. We therefore postulated that receptors containing the γ2 subunit mediate fast, synaptic responses to GABA and the δ subunit mediate tonic, extrasynaptic responses to GABA mediate synchrony in the SCN that affects wheel-running behavior.

Recent evidence for a tonic GABA current in the SCN (22, 25) has renewed interest in a long-standing hypothesis that extrasynaptic GABA_A receptors drive circadian changes in [Cl⁻]_i (26) and play a role in daily rhythms in behavior. Computer simulations suggest that a GABA-mediated tonic current in the SCN could excite, and facilitate synchrony among, circadian cells in the SCN (26–29). Consistent with this, knockdown of the δ subunit reduced circadian amplitude and synchrony in the SCN (28). Using shRNA to specifically knockdown expression in the SCN and mice with mutations in the genes coding for γ2 or δ, we tested the hypothesis that GABA_A receptors containing γ2 or δ subunits differ in their regulation of daily rhythms. Using these two methods to independently reduce the density of either receptor subunit along with overexpression assays, we found, surprisingly, that both γ2 and δ subunits are required to regulate daily rhythms in physiology and behavior.

Results

γ2 and δ GABA_A Receptors Regulate Daily Rhythms in Rest and Locomotor Activity.

To test whether subunits of the GABA_A receptor differentially regulate daily rhythms, we recorded locomotion of mice with genetic knock-down (KD) or knock-in (KI) of γ2 or δ subunits. Specifically, we compared running-wheel activity in mice with shRNA-mediated KD of γ2 or δ in neurons of the SCN, mice with functional mutations in Gabrd, the gene encoding δ receptor subunits, and control mice. Mice with mutant Gabrg2, encoding γ2 subunits, had persistent seizures or died, apparently as a result of reduced channel open probability (30) and, consequently were excluded from locomotor analysis. In LD, mice ran in the dark more than in the light, but those with SCN-targeted knockdown of γ2 or δ subunits had reduced nocturnal activity (Fig. 1 A−D and K). In constant darkness (DD), knockdown of γ2 or δ subunits significantly decreased nocturnal activity (Fig. 1 E and L), increased the ratio of time spent active to inactive (activity/rest ratio; SI Appendix, Fig. S2C), and decreased the amplitude of daily wheel running (SI Appendix, Fig. S2D). Similarly, knock-in mice with the mutant δ subunit showed decreased nocturnal activity (Fig. 1 F–J), increased activity/rest ratio (SI Appendix, Fig. S2 E and G), and decreased running wheel amplitude in LD and DD (Fig. 1 M and N). Previous work has not detected a strong phenotype of the δ subunit point mutation in the hippocampus and cortical interneurons (31–34), but we assume that at least some cells of SCN are affected by reduced current through mutated δ-containing receptors. These results indicate that either mutation of the δ subunit throughout the body or SCN-specific reduction of either γ2 or δ subunits similarly reduced the amplitude of daily rest–activity rhythms.

Fig. 1.

Knock-down or mutation of γ2 or δ GABA_A receptor subunits reduced the amplitude of daily rhythms and decreased nocturnal activity in wheel-running behavior. Representative actograms for mice treated with shRNA nontargeted controls (A, NT, n = 17) or targeting either γ2 (B, KD, n = 19) or δ (C, KD, n = 12) GABA_A receptor subunits and mutation experiments from WT (F, n = 14) and δ KI (G and H, n = 17) recorded in LD for ~10 d and then released in DD for ~11 to 20 d. Activity profiles for group mean ± SEM in LD for KD (D) and mutants (I) and in DD for KD (E) and mutants (J). Locomotor analysis for group mean ± SEM in LD (K, for KD, and M, for KI) and DD (L, for KD, and N, for KI) showed a decrease in activity during the dark phase independent of light conditions. Panel K, F(2,40) = 4.2, P = 0.02 in LD, panel L, F(2,40) = 5.1, P = 0.01 in DD for KD, one-way ANOVA and panel M, t(28) = 4.9, P < 0.0001 in LD and panel N, t(30) = 2.6, P = 0.01 in DD for mutations, t test.

Circadian PER2 Expression of SCN Cells Depends on γ2- and δ-GABA_A Receptor Subunits.

To test whether the reduced daily rhythms in behavior with γ2 or δ deficiency relate to central circadian pacemaker function, we monitored expression of the clock protein, PER2, in SCN explants infected at P4. Two weeks after in vitro infection with targeted shRNA, we found that knockdown of γ2 or δ reduced the peak-to-trough amplitude of daily PER2 rhythms (Fig. 2 A and C, γ2 51.6 ± 6.8%, δ 65.4 ± 4.8%, compared to nontargeted, NT, 84.2 ± 3.5%, mean ± SEM on recording day 3). Similarly, SCN harvested from mice homozygous for the KI mutations of γ2 or δ subunits approximately halved PER2 circadian amplitude (γ2 KI 45.5 ± 4.7%, δ KI 59.9 ± 6.2% compared to WT 82.2 ± 6.6%, mean ± SEM on recording day 3, Fig. 2 B and D). Circadian period of PER2 expression increased in SCN explants following depletion of γ2 or δ subunits by knockdown (NT 24.2 ± 0.2 h, γ2 KD 25.1 ± 0.2 h, δ KD 24.9 ± 0.1 h, one-way ANOVA, F(2,33) = 9.7, P = 0.0005) or mutations (WT 24.6 ± 0.1 h, γ2 KI 25.2 ± 0.2 h, δ KI 25.2 ± 0.1 h, one-way ANOVA, F(2,56) = 8.3, P = 0.0007).

Fig. 2.

Knock-down or mutation of γ2 or δ GABA_A receptor subunits decreased daily rhythms of PER2 expression in the SCN. (A) Bioluminescence recorded from representative SCN explants that had either been infected with virus delivering nontarget (NT, n = 10), γ2 shRNA-mediated KD (n = 6), or δ KD (n = 7) along 14 d of recording. (B) Bioluminescence recorded from representative SCN explants from mice that were WT (n = 21), γ2 knock-in (γ2 KI, n = 15), or δ KI (n = 23) along 6 d of recording. (C) γ2 or δ KD and (D) γ2 or δ mutations reduced PER2 expression amplitude (mean ± SEM) compared to controls after day 2 for KD (panel C, Group F(2,191) = 229.9, P < 0.0001; Days F(11,191) = 50.58, P < 0.0001; Group*Days F(22,191) = 1.48, two-way mixed ANOVA) or after day 1 for mutations (panel D, Group F(2,194) = 45.99, P < 0.0001; Days F(3,194) = 16.41, P < 0.0001; Group*Days F(6,194) = 1.24, P = 0.29), two-way mixed ANOVA.

To evaluate whether the decrease in the ensemble PER2 amplitude resulted from desynchrony among SCN cells or reduced amplitude at the single-cell level, we imaged SCN explants following perturbation of γ2 or δ GABA_A receptor subunits. Based on the time of daily peak PER2 in each cell within an explant, we calculated the Synchrony Index (SI) where 0 indicates cells peaking at random times of day and 1 indicates all cells peaking at the same time of day. We found knockdown of either γ2 or δ decreased SI among cells in SCN explants by approximately two-fold (0.44 ± 0.1 for γ2 KD, 0.42 ± 0.1 for δ KD, mean ± SEM measured on recording day 3) compared to NT controls (0.95 ± 0.01, Fig. 3 A–E), with decreased daily PER2 amplitude in the population (Fig. 3D) and single SCN cells (SI Appendix, Fig. S3A) following γ2 KD (0.3 ± 0.1; mean ± SEM measured on recording day 3) or δ KD (0.2 ± 0.07) compared to NT (0.6 ± 0.1). Increased period distribution of single SCN cells (1.5 h, SD) for γ2 KD, 1.9 h for δ KD, and 1.1 h for NT group, Fig. 3F) and fewer circadian cells (1% arrhythmic in NT; 8% in γ2 KD and 20% in δ KD; SI Appendix, Fig. S4A) were found. Knockdown of γ2 or δ also disrupted the typical dorsal-to-ventral wave of daily PER2 expression and the enrichment of lower amplitude, shorter period cells in the ventral SCN (Movie S1). Similarly, mutation of the genes encoding either γ2 or δ subunits reduced synchronous PER2 daily rhythms among SCN cells (SI of 0.7 ± 0.1 for γ2 KI, 0.6 ± 0.1 for δ KI, 0.95 ± 0.02 for WT; Fig. 3 G–K; mean ± SEM measured on recording day 3), increased their period distributions (0.3 h SD for WT, 2 h for γ2 KI and 1.1 h for δ KI, Fig. 3L), decreased single-cell circadian amplitudes (0.8 ± 0.04 in WT, 0.4 ± 0.02 in γ2 KI and 0.5 ± 0.08 in δ KI, SI Appendix, Fig. S3B) and increased the fraction of arrhythmic cells (1% in WT, 19% in γ2 KI, 21% in δ KI, SI Appendix, Fig. S4B). Again, circadian phase, amplitude, and period maps showed an altered distribution of these variables in the mutant SCN (Movie S2). These results, while surprising compared to prior results using GABA_A receptor antagonists in the SCN, indicate that targeted knockdown or mutation of GABA_A receptor subunits, γ2 or δ, impairs amplification and synchronization of circadian clock gene expression in SCN cells.

Fig. 3.

Knock-down or mutation of γ2 or δ GABA_A receptor subunits reduced synchrony among circadian cells in the SCN. (A, B, and C) Raster plots depicting PER2 expression over 6 d from cells in a representative SCN explant treated either with NT, γ2 KD, and δ KD (n = 6) or cultured from WT, γ2 KI, or δ KI mice, respectively (G, H, and I). Ensemble (mean ± SEM) PER2 expression from SCN explants treated with NT (n = 4), γ2 KD (n = 4), and δ KD (n = 6, panel D) or from WT (n = 4), γ2 KI (n = 5), or δ KI (n = 7, panel J) mice. Loss or mutation of γ2 or δ GABA receptor subunits reduced daily synchrony indexes (E and K; mean ± SEM) and increased the period distributions among circadian cells (F and L) in the isolated SCN. For panel E, Group F(2,60) = 47, P < 0.0001; Days (5,60) = 1.7, P = 0.1; Group*Days F(10,60) = 0.4, P = 0.9. For panel K, Group F(2,48) = 20.5, P < 0.0001; Days F(5,48) = 1.8, P = 0.1; Group*Days F(10,48) = 0.5, P = 0.8, two-way mixed ANOVA. For panel F, F(2, 65) = 45.3 one-way ANOVA; NT period SD 1.3, period CV 5.4; γ2 KD period SD 1.5, period CV 5.7; δKD period SD 1.9, period CV 7.2, Bartlett’s test P < 0.0001. For panel L, F(2, 68) = 41, one-way ANOVA, WT period SD 0.3, period CV 1.3; γ2 KI period SD 1.9, period CV 7.9; δKI, period SD 1.1, period CV 4.4, Barlett’s test P < 0.0001. SD = standard deviation, CV = coefficient of variance.

γ2 or δ GABA_A Receptor Subunits Compensate for Each Other in the SCN.

With evidence that γ2 and δ GABA_A receptor subunits are required to coordinate daily rhythms in the SCN, we next tested whether they act independently by overexpressing (OV) one while impairing the function of the other. Overexpressing γ2 subunits rescued the deficits induced by KD or mutations targeting δ subunits in circadian amplitude, fraction of circadian cells, phase and period synchrony, to control levels. When we overexpressed δ subunits, we rescued the deficits induced by KD or mutations targeting γ2 subunits (Figs. 4 and 5 and Movies S3 and S4). These data reveal that the endogenous γ2 or δ GABA_A receptors can compensate for each other in the SCN.

Fig. 4.

γ2 or δ GABA_A receptor subunit overexpression rescues daily PER2 rhythms in the SCN of γ2- or δ-deficient SCN. (A) PER2 traces from representative SCN explants expressing either NT, δKD, or γ2 OV + δ KD or (B) NT, γ2 KD, or δ OV + γ2 KD. Group (mean ± SEM) normalized amplitudes for KD (C and D; for γ2 OV + δ KD n = 7, for δ OV + γ2 KD n = 7) or mutation experiments (E and F) with their respective normalized amplitudes (G and H, for δ OV + γ2 KI n = 10, for γ2 OV + δ KI n = 10), where overexpression of one receptor rescues decreased amplitude induced by the γ2 or δ KD or mutations in GABA receptors. For panel C (Group F(2,262) = 91.9, P < 0.0001; Days F(11,262) = 40.3, P < 0.0001; Group*Days F(22,262) = 0.8, P = 0.7. For panel D (Group F(2,233) = 126.7, P < 0.0001; Days F(11,233) = 19.7, P < 0.0001; Group*Days F(22,233) = 0.7, P = 0.8. For panel G (Group F = (2,160) = 21.45, P < 0.0001; Days F(3,160) = 12.86, P < 0.0001; Groups*Days F(6,160) = 1.4, P = 0.2. For panel H, (Group F(2,156) = 46.92, P < 0.0001; Days F(3,156) = 4.79, P = 0.003; Group*Days F(6,156) = 0.6, P = 0.8, two-way mixed ANOVA.

Fig. 5.

Both γ2 and δ GABA_A subunits are regulating synchrony among SCN cells. Cell raster plots from a representative single SCN explant for δ OV + γ2 KD and γ2 OV + δ KD (A and B) or δ OV + γ2 KI and γ2 OV + δ KI (F and G). Ensemble mean ± SEM PER2 expression traces from the same color-coded single SCN explants for KD (C; δ OV + γ2 KD, n = 2 and γ2 OV + δ KD, n = 2) and mutations (H; δOV + γ2 KI, n = 3 and γ2 OV + δ KI, n = 3). (D and I) Group mean ± SEM sync indices showing that overexpression of one GABA receptor rescues desynchrony induced by KD or mutations of the complementary GABA receptor. (E and J) Overexpression of γ2 or δ receptors restore period distributions in the KD (E) and mutations (J) groups. Panel D, Group F(2,36) = 1.6, P = 0.2; Days F(5,36) = 0.1, P = 0.9; Group*Days F(10,36) = 0.1, P = 0.9. Panel I, Group F(2,25) = 2.4, P = 0.3; Days F(4,25) = 0.2, P = 0.9; Group*Days F(8,25) = 0.04, P = 0.9, two-way mixed ANOVA.

Knock-Down of γ2 or δ GABA_A Receptor Subunits Increases Excitability in SCN Neurons.

We next tested the hypothesis that γ2 and δ GABA_A receptor subunits independently regulate SCN neuronal firing patterns. Two weeks after in vitro shRNA-mediated knockdown (neurons were infected at P4) of either γ2 or δ GABA_A receptor subunits, we recorded firing rates of dispersed SCN neurons on multielectrode arrays for at least six days. We found that depletion of γ2 or δ increased the daily peak (mean ± SEM, 8.1 ± 0.5 Hz for γ2 KD; 8.1 ± 0.5 Hz for δ KD and 3.6 ± 0.1 Hz for NT, P < 0.001) and trough firing rates (0.6 ± 0.06 Hz for γ2 KD; 1.4 ± 0.2 Hz for δ KD; and 0.2 ± 0.2 Hz for NT, P < 0.001, Fig. 6). This increase in firing suggests increased excitability of individual neurons as a result of decreased GABA receptor density. Given that Gabazine has been reported to minimally alter SCN firing rates while increasing the fraction of neurons with tonic firing (22, 35, 36), this knockdown-induced increase in firing suggests increased excitability of individual neurons as a result of decreased GABA receptor density rather than loss of GABA signaling. Because blocking GABA receptors with bicuculline or gabazine has been reported to convert SCN neurons from irregular to regular firing patterns (36), we analyzed their firing patterns. We found that all recorded neurons (NT, γ2 KD, and δ KD) showed irregular firing. We conclude that whereas GABAR antagonists promote tonic firing, a decrease in GABA γ2 or δ subunits does not change spike patterning.

Fig. 6.

Knock-down of γ2 or δ GABA_A receptor subunits increased excitability in SCN neurons. (A, B, and C) Columns of five representative SCN neurons for NT, γ2 KD, or δ KD recorded for 6 continuous days. (D and E) KD of γ2 or δ receptors increased firing rate at the peak (D = 0.5, P < 0.0001) or the trough (D = 0.7, P < 0.002, Kolmogorov–Smirnov test), respectively. (F) KD of γ2 or receptors also increased the peak–trough amplitude, F(2,267) = 43.5, **** P < 0.0001, one-way ANOVA.

Whole-cell patch-clamp results from SCN neurons from γ2 or δ KI mice recorded during the day supported the conclusion that either receptor subunit substitutes for the other in the SCN. Specifically, we found that GABA_A receptor antagonists, picrotoxin or gabazine, reduced the tonic current by ~three-fold and abolished spontaneous inhibitory postsynaptic currents (sIPSCs) in neurons cultured SCN from WT, γ2 KI or δ KI mice. In addition, neurons from γ2 KI, δ KI, or WT SCN did not differ in their sIPSC amplitudes, frequencies, or decay rates (SI Appendix, Fig. S5).

Discussion

We found that decreasing density of either γ2 or δ GABA_A subunits in the SCN increases the daily duration of time spent running while reducing the amount of nocturnal activity in mice. Consistent with these behavioral consequences, we found that KD of γ2 or δ subunits dramatically broadens the times when SCN cells reach their daily peak of PER2 expression and increases their day and night firing levels. These results are consistent with recent reports that deletion of the vesicular GABA transporter (VGAT) in SCN neurons increases locomotor activity duration and decreases nocturnal locomotion (37, 38). Surprisingly, loss of GABA receptor subunits had much larger effects on SCN firing rate and synchrony than blocking GABA neurotransmission (e.g., with bicuculline or gabazine) or genetic deletion of GAD or VGAT (rate-limiting steps in GABA production or vesicle loading) (35, 37–40). These results are consistent with a recent computational model (28) that predicted, while blocking GABA signaling has modest effects, the density of γ2 or δ subunits can modulate circadian synchrony and amplitude among SCN neurons. The underlying mechanism may relate to daily rhythms in intraneuronal Cl⁻ which peak in the SCN around midday even during GABA_A receptor blockade (41). It may be that, in a 12:12 light:dark cycle, abundant GABA_A receptors and Cl⁻ transporters drive intracellular daytime Cl⁻ high, without need for extracellular GABA, resulting in GABA being excitatory in the SCN during the day and early night (4, 29, 42, 43). It is likely, however, the SCN is further complicated by a nightly increase in extracellular GABA peaking (40) contributing to increased GABA-mediated inhibitory postsynaptic currents around dusk (44). Taken together, these results indicate that reducing the density of GABA_A receptors reduces intracellular Cl⁻ and circadian synchrony among SCN cells with little dependence on extracellular GABA.

Our data indicate that the functions of γ2 and δ GABA_A subunits are redundant in the SCN of mice housed in 12:12 light cycles. KD or mutations of either of the subunits induced similar changes in PER2 amplitude, synchrony, period distribution and running-wheel behavior. Loss of either one reveals their necessity to promote circadian synchrony and amplitude among SCN cells. Furthermore, overexpression of one can rescue deficiencies due to loss of the other, suggesting that dynamic modulation of synchrony can be achieved by changing the levels of either γ2 or δ subunits in the SCN. This overlap in the function of γ2- and δ-containing receptors is atypical but recently found in the hippocampus (32). Because γ2 and δ subunits prefer to associate with distinct α subunits, we suggest that the receptors bearing these subunits substitute functionally for each other rather than one subunit species substituting for the other in individual receptors (33). We posit that changes in GABA_A receptor densities could provide a means to loosen intercellular coupling in the SCN (e.g., with seasonal changes in receptor density following day length) implying that long days could reduce synchrony by decreasing γ2 and δ GABA_A receptor densities.

Our data also implicate γ2 and δ GABA_A subunits levels in firing rate regulation. When we reduced the level of either subunit, we found SCN neurons fire more day and night for the duration of the recording. Because we found no evidence for impaired GABA-mediated IPSCs or tonic currents, we posit that receptor-mediated Cl⁻ or Ca²⁺ signaling contributes to this increase in excitability but not to firing pattern changes. Regardless, this mechanism for increased excitability and impaired circadian synchrony and amplitude in the SCN contrasts with the increased synchrony seen during pharmacological GABA receptor blockade (35) and desynchrony seen with reduced neuronal firing [e.g., TTX, (45)] or neuropeptide signaling [e.g., Vip^−/−, (46)].

These results showed that there is no difference between local KD of γ2 or δ subunits and global mutations of the same, the effects on PER2 amplitude and synchrony and even locomotor behaviors were indistinguishable. The global γ2 or δ subunit mutations used in present experiments have been reported to leave the subunits expressed but insensitive to picrotoxin (31). Here, we find that, at least in the SCN, these mutations can modify the functionality of these subunits inducing desynchrony among SCN neurons and changes in locomotor behavior. Although there are no reports of developmental compensation for the mutations used here, KD or mutations could result in changes in subunit expression level. Future studies should test whether the M2 mutation of GABA_A subunits has greater effects in the SCN than in other brain regions. Because tonic inhibition plays an important role in synaptic plasticity, neurogenesis (47, 48) as well as cognitive functions (49, 50). Any disturbance in phasic or tonic inhibition is associated with many neurological and psychiatric diseases. Thus, modulating these signals has become the basis of drug therapy as well as anesthesia (51–54). Future studies will help to understand how changes in density of GABA_A receptors modulate excitability, specifically how Cl⁻ conductance and Ca⁺⁺ levels are changing, circadian amplitude and synchrony and whether changes in daylength regulate γ2 or δ levels or other subunits forming part of the GABA_A receptors.

Materials and Methods

Animals.

Animal procedures were approved by the Animal Care and Use Committee at Washington University and conducted in accordance with the NIH guidelines for the care and use of laboratory mice. We used homozygous Per2^LUC male and female mice in which the Luciferase gene was fused to the endogenous mouse Per2 gene, resulting in the production of a PERIOD2::LUCIFERASE fusion protein as a bioluminescent reporter of PER2 protein abundance (55), founders generously provided by Dr. J.S. Takahashi, Univ. Texas-SW). Two additional knock-in mouse strains were used where point mutations of the endogenous mouse γ2 (γ2 KI) or δ (δ KI) subunit reduced picrotoxin sensitivity of the receptor without changing sensitivity to GABA (56, 57). These mice (originally on a C57BLx6CBA background) were crossed with Per2^LUC mice to produce male and female mice homozygous for the mutation γ2 or δ KI Per2^LUC homozygous. All mice were maintained on the C57BL/6JN genetic background for at least 4 generations before use in experiments.

Screening for GABA_A γ2-and δ-Targeted knockdown.

We developed an interfering RNA strategy to allow the acute KD of the γ2- or δ- GABA_A receptor subunits in the in vivo or in vitro SCN. The initial screening consisted of overexpressing (OVX) Gabrg2, the GABA_A γ2 (Addgene, #49170) gene, or Gabrd, the gene encoding δ (Applied Biological Materials, #AAV0872429) receptor subunits in HEK-293 cells (ATCC). The Gabrg2 and Gabrd constructs were cloned into a pcDNA 3.1 backbone with a GFP fluorophore and a neomycin/kanamycin selection site by standard cloning techniques (58). We transfected HEK-293 cells (60 to 80% confluent) in 60 mm² dishes (Greiner, CellStar) with 16 µl of Fugene6 and 1 µg of plasmid diluted in OptiMEM (Life Technologies) up to a final volume of 250 µl. We cultured the cells in 2 ml DMEM supplemented with 5% fetal bovine serum (FBS; Gibco/Life Technologies) and 1% penicillin/streptomycin (Gibco/Life Technologies) for two days. On day three, penicillin/streptomycin was replaced with 550 µg/µl of G418 sulfate (ThermoFisher). Selected HEK-293 cells overexpressing γ2 or δ receptor subunits were then transfected with one of five shRNA clones (Broad Institute, St. Louis) against either the γ2 (NM_008073.1-1932s1c1; NM_008073.1-1527s1c1; NM_008073.1-479s1c1; NM_008073.1-984s1c1; NM_008073.1-1520s1c1) or δ subunit (NM_008072.1-1565s1c1; NM_008072.1-164s1c1; NM_008072.1-736s1c1; NM_008072.1-1361s1c1; NM_008072.1-397s1c1).

Quantification of Knockdown by qPCR.

We measured the efficacy of each shRNA 48 h after transfection. We precipitated and isolated total RNA from treated HEK293 cells using TRIzol (Fisher). RNA was purified using the Direct-zol RNA MiniPrep Plus kit (Zymo), and cDNA was generated by RT-PCR using the SuperScript® III First-Strand Synthesis System (Thermo Fisher). Gene expression changes were further probed using iTaq™ Universal SYBR® Green Supermix (Bio-Rad), We performed a two-step qPCR using the SuperScript III first-strand synthesis and SuperScript^TM III Reverse Transcriptase systems (ThermoFisher). We performed PCR amplification (95 °C for 3 min, 39 cycles of 95 °C for 30 s and 63 °C for 30 s, and 72 °C for 1 min) of template (100 ng) in triplicate and obtained the mean and SE. Gapdh expression was used as the internal control for normalizing abundance. Negative controls included cultures transfected with nontargeting shRNA (59) and samples lacking reverse transcriptase or template. Based on their cycle thresholds, we chose γ2 clone NM_008073.1-1527s1c1 and δ clone NM_008072.1–397s1c1 which reduced expression of γ2 by 79% and of δ by 68%, respectively (SI Appendix, Fig. S1 A and B).

Generation of shRNA-Expressing Adeno-Associated Viruses.

The selected γ2 (5’AGAAAAACCCTCTTCTTCGGATG-3’), δ(5’CCAGCATTGACCATATCTCAGAG-3’), and NT shRNA (5’CAACAAGATGAAGAGCACCAA-3’) nucleotide sense sequences were synthesized (Integrated DNA Technologies) into the corresponding 97-nucleotide microRNA-adapted shRNA oligonucleotides, containing sense and antisense sequences linked by a 19-nucleotide hairpin loop. shRNAs were cloned into AAV backbone with a U6 promotor and a fluorescent protein (m-Cherry) and packaged using the AAV8 packaging system by the Hope Center Virus Core Facility at Washington University Medical School.

Virus Injections in the SCN.

Per2^LUC male and female mice (ages 60 to 90 d; P60 to 90) were anesthetized with 2% isoflurane and secured in a stereotaxic head frame (Kopf Instruments). For knockdown experiments, 0.4 µl of the AAV8 carrying a nontargeted (NT), γ2- or δ-targeted shRNA was injected into each side of the bilateral SCN (coordinates relative to Bregma: −0.46 mm posterior, ±0.15 mm lateral, −5.65 mm ventral). We delivered the virus at 0.05 µl/min using a syringe pump (KD Scientific) and 2-µl syringe (Neuros Hamilton). The syringe was left in place for ~5 min after injection to minimize upward reflux during the removal of the needle. Fifteen days after virus injection, mice were placed in a cage for locomotor activity recordings. For behavioral analysis in DD, the activity profiles were generated with ClockLab software normalized to the free-running period to obtain activity counts versus circadian time. After behavioral studies were completed, histological analysis was performed from 50 µm coronal cryosections or 250 µm explants containing the SCN to visualize the injection site and the m-Cherry tag expression. Mice with m-Cherry expression in the bilateral SCN were included in locomotor analysis as shRNA treated (SI Appendix, Fig. S1C).

Real-Time PER2^Luc Gene Expression Recordings.

Coronal slices (300 μm) were prepared on a vibratome (EMS) from neonatal (P4-7; shRNA experiments) or adult mice (P60-P90, γ2 KI or δ KI experiments) and placed in chilled in Hanks’ balanced solution (HBSS), supplemented with 0.01 M HEPES, 100 U/ml penicillin, and 4 mM NaHCO₃. We dissected the SCN with scalpels and cultured them on 0.4 mm membrane inserts (Millipore) with 1 ml HEPES-buffered air-DMEM (1.2 ml) supplemented with 10% new-born calf serum (Invitrogen) and 0.1 mM beetle luciferin (Biosynth). For KD experiments, fresh slices were transduced with either 0.4 µl of the NT shRNA-, or γ2-targeted shRNA- or δ-targeted shRNA-expressing AAV8. We replaced the media after 3 d and exchanged half of the volume every 3 to 4 d thereafter.

After two weeks in vitro, each SCN culture was recorded either with a photomultiplier tube (PMT, H93319-11; Hamamatsu, 10-min bins) or an ultrasensitive CCD camera (Andor Ixon or Ikon; 1 × 1 binning, 1 h exposures) at 36 °C (In Vivo Scientific incubator) on an inverted microscope (Nikon TE2000 or Leica DMi8 fitted with a 20× objective and a 0.5× coupler) in a sealed petri dish with media supplemented with 0.1 mM beetle luciferin (Biosynth).

From the recorded bioluminescence, we measured peak-to-trough amplitude and circadian period using Chronostar (60), for whole SCN explant experiments. SCN explants were considered circadian if Cosinor analysis (Chronostar) found a period between 18 and 30 h and a coefficient of correlation greater than 0.7.

To locate and track cellular bioluminescence, we used a custom Python code (45) which identified cells in each frame using a standard difference of Gaussian blob detector, in addition, videos underwent pixel-based analysis using a custom Python script to measure instantaneous phase, amplitude, and synchronization index (SI). We measured intensities by implementing a multidimensional Gaussian filter in SciPy tools using established methods (61) to generate time series of pixel intensities. Generated time series of pixel intensities were then detrended and smoothed using Sinc filter and the instantaneous phase and amplitudes were calculated using a continuous wavelet transformation analysis implemented in pyBOAT (61). Synchrony indexes were calculated as the first-order Kuramoto (45) parameter $R\left(t\right)=\left|{\sum }_j{e}^{i\varnothing j(t)}\right|$ by vectorial or directional averaging of instantaneous phases on the complex plane (62). We used MetaCycle (63) to quantify the circadian phase, amplitude, and period of each pixel and calculated the percentage of circadian pixels as [number of pixels reported as significantly circadian (BH.Q < 0.01)/number of pixels with nonzero pixel intensities] × 100. The period range to consider a cell circadian was stablished between 18 and 30 h and with a P < 0.05 value from MetaCycle.

Extracellular Multielectrode Array Recordings.

Dispersed SCN neuron cultures were prepared as previously described (64). Briefly, SCN explants were obtained from postnatal day 4 (P4) PER2^LUC mice and coronal slices (300 µm) using a Vibratome (OTS-5000, Electron Microscopy Sciences). Six unilateral SCN punches (1 mm diameter) were obtained and enzymatically dissociated using papain. Dissociated SCN neurons in a volume of 10 µl were incubated with 0.4 µl of the NT, or γ2 shRNA or δ shRNA and plated at a density of ~10,000 neurons/mm² on poly-D-lysine/laminin coated multielectrode arrays (sixty 30-µm diameter electrodes, Multichannel Systems). After 1 h, 1.2 ml of DMEM supplemented with 10% fetal calf serum was added to each dish and maintained at 36 °C in a 95% O₂ / 5% CO₂ incubator. Extracellular action potentials were recorded continuously from dispersed SCN neurons for at least 5 d as previously described (65). Action potentials were digitized in real-time (MC-Rack Software, Multichannel Systems) and single-neuron activity discriminated offline using principal component analysis and a refractory period of greater than 1 ms (Offline Sorter). Firing rate was binned in 10-min intervals (NeuroExplorer). Firing patterns with periods between 18 and 30 h were scored as circadian using MetaCycle analysis (63). The average spontaneous firing rate of each neuron at its daily peak and trough was calculated from the five days of recording. Interspike interval histograms (ISI) for each recorded SCN neuron were obtained (NeuroExplorer) to look for changes in firing patterns after the depletion of the γ2 or δ subunits using similar parameters reported by Kononenko (36). Briefly, ISI histograms were created using bins of 1 or 10 ms over 10-min recordings and analyzed for asymmetry of the ISI distributions.

Whole-Cell Electrophysiology.

Cultured SCN slices (3 to 6 d in vitro) from WT, δ KI, and γ2 KI mice were transferred to a recording chamber with continuous perfusion (2 ml/min, 32 °C) of oxygenated artificial cerebrospinal fluid (in mM: 125 NaCl, 25 glucose, 25 NaHCO₃, 2.5 KCl, 1.25 NaH₂PO₄, 2 CaCl₂, and 1 MgCl₂, equilibrated with 95% oxygen/5% CO₂). Whole-cell recordings were performed with borosilicate glass pipettes (World Precision Instruments) with open tip resistances of 3 to 6 MΩ. Pipettes were filled with internal solution containing the following (in mM): 130 CsCl, 2 NaCl, 0.5 CaCl₂, 5 EGTA, 10 HEPES, 2 MgATP, and 0.5 Na₂GTP, pH 7.39 with CsOH. Currents were confirmed to be GABAergic at the conclusion of recordings based on their response to application of 100 μM gabazine (GBZ, Tocris).

Voltage clamp recordings were performed at −70 mV, and spontaneous phasic GABA_A receptor currents were recorded at baseline and in the presence of picrotoxin (PTX, 100 μM). Tonic GABA_AR currents were recorded at baseline and in the presence of PTX (100 μM) and measured as the SD of the baseline holding current in event-free portions of the recording and expressed relative to the GBZ condition.

Wheel-Running Recordings.

Wheel-running activity was recorded in 6-min bins (ClockLab Actimetrics) for ~10 d in a 12:12 h light–dark (LD) cycle (lights ON 6:00 am, lights OFF 6:00 pm) followed by ~10 d in DD from shRNA-injected and knock-in mice. Bilateral injections aimed at the SCN (0.4 µl × side) of NT shRNA, γ2 shRNA-, or δ shRNA-expressing AAV8 were made in adult Per2^LUC mice (postnatal days 60 to 90; P60 to P90). Injected mice recovered for 5 d before locomotor assays. All mice, including δ-mutant mice, were placed individually in cages equipped with a running wheel in light-tight chambers illuminated with fluorescent bulbs (~2.4 × 10¹⁸ photons/cm²/s; General Electric). The period and amplitude of behavioral rhythmicity of each mouse were determined by χ² periodogram analysis (66) from the last 5 continuous days in LD and in DD.

Statistical Analyses.

Unless otherwise specified, comparisons between control, KD, or KI groups were performed by a one-way or two-way mixed ANOVA (post hoc Tukey’s or Bartlett’s test). We used the Kolmogorov–Smirnov test for firing rate comparisons. Unless otherwise specified, data are presented as mean ± SEM or %. Statistical significance was set to P < 0.05.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Representative raw bioluminescence, phase, amplitude, and period unilateral SCN maps (columns) generated from NT, γ2 KD and δ KD (rows) SCN slices. White pixels represent areas scored as arrhythmic. Maps show a more homogeneous distribution of phases, amplitudes and periods in the control SCN compared to γ2 KD and δ KD.

Movie S2.

Representative raw bioluminescence, phase, amplitude, and period unilateral SCN maps (columns) generated from WT, γ2 KI and δ KI (rows) SCN slices. Maps show a more homogeneous distribution of phases, amplitudes and periods in the control SCN compared to γ2 KI and δ KI mutants.

Movie S3.

Representative raw bioluminescence, phase, amplitude, and period unilateral SCN maps (columns) generated from γ2 KD + δ OVX and δ KD + γ2 KD SCN slices (rows). Maps show a more homogeneous distribution of phases, amplitudes and periods when γ2 or d receptors are overexpressed.

Movie S4.

Representative raw bioluminescence, phase, amplitude, and period unilateral SCN maps (columns) generated from γ2 KI + δ OVX and δ KI + γ2 KD SCN slices (rows). Maps show a more homogeneous distribution of phases, amplitudes and periods when γ2 or d receptors are overexpressed.

Data, Materials, and Software Availability

All study data are included in the article and python scripts data have been deposited in github (https://github.com/herzog-lab/daniel-granadosf-et.-al.-2024) (67).