To investigate how P1 neurons modulate different behaviors, we first aimed to identify distinct GAL4 drivers for labeling and manipulating P1 neurons. We previously used two intersectional methods to target P1 neurons, one using the split-GAL4 system (P1a-splitGAl4: R15A01-AD; R71G01-DBD, Fig. 1A), and the other using the Flip-out system (P1e: R71G01-LexA/LexAop2-FlpL; UAS>stop>myrGFP/dsxGAL4, Fig. 1B) . We later identified two fragment GAL4s (R17D06 and R22D03) that also label male-specific P1 neurons and thus made three additional splitGAL4s (P1b by R15A01-AD; R22D03-DBD, P1c by R17D06-AD; R71G01-DBD, and P1d by R17D06-AD; R22D03-DBD). The four split-GAL4s each labeled 9–12 P1 neurons (Fig. 1C), ~20%–25% of all P1 neurons, as well as a few other cells. Although we hoped to target distinct subsets of P1 neurons with different split-GAL4 drivers, we were not able to discriminate among them. However, when P1a-splitGAl4 that labels 9–10 P1 cells intersected with R17D06-LexA, only ~2 P1 neurons were consistently labeled, suggesting that only two P1 neurons are labeled by both drivers. Thus, it was highly likely that these four P1-splitGAL4 drivers partially overlapped as well as targeting distinct P1 neurons.
We next used these five P1 drivers (P1a–P1d splitGAl4s, and P1e from intersection between R71G01-LexA and dsxGAL4, hereafter referred to simply as P1a–P1e) to investigate how the activation of these neurons via the temperature-sensitive activator dTrpA1  modulates sleep, courtship, and/or feeding behaviors. It has been shown that neurons expressing dTrpA1 begin to fire at 26°C, and higher temperatures further increase their activity . Thus we tested the above behaviors at four temperatures (25.5°C, 27°C, 28.5°C, and 30°C), at which P1 neurons would be increasingly activated, presenting data at 27°C as mild activation and 30°C as stronger activation unless stated otherwise. We found that mild activation of P1a, P1b, P1c, and P1e, but not P1d neurons significantly inhibited sleep (up to 80% sleep loss, Fig. 2A, B). None of these lines induced unilateral wing extension, a key step in courtship rituals (Fig. 2C), nor did they affect feeding behavior in starved males (Fig. 2D). Furthermore, stronger activation of all P1a–P1e neurons at 30°C significantly inhibited sleep (Fig. 2E, F), and such activation of the four sets of P1 neurons (P1a, P1c, P1d, and P1e) was able to induce unilateral wing extension (Fig. 2G). However, only stronger activation of the P1e neurons at 30°C significantly suppressed feeding behavior (Fig. 2H). To test whether stronger activation of P1a–P1d neurons at a higher temperature affected feeding, we performed the feeding experiments at 32°C, and found the same results as those at 30°C. Activation of P1e, but not P1a–P1d neurons, suppressed feeding at 32°C (Fig. S1). The P1e driver labeled ~23 P1 neurons, while the other four P1 drivers each labeled 9–12 P1 neurons. There are at least two possibilities why only activation of P1e but not P1a–P1d neurons suppressed feeding behaviors in starved males: (1) P1e neurons include a set of neurons that are not labeled by other P1 drivers, and these neurons specifically affect feeding behavior; or (2) as the number of P1 neurons labeled by P1e is about double that of P1a–P1d, it is possible that feeding requires the activation of a larger number of P1 neurons than sleep or courtship. Together, these results indicate that sleep, courtship, and feeding behaviors are all affected by activation of P1 neurons, but with different activation thresholds.
We next asked whether the suppression of feeding by activation of P1e neurons was correlated with locomotor activity, as stronger activation of P1e neurons at 30°C indeed increases walking velocity . We assayed walking velocity in all P1a–P1e-activated males at various temperatures (25.5–30°C, Fig. 3A–D) for 24 h, and found that mild activation at 27°C already increased the average velocity in P1e-activated males, as well as in the other P1-activated males. Stronger activation at 30°C further increased the velocity, but that of P1e-activated males was not as high as the other P1-activated males (average velocity of P1a–P1e-activated males: 221.0 ± 8.4, 266.6 ± 19.1, 218.6 ± 19.3, 283.5 ± 17.3, and 183.7 ± 15.0 mm/min, respectively). The two control lines also showed a slightly increased walking velocity at 30°C. As only activation of P1e neurons, but not P1a–P1d, suppressed feeding, this suppression could not be due to increased locomotion. To further test if feeding and walking velocity were negatively correlated, we plotted all the feeding and locomotor data for all P1-activated males and control males at 25.5°C, 27°C, 28.5°C, and 30°C, and found that feeding and walking velocity were not negatively correlated, and even slightly positively correlated (r = 0.37), although the correlation was not significant (P = 0.0503). These results indicate that decreased feeding by P1e activation is not due to increased locomotion. How P1e neurons regulate feeding is still unclear and awaits further investigation.
In summary, we used five independent P1 drivers that potentially labeled partially overlapping and distinct P1 neurons, and systematically investigated how different activation levels (25.5°C, 27°C, 28.5°C, and 30°C) of these targeted P1 neurons affect sleep, courtship, and feeding behaviors. We found that all these behaviors were affected by stronger activation of at least one P1 driver (e.g., P1e), suggesting that P1 activity affects all sleep, courtship, and feeding behaviors. Furthermore, we found that differential activation thresholds for P1 neurons were required to affect these three behaviors (Fig. 4A). First, minimum activation (mild activation of 9–12 P1 neurons) was sufficient to suppress sleep; second, stronger activation of P1 neurons was required to induce courtship behavior; and third, only stronger activation of P1e that included ~23 P1 neurons affected feeding.
Sleep, courtship, and feeding are competing behaviors that are mediated by external sensory cues and internal states. Whether these competing behaviors are regulated by common neural nodes is an intriguing question. P1 neurons have been established as a center that mediates sexual arousal, but their role in regulating other internal states and behaviors has been underestimated. Our findings that P1 neurons mediate sleep, courtship, and feeding behaviors not only reveal a neural node (P1) that participates in all these competing behaviors, but also how P1 neurons modulate these behaviors in a hierarchical manner (Fig. 4B).
There are nearly 50 pairs of P1 neurons in the male fly brain , and we studied here only 20%–50% of them. Given that mild activation of ~10 P1 neurons was sufficient to inhibit sleep, and stronger activation of ~23 (nearly half) suppressed feeding, what if the other half or all P1 neurons were activated? Do P1 neurons regulate behaviors other than sleep, courtship, aggression, and feeding? Is P1 a center for internal states that coordinate different behaviors? To answer these questions, better tools are needed to subdivide P1 populations, with driver lines that target small and distinct subsets of P1 neurons and driver lines targeting all or the majority of P1 neurons.
We also note that, although males and females play distinct roles in sexual behavior, their differences in non-sexual behaviors (e.g., different amounts of sleep or feeding) are relatively smaller and underestimated, and the mechanism underlying these differences is unclear. That P1 neurons are male-specific and regulate sleep, courtship, aggression, and feeding suggests that sexual dimorphism in these behaviors may be greater than we thought, and our results provide a simple model of how a small number of sex-specific neurons can contribute to various sexually-dimorphic behaviors.