Leveraging simplicity to generate fundamental insights into the complex nature of sleep-drives

A recent study in fruit flies (Drosophila melanogaster), published in the Journal of Neuroscience by Satterfield et al. [1], offers new insights into the nature and source of sleep drives that are highly relevant to our general understanding of sleep. Across species, our efforts to understand the control of sleep and wakefulness often focus on delineating the critical brain circuitry that controls these states. Due to the overwhelming complexity of the brain, particularly in mammals, efforts to map embedded control systems can easily become excessively granular neuroanatomical exercises that fail to reveal clear conceptual insights. It can be easy to lose sight of the relative simplicity of the rules that govern sleep control systems and that circuit mapping efforts should endeavor to explain. One of the principle rules of sleep systems is that a loss of sleep will often result in a subsequent increase in sleep time (i.e. a post-deprivation rebound, commonly referred to as sleep homeostasis). This means that the present and future propensity for sleep can critically depend on the previous sleep/wake history of the system. More specifically, this implies that the history of sleep and wake is somehow encoded within the system and that sleep/wake control circuitry is sensitive and responsive to this encoded history. This encoded history forms the basis of the so-called sleep drive. Identifying neurons that relay sleep drive can serve as a critical starting point for mapping sleep control networks that are embedded in complex systems. Starting with sleep-drive neurons, one can trace mechanisms for encoding sleep/wake history and state switching, which in turn integrate with mechanisms of sleep phenomenology and associated sleep functions. Satterfield et al. [1] leverage the simplicity, high throughput, and capacity for intersectional genetic manipulations of Drosophila to reveal fundamental characteristics of sleep drives, including their possible origins outside of the brain. Satterfield et al. [1] use a technique of driving targeted expression of heat-activable channels in select neuronal populations of Drosophila (i.e. thermogenetics) [2], which allows for activation of those neurons through a simple increase in the ambient temperature. Combining this thermogenetic cell activation approach with sleep recording, they are able to systematically characterize the function of genetically defined neuronal populations in the control of sleep dynamics, particularly with respect to post-deprivation sleep rebound. The recent findings of Satterfield et al. [1], build on a previous study conducted by the same group in 2015 [3]. In that study, Seidner et al. thermogenetically activated discrete arousal pathways to produce sleep deprivation and subsequently quantified rebound sleep. Firstly, it is important to consider that when manipulating neuronal activity by means of a thermal stimulus, the temperature change per se could produce a confounding stress response or circadian dysregulation/phase shift that might explain the observed rebound sleep. The reader is directed to Seidner et al. [3], in particular, for critical control experiments that suggest that the rebound sleep they induce is both genuine and dependent on modulation of the target cells. Most importantly, Seidner et al. [3] found that only a select few arousal pathways were associated with sleep rebound following deprivation. This is remarkable because it supports the idea that sleep drive is not related to wakefulness per se, where we would expect any arousing stimulus to potentially illicit a sleep rebound. In other words, the sleep control system does not appear to be strictly tracking the length of the inter-sleep interval globally, rather the encoded history of wakefulness reflects a tracking of select sensory domains of waking experience. Importantly, this insight has important implications for configuration of sleep control systems. A priori, one may hypothesize a global or macro-level feedback system where control circuitry induces macro-brain states and where the encoded history of such states feeds back onto the underlying control circuitry in a diffuse, sensory modality-independent fashion (Figure 1,A). This very simple, perhaps naïve, configuration is challenged by the findings of Seidner et al. [3]. The finding that macro-sleep is not strictly visible to the sleep control system would seem to necessitate, at the very least, a kind of mesoscale feedback system: a sleep control system composed of mesoscale interactions between controllers and discrete brain circuits underlying processing of discrete sensory experiences, where macro-level sleep phenomenology arises as an emergent property of the mesoscale interactions. Identifying sleep-drives originating outside the brain enables a reimagining of the sleep control system. (A) A schematic of a global/macro-level, brain-centric feedback system, in contrast to (B), which depicts an organism-wide, mesoscale, and feed-forward control system. The latter configuration is supported by the findings that global wakefulness is not strictly visible to the sleep-wake control system; rather, specific modalities of peripheral sensory input constitute the encoded history of wakefulness that drives sleep. The findings of Seidner et al. [3] are consistent with an even more radical reconfiguration of the sleep-wake control system. The idea that sleep-wake time is under the control of an encoded record of sleep-wake time itself, at the macro level, predisposes us to think that sleep regulation is a brain-centric phenomenon. After all, if tracking macro-level sleep-wake is assumed to be a system requirement, then where better to perform such tracking than in the brain? However, the findings of Seidner et al. [3] suggest that macro-wakefulness is not strictly visible to the sleep-wake control system; rather, specific information processing modules contribute to the encoded history of wakefulness. This modality specificity introduces the possibility that select sensory inputs from the periphery form the history of waking experience that drives the brain state. This organization would constitute an organism-wide feed-forward control system (Figure 1,B), in contrast to the notion of closed feedback control that is, “of the brain, by the brain and for the brain” (Figure 1,A) [4]. The recent findings of Satterfield et al. [1], provide support for such an organism-wide feed-forward control system (Figure 1,B). They show that a specific group of sleep-drive neurons—ppk neurons, which induce wake and a subsequent compensatory sleep rebound—are not found in the brain, rather they are located peripherally and project to the brain [1, 3]. By examining the overlap between these cells and another genetic marker of sleep drive (i.e. 20B01), the authors were able to identify a more uniquely defined subset of sleep-drive neurons. Cell bodies of these neurons were specifically found in the legs and driving activity in these cells produced wakefulness and subsequent sleep rebound. Therefore, the accrual of waking peripheral input from the limbs to the brain seems sufficient to form a sleep drive. Identifying sources of sleep drive immediately begs questions about the mechanisms of sleep-drive encoding. A previous study from 2016, by Liu et al. [5], claims that the Ellipsoid Body in the Drosophila brain acts as a dedicated sleep-drive encoder. Satterfield et al. [1], reexamined the genetic markers used in that study. Their findings challenge the assertion that the ellipsoid body acts in this capacity. The original study by Lui et al. [5], which used specific genetic markers to drive activity in the ellipsoid body of the brain, neglected to look at the expression of those markers peripherally. Satterfield et al. [1], show that some of these markers do express peripherally and, even more significantly, that they overlap with the ppk neuron population. Satterfield et al. [1], performed a critical intersectional experiment where they thermogenetically activated the co-expressing population and successfully induced arousal, sleep loss, and subsequent rebound. In so doing, they undermined the previous identification of a central sleep-drive encoder, and they reattributed to a peripheral pathway, functional significance that had been associated with a central sleep controller in the brain. One interesting strategy to identify sleep-drive encoding populations is to trace the points of convergence/integration of multiple sleep-drive pathways. Many of the genetic markers studied by Satterfield et al [1] were shown to have an overlapping expression (i.e. they belong to common/intersecting populations); however, they show, immunohistochemically and physiologically, that one previously identified marker of sleep-drive neurons (i.e. 52B10) [5], constitutes an independent sleep-drive pathway. The identification of independent sleep-drive pathways enables a search for drive-encoding circuitry using such a pathway convergence approach. Satterfield et al. [1], do not demonstrate convergent input of these pathways into a shared population that functions as a sleep-drive encoder. However, they do show that these populations both provide input to the subesophageal zone of the Drosophila brain. Future studies are required to determine whether this region is a genuine point of convergence that legitimately contributes to encoding wake history. Nevertheless, as previously discussed, identifying sleep-drive encoders is important because it facilitates description of the encoding mechanisms as well as upstream mechanisms underlying state switching and state phenomenology. However, it would also be meaningful if efforts to identify points of sleep-drive convergence yielded no such intersections. This is because a lack of convergence would be evidence of decentralization in sleep-drive encoding. In the case of convergence, even if the sleep control system is not strictly tracking the length of the inter-sleep-interval globally, and the encoded history of wakefulness is instead an encoded history of select domains of waking sensory experience, the integration of these histories could effectively amount to a nonspecific record of waking activity. In the decentralized case, sleep could be driven by multiple histories acting in parallel. This distinction is non-trivial, because where sleep falls on the spectrum of decentralized versus centralized control ultimately dictates whether sleep is best framed as an intrinsic property of information processing circuits versus a mode of circuit operation that is imposed externally by dedicated controllers. The work of Satterfield et al. [1] contributes to a body of work that provides due consideration to peripheral factors in the regulation of sleep. While the specific identities of those sensory modalities linked to sleep drives will have to be determined—in Drosophila and other species—we are already aware of many peripheral factors that modulate sleep. This includes numerous humoral factors relating to metabolic and immune function (for review see references [6] and [7], respectively). Particularly interesting in the context of the present study is that moderate skeletal muscle activity (e.g. exercise) is known to produce humoral factors which promote sleep (for review [8]). Also, beyond the organism itself, we also know that an organism’s ecological niche will strongly influence its sleep [9]. Taken together, the study by Satterfield et al [1] evokes important questions about the source and nature of sleep drives, encoding of sleep history, macro versus mesoscale mechanisms, feedback versus feed-forward control, and the level of control centralization. These are complex topics that basic sleep research must confront head-on. We should look to studies like that of Satterfield et al. [1], which are able to leverage simple and high throughput models of sleep, to drive our understanding forward. Commentary on: Satterfield LK, De J, Wu M, Qiu T, Joiner WJ (2022) Inputs to the sleep homeostat originate outside the brain. J. Neurosci. 42(29):5695-5704

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.