Anticipatory countering of motor challenges by premovement activation of orexin neurons

Abstract Countering upcoming challenges with anticipatory movements is a fundamental function of the brain, whose neural implementations remain poorly defined. Recently, premovement neural activation was found outside canonical premotor areas, in the hypothalamic hypocretin/orexin neurons (HONs). The purpose of this hypothalamic activation is unknown. By studying precisely defined mouse–robot interactions, here we show that the premovement HON activity correlates with experience-dependent emergence of anticipatory movements that counter imminent motor challenges. Through targeted, bidirectional optogenetic interference, we demonstrate that the premovement HON activation governs the anticipatory movements. These findings advance our understanding of the behavioral and cognitive impact of temporally defined HON signals and may provide important insights into healthy adaptive movements.


Introduction
Brain systems that counter upcoming challenges with anticipatory actions have long been at the forefront of research across diverse academic fields. Studies in neuroscience, artificial intelligence, and engineering proposed general and neural algorithms for anticipation, and a role of premovement cortical activity in anticipatory actions (1)(2)(3)(4). Premovement neural activity refers to neural signals that appear during seconds/subseconds before movement onset, and predict movement initiation. Such signals have been documented in the cortex, basal ganglia, and the lateral hypothalamus (LH) (5-7). The recently described premovement activation of hypothalamic hypocretin/orexin neurons (HONs) (7) is of particular interest, because its purpose in the context of anticipatory movements is unknown.

Results
To examine the role of premovement HON activation in anticipatory movements, we first fiberoptically recorded HON-GCaMP (7) signals in mice trained to pull on a robot-controlled handle (8), where a straight movement (pull) was rewarded (Fig. 1A-C; see "Methods"). To create a challenge during the movement, we programmed the robot to apply a lateral force after a specific point in each pull trajectory, pushing the handle sideways (Fig. 1A). The robot sensed handle position at a high temporal resolution (8). This enabled us to quantify dynamics and kinematics of movements happening before the challenge onset, where challengeanticipatory movements occur ( Fig. 1B and C). Mice adapted to the push challenge by preemptively accelerating in the opposite direction (Fig. 1D). Henceforth, we refer to this as "anticipatory movement" (quantified as acceleration in the opposite direction to the force challenge), because it preceded the challenge, and was not seen in sessions without the challenge (Fig. 1E).
Stronger anticipatory movements were associated with increased premovement HON-GCaMP signals (Fig. 1D). Importantly, in sessions without the push challenge, no such changes in HON activation and movement occurred (Fig. 1E). Statistical quantification revealed that the premovement HON-GCaMP signals stayed steady during sessions without the challenge, but progressively increased across the challenge trials with a similar time-course as the anticipatory movements ( Fig. 1F and G). During the challenge trials, the premovement HON activation was significantly correlated with the anticipatory movement component that directionally opposed the push challenge (Fig. 1H, black). This correlation was not observed in the absence of the challenge (Fig. 1H, gray), or in the other (challenge-perpendicular) component of anticipatory movement (Fig. 1I), confirming that it reflects challenge-anticipatory counter action. At the level of the whole vector of the initial pull, we also observed a correlation with HON activity in the presence but not in the absence of challenge (Fig. 1J), as expected from the individual components of this vector ( Fig. 1H and I).
Based on this correlative evidence linking premovement HON activity and anticipatory movements (Fig. 1H), we hypothesized that the premovement HON activation might play a role in regulating the anticipatory movements. To test this hypothesis, we examined whether the anticipatory movements are altered by optogenetic suppression or augmentation of the premovement HON signals, using the previously developed and validated opsins specifically targeted to HONs (7). Laser light application for optogenetic HON silencing or stimulation (in orexin:: ArchT mice and orexin::C1V1 mice, respectively) was precisely restricted to the premovement epoch where premovement HON signals occur ( Fig. 2A). Fiberoptically delivered, bilateral laser illumination of the LH profoundly suppressed anticipatory movements in orexin::ArchT mice, and potentiated them in orexin::C1V1 mice [movement examples: Fig. 2(A) middle and bottom; quantification: Fig. 2(B)-(D)]. In control mice without the opsins, the same LH laser illuminations did not evoke such effects (Fig. 2E). These results indicate that the premovement HON activity bidirectionally regulates the anticipatory movements.
Finally, we sought to determine whether premovement HON activation nonspecifically regulates movement acceleration, rather than plays a more specialized role in the control of anticipatory movements. To achieve this, we repeated the bidirectional optogenetic manipulation of premovement HON activity without the force field challenge (Fig. 2F). We quantified the acceleration vector during the same epoch where challenge-anticipatory movements were measured earlier. We found this parameter unaffected by the excitatory opto-manipulations of premovement HON activity (Fig. 2F). Thus, premovement HON activity does not regulate acceleration in all scenarios, but is more specifically involved in anticipatory movements.

Discussion
Our findings suggest that a physiological role of premovement HON signals is to modulate the initial, anticipatory aspect of skilled movements. This aspect of movement, also known as predictive or feedforward aspect, is thought to be planned before movement initiation according to internal models of anticipated challenges, as opposed to feedback/reactive movements that are reactions to challenges occurring "on the fly" after the movement is already initiated (1,9,10). Predictive control of movement is important, since feedback control can be too slow to produce fast smooth movements (1,9,10). Thus, we suggest that premovement HON signals contribute to movement planning, as well as adaptation (11). Since the purpose of premovement HON activity was unknown, this advances our understanding of the role of this core, evolutionarily conserved, neural module in adaptive behavior. About 1:1,000 people suffer from narcolepsy, a debilitating disease associated with HON deficiency; these human patients, as well as HON deficient animals, are strikingly unable to generate appropriate muscle tone in certain challenging situations, leading to postural collapse [reviewed in (12)]. It would be interesting to re-evaluate these motor failures as failures of HON-dependent motor planning, which may diversify existing treatments for these debilitating symptoms, for example by targeting anticipatory postural adjustments. In future work, it would also be important to understand the mechanisms through which HONs regulate anticipatory movements. The HONs' well-documented ability to stimulate arousal [reviewed in (13)] could be consistent with the present findings, based on the long-proposed link between arousal and learning (14,15). Premovement HON activation in challenging situations may provide arousal necessary for motor learning, such as internal model updating that is considered key for anticipatory movements (9,10). Deeper knowledge of the behavioral and cognitive impact of temporally defined HON signals will provide important insights into healthy adaptive behavior.

Materials and Methods
See Supporting Information where full methods are given.

Supplementary Material
Supplementary material is available at PNAS Nexus online.

Funding
This work is funded by ETH Zürich. D.P. contributed to data interpretation. D.B. wrote the paper with input from all coauthors.

Data Availability
All data are included in the manuscript and/or supporting information.