A daily body temperature rhythm (BTR) is critical for the maintenance of homeostasis in mammals. Whereas mammals use internal energy to regulate body temperature, ectotherms typically regulate body temperature behaviorally [1]. Some ectotherms maintain homeostasis via a daily temperature preference rhythm (TPR) [2], but the underlying mechanisms are largely unknown. Here, we show that Drosophila exhibit a daily circadian clock-dependent TPR that resembles mammalian BTR. Pacemaker neurons critical for locomotor activity are not necessary for TPR; instead, the dorsal neuron 2 s (DN2s), whose function was previously unknown, is sufficient. This indicates that TPR, like BTR, is controlled independently from locomotor activity. Therefore, the mechanisms controlling temperature fluctuations in fly TPR and mammalian BTR may share parallel features. Taken together, our results reveal the existence of a novel DN2-based circadian neural circuit that specifically regulates TPR; thus, understanding the mechanisms of TPR will shed new light on the function and neural control of circadian rhythms.

The brain of Drosophila melanogaster contains approximately 150 circadian neurons [1] functionally divided into morning and evening cells that control peaks in daily behavioral activity at dawn and dusk, respectively [2, 3]. The PIGMENT DISPERSING-FACTOR (PDF)-positive small ventral lateral neurons (sLN(v)s) promote morning behavior, whereas the PDF-negative sLN(v) and the dorsal lateral neurons (LN(d)s) generate evening activity. Much less is known about the approximately 120 dorsal neurons (DN1, 2, and 3). Using a Clk-GAL4 driver that specifically targets a subset of DN1s, we generated mosaic per(0) flies with clock function restored only in these neurons. We found that the Clk4.1M-GAL4-positive DN1s promote only morning activity under standard (high light intensity) light/dark cycles. Surprisingly, however, these circadian neurons generate a robust evening peak of activity under a temperature cycle in constant darkness. Using different light intensities and ambient temperatures, we resolved this apparent paradox. The DN1 behavioral output is under both photic and thermal regulation. High light intensity suppresses DN1-generated evening activity. Low temperature inhibits morning behavior, but it promotes evening activity under high light intensity. Thus, the Clk4.1M-GAL4-positive DN1s, or the neurons they target, integrate light and temperature inputs to control locomotor rhythms. Our study therefore reveals a novel mechanism contributing to the plasticity of circadian behavior.