Sodium regulates clock time and output via an excitatory GABAergic pathway

The suprachiasmatic nucleus (SCN) serves as the body’s master circadian clock that adaptively coordinates changes in physiology and behaviour in anticipation of changing requirements throughout the 24-h day–night cycle 1 – 4 . For example, the SCN opposes overnight adipsia by driving water intake before sleep 5 , 6 , and by driving the secretion of anti-diuretic hormone 7 , 8 and lowering body temperature 9 , 10 to reduce water loss during sleep 11 . These responses can also be driven by central osmo-sodium sensors to oppose an unscheduled rise in osmolality during the active phase 12 – 16 . However, it is unknown whether osmo-sodium sensors require clock-output networks to drive homeostatic responses. Here we show that a systemic salt injection (hypertonic saline) given at Zeitgeber time 19—a time at which SCN VP (vasopressin) neurons are inactive—excited SCN VP neurons and decreased non-shivering thermogenesis (NST) and body temperature. The effects of hypertonic saline on NST and body temperature were prevented by chemogenetic inhibition of SCN VP neurons and mimicked by optogenetic stimulation of SCN VP neurons in vivo. Combined anatomical and electrophysiological experiments revealed that osmo-sodium-sensing organum vasculosum lamina terminalis (OVLT) neurons expressing glutamic acid decarboxylase (OVLT GAD ) relay this information to SCN VP neurons via an excitatory effect of γ-aminobutyric acid (GABA). Optogenetic activation of OVLT GAD neuron axon terminals excited SCN VP neurons in vitro and mimicked the effects of hypertonic saline on NST and body temperature in vivo. Furthermore, chemogenetic inhibition of OVLT GAD neurons blunted the effects of systemic hypertonic saline on NST and body temperature. Finally, we show that hypertonic saline significantly phase-advanced the circadian locomotor activity onset of mice. This effect was mimicked by optogenetic activation of the OVLT GAD → SCN VP pathway and was prevented by chemogenetic inhibition of OVLT GAD neurons. Collectively, our findings provide demonstration that clock time can be regulated by non-photic physiologically relevant cues, and that such cues can drive unscheduled homeostatic responses via clock-output networks. The authors demonstrate that clock time can be regulated by non-photic physiologically relevant cues and that such cues can drive unscheduled homeostatic responses via clock-output networks.